code
stringlengths
1
1.72M
language
stringclasses
1 value
""" FludServer.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. flud server operations """ import threading, binascii, time, os, stat, httplib, gc, re, sys, logging, sets from twisted.web import server, resource, client from twisted.web.resource import Resource from twisted.internet import reactor, threads, defer from twisted.web import http from twisted.python import threadable, failure from flud.FludCrypto import FludRSA import flud.FludkRouting from ServerPrimitives import * from ServerDHTPrimitives import * from LocalPrimitives import * from FludCommUtil import * threadable.init() class FludServer(threading.Thread): """ This class runs the webserver, responding to all requests. """ def __init__(self, node, port): threading.Thread.__init__(self) self.port = port self.node = node self.clientport = node.config.clientport self.logger = node.logger self.root = ROOT(self) commandmap = node.config.commandmap self.root.putChild(commandmap['ID'], ID(self)) self.root.putChild(commandmap['STORE'], STORE(self)) self.root.putChild(commandmap['RETRIEVE'], RETRIEVE(self)) self.root.putChild(commandmap['VERIFY'], VERIFY(self)) self.root.putChild(commandmap['PROXY'], PROXY(self)) self.root.putChild(commandmap['DELETE'], DELETE(self)) self.root.putChild(commandmap['kFINDNODE'], kFINDNODE(self)) self.root.putChild(commandmap['kFINDVAL'], kFINDVAL(self)) self.root.putChild(commandmap['kSTORE'], kSTORE(self)) self.site = server.Site(self.root) reactor.listenTCP(self.port, self.site) reactor.listenTCP(self.clientport, LocalFactory(node), interface="127.0.0.1") #print "FludServer will listen on port %d, local client on %d"\ # % (self.port, self.clientport) self.logger.log(logging.INFO,\ "FludServer will listen on port %d, local client on %d" % (self.port, self.clientport)) def run(self): self.logger.log(logging.INFO, "FludServer starting") return reactor.run(installSignalHandlers=0) def stop(self): self.logger.log(logging.INFO, "FludServer stopping") reactor.stop()
Python
""" LocalPrimitives.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Protocol for talking to a flud node locally (from client code -- command line, GUI, etc). Each command in the local protocol begins with 4 bytes designating the type of command. These are followed by a status byte, which is either '?'=request, ':'=success response, or '!'=failure response. Arguments to the command follow the status byte. """ import binascii, time, os, stat, httplib, gc, re, sys, logging, sets from twisted.web.resource import Resource from twisted.web import server, resource, client from twisted.internet import protocol, reactor, threads, defer from twisted.protocols import basic from twisted.mail import smtp from twisted.python import failure from Crypto.Cipher import AES from flud.FludCrypto import FludRSA, hashstring, generateRandom import flud.FludkRouting from flud.fencode import fencode, fdecode from flud.FludFileOperations import * from FludCommUtil import * from FludServer import * logger = logging.getLogger("flud.local.server") """ Protocol and Factory for local client/server communication """ MAXCONCURRENT = 300 (CONCURR, MAX, QUEUE) = (0, 1, 2) # indexes into LocalProtocol.commands class LocalProtocol(basic.LineReceiver): authenticated = False commands = {'PUTF': [0, MAXCONCURRENT, []], 'GETF': [0, MAXCONCURRENT, []], 'GETI': [0, MAXCONCURRENT, []], 'FNDN': [0, 1, []], 'STOR': [0, MAXCONCURRENT, []], 'RTRV': [0, MAXCONCURRENT, []], 'VRFY': [0, MAXCONCURRENT, []], 'FNDV': [0, 1, []], 'CRED': [0, 1, []], 'LIST': [0, 1, []], 'GETM': [0, 1, []], 'PUTM': [0, 1, []] } def connectionMade(self): logger.info("client connected") self.authenticated=False def connectionLost(self, reason): self.authenticated=False def doOp(self, command, fname): #print "got command '%s'" % command if command == "PUTF": logger.debug("PUTF %s", fname); return StoreFile(self.factory.node, fname).deferred elif command == "GETI": logger.debug("GETI %s", fname); return RetrieveFile(self.factory.node, fname).deferred elif command == "GETF": logger.debug("GETF %s", fname); return RetrieveFilename(self.factory.node, fname).deferred elif command == "FNDN": logger.debug("FNDN %s" % fname); try: intval = long(fname, 16) except: return defer.fail("fname was not hex") return self.factory.node.client.kFindNode(intval) # The following is for testing aggregation of kFindNode on same key #dl = [] #for i in [1,2,3,4,5]: # d = self.factory.node.client.kFindNode(intval) # dl.append(d) #dlist = defer.DeferredList(dl) #return dlist elif command == "FNDV": logger.debug("FNDV %s", fname); try: intval = long(fname, 16) except: return defer.fail("fname was not hex") return self.factory.node.client.kFindValue(intval) elif command == "CRED": passphrase, email = fdecode(fname) # XXX: allow an optional passphrase hint to be sent in email. passphrase = self.factory.node.config.Kr.decrypt(passphrase) logger.debug("CRED %s to %s", passphrase, email); Kr = self.factory.node.config.Kr.exportPrivateKey() Kr['g'] = self.factory.node.config.groupIDr fKr = fencode(Kr) key = AES.new(binascii.unhexlify(hashstring(passphrase))) fKr = '\x00'*(16-(len(fKr)%16))+fKr efKr = fencode(key.encrypt(fKr)) logger.debug("efKr = %s " % efKr) d = smtp.sendmail('localhost', "your_flud_client@localhost", email, "Subject: Your encrypted flud credentials\n\n" "Hopefully, you'll never need to use this email. Its " "sole purpose is to help you recover your data after a " "catastrophic and complete loss of the original computer " "or hard drive.\n\n" "In that unlucky event, you'll need a copy of your flud " "credentials, which I've included below, sitting between " "the \"---+++---\" markers. These credentials were " "encrypted with a passphrase of your choosing when you " "installed the flud software. I'll only say this " "once:\n\n" "YOU MUST REMEMBER THAT PASSWORD IN ORDER TO RECOVER YOUR " "CREDENTIALS. If you are unable to remember the " "passphrase and your computer fails catastrophically " "(losing its local copy of these credentials), you will " "not be able to recover your data." "\n\n" "Luckily, that's all you should ever need in order to " "recover all your data: your passphrase and these " "credentials." "\n\n" "Please save this email. You may want to print out hard " "copies and store them safely, forward this email to " "other email accounts, etc. Since the credentials are " "encrypted, others won't be able to steal them " "without guessing your passphrase. " "\n\n" "---+++---\n"+efKr+"\n---+++---\n") return d # to decode this email, we search for the '---+++---' markers, make # sure the intervening data is all in one piece (remove any line # breaks \r or \n inserted by email clients) and call this 'cred', # reconstruct the AES key with the H(passphrase) (as above), and # then use the key to .decrypt(fdecode(cred)) and call this dcred, # then fdecode(dcred[dcred.find('d'):]) and call this ddcred, and # finally importPrivateKey(ddcred) and set groupIDr to ddcred['g']. elif command == "LIST": logger.debug("LIST") return defer.succeed(self.factory.config.master) elif command == "GETM": logger.debug("GETM") return RetrieveMasterIndex(self.factory.node).deferred elif command == "PUTM": logger.debug("PUTM") return UpdateMasterIndex(self.factory.node).deferred else: #print "fname is '%s'" % fname host = fname[:fname.find(':')] port = fname[fname.find(':')+1:fname.find(',')] fname = fname[fname.find(',')+1:] print "%s: %s : %s , %s" % (command, host, port, fname) if command == "STOR": logger.debug("STOR"); return self.factory.node.client.sendStore(fname, None, host, int(port)) elif command == "RTRV": logger.debug("RTRV"); return self.factory.node.client.sendRetrieve(fname, host, int(port)) elif command == "VRFY": logger.debug("VRFY"); offset = port[port.find(':')+1:port.find('-')] length = port[port.find('-')+1:] port = port[:port.find(':')] print "%s: %s : %s %s - %s , %s" % (command, host, port, offset, length, fname) return self.factory.node.client.sendVerify(fname, int(offset), int(length), host, int(port)) else: logger.debug("bad op"); return defer.fail("bad op") def serviceQueue(self, command): if len(self.commands[command][QUEUE]) > 0 and \ self.commands[command][CONCURR] <= self.commands[command][MAX]: data = self.commands[command][QUEUE].pop() logger.info("servicing queue['%s'], item %s" % (command, data)) print "taking %s off the queue" % command d = self.doOp(command, data) d.addCallback(self.sendSuccess, command, data) d.addErrback(self.sendFailure, command, data) def sendSuccess(self, resp, command, data, prepend=None): logger.debug("SUCCESS! "+command+":"+data) #logger.debug("response: '%s'" % (resp,)) if prepend: w = "%s:%s %s:%s\r\n" % (prepend, command, fencode(resp), data) else: w = "%s:%s:%s\r\n" % (command, fencode(resp), data) self.transport.write(w) self.commands[command][CONCURR] -= 1 try: self.serviceQueue(command) except: print sys.exec_info() return resp def sendFailure(self, err, command, data, prepend=None): logger.debug("FAILED! %s!%s" % (command, data)) errmsg = err.getErrorMessage() if prepend: w = "%s!%s %s!%s\r\n" % (prepend, command, errmsg, data) else: w = "%s!%s!%s\r\n" % (command, errmsg, data) logger.debug("sending %s" % w) self.transport.write(w) self.commands[command][CONCURR] -= 1 self.serviceQueue(command) return err def lineReceived(self, line): logger.debug("lineReceived: '%s'" % line) # commands: AUTH, PUTF, GETF, VRFY # status: ? = request, : = successful response, ! = failed response command = line[0:4] status = line[4] data = line[5:] #print "data is '%s'" % data if not self.authenticated and command == "AUTH": if status == '?': # asked for AUTH challenge to be sent. send it logger.debug("AUTH challenge requested, sending") echallenge = self.factory.sendChallenge() self.transport.write("AUTH?"+echallenge+"\r\n") elif status == ':' and self.factory.challengeAnswered(data): # sent AUTH response and it passed logger.debug("AUTH challenge successful") self.authenticated = True self.transport.write("AUTH:\r\n") elif status == ':': logger.debug("AUTH challenge failed") self.transport.write("AUTH!\r\n") elif command == "DIAG": if data == "NODE": logger.debug("DIAG NODE") nodetups = self.factory.config.routing.knownExternalNodes() nodes = [] for n in nodetups: node = list(n) if self.factory.config.reputations.has_key(n[2]): node.append(self.factory.config.reputations[n[2]]) else: node.append(0) nodes.append(tuple(node)) self.transport.write("DIAG:NODE%s\r\n" % fencode(nodes)) elif data == "BKTS": logger.debug("DIAG BKTS") bucks = eval("%s" % self.factory.config.routing.kBuckets) self.transport.write("DIAG:BKTS%s\r\n" % fencode(bucks)) else: dcommand = data[:4] ddata = data[5:] logger.debug("DIAG %s %s" % (dcommand, ddata)) self.commands[dcommand][CONCURR] += 1 d = self.doOp(dcommand, ddata) d.addCallback(self.sendSuccess, dcommand, ddata, "DIAG") d.addErrback(self.sendFailure, dcommand, ddata, "DIAG") elif status == '?': # requested an operation to be performed. If we are below our # maximum concurrent ops, do the operation. Otherwise, put it on # the queue to be serviced when current ops finish. Response is # sent back to client when deferreds fire. if self.commands[command][CONCURR] >= self.commands[command][MAX]: #print "putting %s on the queue" % line logger.info("received %s request, enqueuing" % command) self.commands[command][QUEUE].insert(0, data) else: #print "doing %s" % line logger.info("received %s request, executing" % command) print self.commands[command] self.commands[command][CONCURR] += 1 d = self.doOp(command, data) d.addCallback(self.sendSuccess, command, data) d.addErrback(self.sendFailure, command, data) class LocalFactory(protocol.ServerFactory): protocol = LocalProtocol def __init__(self, node): self.node = node self.config = node.config def sendChallenge(self): self.challenge = fencode(generateRandom(challengelength)) echallenge = self.config.Ku.encrypt(self.challenge)[0] echallenge = fencode(echallenge) return echallenge def challengeAnswered(self, resp): return resp == self.challenge
Python
#!/usr/bin/python """ LocalClient.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. LocalClient provides client functions which can be called to send commands to a local FludNode instance. """ import sys, os, time from twisted.internet.protocol import ClientFactory from twisted.protocols.basic import LineReceiver from twisted.internet import reactor from twisted.python import threadable threadable.init() from flud.fencode import fencode, fdecode from LocalPrimitives import * logger = logging.getLogger("flud.local.client") opTimeout = 1200 VALIDOPS = LocalProtocol.commands.keys() + ['AUTH', 'DIAG'] # XXX: print commands should either be raised, or put on factory.msgs class LocalClient(LineReceiver): MAX_LENGTH = 300000 auth=False def connectionMade(self): logger.debug("connection est.") self.auth=False self.sendLine("AUTH?") def lineReceived(self, line): logger.debug("received line '%s'" % line) command = line[0:4] if not command in VALIDOPS: print "error: invalid command op ('%s')-- "\ " are you trying to connect to the wrong port"\ " (local client port is usually external port + 500)?"\ % command return None status = line[4] data = line[5:] if not self.auth: if command == "AUTH" and status == '?': # got challenge, send response logger.debug("got AUTH challenge, sending response") echallenge = data self.sendLine("AUTH:"+self.factory.answerChallenge(echallenge)) return elif command == "AUTH" and status == ':': # response accepted, authenticated logger.debug("AUTH challenge accepted, success") self.auth = True self.factory.clientReady(self) #print "authenticated" else: if command == "AUTH" and status == "!": logger.warn("authentication failed (is FLUDHOME set" " correctly?)") print "authentication failed (is FLUDHOME set correctly?)" else: logger.warn("unknown message received before being" " authenticated:") logger.warn(" %s : %s" % (command, status)) print "unknown message received before being authenticated:" print " %s : %s" % (command, status) self.factory.setDie() elif command == "DIAG": subcommand = data[:4] data = data[4:] if subcommand == "NODE": logger.debug("DIAG NODE: %s" % data) data = fdecode(data) result = "" for i in data: score = i[4] petID = "%064x" % i[2] netID = "%s:%d" % (i[0], i[1]) petID = petID[:(70-len(netID))]+"..." result += "%d %s %s\n" % (score, netID, petID) result += "%d known nodes\n" % len(data) d = self.factory.pending['NODE'].pop('') d.callback(result) return if subcommand == "BKTS": logger.debug("DIAG BKTS") data = fdecode(data) result = "" for i in data: for bucket in i: result += "Bucket %s:\n" % bucket for k in i[bucket]: id = "%064x" % k[2] netID = "%s:%d" % (k[0], k[1]) result += " %s %s...\n" \ % (netID,id[:72-len(netID)]) d = self.factory.pending['BKTS'].pop('') d.callback(result) return elif status == ':': response, data = data.split(status, 1) logger.debug("DIAG %s: success" % subcommand) d = self.factory.pending[subcommand].pop(data) d.callback(fdecode(response)) elif status == "!": response, data = data.split(status, 1) logger.debug("DIAG %s: failure" % subcommand) d = self.factory.pending[subcommand].pop(data) d.errback(failure.DefaultException(response)) elif status == ':': response, data = data.split(status, 1) logger.debug("%s: success" % command) d = self.factory.pending[command].pop(data) d.callback(fdecode(response)) elif status == "!": response, data = data.split(status, 1) logger.debug("%s: failure" % command) if self.factory.pending.has_key(command): if not self.factory.pending[command].has_key(data): print "data key is '%s'" % data print "pending is '%s'" % self.factory.pending[command] if len(self.factory.pending[command]): d = self.factory.pending[command].popitem() d.errback(failure.DefaultException(response)) else: d = self.factory.pending[command].pop(data) d.errback(failure.DefaultException(response)) else: print "failed command '%s' not in pending?" % command print "pending is: %s" % self.factory.pending if command != 'AUTH' and command != 'DIAG' and \ not None in self.factory.pending[command].values(): logger.debug("%s done at %s" % (command, time.ctime())) class LocalClientFactory(ClientFactory): protocol = LocalClient def __init__(self, config): self.config = config self.messageQueue = [] self.client = None self.die = False self.pending = {'PUTF': {}, 'CRED': {}, 'GETI': {}, 'GETF': {}, 'FNDN': {}, 'STOR': {}, 'RTRV': {}, 'VRFY': {}, 'FNDV': {}, 'CRED': {}, 'LIST': {}, 'GETM': {}, 'PUTM': {}, 'NODE': {}, 'BKTS': {}} def clientConnectionFailed(self, connector, reason): #print "connection failed: %s" % reason logger.warn("connection failed: %s" % reason) self.cleanup("connection failed: %s" % reason) def clientConnectionLost(self, connector, reason): #print "connection lost: %s" % reason logger.debug("connection lost: %s" % reason) self.cleanup("connection lost: %s" % reason) def cleanup(self, msg): # override me for cleanup print msg; def clientReady(self, instance): self.client = instance logger.debug("client ready, sending [any] queued msgs") for i in self.messageQueue: self._sendMessage(i) def _sendMessage(self, msg): if self.client: logger.debug("sending msg '%s'" % msg) self.client.sendLine(msg) else: logger.debug("queueing msg '%s'" % msg) self.messageQueue.append(msg) def answerChallenge(self, echallenge): logger.debug("answering challenge") echallenge = (fdecode(echallenge),) challenge = self.config.Kr.decrypt(echallenge) return challenge def expire(self, pending, key): if pending.has_key(fname): logger.debug("timing out operation for %s" % key) #print "timing out operation for %s" % key pending.pop(key) def addFile(self, type, fname): logger.debug("addFile %s %s" % (type, fname)) if not self.pending[type].has_key(fname): d = defer.Deferred() self.pending[type][fname] = d self._sendMessage(type+"?"+fname) return d else: return self.pending[type][fname] def sendPING(self, host, port): logger.debug("sendPING") d = defer.Deferred() d.errback(failure.DefaultException( "ping not yet implemented in FludLocalClient")) return d def sendPUTF(self, fname): logger.debug("sendPUTF %s" % fname) if os.path.isdir(fname): dirlist = os.listdir(fname) dlist = [] for i in dirlist: dlist.append(self.sendPUTF(os.path.join(fname,i))) dl = defer.DeferredList(dlist) return dl elif not self.pending['PUTF'].has_key(fname): d = defer.Deferred() self.pending['PUTF'][fname] = d self._sendMessage("PUTF?"+fname) #reactor.callLater(opTimeout, self.expire, self.pendingPUTF, fname) return d else: return self.pending['PUTF'][fname] def sendCRED(self, passphrase, email): logger.debug("sendCRED") key = fencode((self.config.Ku.encrypt(passphrase)[0], email)) if not self.pending['CRED'].has_key(key): d = defer.Deferred() self.pending['CRED'][key] = d self._sendMessage("CRED?"+key) return d else: return self.pending['CRED'][key] def sendGETI(self, fID): logger.debug("sendGETI") if not self.pending['GETI'].has_key(fID): d = defer.Deferred() self.pending['GETI'][fID] = d self._sendMessage("GETI?"+fID) return d else: return self.pending['GETI'][fID] def sendGETF(self, fname): logger.debug("sendGETF") master = listMeta(self.config) if master.has_key(fname): return self.addFile("GETF",fname) elif fname[-1:] == os.path.sep: dlist = [] for name in master: if fname == name[:len(fname)]: dlist.append(self.addFile("GETF",name)) dl = defer.DeferredList(dlist) return dl def sendFNDN(self, nID): logger.debug("sendFNDN") if not self.pending['FNDN'].has_key(nID): d = defer.Deferred() self.pending['FNDN'][nID] = d self._sendMessage("FNDN?"+nID) return d else: return self.pending['FNDN'][nID] def sendLIST(self): logger.debug("sendLIST") if not self.pending['LIST'].has_key(""): d = defer.Deferred() self.pending['LIST'][''] = d logger.debug("LIST['']=%s" % d) self._sendMessage("LIST?") return d else: return self.pending['LIST'][''] def sendGETM(self): logger.debug("sendGETM") if not self.pending['GETM'].has_key(''): d = defer.Deferred() self.pending['GETM'][''] = d logger.debug("GETM['']=%s" % d) self._sendMessage("GETM?") return d else: return self.pending['GETM'][''] def sendPUTM(self): logger.debug("sendPUTM") if not self.pending['PUTM'].has_key(''): d = defer.Deferred() self.pending['PUTM'][''] = d self._sendMessage("PUTM?") return d else: return self.pending['PUTM'][''] def sendDIAGNODE(self): logger.debug("sendDIAGNODE") if not self.pending['NODE'].has_key(''): d = defer.Deferred() self.pending['NODE'][''] = d self._sendMessage("DIAG?NODE") return d else: return self.pending['NODE'][''] def sendDIAGBKTS(self): logger.debug("sendDIAGBKTS") if not self.pending['BKTS'].has_key(''): d = defer.Deferred() self.pending['BKTS'][''] = d self._sendMessage("DIAG?BKTS") return d else: return self.pending['BKTS'][''] def sendDIAGSTOR(self, command): logger.debug("sendDIAGSTOR") if not self.pending['STOR'].has_key(command): d = defer.Deferred() self.pending['STOR'][command] = d self._sendMessage("DIAG?STOR "+command) return d else: return self.pending['STOR'][command] def sendDIAGRTRV(self, command): logger.debug("sendDIAGRTRV") if not self.pending['RTRV'].has_key(command): d = defer.Deferred() self.pending['RTRV'][command] = d self._sendMessage("DIAG?RTRV "+command) return d else: return self.pending['RTRV'][command] def sendDIAGVRFY(self, command): logger.debug("sendDIAGVRFY") if not self.pending['VRFY'].has_key(command): d = defer.Deferred() self.pending['VRFY'][command] = d self._sendMessage("DIAG?VRFY "+command) return d else: return self.pending['VRFY'][command] def sendDIAGFNDV(self, val): logger.debug("sendDIAGFNDV") if not self.pending['FNDV'].has_key(val): d = defer.Deferred() self.pending['FNDV'][val] = d self._sendMessage("FNDV?"+val) return d else: return self.pending['FNDV'][val] def setDie(self): self.die = True # XXX: this should move into FludNode side of things (LocalClientPrimitives). # anything that calls this should make calls ('LIST', others as necessary) to # get at master metadata, otherwise we could have multiple writer problems. # FludNode should make the file ro while running, too. # And everyone that does anything with the master metadata should do it through # methods of FludConfig, instead of by direct access to the file. def listMeta(config): fmaster = open(os.path.join(config.metadir,config.metamaster), 'r') master = fmaster.read() fmaster.close() if master == "": master = {} else: master = fdecode(master) return master
Python
""" ServerDHTPrimitives.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Primitive server DHT protocol """ import binascii, time, os, stat, httplib, gc, re, sys, logging, random, sets from twisted.web.resource import Resource from twisted.web import server, resource, http, client from twisted.internet import reactor, defer from twisted.python import failure from flud.FludCrypto import FludRSA from flud.fencode import fencode, fdecode from ServerPrimitives import ROOT from FludCommUtil import * logger = logging.getLogger("flud.server.dht") # XXX: move kRouting.insertNode code out of FludConfig. Add a 'addNode' method # to FludProtocol module which calls config.addNode, calls # kRouting.insertNode and if it gets a return value, calls sendGetID with # the callback doing nothing (unless the header comes back in error) and # the errback calling kRouting.replaceNode. Anywhere that we are # currently called config.addNode, call this new method instead. # FUTURE: check flud protocol version for backwards compatibility # XXX: need to make sure we have appropriate timeouts for all comms. # FUTURE: DOS attacks. For now, assume that network hardware can filter these # out (by throttling individual IPs) -- i.e., it isn't our problem. If we # want to defend against this at some point, we need to keep track of who # is generating requests and then ignore them. # XXX: find everywhere we are sending longs and consider sending hex (or our # own base-64) encoded instead # XXX: might want to consider some self-healing for the kademlia layer, as # outlined by this thread: # http://zgp.org/pipermail/p2p-hackers/2003-August/001348.html (should also # consider Zooko's links in the parent to this post) """ The children of ROOT beginning with 'k' are kademlia protocol based. """ # XXX: need to do all the challenge/response jazz in the k classes class kFINDNODE(ROOT): isLeaf = True def render_GET(self, request): """ Return the k closest nodes to the target ID from local k-routing table """ self.setHeaders(request) self.node.DHTtstamp = time.time() try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'key') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by kFINDNODE" logger.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: logger.info("received kFINDNODE request from %s..." % params['nodeID'][:10]) reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) host = getCanonicalIP(request.getClientIP()) #return "{'id': '%s', 'k': %s}"\ # % (self.config.nodeID,\ # self.config.routing.findNode(fdecode(params['key']))) kclosest = self.config.routing.findNode(fdecode(params['key'])) notclose = list(set(self.config.routing.knownExternalNodes()) - set(kclosest)) if len(notclose) > 0 and len(kclosest) > 1: r = random.choice(notclose) #logger.info("**** got some notclose: %s:%d ****" % (r[0],r[1])) kclosest.append(r) #logger.info("returning kFINDNODE response: %s" % kclosest) updateNode(self.node.client, self.config, host, int(params['port']), reqKu, params['nodeID']) return "{'id': '%s', 'k': %s}" % (self.config.nodeID, kclosest) class kSTORE_true(ROOT): # unrestricted kSTORE. Will store any key/value pair, as in generic # kademlia. This should be unregistered in FludServer (can't allow # generic stores). isLeaf = True def render_PUT(self, request): self.setHeaders(request) try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'key', 'val') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by kSTORE_true" logger.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: logger.info("received kSTORE_true request from %s..." % params['nodeID'][:10]) reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) host = getCanonicalIP(request.getClientIP()) updateNode(self.node.client, self.config, host, int(params['port']), reqKu, params['nodeID']) fname = self.config.kstoredir+'/'+params['key'] logger.info("storing dht data to %s" % fname) f = open(fname, "wb") f.write(params['val']) f.close() return "" class kSTORE(ROOT): # XXX: To prevent abuse of the DHT layer, we impose restrictions on its # format. But format alone is not sufficient -- a malicious client # could still format its data in a way that is allowed and gain # arbitrary amounts of freeloading storage space in the DHT. To # prevent this, nodes storing data in the DHT layer must also validate # it. Validation simply requires that the blockIDs described in the # kSTORE actually reside at a significant percentage of the hosts # described in the kSTORE op. In other words, validation requires a # VERIFY op for each block described in the kSTORE op. Validation can # occur randomly sometime after a kSTORE operation, or at the time of # the kSTORE op. The former is better, because it not only allows # purging bad kSTOREs, but prevents them from happening in the first # place (without significant conspiring among all participants). # Since the originator of this request also needs to do a VERIFY, # perhaps we can piggyback these through some means. And, since the # kSTORE is replicated to k other nodes, each of which also should to # a VERIFY, there are several ways to optimize this. One is for the k # nodes to elect a single verifier, and allow the client to learn the # result of the VERIFY op. Another is to allow each k node to do its # own VERIFY, but stagger them in a way such that they can take the # place of the originator's first k VERIFY ops. This could be # coordinated or (perhaps better) allow each k node to randomly pick a # time at which it will VERIFY, distributed over a period for which it # is likely to cover many of the first k VERIFY ops generated by # the originator. The random approach is nice because it is the same # mechanism used by the k nodes to occasionally verify that the DHT # data is valid and should not be purged. isLeaf = True def render_PUT(self, request): self.setHeaders(request) self.node.DHTtstamp = time.time() try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'key', 'val') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by kSTORE" logger.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: logger.info("received kSTORE request from %s..." % params['nodeID'][:10]) reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) host = getCanonicalIP(request.getClientIP()) updateNode(self.node.client, self.config, host, int(params['port']), reqKu, params['nodeID']) fname = self.config.kstoredir+'/'+params['key'] md = fdecode(params['val']) if not self.dataAllowed(params['key'], md, params['nodeID']): msg = "malformed store data" logger.info("bad data was: %s" % md) request.setResponseCode(http.BAD_REQUEST, msg) return msg # XXX: see if there isn't already a 'val' for 'key' present # - if so, compare to val. Metadata can differ. Blocks # shouldn't. However, if blocks do differ, just add the # new values in, up to N (3?) records per key. Flag these # (all N) as ones we want to verify (to storer and storee). # Expunge any blocks that fail verify, and punish storer's # trust. logger.info("storing dht data to %s" % fname) if os.path.exists(fname) and isinstance(md, dict): f = open(fname, "rb") edata = f.read() f.close() md = self.mergeMetadata(md, fdecode(edata)) f = open(fname, "wb") f.write(fencode(md)) f.close() return "" # XXX: return a VERIFY reverse request: segname, offset def dataAllowed(self, key, data, nodeID): # ensures that 'data' is in [one of] the right format[s] (helps prevent # DHT abuse) def validValue(val): if not isinstance(val, long) and not isinstance(val, int): return False # not a valid key/nodeid if val > 2**256 or val < 0: # XXX: magic 2**256, use fludkrouting return False # not a valid key/nodeid return True def validMetadata(blockdata, nodeID): # returns true if the format of data conforms to the standard for # metadata blocks = 0 try: k = blockdata.pop('k') n = blockdata.pop('n') if not isinstance(k, int) or not isinstance(n, int): return False if k != 20 or n != 20: # XXX: magic numbers '20' # XXX: to support other than 20/20, need to constrain an # upper bound and store multiple records with different m/n # under the same key return False m = k+n except: return False for (i, b) in blockdata: if i > m: return False if not validValue(b): #print "%s is invalid key" %i return False location = blockdata[(i,b)] if isinstance(location, list): if len(location) > 5: #print "too many (list) nodeIDs" % j return False for j in location: if not validValue(j): #print "%s is invalid (list) nodeID" % j return False elif not validValue(location): #print "%s is invalid nodeID" % location return False blocks += 1 if blocks != m: return False # not the right number of blocks blockdata['k'] = k blockdata['n'] = n return True def validMasterCAS(key, data, nodeID): # returns true if the data fits the characteristics of a master # metadata CAS key, i.e., if key==nodeID and the data is the right # length. nodeID = fencode(long(nodeID,16)) if key != nodeID: return False # XXX: need to do challange/response on nodeID (just as in the # regular primitives) here, or else imposters can store/replace # this very important data!!! # XXX: do some length stuff - should only be as long as a CAS key return True return (validMetadata(data, nodeID) or validMasterCAS(key, data, nodeID)) def mergeMetadata(self, m1, m2): # merges the data from m1 into m2. After calling, both m1 and m2 # contain the merged data. """ >>> a1 = {'b': {1: (1, 'a', 8), 2: (2, 'b', 8), 5: (1, 'a', 8)}} >>> a2 = {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80)}} >>> mergeit(a1, a2) {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80), (2, 'b', 8)], 10: (10, 't', 80), 5: (1, 'a', 8)}} >>> a1 = {'b': {1: (1, 'a', 8), 2: (2, 'b', 8), 5: (1, 'a', 8)}} >>> a2 = {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80)}} >>> mergeit(a2, a1) {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80), (2, 'b', 8)], 10: (10, 't', 80), 5: (1, 'a', 8)}} >>> a1 = {'b': {1: (1, 'a', 8), 2: [(2, 'b', 8), (7, 'r', 8)], 5: (1, 'a', 8)}} >>> a2 = {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80)}} >>> mergeit(a2, a1) {'b': {1: (1, 'a', 8), 2: [(2, 'b', 8), (7, 'r', 8), (3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80), 5: (1, 'a', 8)}} >>> a1 = {'b': {1: [(1, 'a', 8)], 2: [(2, 'b', 8), (7, 'r', 8)], 5: (1, 'a', 8)}} >>> a2 = {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80)}} >>> mergeit(a1, a2) {'b': {1: (1, 'a', 8), 2: [(2, 'b', 8), (7, 'r', 8), (3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80), 5: (1, 'a', 8)}} """ # first merge blocks ('b' sections) n = {} for i in m2: if m1.has_key(i) and m2[i] != m1[i]: if isinstance(m1[i], list) and len(m1[i]) == 1: m1[i] = m1[i][0] # collapse list of len 1 if isinstance(m2[i], list) and len(m2[i]) == 1: m2[i] = m2[i][0] # collapse list of len 1 # combine if isinstance(m1[i], list) and isinstance(m2[i], list): n[i] = m2[i] n[i].extend(m1[i]) elif isinstance(m2[i], list): n[i] = m2[i] n[i] = n[i].append(m1[i]) elif isinstance(m1[i], list): n[i] = m1[i] n[i] = n[i].append(m2[i]) elif m1[i] == m2[i]: n[i] = m1[i] else: n[i] = [m1[i], m2[i]] else: n[i] = m2[i] for i in m1: if not n.has_key(i): n[i] = m1[i] # now n contains the merged blocks. m1 = m2 = n return m1 class kFINDVAL(ROOT): isLeaf = True def render_GET(self, request): """ Return the value, or if we don't have it, the k closest nodes to the target ID """ self.setHeaders(request) self.node.DHTtstamp = time.time() try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'key') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by kFINDVALUE" logger.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: logger.info("received kFINDVALUE request from %s..." % params['nodeID'][:10]) reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) host = getCanonicalIP(request.getClientIP()) updateNode(self.node.client, self.config, host, int(params['port']), reqKu, params['nodeID']) fname = self.config.kstoredir+'/'+params['key'] if os.path.isfile(fname): f = open(fname, "rb") logger.info("returning data from kFINDVAL") request.setHeader('nodeID',str(self.config.nodeID)) request.setHeader('Content-Type','application/x-flud-data') d = fdecode(f.read()) if isinstance(d, dict) and d.has_key(params['nodeID']): #print d resp = {'b': d['b'], params['nodeID']: d[params['nodeID']]} #resp = {'b': d['b']} #if d.has_key(params['nodeID']): # resp[params['nodeID']] = d[params['nodeID']] else: resp = d request.write(fencode(resp)) f.close() return "" else: # return the following if it isn't there. logger.info("returning nodes from kFINDVAL for %s" % params['key']) request.setHeader('Content-Type','application/x-flud-nodes') return "{'id': '%s', 'k': %s}"\ % (self.config.nodeID,\ self.config.routing.findNode(fdecode(params['key'])))
Python
#!/usr/bin/python """ LocalClient.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. LocalClient provides client functions which can be called to send commands to a local FludNode instance. """ import sys, os, time from twisted.internet.protocol import ClientFactory from twisted.protocols.basic import LineReceiver from twisted.internet import reactor from twisted.python import threadable threadable.init() from flud.fencode import fencode, fdecode from LocalPrimitives import * logger = logging.getLogger("flud.local.client") opTimeout = 1200 VALIDOPS = LocalProtocol.commands.keys() + ['AUTH', 'DIAG'] # XXX: print commands should either be raised, or put on factory.msgs class LocalClient(LineReceiver): MAX_LENGTH = 300000 auth=False def connectionMade(self): logger.debug("connection est.") self.auth=False self.sendLine("AUTH?") def lineReceived(self, line): logger.debug("received line '%s'" % line) command = line[0:4] if not command in VALIDOPS: print "error: invalid command op ('%s')-- "\ " are you trying to connect to the wrong port"\ " (local client port is usually external port + 500)?"\ % command return None status = line[4] data = line[5:] if not self.auth: if command == "AUTH" and status == '?': # got challenge, send response logger.debug("got AUTH challenge, sending response") echallenge = data self.sendLine("AUTH:"+self.factory.answerChallenge(echallenge)) return elif command == "AUTH" and status == ':': # response accepted, authenticated logger.debug("AUTH challenge accepted, success") self.auth = True self.factory.clientReady(self) #print "authenticated" else: if command == "AUTH" and status == "!": logger.warn("authentication failed (is FLUDHOME set" " correctly?)") print "authentication failed (is FLUDHOME set correctly?)" else: logger.warn("unknown message received before being" " authenticated:") logger.warn(" %s : %s" % (command, status)) print "unknown message received before being authenticated:" print " %s : %s" % (command, status) self.factory.setDie() elif command == "DIAG": subcommand = data[:4] data = data[4:] if subcommand == "NODE": logger.debug("DIAG NODE: %s" % data) data = fdecode(data) result = "" for i in data: score = i[4] petID = "%064x" % i[2] netID = "%s:%d" % (i[0], i[1]) petID = petID[:(70-len(netID))]+"..." result += "%d %s %s\n" % (score, netID, petID) result += "%d known nodes\n" % len(data) d = self.factory.pending['NODE'].pop('') d.callback(result) return if subcommand == "BKTS": logger.debug("DIAG BKTS") data = fdecode(data) result = "" for i in data: for bucket in i: result += "Bucket %s:\n" % bucket for k in i[bucket]: id = "%064x" % k[2] netID = "%s:%d" % (k[0], k[1]) result += " %s %s...\n" \ % (netID,id[:72-len(netID)]) d = self.factory.pending['BKTS'].pop('') d.callback(result) return elif status == ':': response, data = data.split(status, 1) logger.debug("DIAG %s: success" % subcommand) d = self.factory.pending[subcommand].pop(data) d.callback(fdecode(response)) elif status == "!": response, data = data.split(status, 1) logger.debug("DIAG %s: failure" % subcommand) d = self.factory.pending[subcommand].pop(data) d.errback(failure.DefaultException(response)) elif status == ':': response, data = data.split(status, 1) logger.debug("%s: success" % command) d = self.factory.pending[command].pop(data) d.callback(fdecode(response)) elif status == "!": response, data = data.split(status, 1) logger.debug("%s: failure" % command) if self.factory.pending.has_key(command): if not self.factory.pending[command].has_key(data): print "data key is '%s'" % data print "pending is '%s'" % self.factory.pending[command] if len(self.factory.pending[command]): d = self.factory.pending[command].popitem() d.errback(failure.DefaultException(response)) else: d = self.factory.pending[command].pop(data) d.errback(failure.DefaultException(response)) else: print "failed command '%s' not in pending?" % command print "pending is: %s" % self.factory.pending if command != 'AUTH' and command != 'DIAG' and \ not None in self.factory.pending[command].values(): logger.debug("%s done at %s" % (command, time.ctime())) class LocalClientFactory(ClientFactory): protocol = LocalClient def __init__(self, config): self.config = config self.messageQueue = [] self.client = None self.die = False self.pending = {'PUTF': {}, 'CRED': {}, 'GETI': {}, 'GETF': {}, 'FNDN': {}, 'STOR': {}, 'RTRV': {}, 'VRFY': {}, 'FNDV': {}, 'CRED': {}, 'LIST': {}, 'GETM': {}, 'PUTM': {}, 'NODE': {}, 'BKTS': {}} def clientConnectionFailed(self, connector, reason): #print "connection failed: %s" % reason logger.warn("connection failed: %s" % reason) self.cleanup("connection failed: %s" % reason) def clientConnectionLost(self, connector, reason): #print "connection lost: %s" % reason logger.debug("connection lost: %s" % reason) self.cleanup("connection lost: %s" % reason) def cleanup(self, msg): # override me for cleanup print msg; def clientReady(self, instance): self.client = instance logger.debug("client ready, sending [any] queued msgs") for i in self.messageQueue: self._sendMessage(i) def _sendMessage(self, msg): if self.client: logger.debug("sending msg '%s'" % msg) self.client.sendLine(msg) else: logger.debug("queueing msg '%s'" % msg) self.messageQueue.append(msg) def answerChallenge(self, echallenge): logger.debug("answering challenge") echallenge = (fdecode(echallenge),) challenge = self.config.Kr.decrypt(echallenge) return challenge def expire(self, pending, key): if pending.has_key(fname): logger.debug("timing out operation for %s" % key) #print "timing out operation for %s" % key pending.pop(key) def addFile(self, type, fname): logger.debug("addFile %s %s" % (type, fname)) if not self.pending[type].has_key(fname): d = defer.Deferred() self.pending[type][fname] = d self._sendMessage(type+"?"+fname) return d else: return self.pending[type][fname] def sendPING(self, host, port): logger.debug("sendPING") d = defer.Deferred() d.errback(failure.DefaultException( "ping not yet implemented in FludLocalClient")) return d def sendPUTF(self, fname): logger.debug("sendPUTF %s" % fname) if os.path.isdir(fname): dirlist = os.listdir(fname) dlist = [] for i in dirlist: dlist.append(self.sendPUTF(os.path.join(fname,i))) dl = defer.DeferredList(dlist) return dl elif not self.pending['PUTF'].has_key(fname): d = defer.Deferred() self.pending['PUTF'][fname] = d self._sendMessage("PUTF?"+fname) #reactor.callLater(opTimeout, self.expire, self.pendingPUTF, fname) return d else: return self.pending['PUTF'][fname] def sendCRED(self, passphrase, email): logger.debug("sendCRED") key = fencode((self.config.Ku.encrypt(passphrase)[0], email)) if not self.pending['CRED'].has_key(key): d = defer.Deferred() self.pending['CRED'][key] = d self._sendMessage("CRED?"+key) return d else: return self.pending['CRED'][key] def sendGETI(self, fID): logger.debug("sendGETI") if not self.pending['GETI'].has_key(fID): d = defer.Deferred() self.pending['GETI'][fID] = d self._sendMessage("GETI?"+fID) return d else: return self.pending['GETI'][fID] def sendGETF(self, fname): logger.debug("sendGETF") master = listMeta(self.config) if master.has_key(fname): return self.addFile("GETF",fname) elif fname[-1:] == os.path.sep: dlist = [] for name in master: if fname == name[:len(fname)]: dlist.append(self.addFile("GETF",name)) dl = defer.DeferredList(dlist) return dl def sendFNDN(self, nID): logger.debug("sendFNDN") if not self.pending['FNDN'].has_key(nID): d = defer.Deferred() self.pending['FNDN'][nID] = d self._sendMessage("FNDN?"+nID) return d else: return self.pending['FNDN'][nID] def sendLIST(self): logger.debug("sendLIST") if not self.pending['LIST'].has_key(""): d = defer.Deferred() self.pending['LIST'][''] = d logger.debug("LIST['']=%s" % d) self._sendMessage("LIST?") return d else: return self.pending['LIST'][''] def sendGETM(self): logger.debug("sendGETM") if not self.pending['GETM'].has_key(''): d = defer.Deferred() self.pending['GETM'][''] = d logger.debug("GETM['']=%s" % d) self._sendMessage("GETM?") return d else: return self.pending['GETM'][''] def sendPUTM(self): logger.debug("sendPUTM") if not self.pending['PUTM'].has_key(''): d = defer.Deferred() self.pending['PUTM'][''] = d self._sendMessage("PUTM?") return d else: return self.pending['PUTM'][''] def sendDIAGNODE(self): logger.debug("sendDIAGNODE") if not self.pending['NODE'].has_key(''): d = defer.Deferred() self.pending['NODE'][''] = d self._sendMessage("DIAG?NODE") return d else: return self.pending['NODE'][''] def sendDIAGBKTS(self): logger.debug("sendDIAGBKTS") if not self.pending['BKTS'].has_key(''): d = defer.Deferred() self.pending['BKTS'][''] = d self._sendMessage("DIAG?BKTS") return d else: return self.pending['BKTS'][''] def sendDIAGSTOR(self, command): logger.debug("sendDIAGSTOR") if not self.pending['STOR'].has_key(command): d = defer.Deferred() self.pending['STOR'][command] = d self._sendMessage("DIAG?STOR "+command) return d else: return self.pending['STOR'][command] def sendDIAGRTRV(self, command): logger.debug("sendDIAGRTRV") if not self.pending['RTRV'].has_key(command): d = defer.Deferred() self.pending['RTRV'][command] = d self._sendMessage("DIAG?RTRV "+command) return d else: return self.pending['RTRV'][command] def sendDIAGVRFY(self, command): logger.debug("sendDIAGVRFY") if not self.pending['VRFY'].has_key(command): d = defer.Deferred() self.pending['VRFY'][command] = d self._sendMessage("DIAG?VRFY "+command) return d else: return self.pending['VRFY'][command] def sendDIAGFNDV(self, val): logger.debug("sendDIAGFNDV") if not self.pending['FNDV'].has_key(val): d = defer.Deferred() self.pending['FNDV'][val] = d self._sendMessage("FNDV?"+val) return d else: return self.pending['FNDV'][val] def setDie(self): self.die = True # XXX: this should move into FludNode side of things (LocalClientPrimitives). # anything that calls this should make calls ('LIST', others as necessary) to # get at master metadata, otherwise we could have multiple writer problems. # FludNode should make the file ro while running, too. # And everyone that does anything with the master metadata should do it through # methods of FludConfig, instead of by direct access to the file. def listMeta(config): fmaster = open(os.path.join(config.metadir,config.metamaster), 'r') master = fmaster.read() fmaster.close() if master == "": master = {} else: master = fdecode(master) return master
Python
""" ConnectionQueue, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. This module manages the connection queue. In order to reduce the probability of the reactor getting tied up servicing requests/responses during periods of extreme busy-ness (and thus 'starving' some ops, causing TimeoutErrors), we throttle the number of outstanding requests that we send to MAXOPS. The rest, we put in the 'waiting' queue, and these are popped off when a spot becomes available. """ import logging MAXOPS = 80 # maximum number of concurrent connections to maintain pending = 0 # number of current connections waiting = [] # queue of waiting connections to make. This queue contains # tuples. The first element of the tuple must have a # startRequest() func that takes the rest of the tuple as # arguments. logger = logging.getLogger("flud.client.connq") def checkWaiting(resp, finishedOne=True): """ This function pops items off the waiting queue and invokes their startRequest() method. It should eventually be called by any process that also calls queueWaiting() (usually as part of callback/errback chain). The 'resp' object passed in will be returned (so that this function can sit transparently in the errback/callback chain). """ numwaiting = len(waiting) logger.debug("in checkWaiting, len(waiting) = %d" % numwaiting) #print "resp = %s..." % fencode(long(resp,16))[:8] #print "resp = %s..." % str(resp) global pending if finishedOne: pending = pending - 1 logger.debug("decremented pending to %s" % pending) if numwaiting > 0 and pending < MAXOPS: saved = waiting.pop(0) Req = saved[0] args = saved[1:] logger.debug("w: %d, p: %d, restoring Request %s(%s)" % (numwaiting, pending, Req.__class__.__name__, str(args))) Req.startRequest(*args) pending += 1 return resp def enqueue(requestTuple): """ Adds a requestTuple to those waiting. The first element of the tuple must have a startRequest() func that takes the rest of the tuple as arguments. This startRequest() function will be called with those arguments when it comes off the queue (via checkWaiting). """ waiting.append(requestTuple) logger.debug("trying to do %s now..." % requestTuple[0].__class__.__name__) checkWaiting(None, finishedOne=False)
Python
""" ClientPrimitives.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Primitive client storage protocol """ from twisted.web import http, client from twisted.internet import reactor, threads, defer, error from twisted.python import failure import time, os, stat, httplib, sys, logging, tarfile, gzip from StringIO import StringIO from flud.FludCrypto import FludRSA from flud.fencode import fencode, fdecode import ConnectionQueue from FludCommUtil import * logger = logging.getLogger("flud.client.op") loggerid = logging.getLogger("flud.client.op.id") loggerstor = logging.getLogger("flud.client.op.stor") loggerstoragg = logging.getLogger("flud.client.op.stor.agg") loggerrtrv = logging.getLogger("flud.client.op.rtrv") loggerdele = logging.getLogger("flud.client.op.dele") loggervrfy = logging.getLogger("flud.client.op.vrfy") loggerauth = logging.getLogger("flud.client.op.auth") MINSTORSIZE = 512000 # anything smaller than this tries to get aggregated TARFILE_TO = 2 # timeout for checking aggregated tar files MAXAUTHRETRY = 4 # number of times to retry auth # FUTURE: check flud protocol version for backwards compatibility # XXX: need to make sure we have appropriate timeouts for all comms. # FUTURE: DOS attacks. For now, assume that network hardware can filter these # out (by throttling individual IPs) -- i.e., it isn't our problem. If we # want to defend against this at some point, we need to keep track of who # is generating requests and then ignore them. # XXX: might want to consider some self-healing for the kademlia layer, as # outlined by this thread: # http://zgp.org/pipermail/p2p-hackers/2003-August/001348.html (should also # consider Zooko's links in the parent to this post) # XXX: disallow requests to self. class REQUEST(object): """ This is a parent class for generating http requests that follow the FludProtocol. """ def __init__(self, host, port, node=None): """ All children should inherit. By convention, subclasses should create a URL and attempt to retrieve it in the constructor. @param node the requestor's node object """ self.host = host self.port = port self.dest = "%s:%d" % (host, port) if node: self.node = node self.config = node.config self.headers = {'Fludprotocol': fludproto_ver, 'User-Agent': 'FludClient 0.1'} class SENDGETID(REQUEST): def __init__(self, node, host, port): """ Send a request to retrive the node's ID. This is a reciprocal request -- must send my own ID in order to get one back. """ host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) Ku = self.node.config.Ku.exportPublicKey() url = "http://"+host+":"+str(port)+"/ID?" url += 'nodeID='+str(self.node.config.nodeID) url += '&port='+str(self.node.config.port) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) #self.nKu = {} self.timeoutcount = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, node, host, port, url)) def startRequest(self, node, host, port, url): loggerid.info("sending SENDGETID to %s" % self.dest) d = self._sendRequest(node, host, port, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) d.addErrback(self._errID, node, host, port, url) def _sendRequest(self, node, host, port, url): factory = getPageFactory(url, timeout=primitive_to, headers=self.headers) d2 = factory.deferred d2.addCallback(self._getID, factory, host, port) d2.addErrback(self._errID, node, host, port, url) return d2 def _getID(self, response, factory, host, port): loggerid.debug( "received ID response: %s" % response) if not hasattr(factory, 'status'): raise failure.DefaultException( "SENDGETID FAILED: no status in factory") if eval(factory.status) != http.OK: raise failure.DefaultException("SENDGETID FAILED to "+self.dest+": " +"server sent status "+factory.status+", '"+response+"'") try: nKu = {} nKu = eval(response) nKu = FludRSA.importPublicKey(nKu) loggerid.info("SENDGETID PASSED to %s" % self.dest) updateNode(self.node.client, self.config, host, port, nKu) return nKu except: raise failure.DefaultException("SENDGETID FAILED to "+self.dest+": " +"received response, but it did not contain valid key") def _errID(self, err, node, host, port, url): if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #print "GETID request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(node, host, port, url) else: #print "not trying again [#%d]" % self.timeoutcount return err # XXX: updateNode #print "_errID: %s" % err #print "_errID: %s" % str(err.stack) return err # XXX: either 1) produce filekey here or 2) send it in as part of API # (similar fixes for SENDRETRIEVE and VERIFY)? Currently filekey is # chosen by caller, and is simply the filename. class SENDSTORE(REQUEST): def __init__(self, nKu, node, host, port, datafile, metadata=None, fsize=0): """ Try to upload a file. """ host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) loggerstor.info("sending STORE request to %s" % self.dest) if not fsize: fsize = os.stat(datafile)[stat.ST_SIZE] Ku = self.node.config.Ku.exportPublicKey() params = [('nodeID', self.node.config.nodeID), ('Ku_e', str(Ku['e'])), ('Ku_n', str(Ku['n'])), ('port', str(self.node.config.port)), ('filekey', os.path.basename(datafile)), ('size', str(fsize))] self.timeoutcount = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, self.headers, nKu, host, port, datafile, metadata, params, True)) #self.deferred = self._sendRequest(self.headers, nKu, host, port, # datafile, params, True) def startRequest(self, headers, nKu, host, port, datafile, metadata, params, skipFile): d = self._sendRequest(headers, nKu, host, port, datafile, metadata, params, skipFile) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, headers, nKu, host, port, datafile, metadata, params, skipfile=False): """ skipfile - set to True if you want to send everything but file data (used to send the unauthorized request before responding to challenge) """ if skipfile: files = [(None, 'filename')] elif metadata: metakey = metadata[0] params.append(('metakey', metakey)) metafile = metadata[1] files = [(datafile, 'filename'), (metafile, 'meta')] else: files = [(datafile, 'filename')] deferred = threads.deferToThread(fileUpload, host, port, '/STORE', files, params, headers=self.headers) deferred.addCallback(self._getSendStore, nKu, host, port, datafile, metadata, params, self.headers) deferred.addErrback(self._errSendStore, "Couldn't upload file %s to %s:%d" % (datafile, host, port), self.headers, nKu, host, port, datafile, metadata, params) return deferred def _getSendStore(self, httpconn, nKu, host, port, datafile, metadata, params, headers): """ Check the response for status. """ deferred2 = threads.deferToThread(httpconn.getresponse) deferred2.addCallback(self._getSendStore2, httpconn, nKu, host, port, datafile, metadata, params, headers) deferred2.addErrback(self._errSendStore, "Couldn't get response", headers, nKu, host, port, datafile, metadata, params, httpconn) return deferred2 def _getSendStore2(self, response, httpconn, nKu, host, port, datafile, metadata, params, headers): httpconn.close() if response.status == http.UNAUTHORIZED: loggerstor.info("SENDSTORE unauthorized, sending credentials") challenge = response.reason d = answerChallengeDeferred(challenge, self.node.config.Kr, self.node.config.groupIDu, nKu.id(), headers) d.addCallback(self._sendRequest, nKu, host, port, datafile, metadata, params) d.addErrback(self._errSendStore, "Couldn't answerChallenge", headers, nKu, host, port, datafile, metadata, params, httpconn) return d elif response.status == http.CONFLICT: result = response.read() # XXX: client should check key before ever sending request raise BadCASKeyException("%s %s" % (response.status, response.reason)) elif response.status != http.OK: result = response.read() raise failure.DefaultException( "received %s in SENDSTORE response: %s" % (response.status, result)) else: result = response.read() updateNode(self.node.client, self.config, host, port, nKu) loggerstor.info("received SENDSTORE response from %s: %s" % (self.dest, str(result))) return result def _errSendStore(self, err, msg, headers, nKu, host, port, datafile, metadata, params, httpconn=None): if err.check('socket.error'): #print "SENDSTORE request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(headers, nKu, host, port, datafile, metadata, params) else: print "Maxtimeouts exceeded: %d" % self.timeoutcount elif err.check(BadCASKeyException): pass else: print "%s: unexpected error in SENDSTORE: %s" % (msg, str(err.getErrorMessage())) # XXX: updateNode if httpconn: httpconn.close() #loggerstor.info(msg+": "+err.getErrorMessage()) return err aggDeferredMap = {} # a map of maps, containing a list of deferreds. The # deferred(s) for file 'x' in tarball 'y' are accessed as # aggDeferredMap['y']['x'] aggTimeoutMap = {} # a map of timout calls for a tarball. The timeout for # tarball 'y' is stored in aggTimeoutMap['y'] class AggregateStore: # XXX: if multiple guys store the same file, we're going to get into bad # cb state (the except clause in errbackTarfiles). Need to catch this # as it happens... (this happens e.g. for small files with the same # filehash, e.g, 0-byte files, file copies etc). Should fix this in # FludClient -- non-agg store has a similar problem (encoded file chunks # get deleted out from under successive STOR ops for the same chunk, i.e. # from two concurrent STORs of the same file contents) def __init__(self, nKu, node, host, port, datafile, metadata): tarfilename = os.path.join(node.config.clientdir,nKu.id())\ +'-'+host+'-'+str(port)+".tar" loggerstoragg.debug("tarfile name is %s" % tarfilename) if not os.path.exists(tarfilename) \ or not aggDeferredMap.has_key(tarfilename): loggerstoragg.debug("creating tarfile %s to append %s" % (tarfilename, datafile)) tar = tarfile.open(tarfilename, "w") tarfileTimeout = reactor.callLater(TARFILE_TO, self.sendTar, tarfilename, nKu, node, host, port) aggDeferredMap[tarfilename] = {} aggTimeoutMap[tarfilename] = tarfileTimeout else: loggerstoragg.debug("opening tarfile %s to append %s" % (tarfilename, datafile)) tar = tarfile.open(tarfilename, "a") if os.path.basename(datafile) not in tar.getnames(): loggerstoragg.info("adding datafile %s to tarball, %s" % (os.path.basename(datafile), tar.getnames())) loggerstoragg.debug("adding data to tarball") tar.add(datafile, os.path.basename(datafile)) else: loggerstoragg.info("skip adding datafile %s to tarball" % datafile) if metadata: metafilename = "%s.%s.meta" % (os.path.basename(datafile), metadata[0]) loggerstoragg.debug("metadata filename is %s" % metafilename) try: if isinstance(metadata[1], StringIO): loggerstoragg.debug("metadata is StringIO") tinfo = tarfile.TarInfo(metafilename) metadata[1].seek(0,2) tinfo.size = metadata[1].tell() metadata[1].seek(0,0) tar.addfile(tinfo, metadata[1]) else: loggerstoragg.debug("metadata is file") tar.add(metadata[1], metafilename) except: import traceback loggerstoragg.debug("exception while adding metadata to" " tarball") print sys.exc_info()[2] traceback.print_exc() tar.close() loggerstoragg.debug("prepping deferred") # XXX: (re)set timeout for tarfilename self.deferred = defer.Deferred() loggerstoragg.debug("adding deferred on %s for %s" % (tarfilename, datafile)) try: aggDeferredMap[tarfilename][os.path.basename(datafile)].append( self.deferred) except KeyError: aggDeferredMap[tarfilename][os.path.basename(datafile)] \ = [self.deferred] self.resetTimeout(aggTimeoutMap[tarfilename], tarfilename) def resetTimeout(self, timeoutFunc, tarball): loggerstoragg.debug("in resetTimeout...") if timeoutFunc.active(): #timeoutFunc.reset(TARFILE_TO) if os.stat(tarball)[stat.ST_SIZE] < MINSTORSIZE: loggerstoragg.debug("...reset") timeoutFunc.reset(TARFILE_TO) return loggerstoragg.debug("...didn't reset") def sendTar(self, tarball, nKu, node, host, port): gtarball = tarball+".gz" loggerstoragg.info( "aggregation op triggered, sending tarfile %s to %s:%d" % (gtarball, host, port)) # XXX: bad blocking io gtar = gzip.GzipFile(gtarball, 'wb') gtar.write(file(tarball, 'r').read()) gtar.close() os.remove(tarball) self.deferred = SENDSTORE(nKu, node, host, port, gtarball).deferred self.deferred.addCallback(self.callbackTarfiles, tarball) self.deferred.addErrback(self.errbackTarfiles, tarball) # XXX: make aggDeferredMap use a non-.tar key, so that we don't have to # keep passing 'tarball' around (since we removed it and are really only # interested in gtarball now, use gtarball at the least) def callbackTarfiles(self, result, tarball): loggerstoragg.debug("callbackTarfiles") gtarball = tarball+".gz" tar = tarfile.open(gtarball, "r:gz") cbs = [] try: for tarinfo in tar: if tarinfo.name[-5:] != '.meta': dlist = aggDeferredMap[tarball].pop(tarinfo.name) loggerstoragg.debug("callingback for %s in %s" " (%d deferreds)" % (tarinfo.name, tarball, len(dlist))) for d in dlist: cbs.append(d) except KeyError: loggerstoragg.warn("aggDeferredMap has keys: %s" % str(aggDeferredMap.keys())) loggerstoragg.warn("aggDeferredMap[%s] has keys: %s" % (tarball, str(aggDeferredMap[tarball].keys()))) tar.close() loggerstoragg.debug("deleting tarball %s" % gtarball) os.remove(gtarball) for cb in cbs: cb.callback(result) def errbackTarfiles(self, failure, tarball): loggerstoragg.debug("errbackTarfiles") gtarball = tarball+".gz" tar = tarfile.open(gtarball, "r:gz") cbs = [] try: for tarinfo in tar: dlist = aggDeferredMap[tarball].pop(tarinfo.name) loggerstoragg.debug("erringback for %s in %s" " (%d deferreds)" % (tarinfo.name, tarball, len(dlist))) for d in dlist: cbs.append(d) except KeyError: loggerstoragg.warn("aggDeferredMap has keys: %s" % str(aggDeferredMap.keys())) loggerstoragg.warn("aggDeferredMap[%s] has keys: %s" % (tarball, str(aggDeferredMap[tarball].keys()))) tar.close() loggerstoragg.debug("NOT deleting tarball %s (for debug)" % gtarball) #os.remove(gtarball) for cb in cbs: cb.errback(failure) class SENDRETRIEVE(REQUEST): def __init__(self, nKu, node, host, port, filekey, metakey=True): """ Try to download a file. """ host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) loggerrtrv.info("sending RETRIEVE request to %s:%s" % (host, str(port))) Ku = self.node.config.Ku.exportPublicKey() url = 'http://'+host+':'+str(port)+'/RETRIEVE/'+filekey+'?' url += 'nodeID='+str(self.node.config.nodeID) url += '&port='+str(self.node.config.port) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) url += "&metakey="+str(metakey) #filename = self.node.config.clientdir+'/'+filekey self.timeoutcount = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, self.headers, nKu, host, port, url)) def startRequest(self, headers, nKu, host, port, url): #print "doing RET: %s" % filename d = self._sendRequest(headers, nKu, host, port, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, headers, nKu, host, port, url): factory = multipartDownloadPageFactory(url, self.node.config.clientdir, headers=headers, timeout=transfer_to) deferred = factory.deferred deferred.addCallback(self._getSendRetrieve, nKu, host, port, factory) deferred.addErrback(self._errSendRetrieve, nKu, host, port, factory, url, headers) return deferred def _getSendRetrieve(self, response, nKu, host, port, factory): if eval(factory.status) == http.OK: # response is None, since it went to file (if a server error # occured, it may be printed in this file) # XXX: need to check that file hashes to key! If we don't do this, # malicious nodes can corrupt entire files without detection! #result = "received SENDRETRIEVE response" loggerrtrv.info(response) updateNode(self.node.client, self.config, host, port, nKu) return response else: raise failure.DefaultException("SENDRETRIEVE FAILED: " +"server sent status "+factory.status+", '"+response+"'") def _errSendRetrieve(self, err, nKu, host, port, factory, url, headers): if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #or \ #err.check('twisted.internet.error.ConnectBindError'): self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "RETR trying again [#%d]..." % self.timeoutcount return self._sendRequest(headers, nKu, host, port, url) else: #print "RETR timeout exceeded: %d" % self.timeoutcount pass elif hasattr(factory, 'status') and \ eval(factory.status) == http.UNAUTHORIZED: loggerrtrv.info("SENDRETRIEVE unauthorized, sending credentials") challenge = err.getErrorMessage()[4:] d = answerChallengeDeferred(challenge, self.node.config.Kr, self.node.config.groupIDu, nKu.id(), headers) d.addCallback(self._sendRequest, nKu, host, port, url) #d.addErrback(self._errSendRetrieve, nKu, host, port, factory, # url, headers) return d #extraheaders = answerChallenge(challenge, self.node.config.Kr, # self.node.config.groupIDu, nKu.id(), self.headers) #return self._sendRequest(nKu, host, port, url, extraheaders) # XXX: these remaining else clauses are really just for debugging... elif hasattr(factory, 'status'): if eval(factory.status) == http.NOT_FOUND: err = NotFoundException(err) elif eval(factory.status) == http.BAD_REQUEST: err = BadRequestException(err) elif err.check('twisted.internet.error.ConnectionRefusedError'): pass # fall through to return err else: print "non-timeout, non-UNAUTH RETR request error: %s" % err # XXX: updateNode loggerrtrv.info("SENDRETRIEVE failed") raise err class SENDDELETE(REQUEST): def __init__(self, nKu, node, host, port, filekey, metakey): """ Try to delete a file. """ host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) loggerdele.info("sending DELETE request to %s:%s" % (host, str(port))) Ku = self.node.config.Ku.exportPublicKey() url = 'http://'+host+':'+str(port)+'/DELETE/'+filekey+'?' url += 'nodeID='+str(self.node.config.nodeID) url += '&port='+str(self.node.config.port) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) url += "&metakey="+str(metakey) self.timeoutcount = 0 self.authRetry = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, self.headers, nKu, host, port, url)) def startRequest(self, headers, nKu, host, port, url): d = self._sendRequest(headers, nKu, host, port, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, headers, nKu, host, port, url): factory = getPageFactory(url, headers=headers, timeout=primitive_to) deferred = factory.deferred deferred.addCallback(self._getSendDelete, nKu, host, port, factory) deferred.addErrback(self._errSendDelete, nKu, host, port, factory, url, headers) return deferred def _getSendDelete(self, response, nKu, host, port, factory): if eval(factory.status) == http.OK: loggerdele.info("received SENDDELETE response") updateNode(self.node.client, self.config, host, port, nKu) return response else: # XXX: updateNode raise failure.DefaultException("SENDDELETE FAILED: " +"server sent status "+factory.status+", '"+response+"'") def _errSendDelete(self, err, nKu, host, port, factory, url, headers): if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #print "DELETE request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(headers, nKu, host, port, url) elif hasattr(factory, 'status') and \ eval(factory.status) == http.UNAUTHORIZED and \ self.authRetry < MAXAUTHRETRY: # XXX: add this authRetry stuff to all the other op classes (so # that we don't DOS ourselves and another node self.authRetry += 1 loggerdele.info("SENDDELETE unauthorized, sending credentials") challenge = err.getErrorMessage()[4:] d = answerChallengeDeferred(challenge, self.node.config.Kr, self.node.config.groupIDu, nKu.id(), headers) d.addCallback(self._sendRequest, nKu, host, port, url) d.addErrback(self._errSendDelete, nKu, host, port, factory, url, headers) return d elif hasattr(factory, 'status'): # XXX: updateNode loggerdele.info("SENDDELETE failed") if eval(factory.status) == http.NOT_FOUND: err = NotFoundException(err) elif eval(factory.status) == http.BAD_REQUEST: err = BadRequestException(err) raise err return err class SENDVERIFY(REQUEST): def __init__(self, nKu, node, host, port, filename, offset, length, meta=None): """ Try to verify a file. If meta is present, it should be a (metakey, filelikeobj) pair. """ host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) filekey = os.path.basename(filename) # XXX: filekey should be hash loggervrfy.info("sending VERIFY request to %s:%s" % (host, str(port))) Ku = self.node.config.Ku.exportPublicKey() url = 'http://'+host+':'+str(port)+'/VERIFY/'+filekey+'?' url += 'nodeID='+str(self.node.config.nodeID) url += '&port='+str(self.node.config.port) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) url += "&offset="+str(offset) url += "&length="+str(length) if meta: url += "&metakey="+str(meta[0]) url += "&meta="+fencode(meta[1].read()) self.timeoutcount = 0 if not isinstance(nKu, FludRSA): raise ValueError("must pass in a FludRSA as nKu to SENDVERIFY") self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, self.headers, nKu, host, port, url)) def startRequest(self, headers, nKu, host, port, url): #loggervrfy.debug("*Doing* VERIFY Request %s" % port) d = self._sendRequest(headers, nKu, host, port, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, headers, nKu, host, port, url): loggervrfy.debug("in VERIFY sendReq %s" % port) factory = getPageFactory(url, headers=headers, timeout=primitive_to) deferred = factory.deferred deferred.addCallback(self._getSendVerify, nKu, host, port, factory) deferred.addErrback(self._errSendVerify, nKu, host, port, factory, url, headers) return deferred def _getSendVerify(self, response, nKu, host, port, factory): loggervrfy.debug("got vrfy response") if eval(factory.status) == http.OK: loggervrfy.info("received SENDVERIFY response") updateNode(self.node.client, self.config, host, port, nKu) return response else: # XXX: updateNode loggervrfy.debug("received non-OK SENDVERIFY response") raise failure.DefaultException("SENDVERIFY FAILED: " +"server sent status "+factory.status+", '"+response+"'") def _errSendVerify(self, err, nKu, host, port, factory, url, headers): loggervrfy.debug("got vrfy err") if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #print "VERIFY request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(headers, nKu, host, port, url) elif hasattr(factory, 'status') and \ eval(factory.status) == http.UNAUTHORIZED: loggervrfy.info("SENDVERIFY unauthorized, sending credentials") challenge = err.getErrorMessage()[4:] d = answerChallengeDeferred(challenge, self.node.config.Kr, self.node.config.groupIDu, nKu.id(), headers) d.addCallback(self._sendRequest, nKu, host, port, url) d.addErrback(self._errVerify, nKu, host, port, factory, url, headers, challenge) #d.addErrback(self._errSendVerify, nKu, host, port, factory, # url, headers) return d elif hasattr(factory, 'status'): # XXX: updateNode loggervrfy.info("SENDVERIFY failed: %s" % err.getErrorMessage()) if eval(factory.status) == http.NOT_FOUND: err = NotFoundException(err) elif eval(factory.status) == http.BAD_REQUEST: err = BadRequestException(err) raise err def _errVerify(self, err, nKu, host, port, factory, url, headers, challenge): # we can get in here after storing the same file as another node when # that data is stored in tarballs under its ID. It was expected that # this would be caught up in _getSendVerify... figure out why it isn't. loggervrfy.debug("factory status=%s" % factory.status) loggervrfy.debug("couldn't answer challenge from %s:%d, WHOOPS: %s" % (host, port, err.getErrorMessage())) loggervrfy.debug("challenge was: '%s'" % challenge) return err def answerChallengeDeferred(challenge, Kr, groupIDu, sID, headers): return threads.deferToThread(answerChallenge, challenge, Kr, groupIDu, sID, headers) def answerChallenge(challenge, Kr, groupIDu, sID, headers={}): loggerauth.debug("got challenge: '%s'" % challenge) sID = binascii.unhexlify(sID) challenge = (fdecode(challenge),) response = fencode(Kr.decrypt(challenge)) # XXX: RSA.decrypt won't restore leading 0's. This causes # some challenges to fail when they shouldn't -- solved for now # on the server side by generating non-0 leading challenges. loggerauth.debug("decrypted challenge to %s" % response) responseID = fdecode(response)[:len(sID)] loggerauth.debug(" response id: %s" % fencode(responseID)) if responseID != sID: # fail the op. # If we don't do this, we may be allowing the server to build a # dictionary useful for attack. The attack is as follows: node A # (server) collects a bunch of un-IDed challenge/response pairs by # issuing challenges to node B (client). Then node A uses those # responses to pose as B to some other server C. This sounds # farfetched, in that such a database would need to be huge, but in # reality, such an attack can happen in real-time, with node A # simultaneously serving requests from B, relaying challenges from C to # B, and then responding with B's responses to C to gain resources # there as an imposter. The ID string prevents this attack. # XXX: trust-- (must go by ip:port, since ID could be innocent) raise ImposterException("node %s is issuing invalid challenges --" " claims to have id=%s" % (fencode(sID), fencode(responseID))) response = fdecode(response)[len(sID):] loggerauth.debug(" challenge response: '%s'" % fencode(response)) response = fencode(response)+":"+groupIDu loggerauth.debug("response:groupIDu=%s" % response) response = binascii.b2a_base64(response) loggerauth.debug("b64(response:groupIDu)=%s" % response) response = "Basic %s" % response headers['Authorization'] = response return headers
Python
""" FludClient.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. flud client ops. """ from twisted.web import client from twisted.internet import error import os, stat, httplib, sys, logging from ClientPrimitives import * from ClientDHTPrimitives import * import FludCommUtil logger = logging.getLogger('flud.client') class FludClient(object): """ This class contains methods which create request objects """ def __init__(self, node): self.node = node self.currentStorOps = {} """ Data storage primitives """ def redoTO(self, f, node, host, port): print "in redoTO: %s" % f #print "in redoTO: %s" % dir(f.getTraceback()) if f.getTraceback().find("error.TimeoutError"): print "retrying........" return self.sendGetID(host, port) else: return f def sendGetID(self, host, port): #return SENDGETID(self.node, host, port).deferred d = SENDGETID(self.node, host, port).deferred #d.addErrback(self.redoTO, self.node, host, port) return d # XXX: we should cache nKu so that we don't do the GETID for all of these # ops every single time def sendStore(self, filename, metadata, host, port, nKu=None): # XXX: need to keep a map of 'filename' to deferreds, in case we are # asked to store the same chunk more than once concurrently (happens # for 0-byte files or from identical copies of the same file, for # example). both SENDSTORE and AggregateStore will choke on this. # if we find a store req in said map, just return that deferred instead # of redoing the op. [note, could mess up node choice... should also do # this on whole-file level in FileOps] # XXX: need to remove from currentStorOps on success or failure key = "%s:%d:%s" % (host, port, filename) if self.currentStorOps.has_key(key): logger.debug("returning saved deferred for %s in sendStore" % filename) return self.currentStorOps[key] def sendStoreWithnKu(nKu, host, port, filename, metadata): return SENDSTORE(nKu, self.node, host, port, filename, metadata).deferred def removeKey(r, key): self.currentStorOps.pop(key) return r if not nKu: # XXX: doesn't do AggregateStore if file is small. Can fix by # moving this AggStore v. SENDSTORE choice into SENDSTORE # proper logger.warn("not doing AggregateStore on small file because" " of missing nKu") print "not doing AggregateStore on small file because" \ " of missing nKu" d = self.sendGetID(host, port) d.addCallback(sendStoreWithnKu, host, port, filename, metadata) self.currentStorOps[key] = d return d fsize = os.stat(filename)[stat.ST_SIZE]; if fsize < MINSTORSIZE: logger.debug("doing AggStore") if metadata: logger.debug("with metadata") d = AggregateStore(nKu, self.node, host, port, filename, metadata).deferred else: logger.debug("SENDSTORE") d = SENDSTORE(nKu, self.node, host, port, filename, metadata).deferred self.currentStorOps[key] = d d.addBoth(removeKey, key) return d # XXX: need a version that takes a metakey, too def sendRetrieve(self, filekey, host, port, nKu=None, metakey=True): def sendRetrieveWithNKu(nKu, host, port, filekey, metakey=True): return SENDRETRIEVE(nKu, self.node, host, port, filekey, metakey).deferred if not nKu: d = self.sendGetID(host, port) d.addCallback(sendRetrieveWithNKu, host, port, filekey, metakey) return d else: return SENDRETRIEVE(nKu, self.node, host, port, filekey, metakey).deferred def sendVerify(self, filekey, offset, length, host, port, nKu=None, meta=None): def sendVerifyWithNKu(nKu, host, port, filekey, offset, length, meta=True): return SENDVERIFY(nKu, self.node, host, port, filekey, offset, length, meta).deferred if not nKu: d = self.sendGetID(host, port) d.addCallback(sendVerifyWithNKu, host, port, filekey, offset, length, meta) return d else: s = SENDVERIFY(nKu, self.node, host, port, filekey, offset, length, meta) return s.deferred def sendDelete(self, filekey, metakey, host, port, nKu=None): def sendDeleteWithNKu(nKu, host, port, filekey, metakey): return SENDDELETE(nKu, self.node, host, port, filekey, metakey).deferred if not nKu: d = self.sendGetID(host, port) d.addCallback(sendDeleteWithNKu, host, port, filekey, metakey) return d else: return SENDDELETE(nKu, self.node, host, port, filekey, metakey).deferred """ DHT single primitives (single call to single peer). These should probably only be called for testing or bootstrapping (sendkFindNode can be used to 'connect' to the flud network via a gateway, for instance). Use the recursive primitives for doing DHT ops. """ def sendkFindNode(self, host, port, key): return SENDkFINDNODE(self.node, host, port, key).deferred def sendkStore(self, host, port, key, val): return SENDkSTORE(self.node, host, port, key, val).deferred def sendkFindValue(self, host, port, key): return SENDkFINDVALUE(self.node, host, port, key).deferred """ DHT recursive primitives (recursive calls to muliple peers) """ def kFindNode(self, key): return kFindNode(self.node, key).deferred def kStore(self, key, val): return kStore(self.node, key, val).deferred def kFindValue(self, key): return kFindValue(self.node, key).deferred
Python
""" ServerPrimitives.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Primitive server storage protocol """ import binascii, time, os, stat, httplib, gc, re, sys, logging, sets import tempfile, tarfile from StringIO import StringIO from twisted.web.resource import Resource from twisted.web import server, resource, client from twisted.internet import reactor, threads, defer from twisted.web import http from twisted.python import failure from flud.FludCrypto import FludRSA, hashstring, hashfile, generateRandom import flud.TarfileUtils as TarfileUtils from flud.fencode import fencode, fdecode import BlockFile from FludCommUtil import * logger = logging.getLogger("flud.server.op") loggerid = logging.getLogger("flud.server.op.id") loggerstor = logging.getLogger("flud.server.op.stor") loggerretr = logging.getLogger("flud.server.op.rtrv") loggervrfy = logging.getLogger("flud.server.op.vrfy") loggerdele = logging.getLogger("flud.server.op.dele") loggerauth = logging.getLogger("flud.server.op.auth") """ These classes represent http requests received by this node, and the actions taken to respond. """ # FUTURE: check flud protocol version for backwards compatibility # XXX: need to refactor challengeResponse stuff so that all share this same # code (put it in REQUEST obj). See if we can do the same for some of # the other bits. # XXX: need to make sure we have appropriate timeouts for all comms. # FUTURE: DOS attacks. For now, assume that network hardware can filter these # out (by throttling individual IPs) -- i.e., it isn't our problem. If we # want to defend against this at some point, we need to keep track of who # is generating requests and then ignore them. # XXX: find everywhere we are sending longs and consider sending hex (or our # own base-64) encoded instead # XXX: might want to consider some self-healing for the kademlia layer, as # outlined by this thread: # http://zgp.org/pipermail/p2p-hackers/2003-August/001348.html (should # also consider Zooko's links in the parent to this post) # XXX: can man-in-the-middle get us with http basic auth, i.e., a node (Bob) # receives a store request from node Alice. Bob 'forwards' this request # to another node, Charlie, posing as Alice. When the challenge comes # back from Charlie, Bob relays it back to Alice, gets Alice's response, # and uses that response to answer Charlie's challenge. Bob then pretends # to store the data, while in reality the data is at Charlie (VERIFYs can # be performed similarly). # [This scenario appears legitimate, but it isn't problematic. Bob can # only store at Charlie if Charlie trusts Bob. And if Bob trust Charlie # so much that he is willing to stake his trust with Alice on Charlie's # reliability, there shouldn't be anything wrong with allowing this. If # Charlie goes down, Bob will go down (partly) with him, as far as trust # is concerned]. # [should disable/remove CHALLENGE and GROUPID so that this only works # for forwarding. Leaving CHALLENGE/GROUPID allows imposters. Limiting # only to forwarding imposters.] # XXX: Auth problem still exists. Http-auth allows servers to know that clients # legitimate, but clients don't know the legitimacy of servers. Proposal: # all client requests are currently responded to with a challenge that is # copied in the header and body. Instead, require that the server sign # the url it was handed by the client, and return this in the body # (challenge remains in the header). Now, the client can verify that the # server is who it says it is. This is still somewhat vulnerable to replay # (if the same url is repeated, which it can be), but this only matters if # the client sends the exact same message again later to an imposter # (imposters can only impersonate the server). Alternatively, could send # a challenge as a param of the request. [I think this is fixed now?] # XXX: disallow requests originating from self. fludproto_ver = '0.1' challengelength = 40 # XXX: is 40 bytes sufficient? class ROOT(Resource): """ Notes on parameters common to most requests: Ku_e - the public key RSA exponent (usually 65537L, but can be other values) Ku_n - the public key RSA modulus. nodeID - the nodeID of the requestor port - the port that the reqestor runs their fludserver on. All REQs will contain a nodeID ('nodeID'), a public key ('Ku_e', 'Ku_n'), and a requestID ('reqID'). [XXX: must be more careful about all this code -- if wrong data is sent, exceptions are raised. Put try blocks around everything.] Client Authentication works as follows: Each client request is responded to with a 401 response in the header, with a challenge in the header message and repeated in the body. The client reissues the response, filling the user 'username' field with the challenge-response and the 'password' field with groupIDu. The client must compute the response to the challenge each time by examining this value. The groupIDu never changes, so the client can simply send this value back with the response. This implies some server-side state; since the server responds with a 401 and expects the client to send a new request with an Authorization header that contains the challenge response, the server has to keep some state for each client. This state expires after a short time, and implies that client must make requests to individual servers serially. """ # FUTURE: might want to switch to nevow's rend.Page as parent... def __init__(self, fludserver): """ All children should inherit. Make sure to call super.__init__ if you override __init__ """ Resource.__init__(self) self.fludserver = fludserver self.node = fludserver.node self.config = fludserver.node.config def getChild(self, name, request): """ should override. """ if name == "": return self return Resource.getChild(self, name, request) def render_GET(self, request): self.setHeaders(request) return "<html>Flud</hrml>" def setHeaders(self, request): request.setHeader('Server','FludServer 0.1') request.setHeader('FludProtocol', fludproto_ver) class ID(ROOT): """ Just received a request to expose my identity. Send public key (from which requestor can determine nodeID). Response codes: 200- OK (default) 204- No Content (returned in case of error or not wanting to divulge ID) """ def render_GET(self, request): self.setHeaders(request) try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port') params = requireParams(request, required) except Exception, inst: msg = "%s in request received by ID" % inst.args[0] loggerid.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: loggerid.info("received ID request from %s..." % params['nodeID'][:10]) loggerid.info("returning ID response") #try: reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) host = getCanonicalIP(request.getClientIP()) updateNode(self.node.client, self.config, host, int(params['port']), reqKu, params['nodeID']) return str(self.config.Ku.exportPublicKey()) #except: # msg = "can't return ID" # loggerid.log(logging.WARN, msg) # request.setResponseCode(http.NO_CONTENT, msg) # return msg class STORE(ROOT): """ A request to store data via http upload. Response codes: 200- OK (default) 400- Bad Request (missing params) 401- Unauthorized (ID hash, CHALLENGE, or GROUPCHALLENGE failed) Each file fragment is stored with its storage key as the file name. The file fragment can be 'touched' (or use last access time if supported) each time it is verified or read, so that we have a way to record age (which also allows a purge strategy). Files are reference-listed (reference count with owners) by the BlockFile object. A preliminary file structure: .flud/ store/ fragmenthash1 fragmenthash2 ... dht/ metadatahash1 metadatahash2 ... meta/ master metadatahash1 metadatahash2 ... dl/ flud.conf fludfile.conf fludrules.init flud.log """ isLeaf = True def render_POST(self, request): self.setHeaders(request) try: required = ('filekey', 'size', 'nodeID', 'Ku_e', 'Ku_n', 'port') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by STORE" loggerstor.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: loggerstor.info("received STORE request from %s..." % params['nodeID'][:10]) host = getCanonicalIP(request.getClientIP()) port = int(params['port']) requestedSize = int(params['size']) reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) #if requestedSize > 10: # msg = "unwilling to enter into storage relationship" # request.setResponseCode(http.PAYMENT_REQUIRED, msg) # return msg return authenticate(request, reqKu, params['nodeID'], host, int(params['port']), self.node.client, self.config, self._storeFile, request, params['filekey'], reqKu, params['nodeID']) def _storeFile(self, request, filekey, reqKu, nodeID): # [XXX: memory management is not happy here. might want to look at # request.registerProducer(). Otherwise, might have to scrap # using the STORE(ROOT(RESOURCE)) deal in favor of # producer/consumer model for STORE ops # (http://itamarst.org/writings/OSCON03/twisted_internet-108.html). # Another option might include subclassing web.resource.Resource # and making this derive from that... Or might be web.Site that # needs to be subclassed... Or maybe web.site.Request - # web.site.Request.process()? Request seems doubly-bad: perhaps a # copy is made somewhere, because memory mushrooms to 2x big # upload, then goes back down to around 1x. # [update: This should be fixable in twisted.web2, but I am informed # that in the current version, there is no workaround] # get the data to a tmp file loggerstor.debug("writing store data to tmpfile") tmpfile = tempfile.mktemp(dir=self.config.storedir) tarball = os.path.join(self.config.storedir,reqKu.id()+".tar") # rename and/or prepend the data appropriately tmpTarMode = None if filekey[-4:] == ".tar": tmpfile = tmpfile+".tar" tmpTarMode = 'r' targetTar = tarball elif filekey[-7:] == ".tar.gz": tmpfile = tmpfile+".tar.gz" tmpTarMode = 'r:gz' targetTar = tarball+".gz" loggerstor.debug("tmpfile is %s" % tmpfile) # XXX: if the server supports both .tar and tar.gz, this is wrong; we'd # need to check *both* for already existing dudes instead of just # choosing one if os.path.exists(tarball+'.gz'): tarball = (tarball+'.gz', 'r:gz') elif os.path.exists(tarball): tarball = (tarball, 'r') else: tarball = None loggerstor.debug("tarball is %s" % str(tarball)) data = request.args.get('filename')[0] # XXX: file in mem! need web2. # XXX: bad blocking stuff here f = open(tmpfile, 'wb') f.write(data) f.close() ftype = os.popen('file %s' % tmpfile) loggerstor.debug("ftype of %s is %s" % (tmpfile, ftype.read())) ftype.close() if tmpTarMode: # client sent a tarball loggerstor.debug("about to chksum %s" % tmpfile) digests = TarfileUtils.verifyHashes(tmpfile, '.meta') loggerstor.debug("chksum returned %s" % digests) ftype = os.popen('file %s' % tmpfile) loggerstor.debug("ftype of %s is %s" % (tmpfile, ftype.read())) ftype.close() if not digests: msg = "Attempted to use non-CAS storage key(s) for" \ " STORE tarball" loggerstor.debug(msg) os.remove(tmpfile) request.setResponseCode(http.CONFLICT, msg) return msg # XXX: add digests to a db of already stored files (for quick # lookup) if tarball: tarname, tarnameMode = tarball loggerstor.debug("concatenating tarfiles %s and %s" % (tarname, tmpfile)) f1 = tarfile.open(tarname, tarnameMode) f2 = tarfile.open(tmpfile, tmpTarMode) f1names = f1.getnames() f2names = f2.getnames() f1.close() f2.close() dupes = [f for f in f1names if f in f2names] TarfileUtils.delete(tmpfile, dupes) ftype = os.popen('file %s' % tarname) loggerstor.debug("ftype of %s is %s" % (tarname, ftype.read())) ftype.close() TarfileUtils.concatenate(tarname, tmpfile) ftype = os.popen('file %s' % tarname) loggerstor.debug("ftype of %s is %s" % (tarname, ftype.read())) ftype.close() else: loggerstor.debug("saving %s as tarfile %s" % (tmpfile, targetTar)) os.rename(tmpfile, targetTar) else: # client sent regular file h = hashfile(tmpfile) if request.args.has_key('meta') and request.args.has_key('metakey'): metakey = request.args.get('metakey')[0] meta = request.args.get('meta')[0] # XXX: file in mem! else: metakey = None meta = None if fencode(long(h, 16)) != filekey: msg = "Attempted to use non-CAS storage key for STORE data " msg += "(%s != %s)" % (filekey, fencode(long(h, 16))) os.remove(tmpfile) request.setResponseCode(http.CONFLICT, msg) return msg fname = os.path.join(self.config.storedir, filekey) if os.path.exists(fname): loggerstor.debug("adding metadata to %s" % fname) f = BlockFile.open(fname,'rb+') if not f.hasNode(nodeID): f.addNode(int(nodeID,16), {metakey: meta}) f.close() os.remove(tmpfile) else: if os.path.exists(nodeID+".tar"): # XXX: need to do something with metadata! print "XXX: need to do something with metadata for tar!" tarball = tarfile.open(tarname, 'r') if fname in tarball.getnames(): loggerstor.debug("%s already stored in tarball" % fname) # if the file is already in the corresponding tarball, # update its timestamp and return success. loggerstor.debug("%s already stored" % filekey) # XXX: update timestamp for filekey in tarball return "Successful STORE" else: loggerstor.debug("tarball for %s, but %s not in tarball" % (nodeID,fname)) if len(data) < 8192 and fname != tarname: #XXX: magic # (blk sz) # If the file is small, move it into the appropriate # tarball. Note that this code is unlikely to ever be # executed if the client is an official flud client, as # they do the tarball aggregation thing already, and all # tarballs will be > 8192. This is, then, really just # defensive coding -- clients aren't required to implement # that tarball aggregation strategy. And it is really only # useful for filesystems with inefficient small file # storage. loggerstor.debug("moving small file '%s' into tarball" % fname) if not os.path.exists(tarname): tarball = tarfile.open(tarname, 'w') else: tarball = tarfile.open(tarname, 'a') # XXX: more bad blocking stuff tarball.add(tmpfile, os.path.basename(fname)) if meta: metafilename = "%s.%s.meta" % (os.path.basename(fname), metakey) loggerstor.debug("adding metadata file to tarball %s" % metafilename) metaio = StringIO(meta) tinfo = tarfile.TarInfo(metafilename) tinfo.size = len(meta) tarball.addfile(tinfo, metaio) tarball.close() os.remove(tmpfile) else: # store the file loggerstor.debug("storing %s" % fname) os.rename(tmpfile, fname) BlockFile.convert(fname, (int(nodeID,16), {metakey: meta})) loggerstor.debug("successful STORE for %s" % filekey) return "Successful STORE" def _storeErr(self, error, request, msg): out = msg+": "+error.getErrorMessage() print "%s" % str(error) loggerstor.info(out) # update trust, routing request.setResponseCode(http.UNAUTHORIZED, "Unauthorized: %s" % msg) # XXX: wrong code return msg def render_GET(self, request): request.setResponseCode(http.BAD_REQUEST, "Bad Request") return "STORE request must be sent using POST" class RETRIEVE(ROOT): """ A request to retrieve data. The file to retrieve is indicated by the URL path, e.g. http://server:port/RETRIEVE/a35cd1339766ef209657a7b Response codes: 200- OK (default) 400- Bad Request (missing params) 401- Unauthorized (ID hash, CHALLENGE, or GROUPCHALLENGE failed) """ isLeaf = True def render_GET(self, request): self.setHeaders(request) try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by RETRIEVE" loggerretr.log(logging.INFO, msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: loggerretr.info("received RETRIEVE request for %s from %s..." % (request.path, params['nodeID'][:10])) host = getCanonicalIP(request.getClientIP()) port = int(params['port']) filekey = re.sub('/RETRIEVE', '', str(request.path)) paths = [p for p in filekey.split(os.path.sep) if p != ''] if len(paths) > 1: msg = "filekey contains illegal path seperator tokens." loggerretr.debug("BAD_REQ: %s" % msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request: %s" % msg) return msg reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) if request.args.has_key('metakey'): returnMeta = request.args['metakey'] if returnMeta == 'True': returnMeta = True else: returnMeta = True return authenticate(request, reqKu, params['nodeID'], host, int(params['port']), self.node.client, self.config, self._sendFile, request, filekey, reqKu, returnMeta) def _sendFile(self, request, filekey, reqKu, returnMeta): fname = self.config.storedir + filekey loggerretr.debug("reading file data from %s" % fname) # XXX: make sure requestor owns the file? tfilekey = filekey[1:] if returnMeta: loggerretr.debug("returnMeta = %s" % returnMeta) request.setHeader('Content-type', 'Multipart/Related') rand_bound = binascii.hexlify(generateRandom(13)) request.setHeader('boundary', rand_bound) if not os.path.exists(fname): # check for tarball for originator tarball = os.path.join(self.config.storedir,reqKu.id()+".tar") tarballs = [] if os.path.exists(tarball+'.gz'): tarballs.append((tarball+'.gz', 'r:gz')) if os.path.exists(tarball): tarballs.append((tarball, 'r')) loggerretr.debug("tarballs = %s" % tarballs) # XXX: does this work? does it close both tarballs if both got # opened? for tarball, openmode in tarballs: tar = tarfile.open(tarball, openmode) try: tinfo = tar.getmember(tfilekey) returnedMeta = False if returnMeta: loggerretr.debug("tar returnMeta %s" % tfilekey) try: metas = [f for f in tar.getnames() if f[:len(tfilekey)] == tfilekey and f[-4:] == 'meta'] loggerretr.debug("tar returnMetas=%s" % metas) for m in metas: minfo = tar.getmember(m) H = [] H.append("--%s" % rand_bound) H.append("Content-Type: " "Application/octet-stream") H.append("Content-ID: %s" % m) H.append("Content-Length: %d" % minfo.size) H.append("") H = '\r\n'.join(H) request.write(H) request.write('\r\n') tarm = tar.extractfile(minfo) loggerretr.debug("successful metadata" " RETRIEVE (from %s)" % tarball) # XXX: bad blocking stuff while 1: buf = tarm.read() if buf == "": break request.write(buf) request.write('\r\n') tarm.close() H = [] H.append("--%s" % rand_bound) H.append("Content-Type: Application/octet-stream") H.append("Content-ID: %s" % tfilekey) H.append("Content-Length: %d" % tinfo.size) H.append("") H = '\r\n'.join(H) request.write(H) request.write('\r\n') returnedMeta = True except: # couldn't find any metadata, just return normal # file loggerretr.debug("no metadata found") pass # XXX: bad blocking stuff tarf = tar.extractfile(tinfo) # XXX: update timestamp on tarf in tarball loggerretr.debug("successful RETRIEVE (from %s)" % tarball) # XXX: bad blocking stuff while 1: buf = tarf.read() if buf == "": break request.write(buf) tarf.close() tar.close() if returnedMeta: T = [] T.append("") T.append("--%s--" % rand_bound) T.append("") T = '\r\n'.join(T) request.write(T) return "" except: tar.close() request.setResponseCode(http.NOT_FOUND, "Not found: %s" % filekey) request.write("Not found: %s" % filekey) else: f = BlockFile.open(fname,"rb") loggerretr.log(logging.INFO, "successful RETRIEVE for %s" % filekey) meta = f.meta(int(reqKu.id(),16)) if returnMeta and meta: loggerretr.debug("returnMeta %s" % filekey) loggerretr.debug("returnMetas=%s" % meta) for m in meta: H = [] H.append("--%s" % rand_bound) H.append("Content-Type: Application/octet-stream") H.append("Content-ID: %s.%s.meta" % (tfilekey, m)) H.append("Content-Length: %d" % len(meta[m])) H.append("") H.append(meta[m]) H = '\r\n'.join(H) request.write(H) request.write('\r\n') H = [] H.append("--%s" % rand_bound) H.append("Content-Type: Application/octet-stream") H.append("Content-ID: %s" % tfilekey) H.append("Content-Length: %d" % f.size()) H.append("") H = '\r\n'.join(H) request.write(H) request.write('\r\n') # XXX: bad blocking stuff while 1: buf = f.read() if buf == "": break request.write(buf) f.close() if returnMeta and meta: T = [] T.append("") T.append("--%s--" % rand_bound) T.append("") request.write('\r\n'.join(T)) return "" def _sendErr(self, error, request, msg): out = msg+": "+error.getErrorMessage() loggerretr.log(logging.INFO, out ) # update trust, routing request.setResponseCode(http.UNAUTHORIZED, "Unauthorized: %s" % msg) request.write(msg) request.finish() class VERIFY(ROOT): """ Just received a storage VERIFY request. Response codes: 200- OK (default) 400- Bad Request (missing params) 401- Unauthorized (ID hash, CHALLENGE, or GROUPCHALLENGE failed) [VERIFY is the most important of the trust-building ops. Issues: 1) how often do we verify. a) each file A) each block of each file 2) proxying hurts trust (see PROXY below) The answer to #1 is, ideally, we check every node who is storing for us every quanta. But doing the accounting for keeping track of who is storing for us is heavy, so instead we just want to check all of our files and hope that gives sufficient coverage, on average, of the nodes we store to. But we have to put some limits on this, and the limits can't be imposed by the sender (since a greedy sender can just modify this), so peer nodes have to do some throttling of these types of requests. But such throttling is tricky, as the requestor must decrease trust when VERIFY ops fail. Also, since we will just randomly select from our files, such a throttling scheme will reduce our accuracy as we store more and more data (we can only verify a smaller percentage). Peers could enforce limits as a ratio of total data stored for a node, but then the peers could act maliciously by artificially lowering this number. In summary, if we don't enforce limits, misbehaving nodes could flood VERIFY requests resulting in effecitve DOS attack. If we do enforce limits, we have to watch out for trust wars where both nodes end up destroying all trust between them. Possible answer: set an agreed-upon threshold a priori. This could be a hardcoded limit, or (better) negotiated between node pairs. If the requestor goes over this limit, he should understand that his trust will be decreased by requestee. If he doesn't understand this, his trust *should* be decreased, and if he decreases his own trust in us as well, we don't care -- he's misbehaving.] """ # XXX: add support for sending multiple verify ops in a single request isLeaf = True def render_GET(self, request): """ A VERIFY contains a file[fragment]id, an offset, and a length. When this message is received, the given file[fragment] should be accessed and its bytes, from offset to offset+length, should be sha256 hashed, and the result returned. [in theory, we respond to all VERIFY requests. In practice, however, we should probably do some throttling of responses to prevent DOS attacks and probing. Or, maybe require that the sender encrypt the challenge with their private key...] [should also enforce some idea of reasonableness on length of bytes to verify] """ self.setHeaders(request) try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'offset', 'length') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by VERIFY" loggervrfy.log(logging.INFO, msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") loggervrfy.debug("BAD REQUEST") return msg else: loggervrfy.log(logging.INFO, "received VERIFY request from %s..." % params['nodeID']) if 'meta' in request.args: params['metakey'] = request.args['metakey'][0] params['meta'] = fdecode(request.args['meta'][0]) loggerretr.info("VERIFY contained meta field with %d chars" % len(params['meta'])) meta = (params['metakey'], params['meta']) else: meta = None host = getCanonicalIP(request.getClientIP()) port = int(params['port']) offset = int(params['offset']) length = int(params['length']) filekey = re.sub('/VERIFY', '', str(request.path)) paths = [p for p in filekey.split(os.path.sep) if p != ''] if len(paths) > 1: msg = "Bad request:"\ " filekey contains illegal path seperator tokens." loggerretr.debug(msg) request.setResponseCode(http.BAD_REQUEST, msg) return msg reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) return authenticate(request, reqKu, params['nodeID'], host, int(params['port']), self.node.client, self.config, self._sendVerify, request, filekey, offset, length, reqKu, params['nodeID'], meta) def _sendVerify(self, request, filekey, offset, length, reqKu, nodeID, meta): fname = self.config.storedir+filekey loggervrfy.debug("request for %s" % fname) if os.path.exists(fname): loggervrfy.debug("looking in regular blockfile for %s" % fname) if meta: f = BlockFile.open(fname, 'rb+') else: f = BlockFile.open(fname, 'rb') else: # check for tarball for originator loggervrfy.debug("checking tarball for %s" % fname) tarballs = [] tarballbase = os.path.join(self.config.storedir, reqKu.id())+".tar" if os.path.exists(tarballbase+".gz"): tarballs.append((tarballbase+".gz", 'r:gz')) if os.path.exists(tarballbase): tarballs.append((tarballbase, 'r')) loggervrfy.debug("tarballs is %s" % tarballs) for tarball, openmode in tarballs: loggervrfy.debug("looking in tarball %s..." % tarball) tar = tarfile.open(tarball, openmode) try: tfilekey = filekey[1:] tarf = tar.extractfile(tfilekey) tari = tar.getmember(tfilekey) # XXX: update timestamp on tarf in tarball fsize = tari.size if offset > fsize or (offset+length) > fsize: # XXX: should limit length loggervrfy.debug("VERIFY response failed (from %s):" " bad offset/length" % tarball) msg = "Bad request: bad offset/length in VERIFY" request.setResponseCode(http.BAD_REQUEST, msg) return msg # XXX: could avoid seek/read if length == 0 tarf.seek(offset) # XXX: bad blocking read data = tarf.read(length) tarf.close() if meta: mfname = "%s.%s.meta" % (tfilekey, meta[0]) loggervrfy.debug("looking for %s" % mfname) if mfname in tar.getnames(): # make sure that the data is the same, if not, # remove it and re-add it tarmf = tar.extractfile(mfname) # XXX: bad blocking read stored_meta = tarmf.read() tarmf.close() if meta[1] != stored_meta: loggervrfy.debug("updating tarball" " metadata for %s.%s" % (tfilekey, meta[0])) tar.close() TarfileUtils.delete(tarball, mfname) if openmode == 'r:gz': tarball = TarfileUtils.gunzipTarball( tarball) tar = tarfile.open(tarball, 'a') metaio = StringIO(meta[1]) tinfo = tarfile.TarInfo(mfname) tinfo.size = len(meta[1]) tar.addfile(tinfo, metaio) tar.close() if openmode == 'r:gz': tarball = TarfileUtils.gzipTarball( tarball) else: loggervrfy.debug("no need to update tarball" " metadata for %s.%s" % (tfilekey, meta[0])) else: # add it loggervrfy.debug("adding tarball metadata" " for %s.%s" % (tfilekey, meta[0])) tar.close() if openmode == 'r:gz': tarball = TarfileUtils.gunzipTarball(tarball) tar = tarfile.open(tarball, 'a') metaio = StringIO(meta[1]) tinfo = tarfile.TarInfo(mfname) tinfo.size = len(meta[1]) tar.addfile(tinfo, metaio) tar.close() if openmode == 'r:gz': tarball = TarfileUtils.gzipTarball( tarball) tar.close() hash = hashstring(data) loggervrfy.info("successful VERIFY (from %s)" % tarball) return hash except: tar.close() loggervrfy.debug("requested file %s doesn't exist" % fname) msg = "Not found: not storing %s" % filekey request.setResponseCode(http.NOT_FOUND, msg) return msg # make sure request is reasonable fsize = os.stat(fname)[stat.ST_SIZE] if offset > fsize or (offset+length) > fsize: # XXX: should limit length loggervrfy.debug("VERIFY response failed (bad offset/length)") msg = "Bad request: bad offset/length in VERIFY" request.setResponseCode(http.BAD_REQUEST, msg) return msg else: # XXX: blocking # XXX: could avoid seek/read if length == 0 (noop for meta update) f.seek(offset) data = f.read(length) if meta: loggervrfy.debug("adding metadata for %s.%s" % (fname, meta[0])) f.addNode(int(nodeID, 16) , {meta[0]: meta[1]}) # XXX: blocking f.close() hash = hashstring(data) loggervrfy.debug("returning VERIFY") return hash def _sendErr(self, error, request, msg): out = "%s:%s" % (msg, error.getErrorMessage()) loggervrfy.info(out) # update trust, routing request.setResponseCode(http.UNAUTHORIZED, "Unauthorized: %s" % msg) request.write(msg) request.finish() class DELETE(ROOT): """ A request to delete data. The file to delete is indicated by the URL path, e.g. http://server:port/DELETE/a35cd1339766ef209657a7b Response codes: 200- OK (default) 400- Bad Request (missing params) 401- Unauthorized (ID hash, CHALLENGE, or GROUPCHALLENGE failed, or nodeID doesn't own this block) """ isLeaf = True def render_GET(self, request): self.setHeaders(request) try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'metakey') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by DELETE" loggerdele.log(logging.INFO, msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: loggerdele.debug("received DELETE request for %s from %s..." % (request.path, params['nodeID'][:10])) host = getCanonicalIP(request.getClientIP()) port = int(params['port']) filekey = re.sub('/DELETE', '', str(request.path)) paths = [p for p in filekey.split(os.path.sep) if p != ''] if len(paths) > 1: msg = "filekey contains illegal path seperator tokens." loggerretr.debug("BAD_REQ: %s" % msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request: %s" % msg) return msg reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) metakey = params['metakey'] return authenticate(request, reqKu, params['nodeID'], host, int(params['port']), self.node.client, self.config, self._deleteFile, request, filekey, metakey, reqKu, params['nodeID']) def _deleteFile(self, request, filekey, metakey, reqKu, reqID): fname = self.config.storedir + filekey loggerdele.debug("reading file data from %s" % fname) if not os.path.exists(fname): # check for tarball for originator tarballs = [] tarballbase = os.path.join(self.config.storedir, reqKu.id())+".tar" if os.path.exists(tarballbase+".gz"): tarballs.append((tarballbase+".gz", 'r:gz')) if os.path.exists(tarballbase): tarballs.append((tarballbase, 'r')) for tarball, openmode in tarballs: tfilekey = filekey[1:] mfilekey = "%s.%s.meta" % (tfilekey, metakey) loggerdele.debug("opening %s, %s for delete..." % (tarball, openmode)) ftype = os.popen('file %s' % tarball) loggerdele.debug("ftype of %s is %s" % (tarball, ftype.read())) ftype.close() tar = tarfile.open(tarball, openmode) mnames = [n for n in tar.getnames() if n[:len(tfilekey)] == tfilekey] tar.close() if len(mnames) > 2: deleted = TarfileUtils.delete(tarball, mfilekey) else: deleted = TarfileUtils.delete(tarball, [tfilekey, mfilekey]) if deleted: loggerdele.info("DELETED %s (from %s)" % (deleted, tarball)) return "" request.setResponseCode(http.NOT_FOUND, "Not found: %s" % filekey) request.write("Not found: %s" % filekey) else: f = BlockFile.open(fname,"rb+") nID = int(reqID, 16) if f.hasNode(nID): # remove this node/metakey from owning this file block f.delNode(nID, metakey) if f.emptyNodes(): # if this was the only owning node, delete it f.close() os.remove(fname) f.close() loggerdele.debug("returning DELETE response") return "" def _sendErr(self, error, request, msg): out = msg+": "+error.getErrorMessage() loggerdele.log(logging.INFO, out ) # update trust, routing request.setResponseCode(http.UNAUTHORIZED, "Unauthorized: %s" % msg) request.write(msg) request.finish() class PROXY(ROOT): """ This is a special request which wraps another request. If received, the node is required to forward the request and keep state to re-wrap when the response returns. This is useful for several reasons: 1) insert a bit of anonymity into VERIFY ops, so that a malicious node can't know to do an early purge data for a missing node. 2) NAT'd boxes will need to use relays for STORE, VERIFY, and RETRIEVE (since all those ops cause the receive to make a connection back) [There must be an incentive for nodes to offer PROXY (trust), otherwise it is advantageous to turn it off.] [disadvantages of proxy: it hurts trust. How do we know that a bad (trust decreasing) op really was caused by the originator? Couldn't the failure be caused by one of the proxies?] [Should consider using GnuNet's "Excess-Based Economic Model" where each request contains a priority which 'spends' some of the trust at the requestee when resources are scarce. In this model, the proxying node[s] charge a fee on the priority, reducing it by a small amount as they forward the request.] """ isLeaf = True def render_GET(self, request): self.setHeaders(request) result = "NOT YET IMPLEMENTED" return result def authenticate(request, reqKu, reqID, host, port, client, config, callable, *callargs): # 1- make sure that reqKu hashes to reqID # 2- send a challenge/groupchallenge to reqID (encrypt with reqKu) challengeResponse = request.getUser() groupResponse = request.getPassword() if reqKu.id() != reqID: msg = "Ku %s does not hash to nodeID %s, failing request" \ % (reqKu.id(), reqID) loggerauth.info(msg) # XXX: update trust, routing request.setResponseCode(http.UNAUTHORIZED, "Unauthorized: %s" % msg) return msg if not challengeResponse or not groupResponse: loggerauth.info("returning challenge for request from %s:%d" \ % (host, port)) return sendChallenge(request, reqKu, config.nodeID) else: if getChallenge(challengeResponse): expireChallenge(challengeResponse) if groupResponse == hashstring( str(reqKu.exportPublicKey()) +str(config.groupIDr)): updateNode(client, config, host, port, reqKu, reqID) return callable(*callargs) else: err = "Group Challenge Failed" loggerauth.info(err) loggerauth.debug("Group Challenge received was %s" \ % groupResponse) # XXX: update trust, routing request.setResponseCode(http.FORBIDDEN, err); return err else: err = "Challenge Failed" loggerauth.info(err) loggerauth.debug("Challenge received was %s" % challengeResponse) # XXX: update trust, routing request.setResponseCode(http.FORBIDDEN, err); return err def sendChallenge(request, reqKu, id): challenge = generateRandom(challengelength) while challenge[0] == '\x00': # make sure we have at least challengelength bytes challenge = generateRandom(challengelength) addChallenge(challenge) loggerauth.debug("unencrypted challenge is %s" % fencode(binascii.unhexlify(id)+challenge)) echallenge = reqKu.encrypt(binascii.unhexlify(id)+challenge)[0] echallenge = fencode(echallenge) loggerauth.debug("echallenge = %s" % echallenge) # since challenges will result in a new req/resp pair being generated, # these could take much longer than the primitive_to. Expire in # 15*primitive_to reactor.callLater(primitive_to*15, expireChallenge, challenge, True) resp = 'challenge = %s' % echallenge loggerauth.debug("resp = %s" % resp) request.setResponseCode(http.UNAUTHORIZED, echallenge) request.setHeader('Connection', 'close') request.setHeader('WWW-Authenticate', 'Basic realm="%s"' % 'default') request.setHeader('Content-Length', str(len(resp))) request.setHeader('Content-Type', 'text/html') request.setHeader('Pragma','claimreserve=5555') # XXX: this doesn't work return resp outstandingChallenges = {} def addChallenge(challenge): outstandingChallenges[challenge] = True loggerauth.debug("added challenge %s" % fencode(challenge)) def expireChallenge(challenge, expired=False): try: challenge = fdecode(challenge) except: pass if outstandingChallenges.has_key(challenge): del outstandingChallenges[challenge] if expired: # XXX: should put these in an expired challenge list so that we # can send a more useful message on failure (must then do # expirations on expired list -- maybe better to just make # these expirations really liberal). loggerauth.debug("expired challenge %s" % fencode(challenge)) else: loggerauth.debug("deleted challenge %s" % fencode(challenge)) def getChallenge(challenge): try: challenge = fdecode(challenge) except: pass if outstandingChallenges.has_key(challenge): return outstandingChallenges[challenge] else: return None
Python
""" FludCommUtil.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Communications routines used by both client and server code. """ from twisted.web import client from twisted.internet import reactor, defer from twisted.python import failure import binascii, httplib, logging, os, stat, random, socket import inspect from flud.FludExceptions import FludException from flud.FludCrypto import FludRSA, generateRandom from flud.HTTPMultipartDownloader import HTTPMultipartDownloader """ Some constants used by the Flud Protocol classes """ fludproto_ver = '0.1' # XXX: when things timeout, bad news. Unintuitive exceptions spewed. Make this # small and fix all issues. primitive_to = 3800 # default timeout for primitives kprimitive_to = primitive_to/2 # default timeout for kademlia primitives #kprimitive_to = 10 # default timeout for kademlia primitives transfer_to = 3600 # 10-hr limit on file transfers MAXTIMEOUTS = 5 # number of times to retry after connection timeout failure CONNECT_TO = 60 CONNECT_TO_VAR = 5 logger = logging.getLogger('flud.comm') class BadCASKeyException(failure.DefaultException): pass class NotFoundException(failure.DefaultException): pass class BadRequestException(failure.DefaultException): pass i = 0 """ Some utility functions used by both client and server. """ def updateNodes(client, config, nodes): if nodes and not isinstance(nodes, list) and not isinstance(nodes, tuple): raise TypeError("updateNodes must be called with node list, tuple," " or kData dict") logger.debug("updateNodes(%s)" % nodes) for i in nodes: host = i[0] port = i[1] nID = i[2] nKu = FludRSA.importPublicKey(i[3]) updateNode(client, config, host, port, nKu, nID) updateNodePendingGETID = {} def updateNode(client, config, host, port, nKu=None, nID=None): """ Updates this node's view of the given node. This includes updating the known-nodes record, trust, and routing table information """ def updateNodeFail(failure, host, port): logging.getLogger('flud').log(logging.INFO, "couldn't get nodeID from %s:%d: %s" % (host, port, failure)) def callUpdateNode(nKu, client, config, host, port, nID): return updateNode(client, config, host, port, nKu, nID) if isinstance(nID, long): nID = "%064x" % nID if nKu is None: #print "updateNode, no nKu" if nID is None: d = client.sendGetID(host, port) d.addCallback(callUpdateNode, client, config, host, port, nID) d.addErrback(updateNodeFail, host, port) else: #print "updateNode, no nKu but got a nID" if config.nodes.has_key(nID): return updateNode(client, config, host, port, FludRSA.importPublicKey(config.nodes[nID]['Ku']), nID) elif updateNodePendingGETID.has_key(nID): pass else: #print "updateNode, sending GETID" updateNodePendingGETID[nID] = True d = client.sendGetID(host, port) d.addCallback(callUpdateNode, client, config, host, port, nID) d.addErrback(updateNodeFail, host, port) elif isinstance(nKu, FludRSA): #print "updateNode with nKu" if updateNodePendingGETID.has_key(nID): del updateNodePendingGETID[nID] if nID == None: nID = nKu.id() elif nID != nKu.id(): raise ValueError("updateNode: given nID doesn't match given nKu." " '%s' != '%s'" % (nID, nKu.id())) # XXX: looks like an imposter -- instead of raising, mark host:port # pair as bad (trust-- on host:port alone, since we don't know id). if config.nodes.has_key(nID) == False: config.addNode(nID, host, port, nKu) # XXX: trust # routing node = (host, port, long(nID, 16), nKu.exportPublicKey()['n']) replacee = config.routing.updateNode(node) #logger.info("knownnodes now: %s" % config.routing.knownNodes()) #print "knownnodes now: %s" % config.routing.knownNodes() if replacee != None: logging.getLogger('flud').info( "determining if replacement in ktable is needed") s = SENDGETID(replacee[0], replacee[1]) s.addErrback(replaceNode, config.routing, replacee, node) else: #print "updateNode nKu=%s, type=%s" % (nKu, type(nKu)) logging.getLogger('flud').warn( "updateNode can't update without a public key or nodeID") frame = inspect.currentframe() # XXX: try/except here for debugging only try: stack = inspect.stack() for i in stack: print "from %s:%d" % (i[1], i[2]) except: print "couldn't get stack trace" raise ValueError("updateNode needs an nKu of type FludRSA" " (received %s) or an nID of type long or str (received %s)" % (type(nKu), type(nID))) # XXX: should really make it impossible to call without one of these... def replaceNode(error, routing, replacee, replacer): routing.replaceNode(replacee, replacer) print "replaced node in ktable" def requireParams(request, paramNames): # Looks for the named parameters in request. If found, returns # a dict of param/value mappings. If any named parameter is missing, # raises an exception params = {} for i in paramNames: try: params[i] = request.args[i][0] except: raise Exception, "missing parameter '"+i+"'" #XXX: use cust Exc return params def getCanonicalIP(IP): # if IP is 'localhost' or '127.0.0.1', use the canonical local hostname. # (this is mostly useful when multiple clients run on the same host) # XXX: could use gethostbyname to get IP addy instead. if IP == '127.0.0.1' or IP == 'localhost': return socket.getfqdn() else: return socket.getfqdn(IP) def getPageFactory(url, contextFactory=None, *args, **kwargs): def failedConnect(reason, factory): try: i = factory.status return reason except: pass #logger.warn("couldn't connect to %s:%d in getPageFactory: %s" # % (factory.host, factory.port, reason)) #logger.warn("state of factory is %s" % factory) #logger.warn("dir() of factory is %s" % dir(factory)) return reason if len(url) >= 16384: raise ValueError( "Too much data sent: twisted server doesn't appear to" " support urls longer than 16384") scheme, host, port, path = client._parse(url) factory = client.HTTPClientFactory(url, *args, **kwargs) factory.deferred.addErrback(failedConnect, factory) to = CONNECT_TO+random.randrange(2+CONNECT_TO_VAR)-CONNECT_TO_VAR if scheme == 'https': from twisted.internet import ssl if contextFactory is None: contextFactory = ssl.ClientContextFactory() reactor.connectSSL(host, port, factory, contextFactory) else: reactor.connectTCP(host, port, factory, timeout=to) return factory def _dlPageFactory(url, target, factoryClass, contextFactory=None, timeout=None, *args, **kwargs): scheme, host, port, path = client._parse(url) if timeout != None: # XXX: do something like http://twistedmatrix.com/pipermail/twisted-python/2003-August/005504.html pass factory = factoryClass(url, target, *args, **kwargs) to = CONNECT_TO+random.randrange(2+CONNECT_TO_VAR)-CONNECT_TO_VAR if scheme == 'https': from twisted.internet import ssl if contextFactory is None: contextFactory = ssl.ClientContextFactory() reactor.connectSSL(host, port, factory, contextFactory) else: reactor.connectTCP(host, port, factory, timeout=to) return factory def downloadPageFactory(url, file, contextFactory=None, timeout=None, *args, **kwargs): return _dlPageFactory(url, file, client.HTTPDownloader, contextFactory, timeout, *args, **kwargs) def multipartDownloadPageFactory(url, dir, contextFactory=None, timeout=None, *args, **kwargs): return _dlPageFactory(url, dir, HTTPMultipartDownloader, contextFactory, timeout, *args, **kwargs) def fileUpload(host, port, selector, files, form=(), headers={}): """ Performs a file upload via http. host - webserver hostname port - webserver listen port selector - the request (relative URL) files - list of files to upload. list contains tuples, with the first entry as filename/file-like obj and the second as form element name. If the first element is a file-like obj, the element will be used as the filename. If the first element is a filename, the filename's basename will be used as the filename on the form. Type will be "application/octet-stream" form (optional) - a list of pairs of additional name/value form elements (param/values). [hopefully, this method goes away in twisted-web2] """ # XXX: set timeout (based on filesize?) port = int(port) rand_bound = binascii.hexlify(generateRandom(13)) boundary = "---------------------------"+rand_bound CRLF = '\r\n' body_content_type = "application/octet-stream" content_type = "multipart/form-data; boundary="+boundary content_length = 0 H = [] for (param, value) in form: H.append('--' + boundary) H.append('Content-Disposition: form-data; name="%s"' % param) H.append('') H.append('%s' % value) form_data = CRLF.join(H)+CRLF content_length = content_length + len(form_data) fuploads = [] for file, element in files: if file == None: file = "/dev/null" # XXX: not portable if 'read' in dir(file): fname = element file.seek(0,2) file_length = file.tell() file.seek(0,0) else: fname = os.path.basename(file) file_length = os.stat(file)[stat.ST_SIZE] #logger.info("upload file %s len is %d" % (fname, file_length)) H = [] # stuff that goes above file data T = [] # stuff that goes below file data H.append('--' + boundary) H.append('Content-Disposition: form-data; name="%s"; filename="%s"' % (element, fname)) H.append('Content-Type: %s\n' % body_content_type) H.append('') file_headers = CRLF.join(H) content_length = content_length + len(file_headers) + file_length fuploads.append((file_headers, file, file_length)) T.append('--'+boundary+'--') T.append('') T.append('') trailer = CRLF.join(T) content_length = content_length + len(trailer) h = httplib.HTTPConnection(host, port) # XXX: blocking h.putrequest('POST', selector) for pageheader in headers: h.putheader(pageheader, headers[pageheader]) h.putheader('Content-Type', content_type) h.putheader('Content-Length', content_length) h.endheaders() h.send(form_data) for fheader, file, flen in fuploads: if 'read' not in dir(file): file = open(file, 'r') h.send(fheader) h.send(file.read(flen)+CRLF) # XXX: blocking file.close() h.send(trailer) return h class ImposterException(FludException): pass
Python
""" ClientDHTPrimitives.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Primitive client DHT protocol """ import time, os, stat, httplib, sys, random, logging from twisted.web import http, client from twisted.internet import reactor, threads, defer from twisted.python import failure import inspect, pdb from flud.FludCrypto import FludRSA import flud.FludkRouting as FludkRouting from flud.fencode import fencode, fdecode import flud.FludDefer as FludDefer import ConnectionQueue from ClientPrimitives import REQUEST from FludCommUtil import * logger = logging.getLogger("flud.client.dht") # FUTURE: check flud protocol version for backwards compatibility # XXX: need to make sure we have appropriate timeouts for all comms. # FUTURE: DOS attacks. For now, assume that network hardware can filter these # out (by throttling individual IPs) -- i.e., it isn't our problem. If we # want to defend against this at some point, we need to keep track of who # is generating requests and then ignore them. # XXX: might want to consider some self-healing for the kademlia layer, as # outlined by this thread: # http://zgp.org/pipermail/p2p-hackers/2003-August/001348.html (should also # consider Zooko's links in the parent to this post). Basic idea: don't # always take the k-closest -- take x random and k-x of the k-closest. # Can alternate each round (k-closest / x + k-x-closest) for a bit more # diversity (as in "Sybil-resistent DHT routing"). # XXX: right now, calls to updateNode are chained. Might want to think about # doing some of this more asynchronously, so that the recursive parts # aren't waiting for remote GETIDs to return before recursing. """ The first set of classes (those beginning with 'k') perform [multiple] queries given a key or key/value pair. They use the second set of classes (those beginning with 'SEND'), which perform a single query to a given node. """ def serviceWaiting(res, key, pending, waiting): # provides a method for calling multiple callbacks on a saved query. # add serviceWaiting as a callback before returning, and pass in the result, # pending dict and waiting dict. All deferreds in the waiting dict will # be called with the result, the waiting dict will be emptied of those # deferreds, and the pending dict will likewise be emptied. if waiting.has_key(key): for d in waiting[key]: #print "offbacking %s" % key d.callback(res) waiting.pop(key) pending.pop(key) return res pendingkFindNodes = {} waitingkFindNodes = {} class kFindNode: """ Perform a kfindnode lookup. """ def __init__(self, node, key): if pendingkFindNodes.has_key(key): d = defer.Deferred() if not waitingkFindNodes.has_key(key): waitingkFindNodes[key] = [] waitingkFindNodes[key].append(d) logger.debug("piggybacking on previous kfindnode for %s" % key) self.deferred = d return self.node = node self.node.DHTtstamp = time.time() self.key = key self.queried = {} self.outstanding = [] self.pending = [] self.kclosest = [] self.abbrvkey = ("%x" % key)[:8]+"..." self.abbrv = "(%s%s)" % (self.abbrvkey, str(self.node.DHTtstamp)[-7:]) self.debugpath = [] self.deferred = self.startQuery(key) def startQuery(self, key): # query self first kclosest = self.node.config.routing.findNode(key) #logger.debug("local kclosest: %s" % kclosest) localhost = getCanonicalIP('localhost') kd = {'id': self.node.config.nodeID, 'k': kclosest} d = self.updateLists(kd, key, localhost, self.node.config.port, long(self.node.config.nodeID, 16)) d.addErrback(self.errkfindnode, key, localhost, self.node.config.port) pendingkFindNodes[key] = d d.addCallback(serviceWaiting, key, pendingkFindNodes, waitingkFindNodes) return d def sendQuery(self, host, port, id, key): self.outstanding.append((host, port, id)) #d = self.node.client.sendkFindNode(host, port, key) d = SENDkFINDNODE(self.node, host, port, key).deferred return d def updateLists(self, response, key, host, port, closestyet, x=0): logger.info("FN: received kfindnode %s response from %s:%d" % (self.abbrv, host, port)) self.debugpath.append("FN: rec. resp from %s:%d" % (host, port)) if not isinstance(response, dict): # a data value is being returned from findval # XXX: moved this bit into findval and call parent for the rest if response == None: logger.warn("got None from key=%s, %s:%d, x=%d, this usually" " means that the host replied None to a findval query" % (key, host, port, x)) # if we found the fencoded value data, return it return defer.succeed(response) logger.debug("updateLists(%s)" % response) if len(response['k']) == 1 and response['k'][0][2] == key: # if we've found the key, don't keep making queries. logger.debug("FN: %s:%d found key %s" % (host, port, key)) self.debugpath.append("FN: %s:%d found key %s" % (host, port, key)) if response['k'][0] not in self.kclosest: self.kclosest.insert(0,response['k'][0]) self.kclosest = self.kclosest[:FludkRouting.k] return defer.succeed(response) #for i in response['k']: # print " res: %s:%d" % (i[0], i[1]) id = long(response['id'], 16) responder = (host, port, id) if responder in self.outstanding: self.outstanding.remove(responder) self.queried[id] = (host, port) knodes = response['k'] for n in knodes: if not self.queried.has_key(n[2])\ and not n in self.pending and not n in self.outstanding: self.pending.append((n[0], n[1], n[2])) if n not in self.kclosest: k = FludkRouting.k # XXX: remove self it in the list? self.kclosest.append(n) self.kclosest.sort( lambda a, b: FludkRouting.kCompare(a[2], b[2], key)) self.kclosest = self.kclosest[:k] #for n in self.outstanding: # if n in self.pending: # self.pending.remove(n) # remove anyone we've sent queries to... self.pending = list(set(self.pending) - set(self.outstanding)) for i in self.queried: n = (self.queried[i][0], self.queried[i][1], i) if n in self.pending: self.pending.remove(n) # ...and anyone who has responded. self.pending.sort(lambda a, b: FludkRouting.kCompare(a[2], b[2], key)) #print "queried: %s" % str(self.queried) #print "outstanding: %s" % str(self.outstanding) #print "pending: %s" % str(self.pending) return self.decideToContinue(response, key, x) def decideToContinue(self, response, key, x): ##print "x is %s" % str(x) ##for i in self.kclosest: ## print " kclosest %s" % str(i) ##for i in self.queried: ## print " queried %s" % str(self.queried[i]) #if len(filter(lambda x: x not in self.queried, self.kclosest)) <= 0: # print "finishing up at round %d" % x # # XXX: never gets in here... # # XXX: remove anything not in self.kclosest from self.pending # self.pending =\ # filter(lambda x: x not in self.kclosest, self.pending) # #self.pending = self.pending[:FludkRouting.k] #else: # return self.makeQueries(key, x) # this is here so that kFindVal can plug-in by overriding return self.makeQueries(key, x) def makeQueries(self, key, x): #print "doing round %d" % x self.debugpath.append("FN: doing round %d" % x) dlist = [] for n in self.pending[:(FludkRouting.a - len(self.outstanding))]: #print " querying %s:%d" % (n[0], n[1]) self.debugpath.append("FN: querying %s:%d" % (n[0], n[1])) d = self.sendQuery(n[0], n[1], n[2], key) d.addCallback(self.updateLists, key, n[0], n[1], self.kclosest[0][2], x+1) d.addErrback(self.errkfindnode, key, n[0], n[1], raiseException=False) dlist.append(d) dl = defer.DeferredList(dlist) dl.addCallback(self.roundDone, key, x) return dl def roundDone(self, responses, key, x): #print "done %d:" % x #print "roundDone: %s" % responses if len(self.pending) != 0 or len(self.outstanding) != 0: # should only get here for nodes that don't accept connections # XXX: updatenode -- decrease trust for i in self.pending: logger.debug("FN: %s couldn't contact node %s (%s:%d)" % (self.abbrv, fencode(i[2]), i[0], i[1])) self.debugpath.append( "FN: %s couldn't contact node %s (%s:%d)" % (self.abbrv, fencode(i[2]), i[0], i[1])) for n in self.kclosest: if (n[0],n[1],n[2]) == i: self.kclosest.remove(n) logger.info("kFindNode %s terminated successfully after %d queries." % (self.abbrv, len(self.queried))) self.debugpath.append("FN: %s terminated successfully after %d queries." % (self.abbrv, len(self.queried))) self.kclosest.sort( lambda a, b: FludkRouting.kCompare(a[2], b[2], key)) result = {} if FludkRouting.k > len(self.kclosest): k = len(self.kclosest) else: k = FludkRouting.k result['k'] = self.kclosest[:k] #print "result: %s" % result #if len(result['k']) > 1: # # if the results (aggregated from multiple responses) contains the # # exact key, just return the correct answer (successful node # # lookup done). # #print "len(result): %d" % len(result['k']) # #print "result[0][2]: %s %d" % (type(result['k'][0][2]), # # result['k'][0][2]) # #print " key: %s %d" % (type(key), key) # if result['k'][0][2] == key: # #print "key matched!" # result['k'] = (result['k'][0],) return result def errkfindnode(self, failure, key, host, port, raiseException=True): logger.info("kFindNode %s request to %s:%d failed -- %s" % (self.abbrv, host, port, failure.getErrorMessage())) # XXX: updateNode-- if raiseException: return failure class kStore(kFindNode): """ Perform a kStore operation. """ def __init__(self, node, key, val): self.node = node self.node.DHTtstamp = time.time() self.key = key self.val = val d = kFindNode(node,key).deferred d.addCallback(self.store) d.addErrback(self._kStoreErr, None, 0) self.deferred = d def store(self, knodes): knodes = knodes['k'] if len(knodes) < 1: raise RuntimeError("can't complete kStore -- no nodes") dlist = [] for knode in knodes: host = knode[0] port = knode[1] deferred = SENDkSTORE(self.node, host, port, self.key, self.val).deferred deferred.addErrback(self._kStoreErr, host, port) dlist.append(deferred) dl = FludDefer.ErrDeferredList(dlist) dl.addCallback(self._kStoreFinished) dl.addErrback(self._kStoreErr, None, 0) return dl def _kStoreFinished(self, response): #print "_kStoreFinished: %s" % response logger.info("kStore finished") return "" def _kStoreErr(self, failure, host, port): logger.info("couldn't store on %s:%d -- %s" % (host, port, failure.getErrorMessage())) print "_kStoreErr was: %s" % failure # XXX: updateNode-- return failure class kFindValue(kFindNode): """ Perform a kFindValue. """ def __init__(self, node, key): self.done = False kFindNode.__init__(self, node, key) def startQuery(self, key): # query self first. We override kFindNode.startQuery here so that # we don't just return the closest nodeID, but the value itself (if # present) localhost = getCanonicalIP('localhost') d = self.sendQuery(localhost, self.node.config.port, long(self.node.config.nodeID, 16), key) d.addCallback(self.updateLists, key, localhost, self.node.config.port, long(self.node.config.nodeID, 16), 0) d.addErrback(self.errkfindnode, key, localhost, self.node.config.port) return d def sendQuery(self, host, port, id, key): # We override sendQuery here in order to call sendkFindValue and handle # its response self.outstanding.append((host, port, id)) d = SENDkFINDVALUE(self.node, host, port, key).deferred d.addCallback(self._handleFindVal, host, port) d.addErrback(self.errkfindnode, key, host, port) return d def _handleFindVal(self, response, host, port): if not isinstance(response, dict): # stop sending out more queries. self.pending = [] self.done = True #print "%s:%d sent value: %s" % (host, port, str(response)[:50]) #f = {} #f['k'] = [] #f['id'] = "0" #f['val'] = response # pass on returned value #return f else: pass #print "%s:%d sent kData: %s" % (host, port, response) return response def decideToContinue(self, response, key, x): if self.done: #if not response.has_key('val'): # logger.warn("response has no 'val', response is: %s" % response) #return response['val'] return response else: return self.makeQueries(key, x) def roundDone(self, responses, key, x): self.debugpath.append("FV: roundDone %d" % x) if self.done: result = {} # see if everyone's responses agreed... for success, resp in responses: # only look at successful non-kData (dict) responses. if success and resp != None and not isinstance(resp, dict): if result.has_key(resp): result[resp] += 1 else: result[resp] = 1 if len(result) == 0: # ... if no one responded, XXX: do something orther than None? logger.info("couldn't get any results") return None elif len(result) == 1: # ... if they did, return the result return result.keys()[0] else: # ... otherwise, return the result of the majority # (other options include returning all results) logger.info("got conflicting results, determining best...") quorumResult = None bestScore = 0 for r in result: #logger.debug("result %s scored %d" % (r, result[r])) if result[r] > bestScore: bestScore = result[r] quorumResult = r #logger.debug("result %s is new best" % r) logger.info("returning result %s", fdecode(quorumResult)) return quorumResult class SENDkFINDNODE(REQUEST): """ Makes one request to a node for its k-closest nodes closest to key """ def __init__(self, node, host, port, key, commandName="kFINDNODE"): """ """ logger.info("sending %s (findnode) for %s... to %s:%d" % (commandName, ("%x" % key)[:10], host, port)) self.commandName = commandName host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) Ku = self.node.config.Ku.exportPublicKey() url = 'http://'+host+':'+str(port)+'/'+self.commandName+'?' url += 'nodeID='+str(self.node.config.nodeID) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) url += '&port='+str(self.node.config.port) url += "&key="+str(fencode(key)) self.timeoutcount = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, node, host, port, key, url)) def startRequest(self, node, host, port, key, url): d = self._sendRequest(node, host, port, key, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, node, host, port, key, url): factory = getPageFactory(url, headers=self.headers, timeout=kprimitive_to) factory.deferred.addCallback(self._gotResponse, factory, node, host, port, key) factory.deferred.addErrback(self._errSendk, factory, node, host, port, key, url) return factory.deferred def _gotResponse(self, response, factory, node, host, port, key): logger.debug("kfindnode._gotResponse()") self._checkStatus(factory.status, response, host, port) response = eval(response) nID = long(response['id'], 16) updateNode(node.client, node.config, host, port, None, nID) updateNodes(node.client, node.config, response['k']) return response def _checkStatus(self, status, response, host, port): logger.debug("kfindnode._checkStatus()") if eval(status) != http.OK: raise failure.DefaultException(self.commandName+" FAILED from " +host+":"+str(port)+": received status "+status+", '" +response+"'") def _errSendk(self, err, factory, node, host, port, key, url): if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #print "GETID request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(node, host, port, key, url) else: #print "not trying again [#%d]" % self.timeoutcount return err logger.info("%s to %s failed -- %s" % (self.commandName, self.dest, err.getErrorMessage())) # XXX: updateNode-- return err class SENDkSTORE(REQUEST): """ Sends a single kSTORE to the given host:port, with key=val """ def __init__(self, node, host, port, key, val): logger.info("sending kSTORE to %s:%d" % (host, port)) REQUEST.__init__(self, host, port, node) Ku = node.config.Ku.exportPublicKey() url = 'http://'+host+':'+str(port)+'/kSTORE?' url += 'nodeID='+str(node.config.nodeID) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) url += '&port='+str(node.config.port) url += "&key="+str(fencode(key)) url += "&val="+str(fencode(val)) # XXX: instead of a single key/val, protocol will take a series of # vals representing the blocks of the coded file and their # locations (by nodeID). The entire thing will be stored under # the given key. Also may need things like signature[s] from # storing node[s], etc. #print "in SENDkSTORE.__init__, len(val)=%d" % len(str(val)) #print "in SENDkSTORE.__init__, len(enc(val))=%d" % len(fencode(val)) #print "in SENDkSTORE.__init__, len(url)=%d" % len(url) self.timeoutcount = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, host, port, url)) def startRequest(self, host, port, url): d = self._sendRequest(host, port, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, host, port, url): factory = getPageFactory(url,\ headers=self.headers, method='PUT', timeout=kprimitive_to) self.deferred.addCallback(self._kStoreFinished, host, port) self.deferred.addErrback(self._storeErr, host, port, url) return factory.deferred def _kStoreFinished(self, response, host, port): logger.info("kSTORE to %s:%d finished" % (host, port)) return response def _storeErr(self, err, host, port, url): if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #print "GETID request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(host, port, url) else: #print "not trying again [#%d]" % self.timeoutcount return err logger.info("kSTORE to %s failed: %s" % (self.dest, err.getErrorMessage())) # XXX: updateNode-- return err class SENDkFINDVALUE(SENDkFINDNODE): """ Issues a single kFINDVALUE request to host:port for the key. If the value is found at host:port, it is returned, otherwise, a 404 response is received and any errbacks are called. """ def __init__(self, node, host, port, key): SENDkFINDNODE.__init__(self, node, host, port, key, "kFINDVAL") def _gotResponse(self, response, factory, node, host, port, key): self._checkStatus(factory.status, response, host, port) # The following 'if' block is the only thing different from kFINDNODE. # If a node returns the value, content-type will be set to x-flud-data # and we should grab the data instead of continuing the recursive # search. if factory.response_headers.has_key('content-type')\ and factory.response_headers['content-type']\ == ['application/x-flud-data']: logger.info("received SENDkFINDVALUE data.") nID = None if factory.response_headers.has_key('nodeid'): nID = factory.response_headers['nodeid'][0] updateNode(node.client, node.config, host, port, None, nID) return response response = eval(response) nID = long(response['id'], 16) updateNode(node.client, node.config, host, port, None, nID) logger.info("received SENDkFINDVALUE nodes") logger.debug("received SENDkFINDVALUE nodes: %s" % response) updateNodes(node.client, node.config, response['k']) return response
Python
""" FludkRouting.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Implements kademlia-style kbuckets (the routing table for the DHT layer). Although this is not a derivative of Khashmir (written by Andrew Loewenstern, Aaron Swartz, et. al.), we would like to give Khashmir a nod for inspiring portions of the design. Khashmir is distributed under the MIT License and is a very nice piece of work. Take a look at http://khashmir.sourceforge.net/ for more information. """ from bisect import * import logging #k = 5 # This is the max depth of a kBucket k = 12 # This is the max depth of a kBucket and the replication factor # XXX: need to split out k into two vars a = 3 # alpha, the system-wide concurrency parameter idspace = 256 # using sha-256 logger = logging.getLogger("flud.k") class NodeCache: """ An LRU cache for nodes """ def __init__(self, size): self.size = size self.cache = {} self.cacheOrder = [] def insertNode(self, node): """ adds a node to the cache. if this displaces a node, the displaced node is returned """ if node[2] not in self.cache: self.cache[node[2]] = node self.cacheOrder.append(node[2]) if len(self.cacheOrder) > self.size: popped = self.cacheOrder.pop(0) self.cache.pop(popped) return popped def getNode(self, nodeID): """ returns a node from the cache, or None """ if nodeID in self.cache: return self.cache[node] return None def removeNode(self, node): """ removes a node from the cache """ if node[2] in self.cache: self.cache.pop(node[2]) self.cacheOrder.pop(node[2]) def nodes(self): return [self.cache[i] for i in self.cache] def kCompare(a, b, target): """ Uses the XOR metric to compare target to a and b (useful for sorting) @param a an integer (or long) value @param b an integer (or long) value @param target the target ID as an integer (or long) value @return 1 if b > a, -1 if a < b, 0 if a == b >>> l = [1, 2, 23, 14, 5, 4, 5, 3, 20] >>> l.sort(lambda a, b: kCompare(a, b, 5)) >>> l [5, 5, 4, 1, 3, 2, 14, 20, 23] """ x, y = target^a, target^b if x == y: return 0 return int((x - y) / abs(x - y)) class kRouting: """ Contains the kBuckets for this node. Provides methods for inserting, updating, and removing nodes. Most importantly, performs kademlia-style routing by returning the node[s] closest to a particular id. >>> table = kRouting(('1.2.3.4', 34, 123456), 20, 5) >>> table.insertNode(('2.2.3.4', 34, 23456)) >>> table.insertNode(('3.2.3.4', 34, 223456)) >>> table.insertNode(('4.2.3.4', 34, 723456)) >>> table.insertNode(('5.2.3.4', 34, 423456)) >>> table.insertNode(('6.2.3.4', 34, 323456)) >>> table.kBuckets [{'0-80000': [('1.2.3.4', 34, 123456), ('2.2.3.4', 34, 23456), ('3.2.3.4', 34, 223456), ('5.2.3.4', 34, 423456), ('6.2.3.4', 34, 323456)]}, {'80001-100000': [('4.2.3.4', 34, 723456)]}] >>> table.findNode(23456) [('2.2.3.4', 34, 23456), ('1.2.3.4', 34, 123456), ('3.2.3.4', 34, 223456), ('6.2.3.4', 34, 323456), ('5.2.3.4', 34, 423456)] >>> table.findNode(55555) [('2.2.3.4', 34, 23456), ('1.2.3.4', 34, 123456), ('3.2.3.4', 34, 223456), ('6.2.3.4', 34, 323456), ('5.2.3.4', 34, 423456)] >>> table.findNode(722222) [('4.2.3.4', 34, 723456), ('3.2.3.4', 34, 223456), ('1.2.3.4', 34, 123456), ('2.2.3.4', 34, 23456), ('5.2.3.4', 34, 423456)] >>> table.insertNode(('7.2.3.4', 34, 733456)) >>> table.insertNode(('8.2.3.4', 34, 743456)) >>> table.insertNode(('9.2.3.4', 34, 753456)) >>> table.insertNode(('10.2.3.4', 34, 763456)) >>> table.insertNode(('11.2.3.4', 34, 773456)) ('4.2.3.4', 34, 723456) >>> table.replaceNode(('4.2.3.4', 34, 723456), ('11.2.3.4', 34, 773456)) >>> table.kBuckets [{'0-80000': [('1.2.3.4', 34, 123456), ('2.2.3.4', 34, 23456), ('3.2.3.4', 34, 223456), ('5.2.3.4', 34, 423456), ('6.2.3.4', 34, 323456)]}, {'80001-100000': [('7.2.3.4', 34, 733456), ('8.2.3.4', 34, 743456), ('9.2.3.4', 34, 753456), ('10.2.3.4', 34, 763456), ('11.2.3.4', 34, 773456)]}] >>> table.removeNode(('1.2.3.4', 34, 123456)) >>> table.kBuckets [{'0-80000': [('2.2.3.4', 34, 23456), ('3.2.3.4', 34, 223456), ('5.2.3.4', 34, 423456), ('6.2.3.4', 34, 323456)]}, {'80001-100000': [('7.2.3.4', 34, 733456), ('8.2.3.4', 34, 743456), ('9.2.3.4', 34, 753456), ('10.2.3.4', 34, 763456), ('11.2.3.4', 34, 773456)]}] >>> table.knownNodes() [('2.2.3.4', 34, 23456), ('3.2.3.4', 34, 223456), ('5.2.3.4', 34, 423456), ('6.2.3.4', 34, 323456), ('7.2.3.4', 34, 733456), ('8.2.3.4', 34, 743456), ('9.2.3.4', 34, 753456), ('10.2.3.4', 34, 763456), ('11.2.3.4', 34, 773456)] >>> table.knownExternalNodes() [('2.2.3.4', 34, 23456), ('3.2.3.4', 34, 223456), ('5.2.3.4', 34, 423456), ('6.2.3.4', 34, 323456), ('7.2.3.4', 34, 733456), ('8.2.3.4', 34, 743456), ('9.2.3.4', 34, 753456), ('10.2.3.4', 34, 763456), ('11.2.3.4', 34, 773456)] """ def __init__(self, node, bits=idspace, depth=k): """ @param node a (ip, port, id) triple, where id is an int (this is needed to know when to split a bucket). """ self.k = depth self.replacementCache = NodeCache(300) self.kBuckets = [kBucket(0, 2**bits, depth),] #self.kBuckets = [kBucket(0, 1, depth),] #for i in xrange(1,bits): # self.kBuckets.append(kBucket(2**i, 2**(i+1)-1, depth)) self.insertNode(node) self.node = node def insertNode(self, node): """ Inserts a node into the appropriate kBucket. If the node already exists in the appropriate kBucket, it is moved to the tail of the list. If the bucket is full, this method returns the oldest node, which the caller should then ping. If the oldest node is alive, the caller does nothing. Otherwise, the caller should call replaceNode. @param node a (ip, port, id) triple, where id is a long. """ if len(node) < 3: raise ValueError("node must be a triple (ip, port, id)") id = node[2] bucket = self._findBucket(id) try: # XXX: need to transfer key/vals that belong to new node? bucket.updateNode(node) self.replacementCache.removeNode(node) except BucketFullException, e: if (bucket.begin <= self.node[2] < bucket.end): # bucket is full /and/ the local node is in this bucket, # split and try adding it again. self._splitBucket(bucket) self.insertNode(node) logger.debug("split and added %x" % node[2]) return # XXX: need to also split for some other cases, see sections 2.4 # and 4.2. else: # bucket is full but we won't split. Return the oldest node # so that the caller can determine if it should be expunged. # If the old node is not reachable, caller should call # replaceNode() logger.debug("didn't add %x" % node[2]) return bucket.contents[0] logger.debug("didn't add %x" % node[2]) return bucket.contents[0] def removeNode(self, node): """ Invalidates a node. """ bucket = self._findBucket(node[2]) bucket.delNode(node) def replaceNode(self, replacee, replacer): """ Expunges replacee from its bucket, making room to add replacer """ # XXX: constraint checks: replacee & replacer belong to the same bucket, # bucket is currently full, adding replacer doesn't overfill, etc. self.removeNode(replacee) self.insertNode(replacer) def findNode(self, nodeID): """ Returns k closest node triples with which the caller may make additional queries. If nodeID is found, it will be the first result. @param nodeID an int """ nodes = [] bucket = self._findBucket(nodeID) #n = bucket.findNode(nodeID) #if n != None: # nodes.append(n) nodes += bucket.contents if len(nodes) < self.k: nextbucket = self._nextbucket(bucket) prevbucket = self._prevbucket(bucket) while len(nodes) < self.k \ and (nextbucket != None or prevbucket != None): if nextbucket != None: nodes += nextbucket.contents if prevbucket != None: nodes += prevbucket.contents nextbucket = self._nextbucket(nextbucket) prevbucket = self._prevbucket(prevbucket) nodes.sort(lambda a, b, n=nodeID: cmp(n ^ a[2], n ^ b[2])) return nodes[:self.k] def getNode(self, nodeID): """ Attempts to get the given node, returning a <ip, port, id> triple. If the node is not found locally, returns None @param nodeID an int """ bucket = self._findBucket(nodeID) n = bucket.findNode(nodeID) if not n: n = self.replacementCache.getNode(nodeID) if n != None: return n return None def updateNode(self, node): """ Call to update a node, i.e., whenever the node has been recently seen @param node a (ip, port, id) triple, where id is an int. """ self.insertNode(node) def knownExternalNodes(self): result = [] for i in self.kBuckets: for j in i.contents: if j[2] != self.node[2]: result.append(j) result += self.replacementCache.nodes() return result def knownNodes(self): result = [] for i in self.kBuckets: for j in i.contents: result.append(j) result += self.replacementCache.nodes() return result def _nextbucket(self, bucket): if bucket == None: return bucket i = self.kBuckets.index(bucket)+1 if i >= len(self.kBuckets): return None return self.kBuckets[i] def _prevbucket(self, bucket): if bucket == None: return bucket i = self.kBuckets.index(bucket)-1 if i < 0: return None return self.kBuckets[i] def _findBucket(self, i): """ returns the bucket which would contain i. @param i an int """ #print "kBuckets = %s" % str(self.kBuckets) bl = bisect_left(self.kBuckets, i) if bl >= len(self.kBuckets): raise Exception( "tried to find an ID that is larger than ID space: %s" % i) return self.kBuckets[bisect_left(self.kBuckets, i)] def _splitBucket(self, bucket): """ This is called for the special case when the bucket is full and this node is a member of the bucket. When this occurs, the bucket should be split into two new buckets. """ halfpoint = (bucket.end - bucket.begin) / 2 newbucket = kBucket(bucket.end - halfpoint + 1, bucket.end, self.k) self.kBuckets.insert(self.kBuckets.index(bucket.begin) + 1, newbucket) bucket.end -= halfpoint for node in bucket.contents[:]: if node[2] > bucket.end: bucket.delNode(node) newbucket.addNode(node) class kBucket: """ A kBucket is a list of <ip, port, id> triples, ordered according to time last seen (most recent at tail). Every kBucket has a begin and end number, indicating the chunk of the id space that it contains. >>> b = kBucket(0,100,5) >>> b {'0-64': []} >>> n1 = ('1.2.3.4', 45, 'd234a53546e4c23') >>> n2 = ('10.20.30.40', 45, 'abcd234a53546e4') >>> n3 = ('10.20.30.4', 5, 'abcd') >>> b.addNode(n1) >>> b {'0-64': [('1.2.3.4', 45, 'd234a53546e4c23')]} >>> b.addNode(n2) >>> b {'0-64': [('1.2.3.4', 45, 'd234a53546e4c23'), ('10.20.30.40', 45, 'abcd234a53546e4')]} >>> b.addNode(n1) >>> b {'0-64': [('10.20.30.40', 45, 'abcd234a53546e4'), ('1.2.3.4', 45, 'd234a53546e4c23')]} >>> b.delNode(n3) >>> b {'0-64': [('10.20.30.40', 45, 'abcd234a53546e4'), ('1.2.3.4', 45, 'd234a53546e4c23')]} >>> b.addNode(n2) >>> b {'0-64': [('1.2.3.4', 45, 'd234a53546e4c23'), ('10.20.30.40', 45, 'abcd234a53546e4')]} >>> b.updateNode(n1) >>> b {'0-64': [('10.20.30.40', 45, 'abcd234a53546e4'), ('1.2.3.4', 45, 'd234a53546e4c23')]} >>> b.delNode(n2) >>> b {'0-64': [('1.2.3.4', 45, 'd234a53546e4c23')]} >>> b.addNode(n3) >>> f = b.findNode(n3[2]) >>> f == n3 True >>> c = kBucket(101,200,5) >>> d = kBucket(150,250,5) # wouldn't really have overlap in practice >>> e = kBucket(251, 2**256,5) >>> buckets = (b, c, d, e) # if not added insort, must sort for bisect >>> b1 = b >>> b1 == b True >>> b1 != b False >>> b == 50 True >>> b == 0 True >>> b == 100 True >>> b == -1 False >>> b > -1 True >>> b == 101 False >>> b < 101 True >>> b <= 90 True >>> b <= 100 True >>> b <= 101 True >>> b < d True >>> b <= c True >>> b > c False >>> bisect_left(buckets, 98) 0 >>> bisect_left(buckets, 198) 1 >>> bisect_left(buckets, 238) 2 >>> bisect_left(buckets, 298) 3 """ def __init__(self, begin, end, depth=k): self.k = depth self.begin = begin self.end = end self.contents = [] def __repr__(self): return "{'%x-%x': %s}" % (self.begin, self.end, self.contents) #return "{'"+repr(self.begin)+'-'+repr(self.end)+"': "\ # +repr(self.contents)+"}" #return "<kBucket "+repr(self.begin)+'-'+repr(self.end)+": "\ # +repr(self.contents)+">" def addNode(self, node): """ adds the given node to this bucket. If the node is already a member of this bucket, its position is updated to the end of the list. If the bucket is full, raises an exception """ if node in self.contents: self.contents.remove(node) self.contents.append(node) elif len(self.contents) >= self.k: raise BucketFullException() else: ids = [x[2] for x in self.contents] if node[2] in ids: # remove the matching node's old contact info self.contents.pop(ids.index(node[2])) self.contents.append(node) def updateNode(self, node): """ Moves the given node to the tail of the list. If the node isn't present in this bucket, this method attempts to add it by calling addNode (which may throw a BucketFullException if bucket is full) """ self.addNode(node) def delNode(self, node): """ removes the given node, if present, from this bucket """ try: self.contents.remove(node) except: pass def findNode(self, nodeID): for i in self.contents: if i[2] == nodeID: return i return None # The following comparators allow us to use list & bisect on the buckets. # integers, longs, and buckets all may be compared to a bucket. def __eq__(self, i): return i >= self.begin and self.end >= i def __ne__(self, i): return i < self.begin or self.end < i def __lt__(self, i): return self.end < i def __le__(self, i): return self.begin <= i def __gt__(self, i): return self.begin > i def __ge__(self, i): return self.end >= i class BucketFullException(Exception): pass def _test(): import doctest doctest.testmod() if __name__ == '__main__': _test()
Python
import zfec import zfec.easyfec as easyfec import zfec.filefec as filefec from pyutil import fileutil from pyutil.mathutil import pad_size, log_ceil import array, os, re, struct, traceback FORMAT_FORMAT = "%%s.%%0%dd_%%0%dd%%s" RE_FORMAT = "%s.[0-9]+_[0-9]+%s" def encode_to_files(inf, fsize, dirname, prefix, k, m, suffix=".fec", overwrite=False, verbose=False): """ Encode inf, writing the shares to specially named, newly created files. @param fsize: calling read() on inf must yield fsize bytes of data and then raise an EOFError @param dirname: the name of the directory into which the sharefiles will be written """ mlen = len(str(m)) format = FORMAT_FORMAT % (mlen, mlen,) padbytes = pad_size(fsize, k) fns = [] fs = [] try: for shnum in range(m): hdr = filefec._build_header(m, k, padbytes, shnum) fn = os.path.join(dirname, format % (prefix, shnum, m, suffix,)) if verbose: print "Creating share file %r..." % (fn,) if overwrite: f = open(fn, "wb") else: flags = os.O_WRONLY|os.O_CREAT|os.O_EXCL | (hasattr(os, 'O_BINARY') and os.O_BINARY) fd = os.open(fn, flags) f = os.fdopen(fd, "wb") f.write(hdr) fs.append(f) fns.append(fn) sumlen = [0] def cb(blocks, length): assert len(blocks) == len(fs) oldsumlen = sumlen[0] sumlen[0] += length if verbose: if int((float(oldsumlen) / fsize) * 10) \ != int((float(sumlen[0]) / fsize) * 10): print str(int((float(sumlen[0]) / fsize) * 10) * 10) \ + "% ...", if sumlen[0] > fsize: raise IOError("Wrong file size -- possibly the size of the" " file changed during encoding. Original size: %d," " observed size at least: %s" % (fsize, sumlen[0],)) for i in range(len(blocks)): data = blocks[i] fs[i].write(data) length -= len(data) filefec.encode_file_stringy_easyfec(inf, cb, k, m, chunksize=4096) except EnvironmentError, le: print "Cannot complete because of exception: " print le print "Cleaning up..." # clean up while fs: f = fs.pop() f.close() ; del f fn = fns.pop() if verbose: print "Cleaning up: trying to remove %r..." % (fn,) fileutil.remove_if_possible(fn) return None if verbose: print print "Done!" return fns # Note: if you really prefer base-2 and you change this code, then please # denote 2^20 as "MiB" instead of "MB" in order to avoid ambiguity. # Thanks. # http://en.wikipedia.org/wiki/Megabyte MILLION_BYTES=10**6 def decode_from_files(outf, infiles, verbose=False): """ Decode from the first k files in infiles, writing the results to outf. """ assert len(infiles) >= 2 infs = [] shnums = [] m = None k = None padlen = None byteswritten = 0 for f in infiles: (nm, nk, npadlen, shnum,) = filefec._parse_header(f) if not (m is None or m == nm): raise CorruptedShareFilesError("Share files were corrupted --" " share file %r said that m was %s but another share file" " previously said that m was %s" % (f.name, nm, m,)) m = nm if not (k is None or k == nk): raise CorruptedShareFilesError("Share files were corrupted --" " share file %r said that k was %s but another share file" " previously said that k was %s" % (f.name, nk, k,)) if k > len(infiles): raise InsufficientShareFilesError(k, len(infiles)) k = nk if not (padlen is None or padlen == npadlen): raise CorruptedShareFilesError("Share files were corrupted --" " share file %r said that pad length was %s but another" " share file previously said that pad length was %s" % (f.name, npadlen, padlen,)) padlen = npadlen infs.append(f) shnums.append(shnum) if len(infs) == k: break dec = easyfec.Decoder(k, m) while True: chunks = [ inf.read(filefec.CHUNKSIZE) for inf in infs ] if [ch for ch in chunks if len(ch) != len(chunks[-1])]: raise CorruptedShareFilesError("Share files were corrupted --" " all share files are required to be the same length," " but they weren't.") if len(chunks[-1]) == filefec.CHUNKSIZE: # Then this was a full read, so we're still in the sharefiles. resultdata = dec.decode(chunks, shnums, padlen=0) outf.write(resultdata) byteswritten += len(resultdata) if verbose: if ((byteswritten - len(resultdata)) / (10*MILLION_BYTES)) \ != (byteswritten / (10*MILLION_BYTES)): print str(byteswritten / MILLION_BYTES) + " MB ...", else: # Then this was a short read, so we've reached the end of the # sharefiles. resultdata = dec.decode(chunks, shnums, padlen) outf.write(resultdata) return True if verbose: print print "Done!" return True # fludfilefec -- modified zfec filefec for use with flud, based on: # # zfec -- fast forward error correction library with Python interface # # Copyright (C) 2007 Allmydata, Inc. # Author: Zooko Wilcox-O'Hearn # # This file is part of zfec. # # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the Free # Software Foundation; either version 2 of the License, or (at your option) # any later version, with the added permission that, if you become obligated # to release a derived work under this licence (as per section 2.b), you may # delay the fulfillment of this obligation for up to 12 months. See the file # COPYING for details. # # If you would like to inquire about a commercial relationship with Allmydata, # Inc., please contact partnerships@allmydata.com and visit # http://allmydata.com/.
Python
from twisted.python import failure from twisted.internet import defer """ FludDefer.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. """ class ErrDeferredList(defer.DeferredList): """ ErrDeferredList acts just like DeferredList, except that if *any* of the Deferreds in the DeferredList errback(), the NewDeferredList also errback()s. This is different from DeferredList(fireOnOneErrback=True) in that if you use that method, you only know about the first failure, and you won't learn of subsequent failures/success in the list. returnOne indicates whether the full result of the DeferredList should be returned, or just the first result (or first error) """ def __init__(self, list, returnOne=False): defer.DeferredList.__init__(self, list, consumeErrors=True) self.returnOne = returnOne self.addCallback(self.wrapResult) def wrapResult(self, result): #print "DEBUG: result= %s" % result for i in result: if i[0] == False: if self.returnOne: raise failure.DefaultException(i[1]) else: raise failure.DefaultException(result) if self.returnOne: #print "DEBUG: returning %s" % str(result[0][1]) return result[0][1] else: #print "DEBUG: returning %s" % result return result
Python
#!/usr/bin/python """ FludNode.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), verison 3. FludNode is the process that runs to talk with other nodes in the flud backup network. """ from twisted.internet import reactor, defer import threading, signal, sys, time, os, random, logging from flud.FludConfig import FludConfig from flud.protocol.FludServer import FludServer from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import getCanonicalIP PINGTIME=60 SYNCTIME=900 class FludNode(object): """ A node in the flud network. A node is both a client and a server. It listens on a network accessible port for both DHT and Storage layer requests, and it listens on a local port for client interface commands. """ def __init__(self, port=None): self._initLogger() self.config = FludConfig() self.logger.removeHandler(self.screenhandler) self.config.load(serverport=port) self.client = FludClient(self) self.DHTtstamp = time.time()+10 def _initLogger(self): logger = logging.getLogger('flud') self.screenhandler = logging.StreamHandler() self.screenhandler.setLevel(logging.INFO) logger.addHandler(self.screenhandler) self.logger = logger def pingRandom(self, tstamp): return # XXX: see pg. 4, Section 2.2 (short) or 2.3 (long) of the Kademlia # paper -- once an hour, each node should check any buckets that # haven't been refreshed and pick a random id within that space # to findnode(id) on, for all buckets. if tstamp < self.DHTtstamp: #r = random.randrange(2**256) n = self.config.routing.knownExternalNodes() if len(n) > 2: n1 = random.choice(n)[2] n2 = random.choice(n)[2] r = (n1+n2)/2 else: r = random.randrange(2**256) def badNode(error): node.logger.warn("Couldn't ping %s:%s" % (sys.argv[1], sys.argv[2])) d = self.client.kFindNode(r) d.addErrback(badNode) pingtime = random.randrange(PINGTIME/2, PINGTIME) reactor.callLater(pingtime, self.pingRandom, time.time()) def syncConfig(self): self.config.save() reactor.callLater(SYNCTIME, self.syncConfig) def start(self, twistd=False): """ starts the reactor in this thread """ self.webserver = FludServer(self, self.config.port) self.logger.log(logging.INFO, "FludServer starting") reactor.callLater(1, self.pingRandom, time.time()) reactor.callLater(random.randrange(10), self.syncConfig) if not twistd: reactor.run() def run(self): """ starts the reactor in its own thread """ #signal.signal(signal.SIGINT, self.sighandler) signal.signal(signal.SIGINT, signal.SIG_DFL) signal.signal(signal.SIGTERM, signal.SIG_DFL) self.webserver = FludServer(self, self.config.port) self.webserver.start() # XXX: need to do save out current config every X seconds # XXX: need to seperate known_nodes from config, and then update this # every X seconds. only update config when it changes. def stop(self): self.logger.log(logging.INFO, "shutting down FludNode") self.webserver.stop() def join(self): self.webserver.join() def sighandler(self, sig, frame): self.logger.log(logging.INFO, "handling signal %s" % sig) def connectViaGateway(self, host, port): def refresh(knodes): def refreshDone(results): self.logger.info("bucket refreshes finished: %s" % results) print "flud node connected and listening on port %d"\ % self.config.port #print "found knodes %s" % knodes dlist = [] for bucket in self.config.routing.kBuckets: #if True: if bucket.begin <= self.config.routing.node[2] < bucket.end: pass #print "passed on bucket %x-%s" % (bucket.begin, bucket.end) else: refreshID = random.randrange(bucket.begin, bucket.end) #print "refreshing bucket %x-%x by finding %x" \ # % (bucket.begin, bucket.end, refreshID) self.logger.info("refreshing bucket %x-%x by finding %x" % (bucket.begin, bucket.end, refreshID)) deferred = self.client.kFindNode(refreshID) dlist.append(deferred) dl = defer.DeferredList(dlist) dl.addCallback(refreshDone) # XXX: do we need to ping newly discovered known nodes? If not, # we could be vulnerable to a poisoning attack (at first # glance, this attack seems rather impotent...) # XXX: need to call refresh about every 60 minutes. Add a # reactor.callLater here to do it. def badGW(error): self.logger.warn(error) self.logger.warn("Couldn't connect to gateway at %s:%s" % (sys.argv[1], sys.argv[2])) self.logger.debug("connectViaGateway %s%d" % (host, port)) deferred = self.client.sendkFindNode(host, port, self.config.routing.node[2]) deferred.addCallback(refresh) deferred.addErrback(badGW) def getPath(): # this is a hack to be able to get the location of FludNode.tac return os.path.dirname(os.path.abspath(__file__))
Python
""" CheckboxState.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. CheckboxState represents the states which a checkbox in DirCtrl can take """ class CheckboxState: (UNSELECTED, SELECTED, SELECTEDCHILD, SELECTEDPARENT, EXCLUDED, EXCLUDEDCHILD) = range(6) def offset(oldstate, newstate): return newstate - oldstate offset = staticmethod(offset)
Python
class FludException(Exception): pass
Python
#!/usr/bin/python """ FludTestGauges.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Provides gauges for visualizing storage for multiple flud nodes running on the same host. This is really only useful for demos and testing. """ import sys, os, signal, stat, random import wx import wx.lib.buttons as buttons from flud.FludConfig import FludConfig dutotal = 0 def visit(arg, top, files): global dutotal for file in files: dutotal += os.lstat("%s" % (os.path.join(top,file)))[stat.ST_SIZE] arg += dutotal def du(dir): global dutotal dutotal = 0 os.path.walk(dir, visit, dutotal) return dutotal # XXX: too much manual layout. should convert to a managed layout to allow for # resizing, etc. SGAUGEWIDTH = 230 # storage gauge DGAUGEWIDTH = 100 # dht gauge GAUGEHEIGHT = 20 ROWHEIGHT = 30 SEP = 5 LABELWIDTH = 20 POWERWIDTH = 70 RATIOBARHEIGHT = 70 COLWIDTH = SGAUGEWIDTH+DGAUGEWIDTH+LABELWIDTH+POWERWIDTH COLGAPFUDGE = 30 class FludTestGauges(wx.Frame): def __init__(self, parent, title, dirroot, dirs): screenHeight = wx.SystemSettings.GetMetric(wx.SYS_SCREEN_Y)-100 rowheight = ROWHEIGHT+SEP height = len(dirs)*(rowheight)+RATIOBARHEIGHT columns = height / screenHeight + 1 width = COLWIDTH*columns if columns > 1: height = (len(dirs)/columns)*(rowheight)+RATIOBARHEIGHT if (len(dirs) % columns) > 0: height += rowheight wx.Frame.__init__(self, parent, wx.ID_ANY, title, size=(width,height), style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE) self.storebarend = 1024 self.smultiplier = 100.0 / self.storebarend self.sdivisor = 1 self.sbytelabel = "" self.dhtbarend = 512 self.dmultiplier = 100.0 / self.dhtbarend self.ddivisor = 1 self.dbytelabel = "" self.storeheading = wx.StaticText(self, -1, "block storage", (LABELWIDTH, 5)) self.totaldht = wx.StaticText(self, -1, "metadata", (LABELWIDTH+SGAUGEWIDTH+SEP, 5)) self.gauges = [] curCol = 0 curRow = 30 for i in range(len(dirs)): self.gauges.append(wx.Gauge(self, -1, 100, (curCol*COLWIDTH+LABELWIDTH, curRow), (SGAUGEWIDTH, GAUGEHEIGHT))) self.gauges[i].SetBezelFace(3) self.gauges[i].SetShadowWidth(3) self.gauges[i].SetValue(0) self.gauges[i].dir = "%s%s" % (dirroot,dirs[i]) os.environ['FLUDHOME'] = self.gauges[i].dir; conf = FludConfig() conf.load(doLogging = False) print "%s" % conf.nodeID self.gauges[i].label = wx.StaticText(self, -1, "%2s" % dirs[i], (curCol*COLWIDTH, curRow+(rowheight/4)), size=(LABELWIDTH, -1)) self.gauges[i].idlabel = wx.StaticText(self, -1, "%s" % conf.nodeID, (curCol*COLWIDTH+LABELWIDTH, curRow+20)) font = self.gauges[i].idlabel.GetFont() font.SetPointSize(6) self.gauges[i].idlabel.SetFont(font) self.gauges[i].dhtgauge = wx.Gauge(self, -1, 100, (curCol*COLWIDTH+LABELWIDTH+SGAUGEWIDTH+SEP, curRow), (SGAUGEWIDTH/3, GAUGEHEIGHT)) self.gauges[i].power = wx.Button(self, i, "turn OFF", (curCol*COLWIDTH +LABELWIDTH+SGAUGEWIDTH+2*SEP+SGAUGEWIDTH/3, curRow), (POWERWIDTH, ROWHEIGHT)) #self.gauges[i].power = buttons.GenBitmapToggleButton(self, i, # None, # (LABELWIDTH+SGAUGEWIDTH+2*SEP+SGAUGEWIDTH/3, curRow), # (POWERWIDTH, ROWHEIGHT)) #self.gauges[i].button.SetBestSize() self.gauges[i].power.SetToolTipString("power on/off") self.Bind(wx.EVT_BUTTON, self.onClick, self.gauges[i].power) curRow += rowheight if curRow > height-RATIOBARHEIGHT: curCol += 1 curRow = 30 self.totalstore = wx.StaticText(self, -1, "total: 0", (LABELWIDTH, height-40)) self.totaldht = wx.StaticText(self, -1, "total: 0", (LABELWIDTH+SGAUGEWIDTH+SEP, height-40)) self.ratiogauge = wx.Gauge(self, -1, 100, (LABELWIDTH, height-20), (SGAUGEWIDTH+SEP+SGAUGEWIDTH/3, 10)) self.ratiogauge.SetValue(0) self.Bind(wx.EVT_IDLE, self.IdleHandler) self.timer = wx.PyTimer(self.update) self.timer.Start(1000) def onClick(self, event): # XXX: note that under our current startNnodes.sh scheme, the first # node spawned doesn't contact anyone, so if that one is powered off # and then powered back on, it will not be part of the node until # another node pings it # XXX: unix-specific proc management stuff follows idx = event.GetId() home = self.gauges[idx].dir pidfile = os.path.join(home, 'twistd.pid') if os.path.exists(pidfile): print "shutting down %s" % home f = open(pidfile) pid = int(f.read()) f.close() # XXX: ps command no worky on windows, and "-ww" may not worker on # oldskool unixes self.gauges[idx].savedCmd = os.popen( "ps f -wwp %d -o args=" % pid).read() procline = os.popen("ps e -wwp %d" % pid).read() self.gauges[idx].savedEnv = [e for e in procline.split() if e[:4] == 'FLUD'] # XXX: os.kill no worky on windows, need something like: #def windowskill(pid): # import win32api # handle = win32api.OpenProcess(1, 0, pid) # return (0 != win32api.TerminateProcess(handle, 0)) os.kill(pid, signal.SIGTERM) self.gauges[idx].power.SetLabel("turn ON") self.gauges[idx].Hide() self.gauges[idx].dhtgauge.Hide() else: print "powering up %s" % home # XXX: this exec no worky on windows fullcmd = "%s %s" % (' '.join(self.gauges[idx].savedEnv), self.gauges[idx].savedCmd) print fullcmd result = os.popen('%s %s' % (' '.join(self.gauges[idx].savedEnv), self.gauges[idx].savedCmd)).readlines() self.gauges[idx].power.SetLabel("turn OFF") self.gauges[idx].Show() self.gauges[idx].dhtgauge.Show() print result def update(self): def sizeclass(num): divisor = 1 bytelabel = "" if num > 1024: divisor = 1024.0 bytelabel = 'K' if num > 1048576: divisor = 1048576.0 bytelabel = 'M' if num > 1073741824: divisor = 1073741824.0 bytelabel = 'G' return (divisor, bytelabel) storelargest = 0 dhtlargest = 0 storetotal = 0 dhttotal = 0 for i in self.gauges: if os.path.isdir(i.dir): i.storebytes = du(os.path.join(i.dir,'store')) if i.storebytes > storelargest: storelargest = i.storebytes storetotal += i.storebytes i.dhtbytes = du(os.path.join(i.dir,'dht')) if i.dhtbytes > dhtlargest: dhtlargest = i.dhtbytes dhttotal += i.dhtbytes else: i.storebytes = 0 i.dhtbytes = 0 i.Disable() i.power.Disable() while storelargest > self.storebarend: self.storebarend = self.storebarend * 2 self.smultiplier = 100.0 / self.storebarend self.sdivisor, self.sbytelabel = sizeclass(storetotal) while dhtlargest > self.dhtbarend: self.dhtbarend = self.dhtbarend * 2 self.dmultiplier = 100.0 / self.dhtbarend self.ddivisor, self.dbytelabel = sizeclass(dhttotal) #print "-----" for i in self.gauges: i.SetValue(i.storebytes*self.smultiplier) i.dhtgauge.SetValue(i.dhtbytes*self.dmultiplier) #print "%.2f, %.2f" % ((float(i.storebytes)/float(i.dhtbytes)), # (float(i.GetValue())/float(i.dhtgauge.GetValue()))) self.totalstore.SetLabel("total: %.1f%s" % (float(storetotal)/self.sdivisor, self.sbytelabel)) self.totaldht.SetLabel("total: %.1f%s" % (float(dhttotal)/self.ddivisor, self.dbytelabel)) if (dhttotal+storetotal == 0): self.ratiogauge.SetValue(0) else: self.ratiogauge.SetValue((storetotal*100/(dhttotal+storetotal))) def updateGauges(self, update): for index, value in update: self.monitors[index].setValue(value) def IdleHandler(self, event): pass def main(): if len(sys.argv) < 2: print "usage: %s dircommon exts" % sys.argv[0] print " where exts will be appended to dircommon" print " e.g., '%s /home/joe/.flud 1,2,3,4,10,15,20'"\ % sys.argv[0] print " or, '%s /home/joe/.flud 1-10,15,20'"\ % sys.argv[0] sys.exit() root = sys.argv[1] exts = [] dirs = [d.strip() for d in sys.argv[2].split(',')] for i in dirs: if i == "_": exts.append('') # undocumented, means "just dircommon" elif i.find('-') >= 0: start, end = i.split('-') for j in range(int(start),int(end)+1): exts.append(j) else: exts.append(int(i)) app = wx.PySimpleApp() t = FludTestGauges(None, 'Flud Test Gauges', root, exts) t.Show(1) app.MainLoop() if __name__ == '__main__': main()
Python
import os, stat, sys, tarfile, tempfile import gzip from flud.FludCrypto import hashstream from flud.fencode import fencode """ TarfileUtils.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Provides additional tarfile functionality (deletion of a member from a tarball, and concatenation of tarballs). """ def delete(tarball, membernames): """ Deletes a member file[s] from a tarball. Returns the names of deleted members if they are removed, False if the file[s] aren't members. If membernames contains all the members in the tarball, the entire tarball is deleted """ gzipped = False if tarball[-7:] == ".tar.gz": gzipped = True f = tarfile.open(tarball, 'r:gz') else: f = tarfile.open(tarball, 'r') if not isinstance(membernames, list): membernames = [membernames] tarnames = f.getnames() for membername in membernames: if not membername in tarnames: membernames.remove(membername) if len(membernames) < 1: f.close() return False if len(tarnames) == len(membernames): f.close() os.remove(tarball) return True f.close() if gzipped: tarball = gunzipTarball(tarball) f = open(tarball, 'r+') tfile = tempfile.mktemp() if gzipped: f2 = gzip.GzipFile(tfile, 'w') else: f2 = open(tfile, 'w') empty = tarfile.BLOCKSIZE * '\0' done = False removednames = [] while not done: bytes = f.read(tarfile.BLOCKSIZE) if bytes == "": done = True elif bytes == empty: f2.write(bytes) else: name = bytes[0:99] name = name[:name.find(chr(0))] size = int(bytes[124:135], 8) blocks = size / tarfile.BLOCKSIZE if (size % tarfile.BLOCKSIZE) > 0: blocks += 1 if name in membernames: f.seek(blocks*tarfile.BLOCKSIZE + f.tell()) removednames.append(name) else: f2.write(bytes) for i in range(blocks): f2.write(f.read(tarfile.BLOCKSIZE)) f2.close() f.close() if gzipped: os.remove(tarball) tarball = tarball+".gz" os.rename(tfile, tarball) return removednames def concatenate(tarfile1, tarfile2): """ Combines tarfile1 and tarfile2 into tarfile1. tarfile1 is modified in the process, and tarfile2 is deleted. """ gzipped = False if tarfile1[-7:] == ".tar.gz": gzipped = True f1 = gzip.GzipFile(tarfile1, 'r') tarfile1 = tarfile1[:-3] f1unzip = file(tarfile1, 'w') f1unzip.write(f1.read()) f1unzip.close() f1.close() os.remove(tarfile1+".gz") f = open(tarfile1, "r+") done = False e = '\0' empty = tarfile.BLOCKSIZE*e emptyblockcount = 0 while not done: header = f.read(tarfile.BLOCKSIZE) if header == "": print "error: end of archive not found" return elif header == empty: emptyblockcount += 1 if emptyblockcount == 2: done = True else: emptyblockcount = 0 fsize = eval(header[124:135]) skip = int(round(float(fsize) / float(tarfile.BLOCKSIZE) + 0.5)) f.seek(skip*tarfile.BLOCKSIZE, 1) # truncate the file to the spot before the end-of-tar marker trueend = f.tell() - (tarfile.BLOCKSIZE*2) f.seek(trueend) f.truncate() # now write the contents of the second tarfile into this spot if tarfile2[-7:] == ".tar.gz": f2 = gzip.GzipFile(tarfile2, 'r') else: f2 = open(tarfile2, "r") done = False while not done: header = f2.read(tarfile.BLOCKSIZE) if header == "": print "error: end of archive not found" f.seek(trueend) f.write(empty*2) return else: f.write(header) if header == empty: emptyblockcount += 1 if emptyblockcount == 2: done = True else: emptyblockcount = 0 fsize = eval(header[124:135]) bsize = int(round(float(fsize) / float(tarfile.BLOCKSIZE) + 0.5)) # XXX: break this up if large data = f2.read(bsize*tarfile.BLOCKSIZE) f.write(data) f2.close() f.close() if gzipped: f2 = gzip.GzipFile(tarfile1+".gz", 'wb') f = file(tarfile1, 'rb') f2.write(f.read()) f2.close() f.close() os.remove(tarfile1) # and delete the second tarfile os.remove(tarfile2) #print "concatenated %s to %s" % (tarfile2, tarfile1) def verifyHashes(tarball, ignoreExt=None): # return all the names of files in this tarball if hash checksum passes, # otherwise return False digests = [] done = False if tarball[-7:] == ".tar.gz": f = gzip.GzipFile(tarball, 'r:gz') else: f = open(tarball, 'r') empty = tarfile.BLOCKSIZE * '\0' while not done: bytes = f.read(tarfile.BLOCKSIZE) if bytes == "": done = True elif bytes == empty: pass else: if bytes[0] == '\0' and bytes[124] == '\0': print "WARNING: read nulls when expecting file header" break name = bytes[0:99] name = name[:name.find(chr(0))] size = int(bytes[124:135], 8) blocks = size / tarfile.BLOCKSIZE if ignoreExt and name[-len(ignoreExt):] == ignoreExt: # gzip doesn't support f.seek(size, 1) f.seek(f.tell()+size) else: digest = hashstream(f, size) digest = fencode(int(digest,16)) if name == digest: #print "%s == %s" % (name, digest) digests.append(name) else: #print "%s != %s" % (name, digest) f.close() return [] if (size % tarfile.BLOCKSIZE) > 0: blocks += 1 f.seek((blocks * tarfile.BLOCKSIZE) - size + f.tell()) f.close() return digests def gzipTarball(tarball): if tarball[-4:] != '.tar': return None f = gzip.GzipFile(tarball+".gz", 'wb') f.write(file(tarball, 'rb').read()) f.close() os.remove(tarball) return tarball+".gz" def gunzipTarball(tarball): if tarball[-3:] != '.gz': return None f = gzip.GzipFile(tarball, 'rb') file(tarball[:-3], 'wb').write(f.read()) f.close() os.remove(tarball) return tarball[:-3] if __name__ == "__main__": if (len(sys.argv) < 4 or sys.argv[1] != "-d") \ and (len(sys.argv) != 4 or sys.argv[1] != "-c") \ and sys.argv[1] != "-v": print "usage: [-d tarfile tarfilemembers]\n"\ +" [-c tarfile1 tarfile2]\n"\ +" [-v tarfile]\n"\ +" -d deletes tarfilemembers from tarfile,\n"\ +" -c concatenates tarfile1 and tarfile2 into tarfile1\n"\ +" -v verifies that the names of files in tarfile are sha256\n" sys.exit(-1) if sys.argv[1] == "-d": deleted = delete(sys.argv[2], sys.argv[3:]) if deleted == sys.argv[3:]: print "%s successfully deleted from %s" % (deleted, sys.argv[2]) else: faileddeletes = [x for x in sys.argv[3:] if x not in deleted] print "could not delete %s from %s" % (faileddeletes, sys.argv[2]) elif sys.argv[1] == "-c": concatenate(sys.argv[2], sys.argv[3]) print "concatenated %s and %s into %s" % (sys.argv[2], sys.argv[3], sys.argv[2]) elif sys.argv[1] == "-v": digests = verifyHashes(sys.argv[2]) if digests: print "verified tarfile member digests for: %s" % digests else: print "some tarfile members failed digest check"
Python
#!/usr/bin/python """ FludNode.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), verison 3. FludNode is the process that runs to talk with other nodes in the flud backup network. """ from twisted.internet import reactor, defer import threading, signal, sys, time, os, random, logging from flud.FludConfig import FludConfig from flud.protocol.FludServer import FludServer from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import getCanonicalIP PINGTIME=60 SYNCTIME=900 class FludNode(object): """ A node in the flud network. A node is both a client and a server. It listens on a network accessible port for both DHT and Storage layer requests, and it listens on a local port for client interface commands. """ def __init__(self, port=None): self._initLogger() self.config = FludConfig() self.logger.removeHandler(self.screenhandler) self.config.load(serverport=port) self.client = FludClient(self) self.DHTtstamp = time.time()+10 def _initLogger(self): logger = logging.getLogger('flud') self.screenhandler = logging.StreamHandler() self.screenhandler.setLevel(logging.INFO) logger.addHandler(self.screenhandler) self.logger = logger def pingRandom(self, tstamp): return # XXX: see pg. 4, Section 2.2 (short) or 2.3 (long) of the Kademlia # paper -- once an hour, each node should check any buckets that # haven't been refreshed and pick a random id within that space # to findnode(id) on, for all buckets. if tstamp < self.DHTtstamp: #r = random.randrange(2**256) n = self.config.routing.knownExternalNodes() if len(n) > 2: n1 = random.choice(n)[2] n2 = random.choice(n)[2] r = (n1+n2)/2 else: r = random.randrange(2**256) def badNode(error): node.logger.warn("Couldn't ping %s:%s" % (sys.argv[1], sys.argv[2])) d = self.client.kFindNode(r) d.addErrback(badNode) pingtime = random.randrange(PINGTIME/2, PINGTIME) reactor.callLater(pingtime, self.pingRandom, time.time()) def syncConfig(self): self.config.save() reactor.callLater(SYNCTIME, self.syncConfig) def start(self, twistd=False): """ starts the reactor in this thread """ self.webserver = FludServer(self, self.config.port) self.logger.log(logging.INFO, "FludServer starting") reactor.callLater(1, self.pingRandom, time.time()) reactor.callLater(random.randrange(10), self.syncConfig) if not twistd: reactor.run() def run(self): """ starts the reactor in its own thread """ #signal.signal(signal.SIGINT, self.sighandler) signal.signal(signal.SIGINT, signal.SIG_DFL) signal.signal(signal.SIGTERM, signal.SIG_DFL) self.webserver = FludServer(self, self.config.port) self.webserver.start() # XXX: need to do save out current config every X seconds # XXX: need to seperate known_nodes from config, and then update this # every X seconds. only update config when it changes. def stop(self): self.logger.log(logging.INFO, "shutting down FludNode") self.webserver.stop() def join(self): self.webserver.join() def sighandler(self, sig, frame): self.logger.log(logging.INFO, "handling signal %s" % sig) def connectViaGateway(self, host, port): def refresh(knodes): def refreshDone(results): self.logger.info("bucket refreshes finished: %s" % results) print "flud node connected and listening on port %d"\ % self.config.port #print "found knodes %s" % knodes dlist = [] for bucket in self.config.routing.kBuckets: #if True: if bucket.begin <= self.config.routing.node[2] < bucket.end: pass #print "passed on bucket %x-%s" % (bucket.begin, bucket.end) else: refreshID = random.randrange(bucket.begin, bucket.end) #print "refreshing bucket %x-%x by finding %x" \ # % (bucket.begin, bucket.end, refreshID) self.logger.info("refreshing bucket %x-%x by finding %x" % (bucket.begin, bucket.end, refreshID)) deferred = self.client.kFindNode(refreshID) dlist.append(deferred) dl = defer.DeferredList(dlist) dl.addCallback(refreshDone) # XXX: do we need to ping newly discovered known nodes? If not, # we could be vulnerable to a poisoning attack (at first # glance, this attack seems rather impotent...) # XXX: need to call refresh about every 60 minutes. Add a # reactor.callLater here to do it. def badGW(error): self.logger.warn(error) self.logger.warn("Couldn't connect to gateway at %s:%s" % (sys.argv[1], sys.argv[2])) self.logger.debug("connectViaGateway %s%d" % (host, port)) deferred = self.client.sendkFindNode(host, port, self.config.routing.node[2]) deferred.addCallback(refresh) deferred.addErrback(badGW) def getPath(): # this is a hack to be able to get the location of FludNode.tac return os.path.dirname(os.path.abspath(__file__))
Python
#!/usr/bin/python """ FludScheduler.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), verison 3. FludScheduler is the process monitors files for changes, and then tells flud to back them up. """ import sys, os, time, stat from twisted.internet import reactor from flud.FludConfig import FludConfig from flud.protocol.LocalClient import * from flud.CheckboxState import CheckboxState CHECKTIME=5 class FludScheduler: def __init__(self, config, factory): self.config = config self.factory = factory self.fileconfigfile = None self.fileconfigfileMTime = 0 self.fileChangeTime = 0 self.fileconfigSelected = set() self.fileconfigExcluded = set() self.getMasterMetadata() def getMasterMetadata(self): d = self.factory.sendLIST() d.addCallback(self.gotMasterMetadata) d.addErrback(self.errMasterMetadata) return d def gotMasterMetadata(self, master): self.mastermetadata = master def errMasterMetadata(self, err): print err reactor.stop() def readFileConfig(self, mtime=None): print "reading FileConfig" file = open(self.fileconfigfile, 'r') self.fileconfig = eval(file.read()) file.close() if mtime: self.fileconfigfileMTime = mtime else: self.fileconfigfileMTime = os.stat( self.fileconfigfile)[stat.ST_MTIME] self.fileconfigSelected = set([f for f in self.fileconfig if self.fileconfig[f] == CheckboxState.SELECTED or self.fileconfig[f] == CheckboxState.SELECTEDCHILD]) self.fileconfigExcluded = set([f for f in self.fileconfig if self.fileconfig[f] == CheckboxState.EXCLUDED or self.fileconfig[f] == CheckboxState.EXCLUDEDCHILD]) # The file[s]ChangeStat are the worst possible way to detect file changes. # Much more efficient to use inotify/dnotify/fam/gamin/etc., as well as # more correct (no way to detect cp -a or -p, for example, with stat). # But, these are a fallback method when those aren't present, and are fine # for testing. def fileChangedStat(self, file, fileChangeTime=None): if os.path.isfile(file) or os.path.isdir(file): mtime = os.stat(file)[stat.ST_MTIME] if not fileChangeTime: fileChangeTime = self.fileChangeTime if file in self.mastermetadata: fileChangeTime = self.mastermetadata[file][1] else: return True print "mtime = %s, ctime = %s (%s)" % (mtime, fileChangeTime, file) if mtime > fileChangeTime: return True return False def filesChangedStat(self, files, fileChangeTime=None): result = [] for f in files: if self.fileChangedStat(f, fileChangeTime): result.append(f) return result # Change these to point to something other than the xxxStat() methods def fileChanged(self, file, fileChangeTime=None): """ >>> now = time.time() >>> f1 = tmpfile.mktemp() >>> """ return self.fileChangedStat(file, fileChangeTime) def filesChanged(self, files, fileChangeTime=None): return self.filesChangedStat(files, fileChangeTime) def checkFileConfig(self): # check config file to see if it has changed, then reparse it if not self.fileconfigfile: # first time through print "checking fileconfigfile (initial)" if os.environ.has_key('FLUDHOME'): fludhome = os.environ['FLUDHOME'] elif os.environ.has_key('HOME'): fludhome = os.environ['HOME']+"/.flud" else: fludhome = ".flud" # XXX: fludfile.conf should be in config self.fileconfigfile = os.path.join(fludhome, "fludfile.conf") if os.path.isfile(self.fileconfigfile): self.readFileConfig() return True else: print "no fileconfigfile to read" elif os.path.isfile(self.fileconfigfile): if self.fileChanged(self.fileconfigfile, self.fileconfigfileMTime): print "fileconfigfile changed" mtime = time.time() self.readFileConfig(mtime) return True return False def checkFilesystem(self): checkedFiles = set() changedFiles = set() def checkList(list): #print "checkList: %s" % list #print "checkedFiles: %s" % checkedFiles for entry in list: # XXX: if entry is in master metadata, and its mtime is not # earlier than the time used by fileChanged, skip it (add 'and' # clause) if entry not in checkedFiles and \ entry not in self.fileconfigExcluded and \ entry not in self.mastermetadata: print "checkFilesystem for %s" % entry if os.path.isdir(entry): #print "dir %s" % entry dirfiles = [os.path.join(entry, i) for i in os.listdir(entry)] checkedFiles.update([entry,]) checkList(dirfiles) elif self.fileChanged(entry): print "%s changed" % entry if os.path.isfile(entry): changedFiles.update([entry,]) #print "file %s changed" % entry else: print "entry ?? %s ?? changed" % entry checkedFiles.update([entry,]) checkList(self.fileconfigSelected) self.fileChangeTime = time.time() return changedFiles def storefileFailed(self, err, file): print "storing %s failed: %s" % (file, err) err.printTraceback() #print dir(err) def storefileYay(self, r, file): print "storing %s success" % file def storeFiles(self, changedFiles): #print "storing %s" % changedFiles dlist = [] for f in changedFiles: print "storing %s" % f deferred = self.factory.sendPUTF(f) deferred.addCallback(self.storefileYay, f) deferred.addErrback(self.storefileFailed, f) dlist.append(deferred) dl = defer.DeferredList(dlist) return dl #return defer.succeed(True) def restartCheckTimer(self, v): print "restarting timer (%d) to call run()" % CHECKTIME reactor.callLater(CHECKTIME, self.run) def updateMasterMetadata(self, v): return self.getMasterMetadata() def run(self): print "run" self.checkFileConfig() changedFiles = self.checkFilesystem() print "%s changed" % changedFiles d = self.storeFiles(changedFiles) d.addBoth(self.updateMasterMetadata) d.addBoth(self.restartCheckTimer)
Python
#!/usr/bin/python """ FludLocalClient.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. FludLocalClient provides a command-line client for interacting with FludNode. """ import sys, os, time from twisted.internet import reactor from flud.FludConfig import FludConfig from flud.fencode import fencode, fdecode from flud.FludCrypto import hashfile from protocol.LocalClient import * logger = logging.getLogger('flud') class CmdClientFactory(LocalClientFactory): def __init__(self, config): LocalClientFactory.__init__(self, config) self.quit = False self.msgs = [] def callFactory(self, func, commands, msgs): # since we can't call factory methods from the promptUser thread, we # use this as a convenience to put those calls back in the event loop reactor.callFromThread(self.doFactoryMethod, func, commands, msgs) def doFactoryMethod(self, func, commands, msgs): d = func() d.addCallback(self.queueResult, msgs, '%s succeeded' % commands) d.addErrback(self.queueError, msgs, '%s failed' % commands) return d def promptUser(self): helpDict = {} command = raw_input("%s> " % time.ctime()) commands = command.split(' ') # XXX: should tokenize on any whitespace commandkey = commands[0][:4] # core client operations helpDict['exit'] = "exit from the client" helpDict['help'] = "display this help message" helpDict['ping'] = "send a GETID() message: 'ping host port'" helpDict['putf'] = "store a file: 'putf canonicalfilepath'" helpDict['getf'] = "retrieve a file: 'getf canonicalfilepath'" helpDict['geti'] = "retrieve a file by CAS key: 'geti fencodedCASkey'" helpDict['fndn'] = "send a FINDNODE() message: 'fndn hexIDstring'" helpDict['list'] = "list stored files (read from local metadata)" helpDict['putm'] = "store master metadata" helpDict['getm'] = "retrieve master metadata" helpDict['cred'] = "send encrypted private credentials: cred"\ " passphrase emailaddress" helpDict['node'] = "list known nodes" helpDict['buck'] = "print k buckets" helpDict['stat'] = "show pending actions" helpDict['stor'] = "store a block to a given node:"\ " 'stor host:port,fname'" helpDict['rtrv'] = "retrieve a block from a given node:"\ " 'rtrv host:port,fname'" helpDict['vrfy'] = "verify a block on a given node:"\ " 'vrfy host:port:offset-length,fname'" helpDict['fndv'] = "retrieve a value from the DHT: 'fndv hexkey'" helpDict['dlet'] = "delete from the stor: '[XXX]'" if commandkey == 'exit' or commandkey == 'quit': self.quit = True elif commandkey == 'help': self.printHelp(helpDict) elif commandkey == 'ping': # ping a host # format: 'ping host port' func = lambda: self.sendPING(commands[1], commands[2]) self.callFactory(func, commands, self.msgs) elif commandkey == 'putf': # store a file # format: 'putf canonicalfilepath' func = lambda: self.sendPUTF(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'getf': # retrieve a file # format: 'getf canonicalfilepath' func = lambda: self.sendGETF(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'geti': # retrieve a file by CAS ID # format: 'geti fencoded_CAS_ID' func = lambda: self.sendGETI(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'fndn': # find a node (or the k-closest nodes) # format: 'fndn hexIDstring' func = lambda: self.sendFNDN(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'list': # list stored files self.callFactory(self.sendLIST, commands, self.msgs) elif commandkey == 'putm': # store master metadata self.callFactory(self.sendPUTM, commands, self.msgs) elif commandkey == 'getm': # retrieve master metadata self.callFactory(self.sendGETM, commands, self.msgs) elif commandkey == 'cred': # send encrypted private credentials to an email address # format: 'cred passphrase emailaddress' func = lambda: self.sendCRED( command[len(commands[0])+1:-len(commands[-1])-1], commands[-1]) self.callFactory(func, commands, self.msgs) # the following are diagnostic operations, debug-only utility elif commandkey == 'node': # list known nodes self.callFactory(self.sendDIAGNODE, commands, self.msgs) elif commandkey == 'buck': # show k-buckets self.callFactory(self.sendDIAGBKTS, commands, self.msgs) elif commandkey == 'stat': # show pending actions print self.pending elif commandkey == 'stor': # stor a block to a given node. format: 'stor host:port,fname' storcommands = commands[1].split(',') try: fileid = int(storcommands[1], 16) except: linkfile = fencode(long(hashfile(storcommands[1]),16)) if (os.path.islink(linkfile)): os.remove(linkfile) os.symlink(storcommands[1], linkfile) storcommands[1] = linkfile # XXX: delete this file when the command finishes commands[1] = "%s,%s" % (storcommands[0], storcommands[1]) func = lambda: self.sendDIAGSTOR(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'rtrv': # retrive a block from a given node. format: 'rtrv host:port,fname' func = lambda: self.sendDIAGRTRV(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'vrfy': # verify a block on a given node. # format: 'vrfy host:port:offset-length,fname' logger.debug("vrfy(%s)" % commands[1]) func = lambda: self.sendDIAGVRFY(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'dlet': print "not yet implemented" elif commandkey == 'fndv': # try to retrieve a value from the DHT # format: 'fndv key' func = lambda: self.sendDIAGFNDV(commands[1]) self.callFactory(func, commands, self.msgs) elif command != "": reactor.callFromThread(self.queueError, None, self.msgs, "illegal command '%s'" % command) def queueResult(self, r, l, msg): logger.debug("got result %s" % msg) l.append((r, msg)) def queueError(self, r, l, msg): logger.debug("got error %s" % msg) if r: l.append((r.getErrorMessage(), msg)) else: l.append((None, msg)) def printHelp(self, helpDict): helpkeys = helpDict.keys() helpkeys.sort() for i in helpkeys: print "%s:\t %s" % (i, helpDict[i]) def promptLoop(self, r): for c in self.pending: for i in self.pending[c].keys(): if self.pending[c][i] == True: print "%s on %s completed successfully" % (c, i) self.pending[c].pop(i) elif self.pending[c][i] == False: print "%s on %s failed" % (c, i) self.pending[c].pop(i) else: print "%s on %s pending" % (c, i) while len(self.msgs) > 0: # this prints in reverse order, perhaps pop() all into a new list, # reverse, then print (errmsg, m) = self.msgs.pop() if errmsg: print "<- %s:\n%s" % (m, errmsg) else: print "<- %s" % m if self.quit: reactor.stop() else: d = threads.deferToThread(self.promptUser) d.addCallback(self.promptLoopDelayed) d.addErrback(self.err) def promptLoopDelayed(self, r): # give the reactor loop time to fire any quick cbs/ebs reactor.callLater(0.1, self.promptLoop, r) def clientConnectionLost(self, connector, reason): if not self.quit: LocalClientFactory.clientConnectionLost(self, connector, reason) def cleanup(self, msg): self.quit = True self.err(msg) def err(self, r): print "bah!: %s" % r reactor.stop() def main(): config = FludConfig() config.load(doLogging=False) logger.setLevel(logging.DEBUG) handler = logging.FileHandler('/tmp/fludclient.log') formatter = logging.Formatter('%(asctime)s %(filename)s:%(lineno)d' ' %(name)s %(levelname)s: %(message)s', datefmt='%H:%M:%S') handler.setFormatter(formatter) logger.addHandler(handler) factory = CmdClientFactory(config) if len(sys.argv) == 2: config.clientport = int(sys.argv[1]) print "connecting to localhost:%d" % config.clientport reactor.connectTCP('localhost', config.clientport, factory) factory.promptLoop(None) reactor.run() if __name__ == '__main__': main()
Python
#!/usr/bin/python """ FludClient.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. FludClient provides a GUI Client for interacting with FludNode. """ #from twisted.internet import wxreactor #wxreactor.install() import sys, os, string, time, glob import wx import wx.lib.mixins.listctrl as listmix import wx.lib.editor.editor from flud.protocol.LocalClient import * from flud.FludConfig import FludConfig from flud.CheckboxState import CheckboxState FLUSHCHECKTIME = 5*60 # s to wait to flush fludfile.conf imgdir = os.path.join(os.path.dirname(os.path.abspath(__file__)),'images') mimeMgr = wx.MimeTypesManager() def getFileIcon(file, il, checkboxes, icondict): ft = mimeMgr.GetFileTypeFromExtension(file[file.rfind('.')+1:]) # XXX: what about from mimetype or magic? if ft == None: return icondict['generic'] else: desc = ft.GetDescription() if icondict.has_key(desc): return icondict[desc] else: icon = ft.GetIcon() if icon == None or not icon.Ok(): #print "couldn't find an icon image for %s" % file icondict[desc] = icondict['generic'] return icondict[desc] bm = wx.BitmapFromIcon(icon) newimages = makeCheckboxBitmaps(bm, checkboxes) #il = self.GetImageList() pos = il.GetImageCount() for i in newimages: il.Add(i) icondict[desc] = pos #print "%s got a %s image" % (file, ft.GetDescription()) return pos def getEmptyBitmapAndDC(width, height): empty = wx.EmptyBitmap(width,height) temp_dc = wx.MemoryDC() temp_dc.SelectObject(empty) temp_dc.Clear() return (empty, temp_dc) def makeCheckboxBitmaps(basebitmap, checkboxes): if basebitmap.GetWidth() != 16 or basebitmap.GetHeight() != 16: img = basebitmap.ConvertToImage() img.Rescale(16, 16) basebitmap = img.ConvertToBitmap() result = [] for i in checkboxes: bm, dc = getEmptyBitmapAndDC(40,16) dc.DrawBitmap(basebitmap, 0, 0, False) dc.DrawBitmap(i, 20, 2, False) result.append(bm) return result def createDefaultImageList(): def getDefaultCheckboxes(): ucbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-unchecked1.png"))) cbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checked1.png"))) ccbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checkedpartial1.png"))) cpbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-parentchecked1.png"))) ebm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excluded1.png"))) ecbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excludedpartial1.png"))) return (ucbm, cbm, ccbm, cpbm, ebm, ecbm) checkboxes = getDefaultCheckboxes() il = wx.ImageList(40, 16) folderimgs = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_FOLDER, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) computer = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_HARDDISK, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) drives = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_HARDDISK, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) cdrom = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_CDROM, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) floppy = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_FLOPPY, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) removable = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_REMOVABLE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) genericfile = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_NORMAL_FILE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) execfile = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_EXECUTABLE_FILE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) j = 0 icondict = {} icondict['folder'] = j for i in folderimgs: il.Add(i) j = j+1 icondict['computer'] = j for i in computer: il.Add(i) j = j+1 icondict['drives'] = j for i in drives: il.Add(i) j = j+1 icondict['cdrom'] = j for i in cdrom: il.Add(i) j = j+1 icondict['floppy'] = j for i in floppy: il.Add(i) j = j+1 icondict['removable'] = j for i in removable: il.Add(i) j = j+1 icondict['generic'] = j for i in genericfile: il.Add(i) j = j+1 icondict['exec'] = j for i in execfile: il.Add(i) j = j+1 return il, checkboxes, icondict class DirCheckboxCtrl(wx.TreeCtrl): def __init__(self, parent, id=-1, dir=None, pos=wx.DefaultPosition, size=wx.DefaultSize, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER), validator=wx.DefaultValidator, name=wx.ControlNameStr, allowExclude=True): self.allowExclude = allowExclude wx.TreeCtrl.__init__(self, parent, id, pos, size, style, validator, name) self.listeners = [] self.parent = parent #self.il = self.GetImageList() #self.checkboxes = self.getDefaultCheckboxes() self.initTree(dir) self.Bind(wx.EVT_TREE_ITEM_EXPANDING, self.onExpand, self) self.Bind(wx.EVT_LEFT_UP, self.onClick, self) self.Bind(wx.EVT_TREE_ITEM_GETTOOLTIP, self.onTooltip, self) self.Bind(wx.EVT_CHAR, self.onChar) def initTree(self, dir): self.expandRoot(dir) # XXX: should expandHome() on first run, then load expanded dir state # from saved state on subsequent runs. self.expandHome(dir) def expandRoot(self, dir): if not os.path.isdir(dir): raise ValueError("%s is not a valid directory path") self.defaultImageList, self.checkboxes, self.icondict \ = createDefaultImageList() self.AssignImageList(self.defaultImageList) self.il = self.GetImageList() if dir == None: self.rootID = self.AddRoot(dir, self.icondict['computer'], -1, wx.TreeItemData((dir, True, False, CheckboxState.UNSELECTED))) # XXX: getTopLevelDirs() and add them as children else: self.rootID = self.AddRoot(dir, self.icondict['folder'], -1, wx.TreeItemData((dir, True, False, CheckboxState.UNSELECTED))) self.expandDir(self.rootID) self.Expand(self.rootID) self.stateChangeTime = time.time() self.flushTime = time.time() reactor.callLater(FLUSHCHECKTIME, self.checkFlush) def expandHome(self, dir): home = os.environ['HOME'] if home: traversal = home.split(os.path.sep)[1:] node = self.rootID for d in traversal: (ipath, isdir, expanded, istate) \ = self.GetItemData(node).GetData() self.expandDir(node) children = self.getChildren(node, False) childrennames = [self.GetItemText(x) for x in children] if d in childrennames: p = childrennames.index(d) node = children[p] self.expandDir(node) self.Expand(node) else: print "couldn't traverse to HOME dir on %s" % d break def checkFlush(self): print "checking for flush" if self.stateChangeTime > self.flushTime: self.flushTime = time.time() print "flushing" self.parent.flushFileConfig() reactor.callLater(FLUSHCHECKTIME, self.checkFlush) def expandDir(self, parentID, hideHidden=False, busycursor=True): def isDriveAvailable(path): if len(path) == 2 and path[1] == ':': path = path.lower() if path[0] == 'a' or path[0] == 'b' or diExists(path): return True else: return False return True (path, isDir, expanded, state) = self.GetItemData(parentID).GetData() if expanded: return if not isDriveAvailable(path): return if busycursor: wx.BusyCursor() try: dirlist = os.listdir(path) except: self.SetItemHasChildren(parentID, False) return if len(dirlist) == 0: self.SetItemHasChildren(parentID, False) return dirs = [] files = [] for i in dirlist: if hideHidden and i[0] == '.': # XXX: dotfile format check is *nix specific # XXX: if this is a hidden file, don't add it. pass elif os.path.isdir(os.path.join(path,i)): dirs.append(i) else: files.append(i) dirs.sort() files.sort() for d in dirs: child = self.AppendItem(parentID, d) self.SetPyData(child, (os.path.join(path,d), True, False, 0)) self.SetItemImage(child, self.icondict['folder'], wx.TreeItemIcon_Normal) self.SetItemHasChildren(child) il = self.GetImageList() for f in files: child = self.AppendItem(parentID, f) # XXX: unicode? self.SetPyData(child, (os.path.join(path,f), False, False, 0)) idx = getFileIcon(os.path.join(path,f), il, self.checkboxes, self.icondict) self.SetItemImage(child, idx, wx.TreeItemIcon_Normal) self.SetPyData(parentID, (path, isDir, True, state)) def getStates(self, node=None): if not node: node = self.rootID states = {} (path, isDir, expanded, state) = self.GetItemData(node).GetData() if state in [CheckboxState.SELECTED, CheckboxState.EXCLUDED]: states[path] = state children = self.getChildren(node, False) for child in children: states.update(self.getStates(child)) return states def setStates(self, states): for i in states: found = self.findNode(i) if found: self.setItemState(found, states[i]) def findNode(self, path): if path[0] == '/': path = path[1:] # XXX: unix only traversal = path.split(os.path.sep) if traversal[0] == '': traversal.remove('') node = self.rootID while True: (ipath, isdir, expanded, istate) = self.GetItemData(node).GetData() if len(traversal) == 0: return node self.expandDir(node) children = self.getChildren(node, False) childrennames = [self.GetItemText(x) for x in children] firstpath = traversal[0] if firstpath in childrennames: p = childrennames.index(firstpath) node = children[p] traversal.remove(firstpath) else: #print " the file %s is no longer present!" % path return None return None def onExpand(self, event): self.expandDir(event.GetItem()) self.renderChildren(event.GetItem(), True) def getFullPath(self, node): path = self.tree.GetItemText(node) n = node while True: n = self.tree.GetItemParent(n) if n and n != self.GetRootItem(): path = os.path.join(self.tree.GetItemText(n),path) else: break return path def renderParents(self, item): if item == self.rootID: return n = item (path, isDir, expanded, state) = self.GetItemData(item).GetData() while True: n = self.GetItemParent(n) (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(n).GetData() #print "parent %s" % parentpath if n and n != self.GetRootItem(): newstate = parentstate if parentstate != CheckboxState.UNSELECTED and \ parentstate != CheckboxState.SELECTEDPARENT: # we only care about changing UNSELECT or SELECTEDPARENT # states break else: if state == CheckboxState.SELECTED or \ state == CheckboxState.SELECTEDCHILD or \ state == CheckboxState.SELECTEDPARENT: # if the item (child) is selected in any way, parent # should be too. newstate = CheckboxState.SELECTEDPARENT elif state == CheckboxState.UNSELECTED or \ state == CheckboxState.EXCLUDED: # if the item (child) is unselected or excluded, the # parent should be too, /unless/ there are other # children at the same level who are selected. children = self.getChildren(n, False) newstate = CheckboxState.UNSELECTED for child in children: (cpath, cisdir, cexp, cstate) = self.GetItemData(child).GetData() if cstate == CheckboxState.SELECTED or \ cstate == CheckboxState.SELECTEDCHILD or \ cstate == CheckboxState.SELECTEDPARENT: newstate = parentstate if newstate == parentstate: break imageidx = self.GetItemImage(n) imageidx += CheckboxState.offset(parentstate, newstate) self.SetPyData(n, (parentpath, parentisDir, parentexpanded, newstate)) self.SetItemImage(n, imageidx) else: break def renderChildren(self, parent, recurse=False): (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(parent).GetData() children = self.getChildren(parent, False) for child in children: #path = self.getFullPath(child) (path, isDir, expanded, state) = self.GetItemData(child).GetData() imageidx = self.GetItemImage(child) newstate = state """ Here are the state transitions for children based on current states: ('-' = no state change, 'x' = should never occur, '!' = should be prevented at the parent, '?' = need to consult children) child unsel sel selch selpar excl exclch unsel - ! unsel x - unsel sel selch - - selch - selch par selch selch - - selch - selch selpar x x unsl?selpr x x x excl exlch ! exlch ! - - exclch exlch - exlch ! - - """ #if parentpath == '/data': # print "/data pstate = %d" % parentstate # print " %s = %d" % (path, state) if state == CheckboxState.UNSELECTED: if parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.SELECTEDCHILD: if parentstate == CheckboxState.UNSELECTED: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTEDPARENT: if self.checkChildrenStates(child, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT]): # XXX: did we need to pass in selections to checkChldSt newstate = CheckboxState.SELECTEDPARENT else: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.SELECTEDPARENT: if parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif state == CheckboxState.EXCLUDEDCHILD: if parentstate == CheckboxState.UNSELECTED: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD imageidx += CheckboxState.offset(state, newstate) self.SetPyData(child, (path, isDir, expanded, newstate)) self.SetItemImage(child, imageidx) if recurse: self.renderChildren(child, recurse) # XXX: why do we renderParents here? It hits the same # 'parent's over and over and over again. If we want to do # this, we need to 'collect up' the parents and just call once # -- this kills performance. #print "renderParents(%s)" % path #self.renderParents(child) def getChildren(self, node, recurse=False): result = [] child, cookie = self.GetFirstChild(node) while child: result.append(child) if recurse: result.extend(self.getChildren(child, recurse)) child, cookie = self.GetNextChild(node, cookie) return result def checkChildrenStates(self, node, states, ignorelist=[]): children = self.getChildren(node) for child in children: if child not in ignorelist: (p, d, e, childstate) = self.GetItemData(child).GetData() for state in states: if state == childstate: #print "%s has state %d" % (p, state) return True if self.checkChildrenStates(child, states, ignorelist): # do this even if it is in ignorelist, because it may have # children which are not in the ignorelist return True return False def getTooltip(self, item): text = self.GetItemText(item) (path, isDir, expanded, state) = self.GetItemData(item).GetData() if state == CheckboxState.SELECTED: if isDir: text = "'%s' is SELECTED for backup\n" \ "ALL files within this folder will be backed up\n" \ "(except those explicitly marked for exclusion)" % text else: text = "'%s' is SELECTED for backup" % text elif state == CheckboxState.UNSELECTED: text = "'%s' is NOT selected for backup" % text elif state == CheckboxState.SELECTEDPARENT: text = "some files within '%s' are selected for backup" % text elif state == CheckboxState.SELECTEDCHILD: text = "'%s' will be backed up\n" \ "(one of its parent folders is selected)" % text elif state == CheckboxState.EXCLUDED: if isDir: text = "'%s' is EXCLUDED from backup\n" \ "No files within this folder will be backed up" % text else: text = "'%s' is EXCLUDED from backup" % text elif state == CheckboxState.EXCLUDEDCHILD: text = "'%s' is EXCLUDED from backup\n" \ "(one of its parent folders is EXCLUDED)" % text return text def onTooltip(self, event): item = event.GetItem() text = self.getTooltip(item) if text: event.SetToolTip(text) else: event.StopPropagation() #print dir(event) def onClick(self, event): point = (event.GetX(), event.GetY()) item, flags = self.HitTest(point) if flags & wx.TREE_HITTEST_ONITEMICON: selections = self.GetSelections() self.changeState(item, selections) def onChar(self, event): if event.KeyCode() == ord('F') and event.ShiftDown() \ and event.ControlDown(): self.flushTime = time.time() print "flushing" self.parent.flushFileConfig() event.Skip() def changeState(self, item, selections=[]): self.stateChangeTime = time.time() (path, isDir, expanded, state) = self.GetItemData(item).GetData() if item == self.rootID: parent = None parentstate = CheckboxState.UNSELECTED else: parent = self.GetItemParent(item) (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(parent).GetData() imageidx = self.GetItemImage(item) # determine newstate from existing state, parent state, and state # of children """ Here are the state transitions for the item based on current states and parent states: ('-' = no state change, 'x' = should never occur, '?' = depends on children state) item unsel sel selch selpar excl exclch unsel sel excl sel sel unsel excl sel sel excl?selpar sel x selch excl par selch x excl sel sel selch excl selpar sel excl x sel unsel excl excl x excl x exclch exclch excl exclch x excl x exclch exclch excl """ newstate = state if state == CheckboxState.UNSELECTED: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTEDPARENT: if parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: # XXX: this should be impossible to reach... newstate = CheckboxState.EXCLUDEDCHILD else: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTED: if self.checkChildrenStates(item, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT], selections): newstate = CheckboxState.SELECTEDPARENT elif self.allowExclude: newstate = CheckboxState.EXCLUDED else: if parent in selections or \ (parentstate == CheckboxState.UNSELECTED or \ parentstate == CheckboxState.SELECTEDPARENT): newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif state == CheckboxState.EXCLUDED: if parent in selections or \ (parentstate == CheckboxState.UNSELECTED or \ parentstate == CheckboxState.SELECTEDPARENT): newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD else: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDED if len(selections) > 1: # if we have multiple selections, the idea is to move all the # selections to the newstate defined above, or to valid # unselected or inherited states if the move to newstate would # be invalid. """ Here are the state transitions for the item based on the newstate as determined by the clicked item and the current states: ('-' = no state change, '?' = consult children) item unsel sel selch selpar excl exclch unsel - unsel - - unsel - sel sel - sel sel sel - newstate selch - unsel - - unsel - selpar - unsel - - unsel - excl excl excl?slpr excl excl - excl exclch - unsel - - unsel - """ for i in selections: (mpath, misDir, mexpanded, mstate) = self.GetItemData( i).GetData() mnewstate = mstate if mstate == CheckboxState.UNSELECTED or \ mstate == CheckboxState.SELECTEDCHILD or \ mstate == CheckboxState.SELECTEDPARENT: if newstate == CheckboxState.SELECTED or \ newstate == CheckboxState.EXCLUDED: mnewstate = newstate elif mstate == CheckboxState.SELECTED: if newstate == CheckboxState.UNSELECTED or \ newstate == CheckboxState.SELECTEDCHILD or \ newstate == CheckboxState.SELECTEDPARENT or \ newstate == CheckboxState.EXCLUDEDCHILD: mnewstate = CheckboxState.UNSELECTED elif newstate == CheckboxState.EXCLUDED: if self.checkChildrenStates(i, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT], selections): mnewstate = CheckboxState.SELECTEDPARENT else: mnewstate = newstate elif mstate == CheckboxState.EXCLUDED: if newstate == CheckboxState.UNSELECTED or \ newstate == CheckboxState.SELECTEDCHILD or \ newstate == CheckboxState.SELECTEDPARENT or \ newstate == CheckboxState.EXCLUDEDCHILD: mnewstate = CheckboxState.UNSELECTED elif newstate == CheckboxState.SELECTED: mnewstate = newstate elif mstate == CheckboxState.EXCLUDEDCHILD: if newstate == CheckboxState.EXCLUDED: mnewstate = newstate self.setItemState(i, mnewstate) self.setItemState(item, newstate, (path, isDir, expanded, state, imageidx)) def setItemState(self, item, newstate, oldData=None): if oldData: (path, isDir, expanded, state, imageidx) = oldData else: (path, isDir, expanded, state) = self.GetItemData(item).GetData() imageidx = self.GetItemImage(item) imageidx += CheckboxState.offset(state, newstate) self.SetPyData(item, (path, isDir, expanded, newstate)) self.SetItemImage(item, imageidx) self.renderChildren(item, True) self.renderParents(item) def getTopLevelDrives(self): sys = platform.system() if sys == 'Windows': # XXX: need to test this all out import win32api, string drives = win32api.GetLogicalDriveStrings() driveletters = string.splitfields(drives,'\000') for d in driveletters: type = win32api.GetDriveType("%s:\\" % d) # XXX: set the appropriate icon return driveletters else: # Unix, OSX, etc. return ['/'] def addListener(self, callback): self.listeners.append(callback) def SetPyData(self, item, data): wx.TreeCtrl.SetPyData(self, item, data) for f in self.listeners: f(item, data) """ Tests for DirCheckboxCtrl A number of unit tests must be performed on the DirCheckboxGUI widget when refactoring. Add to this list so that it becomes comprehensive. Basic Tests: 1. Click on a top-level UNSELECTED object in the tree [should become SELECTED]. - Click again [should become EXCLUDED]. - Click again [should become UNSELECTED]. 2. Click on a non-top-level UNSELECTED object in the tree that has no SELECTED children [should become SELECTED, it's parents should become SELECTEDPARENT and its children SELECTEDCHILD]. - Click again [should become EXCLUDED, it's parents who were SELECTEDPARENT should become UNSELECTED, and it's UNSELECTED children should become EXCLUDED]. - Click again [should become UNSELECTED, and it's children should become UNSELECTED]. 3. Change two children to their SELECTED state [parents should be in SELECTEDPARENT state]. - Click one child to become EXCLUDED [parents should stay in SELECTEDPARENT] - Click the same child to become UNSELECTED [parents should stay in SELECTEDPARENT] - Click the other child to become EXCLUDED [parents should become UNSELECTED] 4. Choose a folder and a child item. - Click the child to become SEL [parent should be SELPAR] - Click the parent [parent should become SEL] - Click the parent again [parent should become SELPAR] 5. Choose a folder and a child item. - Click the parent to become SEL [child should become SELCHILD] - Click the child [child should become SEL] - Click the child again [child should become EXCL] - Click the child again [child should become SELCHILD] 6. Pick a node with children at least two-deep. Change two of the at-least-two-deep children to their SELECTED state [parents should be in SELECTEDPARENT state]. - Click parent closest to SELECTED children to SELECTED [two childen remain in SELECTED, all other children become SELECTEDCHILD. Parent[s] of parent remain SELECTEDPARENT] - Click one child twice to become SELECTEDCHILD [child should not be able to be UNSELECTED, parent states should not change] - Click other child twice to become SELECTEDCHILD [child should not be able to be UNSELECTED, parent states should not change] 7. Pick a node with children at least two-deep. - Click deepest parent to SELECTED [Parent[s] of parent become SELECTEDPARENT] - Click same parent again to become EXCLUDED [Parent[s] of parent become UNSELECTED] - Click same parent again to become UNSELECTED [Parent[s] of parent remain UNSELECTED] 8. Pick a node with children at least two-deep. - Click deepest child to become SELECTED [Parent[s] of parent become SELECTEDPARENT] - Click the topmost parent to become SELECTED [children become SELECTEDCHILD] - Click the topmost parent again to become SELECTEDPARENT [middle child should become SELECTEDPARENT] Multi-Selection Tests: 1. Multi-select three items at the same level and in the same state. Toggle between the three main states [SELECTED, EXCLUDED, UNSELECTED] 2. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the SEL item to see that all three items become EXCL. 3. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the EXCL item to see that all three items become UNSEL. 4. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the UNSEL item to see that all three items become SEL. 5. Choose three items that are nested within each other: a parent folder, one of its children folders, and a file/folder in the child folder. Choose one other item from the child folder. - set the top parent to UNSEL - set the child folder to SEL [parent become SELPAR] - set the child item to SEL - set the other item to EXCL - multi-select all four items - 5A. click on the top parent (which was in SELPAR) [All four items should become SEL, all children of any of these items should become SELCHILD]. Toggle twice more [all selected items should toggle to EXCL, then to UNSEL] - 5B. reset as above, click on the child folder [All four items should become EXCL]. Toggle twice more [all selected items should go to UNSEL, then SEL] - 5C. reset as above, click on the child item [All four items should become EXCL]. Toggle twice more [all selected items should go to UNSEL, then SEL] - 5D. reset as above, click on the other item [All four items should become UNSEL]. Toggle twice more [all selected items should go to SEL, then EXCL] 6. Choose a folder, one if its subfolders, a subfolder of the subfolder, and an item in the deepest subfolder, and an item in the first subfolder, e.g.: [] A [] B [] C [] D [] E - change item 'D' to SEL [parents 'A', 'B', and 'C' should go to SELPAR] - change item 'E' to EXCL - multi-select 'A', 'C', and 'E' - toggle 'E' to UNSEL [all other selections should stay in current state] - toggle 'E' to SEL ['A' and 'B' become SEL, their children become SELCHILD] - toggle 'E' back to EXCL [should get our original multi-select setup back] - toggle 'C' to SEL [all selections to SEL, children to SELCHILD] - toggle 'C' to SELPAR ['A' and 'C' to SELPAR, 'E' to UNSEL] - toggle 'E' twice [should get our original mulit-select setup back] """ class CheckFileListCtrlMixin: # for some insane reason, we can't get EVT_LEFT_DOWN (or _UP) to bind in # FileListCtrl itself. But we are sneaky and can do it by lots of clever # hax0ry, like by using this silly mixin. def __init__(self, toCall): self.Bind(wx.EVT_LEFT_UP, toCall) class FileListCtrl(wx.ListCtrl, listmix.ListCtrlAutoWidthMixin, CheckFileListCtrlMixin): """ Implements a file list control, with a peerctrl that contains the filesystem model. Currently, this peerctrl must implement an addListener(), changeState(), GetItemData(), expandDir(), GetSelections(), and GetChildren() API similar to that implemented by DirCheckBoxCtrl. """ def __init__(self, parent, peerctrl, id=-1, pos=wx.DefaultPosition, size=wx.DefaultSize, style=wx.LC_REPORT, validator=wx.DefaultValidator, name=wx.ListCtrlNameStr): wx.ListCtrl.__init__(self, parent, id, pos, size, style, validator, name) CheckFileListCtrlMixin.__init__(self, self.OnClick) listmix.ListCtrlAutoWidthMixin.__init__(self) self.peerctrl = peerctrl self.peerctrl.addListener(self.itemChanged) self.itemdict = {} # a dict with filepath as key, containing tuples of # (index into ListCtrl, reference to peerctrl object) self.stopsearch = False self.il, self.checkboxes, self.icondict = createDefaultImageList() self.AssignImageList(self.il, wx.IMAGE_LIST_SMALL) self.il = self.GetImageList(wx.IMAGE_LIST_SMALL) self.InsertColumn(0, "Filename") self.InsertColumn(1, "Location") #self.InsertColumn(2, "Last Backup") #self.SetColumnWidth(0, wx.LIST_AUTOSIZE) #self.SetColumnWidth(1, -1) #wx.LIST_AUTOSIZE) self.Bind(wx.EVT_MOTION, self.mouseMotion) self.searchSourceItems = [] def itemChanged(self, item, data): (path, isDir, expanded, state) = data if self.itemdict.has_key(path): item = self.itemdict[path][0] image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) def GetAll(self, excludeStates=[]): result = [] start = -1 for i in range(self.GetItemCount()): item = self.GetNextItem(start, wx.LIST_NEXT_ALL) # XXX: only append if not in excludeStates result.append(item) start = item return result def GetSelections(self): result = [] start = -1 for i in range(self.GetSelectedItemCount()): item = self.GetNextItem(start, wx.LIST_NEXT_ALL, wx.LIST_STATE_SELECTED) result.append(item) start = item return result def GetPeerSelections(self, selections): result = [] for item in selections: path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) if self.itemdict.has_key(path): result.append(self.itemdict[path][1]) return result def mouseMotion(self, event): point = event.GetPosition() item, flags = self.HitTest(point) if flags == wx.LIST_HITTEST_ONITEMICON: path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) text = self.peerctrl.getTooltip(self.itemdict[path][1]) tip = wx.ToolTip(text) self.SetToolTip(tip) #tipwin = tip.GetWindow() #tippos = tipwin.GetPosition() #print "%s vs %s" % (tippos, point) #tipwin.SetPosition(point) def OnClick(self, event): point = event.GetPosition() item, flags = self.HitTest(point) if flags == wx.LIST_HITTEST_ONITEMICON: peerselections = self.GetPeerSelections(self.GetSelections()) path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) ditem = self.itemdict[path][1] # raises if not present self.peerctrl.changeState(ditem, peerselections) (path, isDir, expanded, state) \ = self.peerctrl.GetItemData(ditem).GetData() image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) def searchButtonAction(self, event): selections = self.peerctrl.GetSelections() if len(selections) == 0: return ("Please tell me where to search. Select one or more" " folders in the left-hand panel (hold down SHIFT or" " CTRL for multiple selection), then click the 'find!'" " button again.", None) else: self.DeleteAllItems() self.itemdict = {} b = wx.BusyCursor() searchSourceItems = [] for i in selections: self.addResults(i, event.searchstring) searchSourceItems.append(i) self.searchSourceItems = [self.peerctrl.GetItemData(s).GetData()[0] for s in searchSourceItems] print "sources: %s" % self.searchSourceItems return ("Search results will appear as files that match your" " search are found.", None) return (None, None) def addResults(self, ditem, searchstring): (path, isDir, expanded, state) \ = self.peerctrl.GetItemData(ditem).GetData() position = self.GetItemCount() if isDir: if not expanded: self.peerctrl.expandDir(ditem, busycursor=False) children = self.peerctrl.getChildren(ditem) for c in children: self.addResults(c, searchstring) wx.Yield() if self.stopsearch: break else: terms = [x for x in searchstring.split(' ') if x != ''] for term in terms: print path if path.find(term) > 0: image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state dirname, filename = os.path.split(path) index = self.InsertImageStringItem(position, filename, image) self.SetStringItem(index, 1, dirname) self.SetColumnWidth(0, wx.LIST_AUTOSIZE) self.itemdict[path] = (index, ditem) break def setGroup(self, state): items = self.GetAll() item = items[0] peerselections = self.GetPeerSelections(items) path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) ditem = self.itemdict[path][1] # raises if not present while True: # cycle until the items state matches the desired state self.peerctrl.changeState(ditem, peerselections) # can be slow (path, isDir, expanded, nstate) \ = self.peerctrl.GetItemData(ditem).GetData() if nstate == state: break image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) return self.searchSourceItems class GroupSelectionCheckbox(wx.Panel): def __init__(self, parent, id=-1, setGroupState=None): wx.Panel.__init__(self, parent, id) self.setGroupState = setGroupState self.ubm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-unchecked1.png"))) self.cbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checked1.png"))) self.ebm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excluded1.png"))) self.checkboxButton = wx.BitmapButton(self, -1, self.ubm, style=wx.NO_BORDER) self.Bind(wx.EVT_BUTTON, self.onCheckbox, self.checkboxButton) self.description = wx.StaticText(self, -1, "always BACKUP any files that match these search criteria ") self.state = CheckboxState.UNSELECTED self.gbSizer = wx.GridBagSizer(1,2) self.gbSizer.Add(self.checkboxButton, (0,0), flag=wx.ALIGN_CENTER) self.gbSizer.Add(self.description, (0,1), flag=wx.ALIGN_CENTER) self.gbSizer.AddGrowableRow(1) self.SetSizerAndFit(self.gbSizer) def Enable(self, enable=True): self.checkboxButton.Enable(enable) self.description.Enable(enable) def Disable(self): self.Enable(False) def clear(self): self.checkboxButton.SetBitmapLabel(self.ubm) self.state = CheckboxState.UNSELECTED self.description.SetLabel( "always BACKUP any files that match these search criteria") def setState(self, state): self.state = state if self.state == CheckboxState.UNSELECTED: self.checkboxButton.SetBitmapLabel(self.ubm) self.description.SetLabel( "always BACKUP any files that match these search criteria") elif self.state == CheckboxState.SELECTED: self.checkboxButton.SetBitmapLabel(self.cbm) self.description.SetLabel( "always BACKUP any files that match these search criteria") elif self.state == CheckboxState.EXCLUDED: self.checkboxButton.SetBitmapLabel(self.ebm) self.description.SetLabel( "always EXCLUDE any files that match these search criteria") def onCheckbox(self, event): if self.state == CheckboxState.UNSELECTED: self.checkboxButton.SetBitmapLabel(self.cbm) self.state = CheckboxState.SELECTED if self.setGroupState: self.setGroupState(CheckboxState.SELECTED) elif self.state == CheckboxState.SELECTED: self.checkboxButton.SetBitmapLabel(self.ebm) self.description.SetLabel( "always EXCLUDE any files that match these search criteria") self.state = CheckboxState.EXCLUDED if self.setGroupState: self.setGroupState(CheckboxState.EXCLUDED) elif self.state == CheckboxState.EXCLUDED: self.checkboxButton.SetBitmapLabel(self.ubm) self.description.SetLabel( "always BACKUP any files that match these search criteria") self.state = CheckboxState.UNSELECTED if self.setGroupState: self.setGroupState(CheckboxState.UNSELECTED) class SearchPanel(wx.Panel): def __init__(self, parent, dircheckbox, id=-1, searchButtonAction=None): wx.Panel.__init__(self, parent, id) self.dircheckbox = dircheckbox self.searchButtonAction = searchButtonAction self.SetAutoLayout(False) self.rules = {} # should refer to something from fludrules self.searchField = wx.TextCtrl(self, -1, "search for files to backup here", size=wx.Size(-1,-1), style=wx.TE_PROCESS_ENTER) self.searchField.SetToolTipString('find files within directories' ' selected to the left by entering search terms here') self.searchField.Bind(wx.EVT_TEXT_ENTER, self.onSearchClick) self.searchField.Bind(wx.EVT_LEFT_DOWN, self.selectAllText) self.searchField.Bind(wx.EVT_KILL_FOCUS, self.unfocused) self.searchButton = wx.Button(self, -1, 'find!', name='searchButton') self.Bind(wx.EVT_BUTTON, self.onSearchClick, self.searchButton) self.searchResults = FileListCtrl(self, dircheckbox, -1, name='searchResults', style=wx.SUNKEN_BORDER | wx.LC_REPORT) self.searchResults.SetExtraStyle(0) self.searchResults.SetLabel('found files') self.groupSelection = GroupSelectionCheckbox(self, -1, self.setGroup) self.groupSelection.Disable() self.gbSizer = wx.GridBagSizer(3,2) self.gbSizer.Add(self.searchField, (0,0), flag=wx.EXPAND) self.gbSizer.Add(self.searchButton, (0,1)) self.gbSizer.Add(self.searchResults, (1,0), (1,2), flag=wx.EXPAND|wx.TOP, border=5) self.gbSizer.Add(self.groupSelection, (2,0) ) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) def onSearchClick(self, event): event.searchstring = self.searchField.GetValue() if self.searchButton.GetLabel() == 'stop!': self.searchButton.SetLabel('find!') self.searchResults.stopsearch = True return else: self.groupSelection.clear() self.groupSelection.Disable() self.searchButton.SetLabel('stop!') self.searchButton.Update() err, info = self.searchResults.searchButtonAction(event) selections = self.searchResults.searchSourceItems # see if we should set the checkbox button from a previous rule state = None if len(selections) > 0 and self.rules.has_key(selections[0]): rule = self.rules[selections[0]] if self.rules[selections[0]].has_key(event.searchstring): state = self.rules[selections[0]][event.searchstring] for i in selections: if not self.rules.has_key(i) or self.rules[i] != rule: state = None break #for j in self.rules[i]: if state: print "should restore checkbox to %s" % state self.groupSelection.setState(state) self.searchButton.SetLabel('find!') self.searchResults.stopsearch = False if self.searchButtonAction: self.searchButtonAction(event, errmsg=err, infomsg=info) self.groupSelection.Enable() def selectAllText(self, event): if wx.Window.FindFocus() != self.searchField: self.searchField.SetSelection(-1,-1) self.searchField.SetFocus() else: self.searchField.SetSelection(0,0) event.Skip() def unfocused(self, event): self.searchField.SetSelection(0,0) def setGroup(self, state): b = wx.BusyCursor() selections = self.searchResults.setGroup(state) for s in selections: if not self.rules.has_key(s): self.rules[s] = {} if state == CheckboxState.UNSELECTED: try: self.rules.pop(s) except: pass else: self.rules[s][self.searchField.GetValue()] = state print self.rules class FilePanel(wx.SplitterWindow): def __init__(self, parent, searchButtonAction=None): # Use the WANTS_CHARS style so the panel doesn't eat the Return key. wx.SplitterWindow.__init__(self, parent, -1, style=wx.SP_LIVE_UPDATE | wx.CLIP_CHILDREN | wx.WANTS_CHARS) self.Bind(wx.EVT_SIZE, self.OnSize) self.SetNeedUpdating(True) self.tree = DirCheckboxCtrl(self, -1, dir="/") # XXX: fludrules.init path should be in config self.fludrules = self.getFludHome()+"/fludrules.init" if not os.path.isfile(self.fludrules): # XXX: do the other first time stuff (email encrypted credentials, # etc.) parent.SetMessage("Welcome. This appears to be the first" " time you've run flud. We've automatically selected some" " files for backup. You can make changes by" " selecting/deselecting files and directories. When you are" " done, simply close this window.") src = open('fludrules.init', 'r') dst = open(self.fludrules, 'w') filerules = src.read() dst.write(filerules) dst.close() src.close() filerules = eval(filerules) rulestates = {} for rule in filerules['baserules']: value = filerules['baserules'][rule] rule = glob.glob(os.path.expandvars(rule)) for r in rule: rulestates[r] = value self.tree.setStates(rulestates) # XXX: fludfile.conf path should be in config self.fludfiles = self.getFludHome()+"/fludfile.conf" print self.fludfiles if os.path.isfile(self.fludfiles): file = open(self.fludfiles, 'r') states = eval(file.read()) self.tree.setStates(states) file.close() self.searchPanel = SearchPanel(self, dircheckbox=self.tree, searchButtonAction=searchButtonAction) self.SetMinimumPaneSize(20) self.SplitVertically(self.tree, self.searchPanel) #, 300) def getFludHome(self): if os.environ.has_key('FLUDHOME'): fludhome = os.environ['FLUDHOME'] else: fludhome = os.environ['HOME']+"/.flud" if not os.path.isdir(fludhome): os.mkdir(fludhome, 0700) return fludhome def shutdown(self, event): self.flushFileConfig() event.Skip() def flushFileConfig(self): states = self.tree.getStates() f = open(self.fludfiles, 'w') f.write(str(states)) f.close() for i in states: print "%s %s" % (i, states[i]) def OnSize(self, event): w,h = self.GetClientSizeTuple() if self.tree: self.tree.SetDimensions(0, 0, w, h) event.Skip() class RestoreCheckboxCtrl(DirCheckboxCtrl): # XXX: child/parent selection/deselection isn't quite right still, esp wrt # root node. repro: # -/ # -d1 # -f1 # -d2 # -d3 # -f2 # -f3 # with nothing selected, select d3 and f3, then select root, then deselect # d3 and f3 def __init__(self, parent, id=-1, config=None, pos=wx.DefaultPosition, size=wx.DefaultSize, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER), validator=wx.DefaultValidator, name=wx.ControlNameStr): self.config = config DirCheckboxCtrl.__init__(self, parent, id, config, pos, size, style, validator, name, allowExclude=False) def initTree(self, config): self.expandRoot(config) self.expandUntilMultiple() def expandRoot(self, config): self.defaultImageList, self.checkboxes, self.icondict \ = createDefaultImageList() self.AssignImageList(self.defaultImageList) self.il = self.GetImageList() self.rootID = self.AddRoot("/", self.icondict['computer'], -1, wx.TreeItemData(("", True, False, CheckboxState.UNSELECTED))) self.update() def expandUntilMultiple(self): node = self.rootID while True: (ipath, isdir, expanded, istate) = self.GetItemData(node).GetData() children = self.getChildren(node, False) if len(children) > 1 or len(children) == 0: break; node = children[0] self.Expand(node) def update(self): master = listMeta(self.config) for i in master: if not isinstance(master[i], dict): traversal = i.split(os.path.sep) node = self.rootID path = "/" if traversal[0] == '': traversal.remove('') for n in traversal: path = os.path.join(path, n) children = self.getChildrenDict(node) if n == traversal[-1] and not n in children: child = self.AppendItem(node, n) self.SetPyData(child, (path, False, False, 0)) idx = getFileIcon(i, self.il, self.checkboxes, self.icondict) self.SetItemImage(child, idx, wx.TreeItemIcon_Normal) else: if not n in children: child = self.AppendItem(node, n) self.SetPyData(child, (path, True, False, 0)) self.SetItemImage(child, self.icondict['folder'], wx.TreeItemIcon_Normal) else: child = children[n] node = child self.Expand(self.rootID) def getChildrenDict(self, node): result = {} child, cookie = self.GetFirstChild(node) while child: result[self.GetItemText(child)] = child child, cookie = self.GetNextChild(node, cookie) return result def onExpand(self, event): pass def getSelected(self, startNode=None): if not startNode: startNode = self.rootID children = self.getChildren(startNode) selected = [] for n in children: (path, isDir, expanded, state) = self.GetItemData(n).GetData() if not isDir \ and (state == CheckboxState.SELECTED \ or state == CheckboxState.SELECTEDCHILD): selected.append(n) if isDir and (state == CheckboxState.SELECTED \ or state == CheckboxState.SELECTEDPARENT \ or state == CheckboxState.SELECTEDCHILD): selected += self.getSelected(n) return selected class RestorePanel(wx.Panel): def __init__(self, parent, config, factory): self.config = config self.factory = factory wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) self.tree = RestoreCheckboxCtrl(self, -1, config, #wx.TreeCtrl(self, -1, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER)) self.restoreButton = wx.Button(self, -1, 'restore selected files', name='restoreButton') self.Bind(wx.EVT_BUTTON, self.onRestoreClick, self.restoreButton) self.gbSizer = wx.GridBagSizer(2,1) self.gbSizer.Add(self.tree, (0,0), flag=wx.EXPAND|wx.ALL, border=0) self.gbSizer.Add(self.restoreButton, (1,0), flag=wx.EXPAND|wx.ALL, border=0) self.gbSizer.AddGrowableRow(0) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) def update(self): self.tree.update() def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() def onTooltip(self, event): pass def onRestoreClick(self, event): for n in self.tree.getSelected(): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "restoring %s" % path d = self.factory.sendGETF(path) d.addCallback(self.restored, n) d.addErrback(self.restoreFailed, n) self.tree.UnselectAll() def restored(self, res, n): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "yay, %s" % path self.tree.SetItemTextColour(n, '#005804') self.tree.changeState(n) def restoreFailed(self, err, n): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "boo, %s: %s" % (path, err) self.tree.SetItemTextColour(n, wx.RED) self.tree.changeState(n) class SchedulePanel(wx.Panel): def __init__(self, parent): wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() class FeedbackPanel(wx.Panel): def __init__(self, parent): wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) editor = wx.lib.editor.editor.Editor(parent, -1) def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() class FludNotebook(wx.Notebook): def __init__(self, parent, id=-1, pos=wx.DefaultPosition, size=wx.DefaultSize, style=wx.NB_BOTTOM|wx.NO_BORDER): self.parent = parent self.config = parent.config self.factory = LocalClientFactory(self.config) print "connecting to localhost:%d" % self.config.clientport reactor.connectTCP('localhost', self.config.clientport, self.factory) wx.Notebook.__init__(self, parent, id, pos, style=style) self.filePanel = FilePanel(self, searchButtonAction=parent.searchButtonAction) self.AddPage(self.filePanel, "Backup Files") self.restorePanel = RestorePanel(self, self.config, self.factory) self.AddPage(self.restorePanel, "Restore") self.schedulePanel = SchedulePanel(self) self.AddPage(self.schedulePanel, "Backup Schedule") self.feedbackPanel = FeedbackPanel(self) self.AddPage(self.feedbackPanel, "Feedback") self.Bind(wx.EVT_NOTEBOOK_PAGE_CHANGED, self.changedPage) def shutdown(self, event): self.filePanel.shutdown(event) def changedPage(self, event): page = event.GetSelection() if page == 0: self.SetMessage("Select files and directories for backup" " with the filesystem view on the left, or set up criteria" " for finding files for backup directly with simple" " searches, below right.") elif page == 1: self.SetMessage("Select files/directories to be restored to" " your computer, then click on 'restore!' Files will turn" " green as they arrive.") self.restorePanel.update() elif page == 2: self.SetMessage("Configure how often your computer should backup." "\n (not implemented)") elif page == 3: self.SetMessage("Send feedback to flud programmers. (not" " implemented)") def SetMessage(self, msg): self.parent.SetMessage(msg) class FludLogoPanel(wx.Panel): def __init__(self, parent, id=-1, pos=wx.DefaultPosition, size=wx.Size(10,10), style=wx.TAB_TRAVERSAL, name="logo panel"): wx.Panel.__init__(self, parent, id, pos, size, style, name) self.SetAutoLayout(True) self.SetBackgroundColour(wx.BLACK) self.SetForegroundColour(wx.WHITE) logobmp = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "flud-backup-logo-1-150-nodrop.png"))) pad = 0 self.logowidth = logobmp.GetWidth() self.logoheight = logobmp.GetHeight() self.logo = wx.StaticBitmap(self, -1, logobmp) self.messagePanel = wx.Panel(self, -1) self.messagePanel.SetBackgroundColour(wx.BLACK) self.messagePanel.SetForegroundColour(wx.WHITE) self.message = wx.StaticText(self.messagePanel, -1, "message text area", style=wx.ALIGN_CENTER|wx.ST_NO_AUTORESIZE, size=(-1, self.logoheight-15)) self.message.Bind(wx.EVT_SIZE, self.resizeMessage) self.bsizer = wx.BoxSizer(wx.VERTICAL) self.bsizer.Add(self.message, flag=wx.EXPAND|wx.ALL, border=35) self.bsizer.SetSizeHints(self.messagePanel) self.messagePanel.SetSizer(self.bsizer) self.gbSizer = wx.GridBagSizer(1,2) self.gbSizer.Add(self.logo, (0,0)) self.gbSizer.Add(self.messagePanel, (0,1), flag=wx.EXPAND|wx.ALL) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(1) self.SetSizerAndFit(self.gbSizer) self.SetSize(wx.Size(self.logowidth, self.logoheight)) self.SetSizeHints(self.logowidth, self.logoheight, -1, self.logoheight) def SetMessage(self, msg): (w,h) = self.message.GetSizeTuple() #print "msg area size is %d x %d" % (w,h) self.message.SetLabel(msg) self.message.Wrap(w) #print "msg is '%s'" % self.message.GetLabel() self.message.Center() def resizeMessage(self, evt): # this is mainly to deal with StaticText wonkiness (not calling Wrap() # automatically, not centering properly automatically). It may be # possible to get rid of this with a future wxPython release. (w,h) = self.message.GetSizeTuple() self.message.Wrap(w) m = self.message.GetLabel() m = m.replace('\n',' ') self.message.SetLabel(m) self.message.Wrap(w) self.message.Center() class FludFrame(wx.Frame): def __init__(self, parent, id=wx.ID_ANY, label="flud bakcup client", size=wx.Size(800,600), style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE, config=None): wx.Frame.__init__(self, parent, id, label, size=size, style=style) wx.ToolTip.SetDelay(2000) self.clearMessage = False self.logoPanel = FludLogoPanel(self) self.SetMessage('Welcome.') self.config = config self.notebook = FludNotebook(self) self.operationStatus = wx.StatusBar(name='operationStatus', parent=self, style=0) self.SetStatusBar(self.operationStatus) self.gbSizer = wx.GridBagSizer(2,1) self.gbSizer.Add(self.logoPanel,(0,0), flag=wx.EXPAND) self.gbSizer.Add(self.notebook, (1,0), flag=wx.EXPAND|wx.ALL, border=1) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) self.Bind(wx.EVT_CLOSE, self.shutdown) self.SetSize(size) self.Show(True) def SetMessage(self, message): self.logoPanel.SetMessage(message) def shutdown(self, event): self.notebook.shutdown(event) def searchButtonAction(self, event, errmsg=None, infomsg=None): if errmsg: self.logoPanel.SetMessage(errmsg) self.clearMessage = True elif infomsg: self.logoPanel.SetMessage(infomsg) self.clearMessage = False elif self.clearMessage: self.logoPanel.SetMessage("") #if __name__ == '__main__': # app = wx.PySimpleApp() # # config = FludConfig() # config.load(doLogging=False) # # f = FludFrame(None, wx.ID_ANY, 'flud backup client', size=(795,600), # style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE, # config=config) # # from twisted.internet import reactor # reactor.registerWxApp(app) # reactor.run()
Python
#!/usr/bin/python """ FludTestGauges.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Provides gauges for visualizing storage for multiple flud nodes running on the same host. This is really only useful for demos and testing. """ import sys, os, signal, stat, random import wx import wx.lib.buttons as buttons from flud.FludConfig import FludConfig dutotal = 0 def visit(arg, top, files): global dutotal for file in files: dutotal += os.lstat("%s" % (os.path.join(top,file)))[stat.ST_SIZE] arg += dutotal def du(dir): global dutotal dutotal = 0 os.path.walk(dir, visit, dutotal) return dutotal # XXX: too much manual layout. should convert to a managed layout to allow for # resizing, etc. SGAUGEWIDTH = 230 # storage gauge DGAUGEWIDTH = 100 # dht gauge GAUGEHEIGHT = 20 ROWHEIGHT = 30 SEP = 5 LABELWIDTH = 20 POWERWIDTH = 70 RATIOBARHEIGHT = 70 COLWIDTH = SGAUGEWIDTH+DGAUGEWIDTH+LABELWIDTH+POWERWIDTH COLGAPFUDGE = 30 class FludTestGauges(wx.Frame): def __init__(self, parent, title, dirroot, dirs): screenHeight = wx.SystemSettings.GetMetric(wx.SYS_SCREEN_Y)-100 rowheight = ROWHEIGHT+SEP height = len(dirs)*(rowheight)+RATIOBARHEIGHT columns = height / screenHeight + 1 width = COLWIDTH*columns if columns > 1: height = (len(dirs)/columns)*(rowheight)+RATIOBARHEIGHT if (len(dirs) % columns) > 0: height += rowheight wx.Frame.__init__(self, parent, wx.ID_ANY, title, size=(width,height), style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE) self.storebarend = 1024 self.smultiplier = 100.0 / self.storebarend self.sdivisor = 1 self.sbytelabel = "" self.dhtbarend = 512 self.dmultiplier = 100.0 / self.dhtbarend self.ddivisor = 1 self.dbytelabel = "" self.storeheading = wx.StaticText(self, -1, "block storage", (LABELWIDTH, 5)) self.totaldht = wx.StaticText(self, -1, "metadata", (LABELWIDTH+SGAUGEWIDTH+SEP, 5)) self.gauges = [] curCol = 0 curRow = 30 for i in range(len(dirs)): self.gauges.append(wx.Gauge(self, -1, 100, (curCol*COLWIDTH+LABELWIDTH, curRow), (SGAUGEWIDTH, GAUGEHEIGHT))) self.gauges[i].SetBezelFace(3) self.gauges[i].SetShadowWidth(3) self.gauges[i].SetValue(0) self.gauges[i].dir = "%s%s" % (dirroot,dirs[i]) os.environ['FLUDHOME'] = self.gauges[i].dir; conf = FludConfig() conf.load(doLogging = False) print "%s" % conf.nodeID self.gauges[i].label = wx.StaticText(self, -1, "%2s" % dirs[i], (curCol*COLWIDTH, curRow+(rowheight/4)), size=(LABELWIDTH, -1)) self.gauges[i].idlabel = wx.StaticText(self, -1, "%s" % conf.nodeID, (curCol*COLWIDTH+LABELWIDTH, curRow+20)) font = self.gauges[i].idlabel.GetFont() font.SetPointSize(6) self.gauges[i].idlabel.SetFont(font) self.gauges[i].dhtgauge = wx.Gauge(self, -1, 100, (curCol*COLWIDTH+LABELWIDTH+SGAUGEWIDTH+SEP, curRow), (SGAUGEWIDTH/3, GAUGEHEIGHT)) self.gauges[i].power = wx.Button(self, i, "turn OFF", (curCol*COLWIDTH +LABELWIDTH+SGAUGEWIDTH+2*SEP+SGAUGEWIDTH/3, curRow), (POWERWIDTH, ROWHEIGHT)) #self.gauges[i].power = buttons.GenBitmapToggleButton(self, i, # None, # (LABELWIDTH+SGAUGEWIDTH+2*SEP+SGAUGEWIDTH/3, curRow), # (POWERWIDTH, ROWHEIGHT)) #self.gauges[i].button.SetBestSize() self.gauges[i].power.SetToolTipString("power on/off") self.Bind(wx.EVT_BUTTON, self.onClick, self.gauges[i].power) curRow += rowheight if curRow > height-RATIOBARHEIGHT: curCol += 1 curRow = 30 self.totalstore = wx.StaticText(self, -1, "total: 0", (LABELWIDTH, height-40)) self.totaldht = wx.StaticText(self, -1, "total: 0", (LABELWIDTH+SGAUGEWIDTH+SEP, height-40)) self.ratiogauge = wx.Gauge(self, -1, 100, (LABELWIDTH, height-20), (SGAUGEWIDTH+SEP+SGAUGEWIDTH/3, 10)) self.ratiogauge.SetValue(0) self.Bind(wx.EVT_IDLE, self.IdleHandler) self.timer = wx.PyTimer(self.update) self.timer.Start(1000) def onClick(self, event): # XXX: note that under our current startNnodes.sh scheme, the first # node spawned doesn't contact anyone, so if that one is powered off # and then powered back on, it will not be part of the node until # another node pings it # XXX: unix-specific proc management stuff follows idx = event.GetId() home = self.gauges[idx].dir pidfile = os.path.join(home, 'twistd.pid') if os.path.exists(pidfile): print "shutting down %s" % home f = open(pidfile) pid = int(f.read()) f.close() # XXX: ps command no worky on windows, and "-ww" may not worker on # oldskool unixes self.gauges[idx].savedCmd = os.popen( "ps f -wwp %d -o args=" % pid).read() procline = os.popen("ps e -wwp %d" % pid).read() self.gauges[idx].savedEnv = [e for e in procline.split() if e[:4] == 'FLUD'] # XXX: os.kill no worky on windows, need something like: #def windowskill(pid): # import win32api # handle = win32api.OpenProcess(1, 0, pid) # return (0 != win32api.TerminateProcess(handle, 0)) os.kill(pid, signal.SIGTERM) self.gauges[idx].power.SetLabel("turn ON") self.gauges[idx].Hide() self.gauges[idx].dhtgauge.Hide() else: print "powering up %s" % home # XXX: this exec no worky on windows fullcmd = "%s %s" % (' '.join(self.gauges[idx].savedEnv), self.gauges[idx].savedCmd) print fullcmd result = os.popen('%s %s' % (' '.join(self.gauges[idx].savedEnv), self.gauges[idx].savedCmd)).readlines() self.gauges[idx].power.SetLabel("turn OFF") self.gauges[idx].Show() self.gauges[idx].dhtgauge.Show() print result def update(self): def sizeclass(num): divisor = 1 bytelabel = "" if num > 1024: divisor = 1024.0 bytelabel = 'K' if num > 1048576: divisor = 1048576.0 bytelabel = 'M' if num > 1073741824: divisor = 1073741824.0 bytelabel = 'G' return (divisor, bytelabel) storelargest = 0 dhtlargest = 0 storetotal = 0 dhttotal = 0 for i in self.gauges: if os.path.isdir(i.dir): i.storebytes = du(os.path.join(i.dir,'store')) if i.storebytes > storelargest: storelargest = i.storebytes storetotal += i.storebytes i.dhtbytes = du(os.path.join(i.dir,'dht')) if i.dhtbytes > dhtlargest: dhtlargest = i.dhtbytes dhttotal += i.dhtbytes else: i.storebytes = 0 i.dhtbytes = 0 i.Disable() i.power.Disable() while storelargest > self.storebarend: self.storebarend = self.storebarend * 2 self.smultiplier = 100.0 / self.storebarend self.sdivisor, self.sbytelabel = sizeclass(storetotal) while dhtlargest > self.dhtbarend: self.dhtbarend = self.dhtbarend * 2 self.dmultiplier = 100.0 / self.dhtbarend self.ddivisor, self.dbytelabel = sizeclass(dhttotal) #print "-----" for i in self.gauges: i.SetValue(i.storebytes*self.smultiplier) i.dhtgauge.SetValue(i.dhtbytes*self.dmultiplier) #print "%.2f, %.2f" % ((float(i.storebytes)/float(i.dhtbytes)), # (float(i.GetValue())/float(i.dhtgauge.GetValue()))) self.totalstore.SetLabel("total: %.1f%s" % (float(storetotal)/self.sdivisor, self.sbytelabel)) self.totaldht.SetLabel("total: %.1f%s" % (float(dhttotal)/self.ddivisor, self.dbytelabel)) if (dhttotal+storetotal == 0): self.ratiogauge.SetValue(0) else: self.ratiogauge.SetValue((storetotal*100/(dhttotal+storetotal))) def updateGauges(self, update): for index, value in update: self.monitors[index].setValue(value) def IdleHandler(self, event): pass def main(): if len(sys.argv) < 2: print "usage: %s dircommon exts" % sys.argv[0] print " where exts will be appended to dircommon" print " e.g., '%s /home/joe/.flud 1,2,3,4,10,15,20'"\ % sys.argv[0] print " or, '%s /home/joe/.flud 1-10,15,20'"\ % sys.argv[0] sys.exit() root = sys.argv[1] exts = [] dirs = [d.strip() for d in sys.argv[2].split(',')] for i in dirs: if i == "_": exts.append('') # undocumented, means "just dircommon" elif i.find('-') >= 0: start, end = i.split('-') for j in range(int(start),int(end)+1): exts.append(j) else: exts.append(int(i)) app = wx.PySimpleApp() t = FludTestGauges(None, 'Flud Test Gauges', root, exts) t.Show(1) app.MainLoop() if __name__ == '__main__': main()
Python
#!/usr/bin/python """ FludNode.tac (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL). This is the application file used by twistd to daemonize FludNode. """ import os from twisted.application import service, internet import flud.FludNode from flud.protocol.FludCommUtil import getCanonicalIP port = None gwhost = None gwport = None if 'FLUDPORT' in os.environ: port = int(os.environ['FLUDPORT']) if 'FLUDGWHOST' in os.environ: gwhost = getCanonicalIP(os.environ['FLUDGWHOST']) if 'FLUDGWPORT' in os.environ: gwport = int(os.environ['FLUDGWPORT']) node = flud.FludNode.FludNode(port) if gwhost and gwport: node.connectViaGateway(gwhost, gwport) application = service.Application("flud.FludNode") service = node.start(twistd=True) #service.setServiceParent(application)
Python
import base64 """ fencode.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Provides efficient urlsafe base64 encoding of python types (int, long, string, None, dict, tuple, list) -- in the same vein as BitTorrent's bencode or MNet's mencode. """ class Fencoded: # provides a typed wrapper for previously fencoded data, so that we can # nest fencoded data inside other fencoded structures without bloating it # (note that this could also be used to store strings without b64 bloating, # but that forgoes url safety). See doctests in fencode() for usage # examples. def __init__(self, data): self.data = data def __eq__(self, i): if not isinstance(i, Fencoded): return False return self.data == i.data def fencode(d, lenField=False): """ Takes string data or a number and encodes it to an efficient URL-friendly format. >>> n = None >>> i = 123455566 >>> I = 1233433243434343434343434343434343509669586958695869L >>> s = "hello there, everyone" >>> s2 = "long text ............................................................................... AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAABBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC" >>> d = {'a': 'adfasdfasd', 'aaa': 'rrreeeettt', 'f': 'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa'} >>> d2 = {'a': 123, 'b': 'xyz'} >>> d3 = {'a': 123, 'b': 'xyz', 'c': {'x': 456, 'y': 'abc'}} >>> d3 = {'a': 123, 'b': 'xyz', 'c': d} >>> d4 = {} >>> l = [1,2,3,4,'a','b','cde'] >>> l2 = [i,I,s,d] >>> l3 = [] >>> l4 = [n] >>> t = (1,2,3,4,'a','b','cde') >>> t2 = (i,I,s,d) >>> t3 = () >>> d5 = {t: s, 'n': n, 'a': i, i: "a", 'd': d3, s2: s2} >>> l5 = [1,[],2,[],(),{},3,{t: s, 'n': ()}] >>> fdecode(fencode(n)) == n True >>> fdecode(fencode(i)) == i True >>> fdecode(fencode(I)) == I True >>> fdecode(fencode(-i)) == -i True >>> fdecode(fencode(-I)) == -I True >>> fdecode(fencode(s)) == s True >>> fdecode(fencode(d)) == d True >>> fdecode(fencode(d2)) == d2 True >>> fdecode(fencode(d3)) == d3 True >>> fdecode(fencode(d4)) == d4 True >>> fdecode(fencode(l)) == l True >>> fdecode(fencode(l2)) == l2 True >>> fdecode(fencode(l3)) == l3 True >>> fdecode(fencode(l4)) == l4 True >>> fdecode(fencode(t)) == t True >>> fdecode(fencode(t2)) == t2 True >>> fdecode(fencode(t3)) == t3 True >>> fdecode(fencode(d5)) == d5 True >>> fdecode(fencode(l5)) == l5 True >>> f = Fencoded(fencode(s)) >>> fdecode(fencode(f)) == f True >>> fdecode(fdecode(fencode(f))) == s True >>> fdecode(fencode({i: f})) == {i: f} True >>> fdecode(fdecode(fencode({i: f}))[i]) == s True >>> fdecode(fdecode(fencode({i: f, I: f}))[i]) == s True >>> fdecode(fencode(f), recurse=True) == s True >>> fdecode(fencode(f), recurse=2) == s True >>> f2 = Fencoded(fencode(f)) >>> f3 = Fencoded(fencode(f2)) >>> fdecode(fencode(f3), recurse=True) == s True >>> fdecode(fencode(f3), recurse=3) == f True >>> fdecode(fencode({i: f3, I: f2})) == {i: f3, I: f2} True >>> fdecode(fencode({i: f3, I: f2}), recurse=1) == {i: f3, I: f2} True >>> fdecode(fencode({i: f3, I: f2}), recurse=2) == {i: f2, I: f} True >>> fdecode(fencode({i: f3, I: f2}), recurse=3) == {i: f, I: s} True >>> fdecode(fencode({i: f3, I: f2}), recurse=4) == {i: s, I: s} True >>> fdecode(fencode({i: f3, I: f2}), recurse=True) == {i: s, I: s} True """ def makeLen(i): """ Returns the integer i as a three-byte length value. >>> makeLen(255) '\\x00\\xff' >>> makeLen(65535) '\\xff\\xff' """ if i > 65535 or i < 0: raise ValueError("illegal length for fencoded data" "(0 < x <= 65535)") return fencode(i)[1:-1] if isinstance(d, int) or isinstance(d, long): val = "%x" % d neg = False c = 'i' if isinstance(d, long): c = 'o' if d < 0: neg = True val = val[1:] c = c.upper() if len(val) % 2 != 0: val = "0%s" % val val = val.decode('hex') if len(val) % 2 != 0: val = '\x00' + val val = base64.urlsafe_b64encode(val) if lenField: if len(val) > 65535: raise ValueError("value to large for encode") return c+makeLen(len(val))+val else: return c+val elif isinstance(d, str): # String data may contain characters outside the allowed charset. # urlsafe b64encoding ensures that data can be used inside http urls # (and other plaintext representations). val = base64.urlsafe_b64encode(d) if lenField: if len(val) > 65535: raise ValueError("value to large for encode") return 's'+makeLen(len(val))+val else: return 's'+val elif isinstance(d, dict): result = 'd' contents = "" for i in d: contents = contents + fencode(i,True) + fencode(d[i],True) if lenField: result = result+makeLen(len(contents))+contents else: result = result+contents return result elif isinstance(d, list): result = 'l' contents = '' for i in d: contents = contents + fencode(i,True) if lenField: result = result+makeLen(len(contents))+contents else: result = result+contents return result elif isinstance(d, tuple): result = 't' contents = '' for i in d: contents = contents + fencode(i,True) if lenField: result = result+makeLen(len(contents))+contents else: result = result+contents return result elif isinstance(d, Fencoded): result = 'f' contents = d.data if lenField: result = result+makeLen(len(d.data))+contents else: result = result+contents return result elif d == None: if lenField: return 'n'+makeLen(1)+'0' else: return 'n0' else: raise ValueError("invalid value passed to fencode: %s" % type(d)) def fdecode(d, lenField=False, recurse=1): """ Takes previously fencoded data and decodes it into its python type(s). 'lenField' is used internally, and indicates that the fencoded data has length fields (used for compositiong of tuples, lists, dicts, etc). 'recurse' indicates that fdecode should recursively fdecode Fencoded objects if set to True, or that it should recurse to a depth of 'recurse' when encountering Fencoded objects if it is an integer value. """ def getLen(s): if len(s) != 3 or not isinstance(s, str): raise ValueError("fdecode length strings must be 3 bytes long: '%s'" % s) return fdecode('i'+s+'=', recurse=recurse) def scanval(valstring, lenField=False): """ scans the given valstring and returns a value and the offset where that value ended (as a tuple). If valstring contains more than one value, only the length of the first is returned. Otherwise, the entire length is returned. """ valtype = valstring[0] if lenField: start = 4 end = start+getLen(valstring[1:4]) else: start = 1 end = len(valstring)-1 #print " scanval calling fdecode on val[%d:%d]=%s" % (0, end, valstring) return (fdecode(valstring[0:end], True, recurse=recurse), end) if isinstance(d, Fencoded): return fdecode(d.data, recurse=recurse) if not isinstance(d, str): raise ValueError("decode takes string data or Fencoded object only," " got %s" % type(d)) valtype = d[0] if lenField: length = getLen(d[1:4]) val = d[4:] else: val = d[1:len(d)] if valtype == 'i': val = base64.urlsafe_b64decode(val) val = val.encode('hex') return int(val, 16) elif valtype == 'I': val = base64.urlsafe_b64decode(val) val = val.encode('hex') return -int(val, 16) elif valtype == 'o': val = base64.urlsafe_b64decode(val) val = val.encode('hex') return long(val, 16) elif valtype == 'O': val = base64.urlsafe_b64decode(val) val = val.encode('hex') return -long(val, 16) elif valtype == 's': return base64.urlsafe_b64decode(val) elif valtype == 'd': result = {} while len(val) != 0: #print "string is: %s (len=%d)" % (val, len(val)) (key,l1) = scanval(val, True) #print "got key '%s' of length %d" % (key,l1) (value,l2) = scanval(val[l1:len(val)], True) #print "got value '%s' of length %d" % (value,l2) result[key] = value val = val[l1+l2:] return result elif valtype == 'l': result = [] if lenField: pass while len(val) != 0: (v,l) = scanval(val, True) result.append(v) val = val[l:] return result elif valtype == 't': result = [] if lenField: pass while len(val) != 0: (v,l) = scanval(val, True) result.append(v) val = val[l:] return tuple(result) elif valtype == 'f': if not isinstance(recurse, bool): recurse = recurse-1 if recurse > 0: return fdecode(val, recurse=recurse) return Fencoded(val) elif valtype == 'n': return None else: raise ValueError("invalid value passed to fdecode" " -- cannot fdecode data that wasn't previously fencoded: '%s'" % d[:400]) if __name__ == '__main__': import doctest doctest.testmod()
Python
# XXX: this class goes away class Reputation: """ Each node maintains a list of reputation objects corresponding to reputations of other nodes. Reputations may be self-generated (in which case the originator is this node itself), or may be relayed (in which case some other node is the originator). Self-generated reputations are vastly more reliable than those relayed -- relayed reputations are second-hand information, and are more likely to have false data. """ def __init__(self, ID, originator): """ Variables designated as '%' have values between 0 and 100. """ self.ID = ID self.originator = originator self.confidence = 0 # % originator data stored / nodes the # originator stores to self.verifiability = 0 # % originator data verified success/failed self.availability = 0 # % originator contact attempts success/fail self.bandwidth = 0 # avg bandwidth observed from orig. to ID. self.age = 0 # age of reputation in days def score(self): """ Returns a score for this reputation based in member variables. The reputation must be a local reputation, i.e., the originator must be equal to the global myNodeID. Otherwise, call scoreRelay() >>> myNodeID = "self" >>> rep = Reputation("somenode","self") >>> rep.availability = 50 >>> rep.verifiability = 50 >>> rep.score() 33 >>> rep = Reputation("somenode","someothernode") >>> rep.availability = 30 >>> rep.score() -1 """ # should find a good adjustment of weights (XXX: machine learning?) if self.originator != myNodeID: return -1 return (self.confidence + self.verifiability + self.availability) / 3 # XXX: should also include age and bandwidth def scoreRelay(self): """ Returns a score for this reputation based in member variables. The reputation must be a remote reputation, i.e., the originator must not be equal to the global myNodeID. Otherwise, call score() >>> myNodeID = "self" >>> rep = Reputation("somenode","self") >>> rep.availability = 50 >>> rep.verifiability = 50 >>> rep.scoreRelay() -1 >>> rep = Reputation("somenode","someothernode") >>> rep.availability = 30 >>> rep.scoreRelay() 10 """ if self.originator == myNodeID: return -1 return (self.confidence + self.verifiability + self.availability) / 3 # XXX: should also include age and bandwidth def updateConfidence(self, totalDataStored, totalNodesStoredTo): self.confidence = totalDataStored / totalNodesStoredTo; def _test(self): import doctest doctest.testmod() if __name__ == '__main__': myNodeID = "self" rep = Reputation("other", "self") rep._test()
Python
#!/usr/bin/python """ FludLocalClient.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. FludLocalClient provides a command-line client for interacting with FludNode. """ import sys, os, time from twisted.internet import reactor from flud.FludConfig import FludConfig from flud.fencode import fencode, fdecode from flud.FludCrypto import hashfile from protocol.LocalClient import * logger = logging.getLogger('flud') class CmdClientFactory(LocalClientFactory): def __init__(self, config): LocalClientFactory.__init__(self, config) self.quit = False self.msgs = [] def callFactory(self, func, commands, msgs): # since we can't call factory methods from the promptUser thread, we # use this as a convenience to put those calls back in the event loop reactor.callFromThread(self.doFactoryMethod, func, commands, msgs) def doFactoryMethod(self, func, commands, msgs): d = func() d.addCallback(self.queueResult, msgs, '%s succeeded' % commands) d.addErrback(self.queueError, msgs, '%s failed' % commands) return d def promptUser(self): helpDict = {} command = raw_input("%s> " % time.ctime()) commands = command.split(' ') # XXX: should tokenize on any whitespace commandkey = commands[0][:4] # core client operations helpDict['exit'] = "exit from the client" helpDict['help'] = "display this help message" helpDict['ping'] = "send a GETID() message: 'ping host port'" helpDict['putf'] = "store a file: 'putf canonicalfilepath'" helpDict['getf'] = "retrieve a file: 'getf canonicalfilepath'" helpDict['geti'] = "retrieve a file by CAS key: 'geti fencodedCASkey'" helpDict['fndn'] = "send a FINDNODE() message: 'fndn hexIDstring'" helpDict['list'] = "list stored files (read from local metadata)" helpDict['putm'] = "store master metadata" helpDict['getm'] = "retrieve master metadata" helpDict['cred'] = "send encrypted private credentials: cred"\ " passphrase emailaddress" helpDict['node'] = "list known nodes" helpDict['buck'] = "print k buckets" helpDict['stat'] = "show pending actions" helpDict['stor'] = "store a block to a given node:"\ " 'stor host:port,fname'" helpDict['rtrv'] = "retrieve a block from a given node:"\ " 'rtrv host:port,fname'" helpDict['vrfy'] = "verify a block on a given node:"\ " 'vrfy host:port:offset-length,fname'" helpDict['fndv'] = "retrieve a value from the DHT: 'fndv hexkey'" helpDict['dlet'] = "delete from the stor: '[XXX]'" if commandkey == 'exit' or commandkey == 'quit': self.quit = True elif commandkey == 'help': self.printHelp(helpDict) elif commandkey == 'ping': # ping a host # format: 'ping host port' func = lambda: self.sendPING(commands[1], commands[2]) self.callFactory(func, commands, self.msgs) elif commandkey == 'putf': # store a file # format: 'putf canonicalfilepath' func = lambda: self.sendPUTF(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'getf': # retrieve a file # format: 'getf canonicalfilepath' func = lambda: self.sendGETF(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'geti': # retrieve a file by CAS ID # format: 'geti fencoded_CAS_ID' func = lambda: self.sendGETI(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'fndn': # find a node (or the k-closest nodes) # format: 'fndn hexIDstring' func = lambda: self.sendFNDN(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'list': # list stored files self.callFactory(self.sendLIST, commands, self.msgs) elif commandkey == 'putm': # store master metadata self.callFactory(self.sendPUTM, commands, self.msgs) elif commandkey == 'getm': # retrieve master metadata self.callFactory(self.sendGETM, commands, self.msgs) elif commandkey == 'cred': # send encrypted private credentials to an email address # format: 'cred passphrase emailaddress' func = lambda: self.sendCRED( command[len(commands[0])+1:-len(commands[-1])-1], commands[-1]) self.callFactory(func, commands, self.msgs) # the following are diagnostic operations, debug-only utility elif commandkey == 'node': # list known nodes self.callFactory(self.sendDIAGNODE, commands, self.msgs) elif commandkey == 'buck': # show k-buckets self.callFactory(self.sendDIAGBKTS, commands, self.msgs) elif commandkey == 'stat': # show pending actions print self.pending elif commandkey == 'stor': # stor a block to a given node. format: 'stor host:port,fname' storcommands = commands[1].split(',') try: fileid = int(storcommands[1], 16) except: linkfile = fencode(long(hashfile(storcommands[1]),16)) if (os.path.islink(linkfile)): os.remove(linkfile) os.symlink(storcommands[1], linkfile) storcommands[1] = linkfile # XXX: delete this file when the command finishes commands[1] = "%s,%s" % (storcommands[0], storcommands[1]) func = lambda: self.sendDIAGSTOR(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'rtrv': # retrive a block from a given node. format: 'rtrv host:port,fname' func = lambda: self.sendDIAGRTRV(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'vrfy': # verify a block on a given node. # format: 'vrfy host:port:offset-length,fname' logger.debug("vrfy(%s)" % commands[1]) func = lambda: self.sendDIAGVRFY(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'dlet': print "not yet implemented" elif commandkey == 'fndv': # try to retrieve a value from the DHT # format: 'fndv key' func = lambda: self.sendDIAGFNDV(commands[1]) self.callFactory(func, commands, self.msgs) elif command != "": reactor.callFromThread(self.queueError, None, self.msgs, "illegal command '%s'" % command) def queueResult(self, r, l, msg): logger.debug("got result %s" % msg) l.append((r, msg)) def queueError(self, r, l, msg): logger.debug("got error %s" % msg) if r: l.append((r.getErrorMessage(), msg)) else: l.append((None, msg)) def printHelp(self, helpDict): helpkeys = helpDict.keys() helpkeys.sort() for i in helpkeys: print "%s:\t %s" % (i, helpDict[i]) def promptLoop(self, r): for c in self.pending: for i in self.pending[c].keys(): if self.pending[c][i] == True: print "%s on %s completed successfully" % (c, i) self.pending[c].pop(i) elif self.pending[c][i] == False: print "%s on %s failed" % (c, i) self.pending[c].pop(i) else: print "%s on %s pending" % (c, i) while len(self.msgs) > 0: # this prints in reverse order, perhaps pop() all into a new list, # reverse, then print (errmsg, m) = self.msgs.pop() if errmsg: print "<- %s:\n%s" % (m, errmsg) else: print "<- %s" % m if self.quit: reactor.stop() else: d = threads.deferToThread(self.promptUser) d.addCallback(self.promptLoopDelayed) d.addErrback(self.err) def promptLoopDelayed(self, r): # give the reactor loop time to fire any quick cbs/ebs reactor.callLater(0.1, self.promptLoop, r) def clientConnectionLost(self, connector, reason): if not self.quit: LocalClientFactory.clientConnectionLost(self, connector, reason) def cleanup(self, msg): self.quit = True self.err(msg) def err(self, r): print "bah!: %s" % r reactor.stop() def main(): config = FludConfig() config.load(doLogging=False) logger.setLevel(logging.DEBUG) handler = logging.FileHandler('/tmp/fludclient.log') formatter = logging.Formatter('%(asctime)s %(filename)s:%(lineno)d' ' %(name)s %(levelname)s: %(message)s', datefmt='%H:%M:%S') handler.setFormatter(formatter) logger.addHandler(handler) factory = CmdClientFactory(config) if len(sys.argv) == 2: config.clientport = int(sys.argv[1]) print "connecting to localhost:%d" % config.clientport reactor.connectTCP('localhost', config.clientport, factory) factory.promptLoop(None) reactor.run() if __name__ == '__main__': main()
Python
import urlparse, os, types from twisted.web import client from twisted.internet import reactor, defer from twisted.python import failure """ HTTPMultipartDownloader.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. HTTPMultipartDownloader will download mulitple files from a multipart/related. Note that it does this by using the Content-ID and Content-Length headers in each multipart, and will fail if those are not present (this could be genericized to operate without those fields without too much effort) This code is modeled after twisted.web.client.HTTPDownloader, which is copyright 2001-2004 Twisted Matrix Laboratories, MIT licensed. """ class HTTPMultipartDownloader(client.HTTPDownloader): """Download multiple files, via multipart/related.""" protocol = client.HTTPPageDownloader value = None def __init__(self, url, dir, method='GET', postdata=None, headers=None, agent="Flud client", supportPartial=0): self.requestedPartial = 0 self.filenames = [] self.dir = dir client.HTTPClientFactory.__init__(self, url, method=method, postdata=postdata, headers=headers, agent=agent) self.deferred = defer.Deferred() self.waiting = 1 def gotHeaders(self, headers): if self.requestedPartial: contentRange = headers.get("content-range", None) if not contentRange: # server doesn't support partial requests, oh well self.requestedPartial = 0 return start, end, realLength = http.parseContentRange(contentRange[0]) if start != self.requestedPartial: # server is acting wierdly self.requestedPartial = 0 def openFile(self, partialContent): if partialContent: file = open(self.filename, 'rb+') file.seek(0, 2) else: file = open(self.filename, 'wb') self.filenames.append(self.filename) return file def pageStart(self, partialContent): """Called on page download start. @param partialContent: tells us if the download is partial download we requested. """ if partialContent and not self.requestedPartial: raise ValueError, "we shouldn't get partial content response if"\ " we didn't want it!" self.partialContent = partialContent if self.waiting: self.waiting = 0 self.inSubHeader = True self.file = None self.boundary = None def getSubHeader(self, data): newboundary = data[:data.find('\r\n')] data = data[len(newboundary)+2:] if not self.boundary: self.boundary = newboundary if self.boundary != newboundary: if self.boundary+"--" == newboundary: # end of multiparts return else: raise ValueError, "found illegal boundary" # XXX: print some of newboundary *safely* #raise ValueError, "found illegal boundary: %s, was %s" \ # % (newboundary[:80], self.boundary) headerEnd = data.find('\r\n\r\n') if headerEnd != -1: self.inSubHeader = False self.subHeaders = {} headers = data[:headerEnd].split('\r\n') for header in headers: k, v = header.split(':',1) self.subHeaders[k.lower()] = v.lstrip(' ') if not self.subHeaders.has_key('content-id'): raise ValueError, "no Content-ID field in multipart,"\ " can't continue" # XXX: need to check for badness (e.g, "../../) in content-id self.filename = os.path.join(self.dir, self.subHeaders['content-id']) self.file = self.openFile(self.partialContent) if not self.subHeaders.has_key('content-length'): raise ValueError, "no Content-Length field in multipart,"\ " can't continue" self.filesizeRemaining = int(self.subHeaders['content-length']) self.pagePart(data[headerEnd+4:]) def pagePart(self, data): if self.inSubHeader: self.getSubHeader(data) else: if not self.file: raise ValueError, "file %s not open for output" % self.filename try: if self.filesizeRemaining > len(data): self.file.write(data) self.filesizeRemaining -= len(data) else: self.file.write(data[:self.filesizeRemaining]) skipto = self.filesizeRemaining self.filesizeRemaining = 0 self.file.close() self.file = None self.inSubHeader = True self.getSubHeader(data[skipto+2:]) except IOError: #raise self.file = None self.deferred.errback(failure.Failure()) def pageEnd(self): if self.file: try: self.file.close() except IOError: self.deferred.errback(failure.Failure()) return self.deferred.callback(self.filenames) def doit(): factory = HTTPMultipartDownloader("/ret", "/tmp/") reactor.connectTCP('localhost', 1080, factory) return factory.deferred def didit(r): print "didit: %s" % str(r) reactor.stop() if __name__ == "__main__": # tries to request http://localhost:1080/ret, which it expects to be # multipart/related with Content-Length headers d = doit() d.addBoth(didit) reactor.run()
Python
#!/usr/bin/python """ FludClient.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. FludClient provides a GUI Client for interacting with FludNode. """ #from twisted.internet import wxreactor #wxreactor.install() import sys, os, string, time, glob import wx import wx.lib.mixins.listctrl as listmix import wx.lib.editor.editor from flud.protocol.LocalClient import * from flud.FludConfig import FludConfig from flud.CheckboxState import CheckboxState FLUSHCHECKTIME = 5*60 # s to wait to flush fludfile.conf imgdir = os.path.join(os.path.dirname(os.path.abspath(__file__)),'images') mimeMgr = wx.MimeTypesManager() def getFileIcon(file, il, checkboxes, icondict): ft = mimeMgr.GetFileTypeFromExtension(file[file.rfind('.')+1:]) # XXX: what about from mimetype or magic? if ft == None: return icondict['generic'] else: desc = ft.GetDescription() if icondict.has_key(desc): return icondict[desc] else: icon = ft.GetIcon() if icon == None or not icon.Ok(): #print "couldn't find an icon image for %s" % file icondict[desc] = icondict['generic'] return icondict[desc] bm = wx.BitmapFromIcon(icon) newimages = makeCheckboxBitmaps(bm, checkboxes) #il = self.GetImageList() pos = il.GetImageCount() for i in newimages: il.Add(i) icondict[desc] = pos #print "%s got a %s image" % (file, ft.GetDescription()) return pos def getEmptyBitmapAndDC(width, height): empty = wx.EmptyBitmap(width,height) temp_dc = wx.MemoryDC() temp_dc.SelectObject(empty) temp_dc.Clear() return (empty, temp_dc) def makeCheckboxBitmaps(basebitmap, checkboxes): if basebitmap.GetWidth() != 16 or basebitmap.GetHeight() != 16: img = basebitmap.ConvertToImage() img.Rescale(16, 16) basebitmap = img.ConvertToBitmap() result = [] for i in checkboxes: bm, dc = getEmptyBitmapAndDC(40,16) dc.DrawBitmap(basebitmap, 0, 0, False) dc.DrawBitmap(i, 20, 2, False) result.append(bm) return result def createDefaultImageList(): def getDefaultCheckboxes(): ucbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-unchecked1.png"))) cbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checked1.png"))) ccbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checkedpartial1.png"))) cpbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-parentchecked1.png"))) ebm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excluded1.png"))) ecbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excludedpartial1.png"))) return (ucbm, cbm, ccbm, cpbm, ebm, ecbm) checkboxes = getDefaultCheckboxes() il = wx.ImageList(40, 16) folderimgs = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_FOLDER, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) computer = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_HARDDISK, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) drives = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_HARDDISK, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) cdrom = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_CDROM, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) floppy = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_FLOPPY, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) removable = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_REMOVABLE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) genericfile = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_NORMAL_FILE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) execfile = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_EXECUTABLE_FILE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) j = 0 icondict = {} icondict['folder'] = j for i in folderimgs: il.Add(i) j = j+1 icondict['computer'] = j for i in computer: il.Add(i) j = j+1 icondict['drives'] = j for i in drives: il.Add(i) j = j+1 icondict['cdrom'] = j for i in cdrom: il.Add(i) j = j+1 icondict['floppy'] = j for i in floppy: il.Add(i) j = j+1 icondict['removable'] = j for i in removable: il.Add(i) j = j+1 icondict['generic'] = j for i in genericfile: il.Add(i) j = j+1 icondict['exec'] = j for i in execfile: il.Add(i) j = j+1 return il, checkboxes, icondict class DirCheckboxCtrl(wx.TreeCtrl): def __init__(self, parent, id=-1, dir=None, pos=wx.DefaultPosition, size=wx.DefaultSize, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER), validator=wx.DefaultValidator, name=wx.ControlNameStr, allowExclude=True): self.allowExclude = allowExclude wx.TreeCtrl.__init__(self, parent, id, pos, size, style, validator, name) self.listeners = [] self.parent = parent #self.il = self.GetImageList() #self.checkboxes = self.getDefaultCheckboxes() self.initTree(dir) self.Bind(wx.EVT_TREE_ITEM_EXPANDING, self.onExpand, self) self.Bind(wx.EVT_LEFT_UP, self.onClick, self) self.Bind(wx.EVT_TREE_ITEM_GETTOOLTIP, self.onTooltip, self) self.Bind(wx.EVT_CHAR, self.onChar) def initTree(self, dir): self.expandRoot(dir) # XXX: should expandHome() on first run, then load expanded dir state # from saved state on subsequent runs. self.expandHome(dir) def expandRoot(self, dir): if not os.path.isdir(dir): raise ValueError("%s is not a valid directory path") self.defaultImageList, self.checkboxes, self.icondict \ = createDefaultImageList() self.AssignImageList(self.defaultImageList) self.il = self.GetImageList() if dir == None: self.rootID = self.AddRoot(dir, self.icondict['computer'], -1, wx.TreeItemData((dir, True, False, CheckboxState.UNSELECTED))) # XXX: getTopLevelDirs() and add them as children else: self.rootID = self.AddRoot(dir, self.icondict['folder'], -1, wx.TreeItemData((dir, True, False, CheckboxState.UNSELECTED))) self.expandDir(self.rootID) self.Expand(self.rootID) self.stateChangeTime = time.time() self.flushTime = time.time() reactor.callLater(FLUSHCHECKTIME, self.checkFlush) def expandHome(self, dir): home = os.environ['HOME'] if home: traversal = home.split(os.path.sep)[1:] node = self.rootID for d in traversal: (ipath, isdir, expanded, istate) \ = self.GetItemData(node).GetData() self.expandDir(node) children = self.getChildren(node, False) childrennames = [self.GetItemText(x) for x in children] if d in childrennames: p = childrennames.index(d) node = children[p] self.expandDir(node) self.Expand(node) else: print "couldn't traverse to HOME dir on %s" % d break def checkFlush(self): print "checking for flush" if self.stateChangeTime > self.flushTime: self.flushTime = time.time() print "flushing" self.parent.flushFileConfig() reactor.callLater(FLUSHCHECKTIME, self.checkFlush) def expandDir(self, parentID, hideHidden=False, busycursor=True): def isDriveAvailable(path): if len(path) == 2 and path[1] == ':': path = path.lower() if path[0] == 'a' or path[0] == 'b' or diExists(path): return True else: return False return True (path, isDir, expanded, state) = self.GetItemData(parentID).GetData() if expanded: return if not isDriveAvailable(path): return if busycursor: wx.BusyCursor() try: dirlist = os.listdir(path) except: self.SetItemHasChildren(parentID, False) return if len(dirlist) == 0: self.SetItemHasChildren(parentID, False) return dirs = [] files = [] for i in dirlist: if hideHidden and i[0] == '.': # XXX: dotfile format check is *nix specific # XXX: if this is a hidden file, don't add it. pass elif os.path.isdir(os.path.join(path,i)): dirs.append(i) else: files.append(i) dirs.sort() files.sort() for d in dirs: child = self.AppendItem(parentID, d) self.SetPyData(child, (os.path.join(path,d), True, False, 0)) self.SetItemImage(child, self.icondict['folder'], wx.TreeItemIcon_Normal) self.SetItemHasChildren(child) il = self.GetImageList() for f in files: child = self.AppendItem(parentID, f) # XXX: unicode? self.SetPyData(child, (os.path.join(path,f), False, False, 0)) idx = getFileIcon(os.path.join(path,f), il, self.checkboxes, self.icondict) self.SetItemImage(child, idx, wx.TreeItemIcon_Normal) self.SetPyData(parentID, (path, isDir, True, state)) def getStates(self, node=None): if not node: node = self.rootID states = {} (path, isDir, expanded, state) = self.GetItemData(node).GetData() if state in [CheckboxState.SELECTED, CheckboxState.EXCLUDED]: states[path] = state children = self.getChildren(node, False) for child in children: states.update(self.getStates(child)) return states def setStates(self, states): for i in states: found = self.findNode(i) if found: self.setItemState(found, states[i]) def findNode(self, path): if path[0] == '/': path = path[1:] # XXX: unix only traversal = path.split(os.path.sep) if traversal[0] == '': traversal.remove('') node = self.rootID while True: (ipath, isdir, expanded, istate) = self.GetItemData(node).GetData() if len(traversal) == 0: return node self.expandDir(node) children = self.getChildren(node, False) childrennames = [self.GetItemText(x) for x in children] firstpath = traversal[0] if firstpath in childrennames: p = childrennames.index(firstpath) node = children[p] traversal.remove(firstpath) else: #print " the file %s is no longer present!" % path return None return None def onExpand(self, event): self.expandDir(event.GetItem()) self.renderChildren(event.GetItem(), True) def getFullPath(self, node): path = self.tree.GetItemText(node) n = node while True: n = self.tree.GetItemParent(n) if n and n != self.GetRootItem(): path = os.path.join(self.tree.GetItemText(n),path) else: break return path def renderParents(self, item): if item == self.rootID: return n = item (path, isDir, expanded, state) = self.GetItemData(item).GetData() while True: n = self.GetItemParent(n) (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(n).GetData() #print "parent %s" % parentpath if n and n != self.GetRootItem(): newstate = parentstate if parentstate != CheckboxState.UNSELECTED and \ parentstate != CheckboxState.SELECTEDPARENT: # we only care about changing UNSELECT or SELECTEDPARENT # states break else: if state == CheckboxState.SELECTED or \ state == CheckboxState.SELECTEDCHILD or \ state == CheckboxState.SELECTEDPARENT: # if the item (child) is selected in any way, parent # should be too. newstate = CheckboxState.SELECTEDPARENT elif state == CheckboxState.UNSELECTED or \ state == CheckboxState.EXCLUDED: # if the item (child) is unselected or excluded, the # parent should be too, /unless/ there are other # children at the same level who are selected. children = self.getChildren(n, False) newstate = CheckboxState.UNSELECTED for child in children: (cpath, cisdir, cexp, cstate) = self.GetItemData(child).GetData() if cstate == CheckboxState.SELECTED or \ cstate == CheckboxState.SELECTEDCHILD or \ cstate == CheckboxState.SELECTEDPARENT: newstate = parentstate if newstate == parentstate: break imageidx = self.GetItemImage(n) imageidx += CheckboxState.offset(parentstate, newstate) self.SetPyData(n, (parentpath, parentisDir, parentexpanded, newstate)) self.SetItemImage(n, imageidx) else: break def renderChildren(self, parent, recurse=False): (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(parent).GetData() children = self.getChildren(parent, False) for child in children: #path = self.getFullPath(child) (path, isDir, expanded, state) = self.GetItemData(child).GetData() imageidx = self.GetItemImage(child) newstate = state """ Here are the state transitions for children based on current states: ('-' = no state change, 'x' = should never occur, '!' = should be prevented at the parent, '?' = need to consult children) child unsel sel selch selpar excl exclch unsel - ! unsel x - unsel sel selch - - selch - selch par selch selch - - selch - selch selpar x x unsl?selpr x x x excl exlch ! exlch ! - - exclch exlch - exlch ! - - """ #if parentpath == '/data': # print "/data pstate = %d" % parentstate # print " %s = %d" % (path, state) if state == CheckboxState.UNSELECTED: if parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.SELECTEDCHILD: if parentstate == CheckboxState.UNSELECTED: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTEDPARENT: if self.checkChildrenStates(child, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT]): # XXX: did we need to pass in selections to checkChldSt newstate = CheckboxState.SELECTEDPARENT else: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.SELECTEDPARENT: if parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif state == CheckboxState.EXCLUDEDCHILD: if parentstate == CheckboxState.UNSELECTED: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD imageidx += CheckboxState.offset(state, newstate) self.SetPyData(child, (path, isDir, expanded, newstate)) self.SetItemImage(child, imageidx) if recurse: self.renderChildren(child, recurse) # XXX: why do we renderParents here? It hits the same # 'parent's over and over and over again. If we want to do # this, we need to 'collect up' the parents and just call once # -- this kills performance. #print "renderParents(%s)" % path #self.renderParents(child) def getChildren(self, node, recurse=False): result = [] child, cookie = self.GetFirstChild(node) while child: result.append(child) if recurse: result.extend(self.getChildren(child, recurse)) child, cookie = self.GetNextChild(node, cookie) return result def checkChildrenStates(self, node, states, ignorelist=[]): children = self.getChildren(node) for child in children: if child not in ignorelist: (p, d, e, childstate) = self.GetItemData(child).GetData() for state in states: if state == childstate: #print "%s has state %d" % (p, state) return True if self.checkChildrenStates(child, states, ignorelist): # do this even if it is in ignorelist, because it may have # children which are not in the ignorelist return True return False def getTooltip(self, item): text = self.GetItemText(item) (path, isDir, expanded, state) = self.GetItemData(item).GetData() if state == CheckboxState.SELECTED: if isDir: text = "'%s' is SELECTED for backup\n" \ "ALL files within this folder will be backed up\n" \ "(except those explicitly marked for exclusion)" % text else: text = "'%s' is SELECTED for backup" % text elif state == CheckboxState.UNSELECTED: text = "'%s' is NOT selected for backup" % text elif state == CheckboxState.SELECTEDPARENT: text = "some files within '%s' are selected for backup" % text elif state == CheckboxState.SELECTEDCHILD: text = "'%s' will be backed up\n" \ "(one of its parent folders is selected)" % text elif state == CheckboxState.EXCLUDED: if isDir: text = "'%s' is EXCLUDED from backup\n" \ "No files within this folder will be backed up" % text else: text = "'%s' is EXCLUDED from backup" % text elif state == CheckboxState.EXCLUDEDCHILD: text = "'%s' is EXCLUDED from backup\n" \ "(one of its parent folders is EXCLUDED)" % text return text def onTooltip(self, event): item = event.GetItem() text = self.getTooltip(item) if text: event.SetToolTip(text) else: event.StopPropagation() #print dir(event) def onClick(self, event): point = (event.GetX(), event.GetY()) item, flags = self.HitTest(point) if flags & wx.TREE_HITTEST_ONITEMICON: selections = self.GetSelections() self.changeState(item, selections) def onChar(self, event): if event.KeyCode() == ord('F') and event.ShiftDown() \ and event.ControlDown(): self.flushTime = time.time() print "flushing" self.parent.flushFileConfig() event.Skip() def changeState(self, item, selections=[]): self.stateChangeTime = time.time() (path, isDir, expanded, state) = self.GetItemData(item).GetData() if item == self.rootID: parent = None parentstate = CheckboxState.UNSELECTED else: parent = self.GetItemParent(item) (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(parent).GetData() imageidx = self.GetItemImage(item) # determine newstate from existing state, parent state, and state # of children """ Here are the state transitions for the item based on current states and parent states: ('-' = no state change, 'x' = should never occur, '?' = depends on children state) item unsel sel selch selpar excl exclch unsel sel excl sel sel unsel excl sel sel excl?selpar sel x selch excl par selch x excl sel sel selch excl selpar sel excl x sel unsel excl excl x excl x exclch exclch excl exclch x excl x exclch exclch excl """ newstate = state if state == CheckboxState.UNSELECTED: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTEDPARENT: if parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: # XXX: this should be impossible to reach... newstate = CheckboxState.EXCLUDEDCHILD else: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTED: if self.checkChildrenStates(item, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT], selections): newstate = CheckboxState.SELECTEDPARENT elif self.allowExclude: newstate = CheckboxState.EXCLUDED else: if parent in selections or \ (parentstate == CheckboxState.UNSELECTED or \ parentstate == CheckboxState.SELECTEDPARENT): newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif state == CheckboxState.EXCLUDED: if parent in selections or \ (parentstate == CheckboxState.UNSELECTED or \ parentstate == CheckboxState.SELECTEDPARENT): newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD else: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDED if len(selections) > 1: # if we have multiple selections, the idea is to move all the # selections to the newstate defined above, or to valid # unselected or inherited states if the move to newstate would # be invalid. """ Here are the state transitions for the item based on the newstate as determined by the clicked item and the current states: ('-' = no state change, '?' = consult children) item unsel sel selch selpar excl exclch unsel - unsel - - unsel - sel sel - sel sel sel - newstate selch - unsel - - unsel - selpar - unsel - - unsel - excl excl excl?slpr excl excl - excl exclch - unsel - - unsel - """ for i in selections: (mpath, misDir, mexpanded, mstate) = self.GetItemData( i).GetData() mnewstate = mstate if mstate == CheckboxState.UNSELECTED or \ mstate == CheckboxState.SELECTEDCHILD or \ mstate == CheckboxState.SELECTEDPARENT: if newstate == CheckboxState.SELECTED or \ newstate == CheckboxState.EXCLUDED: mnewstate = newstate elif mstate == CheckboxState.SELECTED: if newstate == CheckboxState.UNSELECTED or \ newstate == CheckboxState.SELECTEDCHILD or \ newstate == CheckboxState.SELECTEDPARENT or \ newstate == CheckboxState.EXCLUDEDCHILD: mnewstate = CheckboxState.UNSELECTED elif newstate == CheckboxState.EXCLUDED: if self.checkChildrenStates(i, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT], selections): mnewstate = CheckboxState.SELECTEDPARENT else: mnewstate = newstate elif mstate == CheckboxState.EXCLUDED: if newstate == CheckboxState.UNSELECTED or \ newstate == CheckboxState.SELECTEDCHILD or \ newstate == CheckboxState.SELECTEDPARENT or \ newstate == CheckboxState.EXCLUDEDCHILD: mnewstate = CheckboxState.UNSELECTED elif newstate == CheckboxState.SELECTED: mnewstate = newstate elif mstate == CheckboxState.EXCLUDEDCHILD: if newstate == CheckboxState.EXCLUDED: mnewstate = newstate self.setItemState(i, mnewstate) self.setItemState(item, newstate, (path, isDir, expanded, state, imageidx)) def setItemState(self, item, newstate, oldData=None): if oldData: (path, isDir, expanded, state, imageidx) = oldData else: (path, isDir, expanded, state) = self.GetItemData(item).GetData() imageidx = self.GetItemImage(item) imageidx += CheckboxState.offset(state, newstate) self.SetPyData(item, (path, isDir, expanded, newstate)) self.SetItemImage(item, imageidx) self.renderChildren(item, True) self.renderParents(item) def getTopLevelDrives(self): sys = platform.system() if sys == 'Windows': # XXX: need to test this all out import win32api, string drives = win32api.GetLogicalDriveStrings() driveletters = string.splitfields(drives,'\000') for d in driveletters: type = win32api.GetDriveType("%s:\\" % d) # XXX: set the appropriate icon return driveletters else: # Unix, OSX, etc. return ['/'] def addListener(self, callback): self.listeners.append(callback) def SetPyData(self, item, data): wx.TreeCtrl.SetPyData(self, item, data) for f in self.listeners: f(item, data) """ Tests for DirCheckboxCtrl A number of unit tests must be performed on the DirCheckboxGUI widget when refactoring. Add to this list so that it becomes comprehensive. Basic Tests: 1. Click on a top-level UNSELECTED object in the tree [should become SELECTED]. - Click again [should become EXCLUDED]. - Click again [should become UNSELECTED]. 2. Click on a non-top-level UNSELECTED object in the tree that has no SELECTED children [should become SELECTED, it's parents should become SELECTEDPARENT and its children SELECTEDCHILD]. - Click again [should become EXCLUDED, it's parents who were SELECTEDPARENT should become UNSELECTED, and it's UNSELECTED children should become EXCLUDED]. - Click again [should become UNSELECTED, and it's children should become UNSELECTED]. 3. Change two children to their SELECTED state [parents should be in SELECTEDPARENT state]. - Click one child to become EXCLUDED [parents should stay in SELECTEDPARENT] - Click the same child to become UNSELECTED [parents should stay in SELECTEDPARENT] - Click the other child to become EXCLUDED [parents should become UNSELECTED] 4. Choose a folder and a child item. - Click the child to become SEL [parent should be SELPAR] - Click the parent [parent should become SEL] - Click the parent again [parent should become SELPAR] 5. Choose a folder and a child item. - Click the parent to become SEL [child should become SELCHILD] - Click the child [child should become SEL] - Click the child again [child should become EXCL] - Click the child again [child should become SELCHILD] 6. Pick a node with children at least two-deep. Change two of the at-least-two-deep children to their SELECTED state [parents should be in SELECTEDPARENT state]. - Click parent closest to SELECTED children to SELECTED [two childen remain in SELECTED, all other children become SELECTEDCHILD. Parent[s] of parent remain SELECTEDPARENT] - Click one child twice to become SELECTEDCHILD [child should not be able to be UNSELECTED, parent states should not change] - Click other child twice to become SELECTEDCHILD [child should not be able to be UNSELECTED, parent states should not change] 7. Pick a node with children at least two-deep. - Click deepest parent to SELECTED [Parent[s] of parent become SELECTEDPARENT] - Click same parent again to become EXCLUDED [Parent[s] of parent become UNSELECTED] - Click same parent again to become UNSELECTED [Parent[s] of parent remain UNSELECTED] 8. Pick a node with children at least two-deep. - Click deepest child to become SELECTED [Parent[s] of parent become SELECTEDPARENT] - Click the topmost parent to become SELECTED [children become SELECTEDCHILD] - Click the topmost parent again to become SELECTEDPARENT [middle child should become SELECTEDPARENT] Multi-Selection Tests: 1. Multi-select three items at the same level and in the same state. Toggle between the three main states [SELECTED, EXCLUDED, UNSELECTED] 2. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the SEL item to see that all three items become EXCL. 3. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the EXCL item to see that all three items become UNSEL. 4. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the UNSEL item to see that all three items become SEL. 5. Choose three items that are nested within each other: a parent folder, one of its children folders, and a file/folder in the child folder. Choose one other item from the child folder. - set the top parent to UNSEL - set the child folder to SEL [parent become SELPAR] - set the child item to SEL - set the other item to EXCL - multi-select all four items - 5A. click on the top parent (which was in SELPAR) [All four items should become SEL, all children of any of these items should become SELCHILD]. Toggle twice more [all selected items should toggle to EXCL, then to UNSEL] - 5B. reset as above, click on the child folder [All four items should become EXCL]. Toggle twice more [all selected items should go to UNSEL, then SEL] - 5C. reset as above, click on the child item [All four items should become EXCL]. Toggle twice more [all selected items should go to UNSEL, then SEL] - 5D. reset as above, click on the other item [All four items should become UNSEL]. Toggle twice more [all selected items should go to SEL, then EXCL] 6. Choose a folder, one if its subfolders, a subfolder of the subfolder, and an item in the deepest subfolder, and an item in the first subfolder, e.g.: [] A [] B [] C [] D [] E - change item 'D' to SEL [parents 'A', 'B', and 'C' should go to SELPAR] - change item 'E' to EXCL - multi-select 'A', 'C', and 'E' - toggle 'E' to UNSEL [all other selections should stay in current state] - toggle 'E' to SEL ['A' and 'B' become SEL, their children become SELCHILD] - toggle 'E' back to EXCL [should get our original multi-select setup back] - toggle 'C' to SEL [all selections to SEL, children to SELCHILD] - toggle 'C' to SELPAR ['A' and 'C' to SELPAR, 'E' to UNSEL] - toggle 'E' twice [should get our original mulit-select setup back] """ class CheckFileListCtrlMixin: # for some insane reason, we can't get EVT_LEFT_DOWN (or _UP) to bind in # FileListCtrl itself. But we are sneaky and can do it by lots of clever # hax0ry, like by using this silly mixin. def __init__(self, toCall): self.Bind(wx.EVT_LEFT_UP, toCall) class FileListCtrl(wx.ListCtrl, listmix.ListCtrlAutoWidthMixin, CheckFileListCtrlMixin): """ Implements a file list control, with a peerctrl that contains the filesystem model. Currently, this peerctrl must implement an addListener(), changeState(), GetItemData(), expandDir(), GetSelections(), and GetChildren() API similar to that implemented by DirCheckBoxCtrl. """ def __init__(self, parent, peerctrl, id=-1, pos=wx.DefaultPosition, size=wx.DefaultSize, style=wx.LC_REPORT, validator=wx.DefaultValidator, name=wx.ListCtrlNameStr): wx.ListCtrl.__init__(self, parent, id, pos, size, style, validator, name) CheckFileListCtrlMixin.__init__(self, self.OnClick) listmix.ListCtrlAutoWidthMixin.__init__(self) self.peerctrl = peerctrl self.peerctrl.addListener(self.itemChanged) self.itemdict = {} # a dict with filepath as key, containing tuples of # (index into ListCtrl, reference to peerctrl object) self.stopsearch = False self.il, self.checkboxes, self.icondict = createDefaultImageList() self.AssignImageList(self.il, wx.IMAGE_LIST_SMALL) self.il = self.GetImageList(wx.IMAGE_LIST_SMALL) self.InsertColumn(0, "Filename") self.InsertColumn(1, "Location") #self.InsertColumn(2, "Last Backup") #self.SetColumnWidth(0, wx.LIST_AUTOSIZE) #self.SetColumnWidth(1, -1) #wx.LIST_AUTOSIZE) self.Bind(wx.EVT_MOTION, self.mouseMotion) self.searchSourceItems = [] def itemChanged(self, item, data): (path, isDir, expanded, state) = data if self.itemdict.has_key(path): item = self.itemdict[path][0] image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) def GetAll(self, excludeStates=[]): result = [] start = -1 for i in range(self.GetItemCount()): item = self.GetNextItem(start, wx.LIST_NEXT_ALL) # XXX: only append if not in excludeStates result.append(item) start = item return result def GetSelections(self): result = [] start = -1 for i in range(self.GetSelectedItemCount()): item = self.GetNextItem(start, wx.LIST_NEXT_ALL, wx.LIST_STATE_SELECTED) result.append(item) start = item return result def GetPeerSelections(self, selections): result = [] for item in selections: path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) if self.itemdict.has_key(path): result.append(self.itemdict[path][1]) return result def mouseMotion(self, event): point = event.GetPosition() item, flags = self.HitTest(point) if flags == wx.LIST_HITTEST_ONITEMICON: path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) text = self.peerctrl.getTooltip(self.itemdict[path][1]) tip = wx.ToolTip(text) self.SetToolTip(tip) #tipwin = tip.GetWindow() #tippos = tipwin.GetPosition() #print "%s vs %s" % (tippos, point) #tipwin.SetPosition(point) def OnClick(self, event): point = event.GetPosition() item, flags = self.HitTest(point) if flags == wx.LIST_HITTEST_ONITEMICON: peerselections = self.GetPeerSelections(self.GetSelections()) path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) ditem = self.itemdict[path][1] # raises if not present self.peerctrl.changeState(ditem, peerselections) (path, isDir, expanded, state) \ = self.peerctrl.GetItemData(ditem).GetData() image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) def searchButtonAction(self, event): selections = self.peerctrl.GetSelections() if len(selections) == 0: return ("Please tell me where to search. Select one or more" " folders in the left-hand panel (hold down SHIFT or" " CTRL for multiple selection), then click the 'find!'" " button again.", None) else: self.DeleteAllItems() self.itemdict = {} b = wx.BusyCursor() searchSourceItems = [] for i in selections: self.addResults(i, event.searchstring) searchSourceItems.append(i) self.searchSourceItems = [self.peerctrl.GetItemData(s).GetData()[0] for s in searchSourceItems] print "sources: %s" % self.searchSourceItems return ("Search results will appear as files that match your" " search are found.", None) return (None, None) def addResults(self, ditem, searchstring): (path, isDir, expanded, state) \ = self.peerctrl.GetItemData(ditem).GetData() position = self.GetItemCount() if isDir: if not expanded: self.peerctrl.expandDir(ditem, busycursor=False) children = self.peerctrl.getChildren(ditem) for c in children: self.addResults(c, searchstring) wx.Yield() if self.stopsearch: break else: terms = [x for x in searchstring.split(' ') if x != ''] for term in terms: print path if path.find(term) > 0: image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state dirname, filename = os.path.split(path) index = self.InsertImageStringItem(position, filename, image) self.SetStringItem(index, 1, dirname) self.SetColumnWidth(0, wx.LIST_AUTOSIZE) self.itemdict[path] = (index, ditem) break def setGroup(self, state): items = self.GetAll() item = items[0] peerselections = self.GetPeerSelections(items) path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) ditem = self.itemdict[path][1] # raises if not present while True: # cycle until the items state matches the desired state self.peerctrl.changeState(ditem, peerselections) # can be slow (path, isDir, expanded, nstate) \ = self.peerctrl.GetItemData(ditem).GetData() if nstate == state: break image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) return self.searchSourceItems class GroupSelectionCheckbox(wx.Panel): def __init__(self, parent, id=-1, setGroupState=None): wx.Panel.__init__(self, parent, id) self.setGroupState = setGroupState self.ubm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-unchecked1.png"))) self.cbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checked1.png"))) self.ebm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excluded1.png"))) self.checkboxButton = wx.BitmapButton(self, -1, self.ubm, style=wx.NO_BORDER) self.Bind(wx.EVT_BUTTON, self.onCheckbox, self.checkboxButton) self.description = wx.StaticText(self, -1, "always BACKUP any files that match these search criteria ") self.state = CheckboxState.UNSELECTED self.gbSizer = wx.GridBagSizer(1,2) self.gbSizer.Add(self.checkboxButton, (0,0), flag=wx.ALIGN_CENTER) self.gbSizer.Add(self.description, (0,1), flag=wx.ALIGN_CENTER) self.gbSizer.AddGrowableRow(1) self.SetSizerAndFit(self.gbSizer) def Enable(self, enable=True): self.checkboxButton.Enable(enable) self.description.Enable(enable) def Disable(self): self.Enable(False) def clear(self): self.checkboxButton.SetBitmapLabel(self.ubm) self.state = CheckboxState.UNSELECTED self.description.SetLabel( "always BACKUP any files that match these search criteria") def setState(self, state): self.state = state if self.state == CheckboxState.UNSELECTED: self.checkboxButton.SetBitmapLabel(self.ubm) self.description.SetLabel( "always BACKUP any files that match these search criteria") elif self.state == CheckboxState.SELECTED: self.checkboxButton.SetBitmapLabel(self.cbm) self.description.SetLabel( "always BACKUP any files that match these search criteria") elif self.state == CheckboxState.EXCLUDED: self.checkboxButton.SetBitmapLabel(self.ebm) self.description.SetLabel( "always EXCLUDE any files that match these search criteria") def onCheckbox(self, event): if self.state == CheckboxState.UNSELECTED: self.checkboxButton.SetBitmapLabel(self.cbm) self.state = CheckboxState.SELECTED if self.setGroupState: self.setGroupState(CheckboxState.SELECTED) elif self.state == CheckboxState.SELECTED: self.checkboxButton.SetBitmapLabel(self.ebm) self.description.SetLabel( "always EXCLUDE any files that match these search criteria") self.state = CheckboxState.EXCLUDED if self.setGroupState: self.setGroupState(CheckboxState.EXCLUDED) elif self.state == CheckboxState.EXCLUDED: self.checkboxButton.SetBitmapLabel(self.ubm) self.description.SetLabel( "always BACKUP any files that match these search criteria") self.state = CheckboxState.UNSELECTED if self.setGroupState: self.setGroupState(CheckboxState.UNSELECTED) class SearchPanel(wx.Panel): def __init__(self, parent, dircheckbox, id=-1, searchButtonAction=None): wx.Panel.__init__(self, parent, id) self.dircheckbox = dircheckbox self.searchButtonAction = searchButtonAction self.SetAutoLayout(False) self.rules = {} # should refer to something from fludrules self.searchField = wx.TextCtrl(self, -1, "search for files to backup here", size=wx.Size(-1,-1), style=wx.TE_PROCESS_ENTER) self.searchField.SetToolTipString('find files within directories' ' selected to the left by entering search terms here') self.searchField.Bind(wx.EVT_TEXT_ENTER, self.onSearchClick) self.searchField.Bind(wx.EVT_LEFT_DOWN, self.selectAllText) self.searchField.Bind(wx.EVT_KILL_FOCUS, self.unfocused) self.searchButton = wx.Button(self, -1, 'find!', name='searchButton') self.Bind(wx.EVT_BUTTON, self.onSearchClick, self.searchButton) self.searchResults = FileListCtrl(self, dircheckbox, -1, name='searchResults', style=wx.SUNKEN_BORDER | wx.LC_REPORT) self.searchResults.SetExtraStyle(0) self.searchResults.SetLabel('found files') self.groupSelection = GroupSelectionCheckbox(self, -1, self.setGroup) self.groupSelection.Disable() self.gbSizer = wx.GridBagSizer(3,2) self.gbSizer.Add(self.searchField, (0,0), flag=wx.EXPAND) self.gbSizer.Add(self.searchButton, (0,1)) self.gbSizer.Add(self.searchResults, (1,0), (1,2), flag=wx.EXPAND|wx.TOP, border=5) self.gbSizer.Add(self.groupSelection, (2,0) ) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) def onSearchClick(self, event): event.searchstring = self.searchField.GetValue() if self.searchButton.GetLabel() == 'stop!': self.searchButton.SetLabel('find!') self.searchResults.stopsearch = True return else: self.groupSelection.clear() self.groupSelection.Disable() self.searchButton.SetLabel('stop!') self.searchButton.Update() err, info = self.searchResults.searchButtonAction(event) selections = self.searchResults.searchSourceItems # see if we should set the checkbox button from a previous rule state = None if len(selections) > 0 and self.rules.has_key(selections[0]): rule = self.rules[selections[0]] if self.rules[selections[0]].has_key(event.searchstring): state = self.rules[selections[0]][event.searchstring] for i in selections: if not self.rules.has_key(i) or self.rules[i] != rule: state = None break #for j in self.rules[i]: if state: print "should restore checkbox to %s" % state self.groupSelection.setState(state) self.searchButton.SetLabel('find!') self.searchResults.stopsearch = False if self.searchButtonAction: self.searchButtonAction(event, errmsg=err, infomsg=info) self.groupSelection.Enable() def selectAllText(self, event): if wx.Window.FindFocus() != self.searchField: self.searchField.SetSelection(-1,-1) self.searchField.SetFocus() else: self.searchField.SetSelection(0,0) event.Skip() def unfocused(self, event): self.searchField.SetSelection(0,0) def setGroup(self, state): b = wx.BusyCursor() selections = self.searchResults.setGroup(state) for s in selections: if not self.rules.has_key(s): self.rules[s] = {} if state == CheckboxState.UNSELECTED: try: self.rules.pop(s) except: pass else: self.rules[s][self.searchField.GetValue()] = state print self.rules class FilePanel(wx.SplitterWindow): def __init__(self, parent, searchButtonAction=None): # Use the WANTS_CHARS style so the panel doesn't eat the Return key. wx.SplitterWindow.__init__(self, parent, -1, style=wx.SP_LIVE_UPDATE | wx.CLIP_CHILDREN | wx.WANTS_CHARS) self.Bind(wx.EVT_SIZE, self.OnSize) self.SetNeedUpdating(True) self.tree = DirCheckboxCtrl(self, -1, dir="/") # XXX: fludrules.init path should be in config self.fludrules = self.getFludHome()+"/fludrules.init" if not os.path.isfile(self.fludrules): # XXX: do the other first time stuff (email encrypted credentials, # etc.) parent.SetMessage("Welcome. This appears to be the first" " time you've run flud. We've automatically selected some" " files for backup. You can make changes by" " selecting/deselecting files and directories. When you are" " done, simply close this window.") src = open('fludrules.init', 'r') dst = open(self.fludrules, 'w') filerules = src.read() dst.write(filerules) dst.close() src.close() filerules = eval(filerules) rulestates = {} for rule in filerules['baserules']: value = filerules['baserules'][rule] rule = glob.glob(os.path.expandvars(rule)) for r in rule: rulestates[r] = value self.tree.setStates(rulestates) # XXX: fludfile.conf path should be in config self.fludfiles = self.getFludHome()+"/fludfile.conf" print self.fludfiles if os.path.isfile(self.fludfiles): file = open(self.fludfiles, 'r') states = eval(file.read()) self.tree.setStates(states) file.close() self.searchPanel = SearchPanel(self, dircheckbox=self.tree, searchButtonAction=searchButtonAction) self.SetMinimumPaneSize(20) self.SplitVertically(self.tree, self.searchPanel) #, 300) def getFludHome(self): if os.environ.has_key('FLUDHOME'): fludhome = os.environ['FLUDHOME'] else: fludhome = os.environ['HOME']+"/.flud" if not os.path.isdir(fludhome): os.mkdir(fludhome, 0700) return fludhome def shutdown(self, event): self.flushFileConfig() event.Skip() def flushFileConfig(self): states = self.tree.getStates() f = open(self.fludfiles, 'w') f.write(str(states)) f.close() for i in states: print "%s %s" % (i, states[i]) def OnSize(self, event): w,h = self.GetClientSizeTuple() if self.tree: self.tree.SetDimensions(0, 0, w, h) event.Skip() class RestoreCheckboxCtrl(DirCheckboxCtrl): # XXX: child/parent selection/deselection isn't quite right still, esp wrt # root node. repro: # -/ # -d1 # -f1 # -d2 # -d3 # -f2 # -f3 # with nothing selected, select d3 and f3, then select root, then deselect # d3 and f3 def __init__(self, parent, id=-1, config=None, pos=wx.DefaultPosition, size=wx.DefaultSize, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER), validator=wx.DefaultValidator, name=wx.ControlNameStr): self.config = config DirCheckboxCtrl.__init__(self, parent, id, config, pos, size, style, validator, name, allowExclude=False) def initTree(self, config): self.expandRoot(config) self.expandUntilMultiple() def expandRoot(self, config): self.defaultImageList, self.checkboxes, self.icondict \ = createDefaultImageList() self.AssignImageList(self.defaultImageList) self.il = self.GetImageList() self.rootID = self.AddRoot("/", self.icondict['computer'], -1, wx.TreeItemData(("", True, False, CheckboxState.UNSELECTED))) self.update() def expandUntilMultiple(self): node = self.rootID while True: (ipath, isdir, expanded, istate) = self.GetItemData(node).GetData() children = self.getChildren(node, False) if len(children) > 1 or len(children) == 0: break; node = children[0] self.Expand(node) def update(self): master = listMeta(self.config) for i in master: if not isinstance(master[i], dict): traversal = i.split(os.path.sep) node = self.rootID path = "/" if traversal[0] == '': traversal.remove('') for n in traversal: path = os.path.join(path, n) children = self.getChildrenDict(node) if n == traversal[-1] and not n in children: child = self.AppendItem(node, n) self.SetPyData(child, (path, False, False, 0)) idx = getFileIcon(i, self.il, self.checkboxes, self.icondict) self.SetItemImage(child, idx, wx.TreeItemIcon_Normal) else: if not n in children: child = self.AppendItem(node, n) self.SetPyData(child, (path, True, False, 0)) self.SetItemImage(child, self.icondict['folder'], wx.TreeItemIcon_Normal) else: child = children[n] node = child self.Expand(self.rootID) def getChildrenDict(self, node): result = {} child, cookie = self.GetFirstChild(node) while child: result[self.GetItemText(child)] = child child, cookie = self.GetNextChild(node, cookie) return result def onExpand(self, event): pass def getSelected(self, startNode=None): if not startNode: startNode = self.rootID children = self.getChildren(startNode) selected = [] for n in children: (path, isDir, expanded, state) = self.GetItemData(n).GetData() if not isDir \ and (state == CheckboxState.SELECTED \ or state == CheckboxState.SELECTEDCHILD): selected.append(n) if isDir and (state == CheckboxState.SELECTED \ or state == CheckboxState.SELECTEDPARENT \ or state == CheckboxState.SELECTEDCHILD): selected += self.getSelected(n) return selected class RestorePanel(wx.Panel): def __init__(self, parent, config, factory): self.config = config self.factory = factory wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) self.tree = RestoreCheckboxCtrl(self, -1, config, #wx.TreeCtrl(self, -1, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER)) self.restoreButton = wx.Button(self, -1, 'restore selected files', name='restoreButton') self.Bind(wx.EVT_BUTTON, self.onRestoreClick, self.restoreButton) self.gbSizer = wx.GridBagSizer(2,1) self.gbSizer.Add(self.tree, (0,0), flag=wx.EXPAND|wx.ALL, border=0) self.gbSizer.Add(self.restoreButton, (1,0), flag=wx.EXPAND|wx.ALL, border=0) self.gbSizer.AddGrowableRow(0) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) def update(self): self.tree.update() def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() def onTooltip(self, event): pass def onRestoreClick(self, event): for n in self.tree.getSelected(): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "restoring %s" % path d = self.factory.sendGETF(path) d.addCallback(self.restored, n) d.addErrback(self.restoreFailed, n) self.tree.UnselectAll() def restored(self, res, n): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "yay, %s" % path self.tree.SetItemTextColour(n, '#005804') self.tree.changeState(n) def restoreFailed(self, err, n): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "boo, %s: %s" % (path, err) self.tree.SetItemTextColour(n, wx.RED) self.tree.changeState(n) class SchedulePanel(wx.Panel): def __init__(self, parent): wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() class FeedbackPanel(wx.Panel): def __init__(self, parent): wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) editor = wx.lib.editor.editor.Editor(parent, -1) def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() class FludNotebook(wx.Notebook): def __init__(self, parent, id=-1, pos=wx.DefaultPosition, size=wx.DefaultSize, style=wx.NB_BOTTOM|wx.NO_BORDER): self.parent = parent self.config = parent.config self.factory = LocalClientFactory(self.config) print "connecting to localhost:%d" % self.config.clientport reactor.connectTCP('localhost', self.config.clientport, self.factory) wx.Notebook.__init__(self, parent, id, pos, style=style) self.filePanel = FilePanel(self, searchButtonAction=parent.searchButtonAction) self.AddPage(self.filePanel, "Backup Files") self.restorePanel = RestorePanel(self, self.config, self.factory) self.AddPage(self.restorePanel, "Restore") self.schedulePanel = SchedulePanel(self) self.AddPage(self.schedulePanel, "Backup Schedule") self.feedbackPanel = FeedbackPanel(self) self.AddPage(self.feedbackPanel, "Feedback") self.Bind(wx.EVT_NOTEBOOK_PAGE_CHANGED, self.changedPage) def shutdown(self, event): self.filePanel.shutdown(event) def changedPage(self, event): page = event.GetSelection() if page == 0: self.SetMessage("Select files and directories for backup" " with the filesystem view on the left, or set up criteria" " for finding files for backup directly with simple" " searches, below right.") elif page == 1: self.SetMessage("Select files/directories to be restored to" " your computer, then click on 'restore!' Files will turn" " green as they arrive.") self.restorePanel.update() elif page == 2: self.SetMessage("Configure how often your computer should backup." "\n (not implemented)") elif page == 3: self.SetMessage("Send feedback to flud programmers. (not" " implemented)") def SetMessage(self, msg): self.parent.SetMessage(msg) class FludLogoPanel(wx.Panel): def __init__(self, parent, id=-1, pos=wx.DefaultPosition, size=wx.Size(10,10), style=wx.TAB_TRAVERSAL, name="logo panel"): wx.Panel.__init__(self, parent, id, pos, size, style, name) self.SetAutoLayout(True) self.SetBackgroundColour(wx.BLACK) self.SetForegroundColour(wx.WHITE) logobmp = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "flud-backup-logo-1-150-nodrop.png"))) pad = 0 self.logowidth = logobmp.GetWidth() self.logoheight = logobmp.GetHeight() self.logo = wx.StaticBitmap(self, -1, logobmp) self.messagePanel = wx.Panel(self, -1) self.messagePanel.SetBackgroundColour(wx.BLACK) self.messagePanel.SetForegroundColour(wx.WHITE) self.message = wx.StaticText(self.messagePanel, -1, "message text area", style=wx.ALIGN_CENTER|wx.ST_NO_AUTORESIZE, size=(-1, self.logoheight-15)) self.message.Bind(wx.EVT_SIZE, self.resizeMessage) self.bsizer = wx.BoxSizer(wx.VERTICAL) self.bsizer.Add(self.message, flag=wx.EXPAND|wx.ALL, border=35) self.bsizer.SetSizeHints(self.messagePanel) self.messagePanel.SetSizer(self.bsizer) self.gbSizer = wx.GridBagSizer(1,2) self.gbSizer.Add(self.logo, (0,0)) self.gbSizer.Add(self.messagePanel, (0,1), flag=wx.EXPAND|wx.ALL) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(1) self.SetSizerAndFit(self.gbSizer) self.SetSize(wx.Size(self.logowidth, self.logoheight)) self.SetSizeHints(self.logowidth, self.logoheight, -1, self.logoheight) def SetMessage(self, msg): (w,h) = self.message.GetSizeTuple() #print "msg area size is %d x %d" % (w,h) self.message.SetLabel(msg) self.message.Wrap(w) #print "msg is '%s'" % self.message.GetLabel() self.message.Center() def resizeMessage(self, evt): # this is mainly to deal with StaticText wonkiness (not calling Wrap() # automatically, not centering properly automatically). It may be # possible to get rid of this with a future wxPython release. (w,h) = self.message.GetSizeTuple() self.message.Wrap(w) m = self.message.GetLabel() m = m.replace('\n',' ') self.message.SetLabel(m) self.message.Wrap(w) self.message.Center() class FludFrame(wx.Frame): def __init__(self, parent, id=wx.ID_ANY, label="flud bakcup client", size=wx.Size(800,600), style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE, config=None): wx.Frame.__init__(self, parent, id, label, size=size, style=style) wx.ToolTip.SetDelay(2000) self.clearMessage = False self.logoPanel = FludLogoPanel(self) self.SetMessage('Welcome.') self.config = config self.notebook = FludNotebook(self) self.operationStatus = wx.StatusBar(name='operationStatus', parent=self, style=0) self.SetStatusBar(self.operationStatus) self.gbSizer = wx.GridBagSizer(2,1) self.gbSizer.Add(self.logoPanel,(0,0), flag=wx.EXPAND) self.gbSizer.Add(self.notebook, (1,0), flag=wx.EXPAND|wx.ALL, border=1) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) self.Bind(wx.EVT_CLOSE, self.shutdown) self.SetSize(size) self.Show(True) def SetMessage(self, message): self.logoPanel.SetMessage(message) def shutdown(self, event): self.notebook.shutdown(event) def searchButtonAction(self, event, errmsg=None, infomsg=None): if errmsg: self.logoPanel.SetMessage(errmsg) self.clearMessage = True elif infomsg: self.logoPanel.SetMessage(infomsg) self.clearMessage = False elif self.clearMessage: self.logoPanel.SetMessage("") #if __name__ == '__main__': # app = wx.PySimpleApp() # # config = FludConfig() # config.load(doLogging=False) # # f = FludFrame(None, wx.ID_ANY, 'flud backup client', size=(795,600), # style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE, # config=config) # # from twisted.internet import reactor # reactor.registerWxApp(app) # reactor.run()
Python
""" FludConfig.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. manages configuration file for flud backup. """ import os, sys, socket, re, logging import ConfigParser import flud.FludCrypto as FludCrypto from flud.FludCrypto import FludRSA from flud.FludkRouting import kRouting from flud.fencode import fencode, fdecode """ default mapping of relative URLs """ def_commandmap = { 'ID': 'ID', 'GROUPID': 'GROUPID', 'STORE': 'STORE', 'RETRIEVE': 'RETRIEVE', 'VERIFY': 'VERIFY', 'PROXY': 'PROXY', 'DELETE': 'DELETE', 'kFINDNODE': 'kFINDNODE', 'kFINDVAL': 'kFINDVAL', 'kSTORE': 'kSTORE' } logger = logging.getLogger('flud') CLIENTPORTOFFSET = 500 def_trustdeltas = { 'NODE_INITIAL_SCORE': 1, 'NODE_POSITIVE_CAP': 100, 'NODE_NEGATIVE_CAP': -100, 'PUT_SUCCEED_REWARD': 2, 'GET_SUCCEED_REWARD': 4, 'VRFY_SUCCEED_REWARD': 4, 'FNDN_FAIL_PENALTY': -1, 'PUT_FAIL_PENALTY': -2, 'GET_FAIL_PENALTY': -10, 'VRFY_FAIL_PENALTY': -10 } class FludDebugLogFilter(logging.Filter): """ Keeps all logging levels defined by loggers, but ups level to DEBUG for loggers whose namespaces match patterns given by wildcards. """ # XXX: doesn't really interact with all logging levels by all loggers, only # with the one defined by the root logger. If children have stricter # loglevels set, this filter won't ever get called on them. def __init__(self, wildcardStrings): self.setWildcards(wildcardStrings) root = logging.getLogger("") if hasattr(root, 'fludDebugLogLevel'): self.effectiveLevel = root.fludDebugLogLevel else: self.effectiveLevel = root.getEffectiveLevel() self.fludDebugLogLevel = root.getEffectiveLevel() root.setLevel(logging.NOTSET) def setWildcards(self, wildcardStrings): self.wildcards = [] if not isinstance(wildcardStrings, list): wildcardStrings = [wildcardStrings] for s in wildcardStrings: self.setWildcard(s) def setWildcard(self, wildcardString): fields = wildcardString.split('.') for i, s in enumerate(fields): #print "%s:%s" % (i, s) if "*" == s: fields[i] = r'[\w.]*' else: try: if s.index(s, '*') > 0: fields[i] = s.replace('*', r'[\w]*') except: pass regex = "^%s$" % r'\.'.join(fields) self.wildcards.append(re.compile(regex)) def filter(self, record): if record.levelno >= self.effectiveLevel: return 1 for w in self.wildcards: m = w.match(record.name) if m: return 1 return 0 # XXX: refactor out the try/except stuff that could be done with has_key() class FludConfig: """ Handles configuration for Flud nodes. Most persistent settings live in this class. Configuration is kept in the directory specified by FLUDHOME if this value is set in the environment, otherwise in HOME/.flud/. If no existing configuration exists, this object will create a configuration with sane default values. """ def __init__(self): self.Kr = 0 self.Ku = 0 self.nodeID = 0 self.groupIDr = 0 self.groupIDu = 0 self.port = -1 self.commandmap = {} self.trustdeltas = {} self.reputations = {} self.nodes = {} try: self.fludhome = os.environ['FLUDHOME'] except: try: home = os.environ['HOME'] self.fludhome = home+"/.flud" except: logger.warn("cannot determine FLUDHOME.") logger.warn("Please set HOME or FLUDHOME environment variable") if os.path.isdir(self.fludhome) == False: os.mkdir(self.fludhome, 0700) self.fludconfig = self.fludhome+"/flud.conf" self.configParser = ConfigParser.ConfigParser() if os.path.isfile(self.fludconfig) == False: conffile = file(self.fludconfig, "w") else: conffile = file(self.fludconfig, "r") self.configParser.readfp(conffile) conffile.close() logger.info('fludhome = %s' % self.fludhome) logger.info('fludconfig = %s' % self.fludconfig) def load(self, serverport=None, doLogging=True): """ If serverport is given, it overrides any value that may be in the configuration file """ self.logfile, self.loglevel = self._getLoggingConf() if doLogging: if os.path.isfile(self.logfile): os.remove(self.logfile) handler = logging.FileHandler(self.logfile) formatter = logging.Formatter('%(asctime)s %(filename)s:%(lineno)d' ' %(name)s %(levelname)s: %(message)s', datefmt='%H:%M:%S') handler.setFormatter(formatter) logger.addHandler(handler) logging.getLogger("").setLevel(self.loglevel) #logger.setLevel(self.loglevel) #logger.setLevel(logging.WARNING) # XXX: overrides user prefs #logger.setLevel(logging.DEBUG) # XXX: overrides user prefs if os.environ.has_key("LOGFILTER"): self.filter = FludDebugLogFilter( os.environ["LOGFILTER"].split(' ')) handler.addFilter(self.filter) # XXX: add a LocalPrimitive that can be called dynamically to # invoke filter.setWildcards() self.Kr, self.Ku, self.nodeID, self.groupIDr, self.groupIDu \ = self._getID() logger.debug('Kr = %s' % self.Kr.exportPrivateKey()) logger.debug('Ku = %s' % self.Ku.exportPublicKey()) logger.debug('nodeID = %s' % self.nodeID) logger.debug('groupIDr = %s' % self.groupIDr) logger.debug('groupIDu = %s' % self.groupIDu) self.port, self.clientport, self.commandmap, self.trustdeltas \ = self._getServerConf() if serverport != None: self.port = serverport self.clientport = serverport + CLIENTPORTOFFSET self.configParser.set("server","port",self.port) self.configParser.set("server","clientport",self.clientport) logger.debug('port = %s' % self.port) logger.debug('clientport = %s' % self.clientport) logger.debug('commandmap = %s' % self.commandmap) self.routing = kRouting((socket.getfqdn(), self.port, long(self.nodeID, 16), self.Ku.exportPublicKey()['n'])) self.storedir, self.generosity, self.minoffer = self._getStoreConf() if os.path.isdir(self.storedir) == False: os.mkdir(self.storedir) os.chmod(self.storedir, 0700) logger.debug('storedir = %s' % self.storedir) self.kstoredir = self._getkStoreConf() if os.path.isdir(self.kstoredir) == False: os.mkdir(self.kstoredir) os.chmod(self.kstoredir, 0700) logger.debug('kstoredir = %s' % self.kstoredir) self.clientdir = self._getClientConf() if os.path.isdir(self.clientdir) == False: os.mkdir(self.clientdir) os.chmod(self.clientdir, 0700) logger.debug('clientdir = %s' % self.clientdir) self.metadir, self.metamaster = self._getMetaConf() if os.path.isdir(self.metadir) == False: os.mkdir(self.metadir) os.chmod(self.metadir, 0700) logger.debug('metadir = %s' % self.metadir) self.reputations = self._getReputations() logger.debug("reputations = %s" % str(self.reputations)) self.nodes = self._getKnownNodes() logger.debug("known nodes = %s" % str(self.nodes)) self.save() os.chmod(self.fludconfig, 0600) self.loadMasterMeta() def save(self): """ saves configuration """ conffile = file(self.fludconfig, "w") self.configParser.write(conffile) conffile.close() def _getLoggingConf(self): """ Returns logging configuration: logfile and loglevel """ if (self.configParser.has_section("logging") == False): self.configParser.add_section("logging") try: logfile = int(self.configParser.get("logging","logfile")) except: logger.debug("no logfile specified, using default") logfile = self.fludhome+'/flud.log' self.configParser.set("logging", "logfile", logfile) try: loglevel = int(self.configParser.get("logging","loglevel")) #loglevel = logging.WARNING # XXX: remove me except: logger.debug("no loglevel specified, using default") loglevel = logging.WARNING self.configParser.set("logging", "loglevel", loglevel) return logfile, loglevel def _getID(self): """ Returns a tuple: private key, public key, nodeID, private group ID, and public group ID from config. If these values don't exist in conf file, they are generated and added. """ # get the keys and IDs from the config file. # If these values don't exist, generate a pub/priv key pair, nodeID, # and groupIDs. if (self.configParser.has_section("identification") == False): self.configParser.add_section("identification") try: privkey = FludRSA.importPrivateKey( eval(self.configParser.get("identification","Kr"))) except: pubkey, privkey = FludCrypto.generateKeys() else: try: pubkey = FludRSA.importPublicKey( eval(self.configParser.get("identification","Ku"))) except: pubkey = privkey.publickey() try: nodeID = self.configParser.get("identification","nodeID") except: #nodeID = FludCrypto.hashstring(str(pubkey.exportPublicKey())) nodeID = pubkey.id() try: privgroupID = self.configParser.get("identification", "groupIDr")[:64] except: privgroupID = 'fludtest' # default groupID hardcoded try: pubgroupID = self.configParser.get("identification","groupIDu") except: pubgroupID = FludCrypto.hashstring(str(pubkey.exportPublicKey()) +privgroupID) # write the settings back out to config object self.configParser.set("identification","Kr",privkey.exportPrivateKey()) self.configParser.set("identification","Ku",pubkey.exportPublicKey()) self.configParser.set("identification","nodeID",nodeID) self.configParser.set("identification","groupIDr",privgroupID) self.configParser.set("identification","groupIDu",pubgroupID) # return the values return privkey, pubkey, nodeID, privgroupID, pubgroupID def _getServerConf(self): """ Returns server configuration: port number """ if (self.configParser.has_section("server") == False): self.configParser.add_section("server") try: port = int(self.configParser.get("server","port")) except: logger.debug("no port specified, using default") port = 8080 # XXX: default should be defined elsewhere. # Should prefer 80. If in use, use 8080+ try: clientport = int(self.configParser.get("server","clientport")) except: logger.debug("no clientport specified, using default") clientport = port+CLIENTPORTOFFSET try: commandmap = eval(self.configParser.get("server","commandmap")) except: logger.debug("no commandmap specified, using default") commandmap = def_commandmap for i in def_commandmap: # ensure that commandmap covers all keys if not commandmap.has_key(i): commandmap[i] = def_commandmap[i] trustdeltas = {} for i in def_trustdeltas: if not trustdeltas.has_key(i): trustdeltas[i] = def_trustdeltas[i] # XXX: could also do a 'parammap' self.configParser.set("server","port",port) self.configParser.set("server","clientport",clientport) self.configParser.set("server","commandmap",commandmap) return port, clientport, commandmap, trustdeltas def _getDirConf(self, configParser, section, default): """ Returns directory configuration """ if (configParser.has_section(section) == False): configParser.add_section(section) try: dir = int(self.configParser.get(section,"dir")) except: logger.debug("no %s directory specified, using default", section) dir = self.fludhome+'/'+default if not os.path.isdir(dir): os.makedirs(dir) self.configParser.set(section,"dir",dir) return dir def _getClientConf(self): """ Returns client configuration: download directory """ return self._getDirConf(self.configParser, "client", "dl") def _getStoreConf(self): """ Returns data store configuration """ storedir = self._getDirConf(self.configParser, "store", "store") try: generosity = self.configParser.get("store", "generosity") except: logger.debug("no generosity specified, using default") generosity = 1.5 try: minoffer = self.configParser.get("store", "minoffer") except: logger.debug("no minoffer specified, using default") minoffer = 1024 return storedir, generosity, minoffer def _getkStoreConf(self): """ Returns dht data store configuration """ return self._getDirConf(self.configParser, "kstore", "dht") def _getMetaConf(self): """ Returns metadata configuration: metadata directory """ metadir = self._getDirConf(self.configParser, "metadata", "meta") try: master = self.configParser.get("meta","master") except: logger.debug("no meta master file specified, using default") master = "master" if not os.path.isfile(metadir+'/'+master): f = open(metadir+'/'+master, 'w') f.close() return (metadir, master) def _getReputations(self): """ Returns dict of reputations known to this node """ # XXX: should probably just throw these in with 'nodes' (for efficiency) return self._getDict(self.configParser, "reputations") def _getKnownNodes(self): """ Returns dict of nodes known to this node """ return {} # XXX: don't read known nodes for now result = self._getDict(self.configParser, "nodes") for i in result: print str(i) self.routing.insertNode( (result[i]['host'], result[i]['port'], long(i, 16), result[i]['nKu'])) return result def _getDict(self, configParser, section): """ creates a dictionary from the list of pairs given by ConfigParser.items(section). Requires that the right-hand side of the config file's "=" operator be a valid python type, as eval() will be invoked on it """ if (configParser.has_section(section) == False): configParser.add_section(section) try: items = configParser.items(section) result = {} for item in items: #print item try: result[str(item[0])] = eval(item[1]) configParser.set(section, item[0], item[1]) except: logger.warn("item '%s' in section '%s'" " of the config file has an unreadable format" % str(item[0]), str(section)) except: logger.warn("Couldn't read %s from config file:" % section) return result def addNode(self, nodeID, host, port, Ku, mygroup=None): """ Convenience method for adding a node to the known. If a node with nodeID already exists, nothing changes. This method /does not/ save the new configuration to file, """ if mygroup == None: mygroup = self.groupIDu if self.nodes.has_key(nodeID) == False: self.nodes[nodeID] = {'host': host, 'port': port, 'Ku': Ku.exportPublicKey(), 'mygroup': mygroup} #logger.log(logging.DEBUG, "nodes: " % str(self.nodes)) # XXX: disabled nodes saving #for k in self.nodes: # self.configParser.set('nodes', k, self.nodes[k]) n = self.routing.insertNode((host, int(port), long(nodeID, 16), Ku.exportPublicKey()['n'])) if n != None: logger.warn("need to ping %s for LRU in routing table!" % str(n)) # XXX: instead of pinging, put it in a replacement cache table # and when one of the nodes needs replaced (future query) # replace it with one of these. Sec 4.1 self.routing.replacementCache.insertNode( (host, int(port), long(nodeID, 16), Ku.exportPublicKey()['n'])) self.reputations[long(nodeID,16)] \ = self.trustdeltas['NODE_INITIAL_SCORE'] # XXX: no management of reputations size: need to manage as a cache def modifyReputation(self, nodeID, delta): """ change reputation of nodeID by delta """ logger.info("modify %s %s" % (nodeID, delta)) if isinstance(nodeID, str): nodeID = long(nodeID,16) if not self.reputations.has_key(nodeID): self.reputations[nodeID] \ = self.trustdeltas['NODE_INITIAL_SCORE'] # XXX: no management of reputations size: need to manage as a cache self.reputations[nodeID] += delta logger.debug("reputation for %d now %d", nodeID, self.reputations[nodeID]) def getOrderedNodes(self, num=None, exclude=None): """ Get nodes ordered by reputation. If num is passed in, return the first 'num' nodes, otherwise all. If exclude list is passed in, try to return nodes not on this list (but do return some excluded if nodes are exhausted, i.e., there aren't num nodes available). """ # XXX: O(n) each time this is called. Better performance if we # maintain sorted list when modified (modifyReputation, addNode), at a # bit higher mem expense. items = self.reputations.items() numitems = len(items) logger.debug("%d items in reps" % numitems) if exclude: items = [(v,k) for (k,v) in items if k not in exclude] if num and len(items) < num and numitems >= num: exitems = [(v,k) for (k,v) in items if k in exclude] items += exitems[num-len(item):] logger.debug("%d items now in reps" % len(items)) else: items = [(v,k) for (k,v) in items] items.sort() items.reverse() items = [(k,v) for (v,k) in items] # need to call routing.getNode() to get node triple and return those if num: logger.debug("returning %d of the %d items" % (num, len(items))) return [self.routing.getNode(f) for (f,v) in items[:num]] else: logger.debug("returning all %d of the items" % len(items)) return [self.routing.getNode(f) for (f,v) in items] # XXX: note that this master metadata all-in-mem scheme doesn't really work # long term; these methods should eventually go to a local db or db-like # something def updateMasterMeta(self, fname, val): """ update fname with val (sK) """ self.master[fname] = val def getFromMasterMeta(self, fname): """ get val (sK) for fname """ try: return self.master[fname] except: return None def deleteFromMasterMeta(self, fname): """ remove fname """ try: self.master.pop(fname) except: pass def loadMasterMeta(self): """ loads fname->sK mappings from file """ fmaster = open(os.path.join(self.metadir, self.metamaster), 'r') master = fmaster.read() fmaster.close() if master == "": master = {} else: master = fdecode(master) self.master = master def syncMasterMeta(self): """ sync in-mem fname->sK mappings to disk """ master = fencode(self.master) fmaster = open(os.path.join(self.metadir, self.metamaster), 'w') fmaster.write(master) fmaster.close() def _test(self): import doctest doctest.testmod() if __name__ == '__main__': fludConfig = FludConfig() fludConfig._test()
Python
#!/usr/bin/python """ FludNode.tac (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL). This is the application file used by twistd to daemonize FludNode. """ import os from twisted.application import service, internet import flud.FludNode from flud.protocol.FludCommUtil import getCanonicalIP port = None gwhost = None gwport = None if 'FLUDPORT' in os.environ: port = int(os.environ['FLUDPORT']) if 'FLUDGWHOST' in os.environ: gwhost = getCanonicalIP(os.environ['FLUDGWHOST']) if 'FLUDGWPORT' in os.environ: gwport = int(os.environ['FLUDGWPORT']) node = flud.FludNode.FludNode(port) if gwhost and gwport: node.connectViaGateway(gwhost, gwport) application = service.Application("flud.FludNode") service = node.start(twistd=True) #service.setServiceParent(application)
Python
#!/usr/bin/python """ FludScheduler.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), verison 3. FludScheduler is the process monitors files for changes, and then tells flud to back them up. """ import sys, os, time, stat from twisted.internet import reactor from flud.FludConfig import FludConfig from flud.protocol.LocalClient import * from flud.CheckboxState import CheckboxState CHECKTIME=5 class FludScheduler: def __init__(self, config, factory): self.config = config self.factory = factory self.fileconfigfile = None self.fileconfigfileMTime = 0 self.fileChangeTime = 0 self.fileconfigSelected = set() self.fileconfigExcluded = set() self.getMasterMetadata() def getMasterMetadata(self): d = self.factory.sendLIST() d.addCallback(self.gotMasterMetadata) d.addErrback(self.errMasterMetadata) return d def gotMasterMetadata(self, master): self.mastermetadata = master def errMasterMetadata(self, err): print err reactor.stop() def readFileConfig(self, mtime=None): print "reading FileConfig" file = open(self.fileconfigfile, 'r') self.fileconfig = eval(file.read()) file.close() if mtime: self.fileconfigfileMTime = mtime else: self.fileconfigfileMTime = os.stat( self.fileconfigfile)[stat.ST_MTIME] self.fileconfigSelected = set([f for f in self.fileconfig if self.fileconfig[f] == CheckboxState.SELECTED or self.fileconfig[f] == CheckboxState.SELECTEDCHILD]) self.fileconfigExcluded = set([f for f in self.fileconfig if self.fileconfig[f] == CheckboxState.EXCLUDED or self.fileconfig[f] == CheckboxState.EXCLUDEDCHILD]) # The file[s]ChangeStat are the worst possible way to detect file changes. # Much more efficient to use inotify/dnotify/fam/gamin/etc., as well as # more correct (no way to detect cp -a or -p, for example, with stat). # But, these are a fallback method when those aren't present, and are fine # for testing. def fileChangedStat(self, file, fileChangeTime=None): if os.path.isfile(file) or os.path.isdir(file): mtime = os.stat(file)[stat.ST_MTIME] if not fileChangeTime: fileChangeTime = self.fileChangeTime if file in self.mastermetadata: fileChangeTime = self.mastermetadata[file][1] else: return True print "mtime = %s, ctime = %s (%s)" % (mtime, fileChangeTime, file) if mtime > fileChangeTime: return True return False def filesChangedStat(self, files, fileChangeTime=None): result = [] for f in files: if self.fileChangedStat(f, fileChangeTime): result.append(f) return result # Change these to point to something other than the xxxStat() methods def fileChanged(self, file, fileChangeTime=None): """ >>> now = time.time() >>> f1 = tmpfile.mktemp() >>> """ return self.fileChangedStat(file, fileChangeTime) def filesChanged(self, files, fileChangeTime=None): return self.filesChangedStat(files, fileChangeTime) def checkFileConfig(self): # check config file to see if it has changed, then reparse it if not self.fileconfigfile: # first time through print "checking fileconfigfile (initial)" if os.environ.has_key('FLUDHOME'): fludhome = os.environ['FLUDHOME'] elif os.environ.has_key('HOME'): fludhome = os.environ['HOME']+"/.flud" else: fludhome = ".flud" # XXX: fludfile.conf should be in config self.fileconfigfile = os.path.join(fludhome, "fludfile.conf") if os.path.isfile(self.fileconfigfile): self.readFileConfig() return True else: print "no fileconfigfile to read" elif os.path.isfile(self.fileconfigfile): if self.fileChanged(self.fileconfigfile, self.fileconfigfileMTime): print "fileconfigfile changed" mtime = time.time() self.readFileConfig(mtime) return True return False def checkFilesystem(self): checkedFiles = set() changedFiles = set() def checkList(list): #print "checkList: %s" % list #print "checkedFiles: %s" % checkedFiles for entry in list: # XXX: if entry is in master metadata, and its mtime is not # earlier than the time used by fileChanged, skip it (add 'and' # clause) if entry not in checkedFiles and \ entry not in self.fileconfigExcluded and \ entry not in self.mastermetadata: print "checkFilesystem for %s" % entry if os.path.isdir(entry): #print "dir %s" % entry dirfiles = [os.path.join(entry, i) for i in os.listdir(entry)] checkedFiles.update([entry,]) checkList(dirfiles) elif self.fileChanged(entry): print "%s changed" % entry if os.path.isfile(entry): changedFiles.update([entry,]) #print "file %s changed" % entry else: print "entry ?? %s ?? changed" % entry checkedFiles.update([entry,]) checkList(self.fileconfigSelected) self.fileChangeTime = time.time() return changedFiles def storefileFailed(self, err, file): print "storing %s failed: %s" % (file, err) err.printTraceback() #print dir(err) def storefileYay(self, r, file): print "storing %s success" % file def storeFiles(self, changedFiles): #print "storing %s" % changedFiles dlist = [] for f in changedFiles: print "storing %s" % f deferred = self.factory.sendPUTF(f) deferred.addCallback(self.storefileYay, f) deferred.addErrback(self.storefileFailed, f) dlist.append(deferred) dl = defer.DeferredList(dlist) return dl #return defer.succeed(True) def restartCheckTimer(self, v): print "restarting timer (%d) to call run()" % CHECKTIME reactor.callLater(CHECKTIME, self.run) def updateMasterMetadata(self, v): return self.getMasterMetadata() def run(self): print "run" self.checkFileConfig() changedFiles = self.checkFilesystem() print "%s changed" % changedFiles d = self.storeFiles(changedFiles) d.addBoth(self.updateMasterMetadata) d.addBoth(self.restartCheckTimer)
Python
""" FludFileOperations.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Implements file storage and retrieval operations using flud primitives. """ import os, stat, sys, logging, binascii, random, time from zlib import crc32 from StringIO import StringIO from twisted.internet import defer, threads from Crypto.Cipher import AES from flud.FludCrypto import FludRSA, hashstring, hashfile from flud.protocol.FludCommUtil import * from flud.fencode import fencode, fdecode import fludfilefec logger = logging.getLogger('flud.fileops') # erasure coding constants code_k = 20 # data blocks code_n = 20 # parity blocks code_m = code_k+code_n # total blocks # temp filenaming defaults appendEncrypt = ".crypt" appendFsMeta = ".nmeta" # XXX: could remove trailing '=' from all stored sha256s (dht keys, storage # keys, etc) and add them back implicitly class Ctx(object): """ Object to encapsulate operation context for logger """ def __init__(self, ctx): self.ctx = ctx self.formatstr = "" self.args = () def msg(self, formatstr, *args): self.formatstr = formatstr self.args = args s = self.formatstr % self.args return self def __repr__(self): return str(self.ctx)+": "+(self.formatstr % self.args) def pathsplit(fname): par, chld = os.path.split(fname) if chld == "": res = [] res.append(par) return res else: res = pathsplit(par) res.append(os.path.join(par,chld)) return res def filemetadata(fname): fstat = os.stat(fname) return {'path' : fname, 'mode' : fstat[stat.ST_MODE], 'uid' : fstat[stat.ST_UID], 'gid' : fstat[stat.ST_GID], 'atim' : fstat[stat.ST_ATIME], 'mtim' : fstat[stat.ST_MTIME], 'ctim' : fstat[stat.ST_CTIME]} class StoreFile: """ Implements the meta operations of storing, retrieving, and verifying files. In addition to using the data primitives laid out in the protocol/ directory, manages the creation of file metadata for local and remote storage. Stores a file to flud by: 1. Create storage and encryption keys for file: Hashes the file once to create an encryption key, eK=H(file), and then again to create the storage key for the file metadata sK=H(eK)=H(H(file)). 2. Create local filesystem file metadata: Encrypts the storage key asymetrically with public key as eeK=e_Ku(eK), and creates local copy of fs metadata with eeK and other file metadata (ownership, name, timestamps). Encrypts this metadata with Ku. (flud file metadata consists of eeK and Ku(fs metadata)). 3. Create data-specific file metadata: Symmetrically encrypt the file with e_file=eK(file). Code e_file into k+m blocks. Perform H(block) on each k+m block. 4. Query DHT for sK. If it exists, grab the metadata record (call it 'storedMetadata') for comparison to one we are generated. Compare data-specific metadata to storedMetadata (if it exists). If it doesn't already exist, this step can be ignored. If it exists and the data doesn't match, either the hash function is broken or a malicious node has poisoned the DHT -- return failure (client can attempt restore under a new key, or challenge the existing stored metadata). 5. Store m+n blocks. If stored metadata exists, can do VERIFYs instead, and only store when VERIFY fails. For each failed VERIFY, must update the data-specific part of the metadata record. Since we don't want to be susceptible to replacement attack, the update is treated as an 'append' to the specific block metadata by the store target. (The store target can monitor its data-specific metadata records and for any that grow beyond a certain size, can shrink all block records to a size of 1 by performing RETRIEVE ops on the block, verfiying hash, and removing any failures. If this fails for all block entries, can prune all but the /first and last/ entries -- keep these to prevent damage by malicious nodes. This strategy is beneficial to the node that performs it, because it prevents storage of arbitrary data in the DHT layer, which is in turn good for the system as a whole) 6. Store file metadata (both node-specific and data-specific) to the DHT. Keep a local copy as well. 7. Update the master file record index for this node, and store it to the DHT layer. """ # XXX: instead of doing all debugging with '"%s ...", self.mkey, ...', make # some props (debug(), warn(), etc) that add the mkey to whatever msg is # passed # XXX: should follow this currentOps model for the other FludFileOps currentOps = {} def __init__(self, node, filename): self.node = node self.filename = filename self.mkey = crc32(self.filename) self.ctx = Ctx(self.mkey).msg self.config = node.config self.Ku = node.config.Ku self.routing = self.config.routing self.metadir = self.config.metadir self.parentcodedir = self.config.clientdir # XXX: clientdir? #self.nodeChoices = self.routing.knownExternalNodes() # ask for code_m + X nodes (we prefer a pool slightly larger than # code_m). XXX: 20 is magic self.nodeChoices = self.config.getOrderedNodes(code_m+20) self.usedNodes = {} self.deferred = self._storeFile() def _storeFile(self): if not os.path.isfile(self.filename): return defer.fail(ValueError("%s is not a file" % self.filename)) # 1: create encryption key (eK) and storage key (sK). Query DHT using # sK self.eK = hashfile(self.filename) logger.debug(self.ctx("_storefile %s (%s)", self.filename, self.eK)) self.sK = long(hashstring(self.eK), 16) self.eeK = self.Ku.encrypt(binascii.unhexlify(self.eK)) self.eKey = AES.new(binascii.unhexlify(self.eK)) #logger.debug(self.ctx("file %s eK:%s, storage key:%d" # % (self.filename, self.eK, self.sK))) # 2: create filesystem metadata locally. sbody = filemetadata(self.filename) sbody = fencode(sbody) self.eNodeFileMetadata = "" for i in range(0, len(sbody), 128): self.eNodeFileMetadata += self.Ku.encrypt(sbody[i:i+128])[0] fsMetadata = fencode({'eeK' : fencode(self.eeK[0]), 'meta' : fencode(self.eNodeFileMetadata)}) # erasure code the metadata # XXX: bad blocking stuff, move into thread self.flatname = fencode(generateRandom(16)) self.mfilename = os.path.join(self.metadir, self.flatname+appendFsMeta) self.encodedir = os.path.join(self.parentcodedir, self.flatname) try: os.mkdir(self.encodedir) except: return defer.fail(failure.DefaultException( "%s already requested" % self.filename)) # XXX: mfiles should be held in mem, as StringIOs (when coder supports # this) self.mfiles = fludfilefec.encode_to_files(StringIO(fsMetadata), len(fsMetadata), self.encodedir, self.mfilename, code_k, code_m) # XXX: piggybacking doesn't work with new metadata scheme, must fix it # to append metadata, or if already in progress, redo via verify ops # if already storing identical file by CAS, piggyback on it if self.currentOps.has_key(self.eK): logger.debug(self.ctx("reusing callback on %s", self.eK)) (d, counter) = self.currentOps[self.eK] self.currentOps[self.eK] = (d, counter+1) # setting sfile, encodedir, and efilename to empty vals is kinda # hokey -- could split _storeMetadata into two funcs instead (the # cleanup part and the store part; see _storeMetadata) self.sfiles = [] self.encodedir = None self.efilename = None d.addCallback(self._piggybackStoreMetadata) return d # 3: encrypt and encode the file locally. self.efilename = os.path.join(self.metadir,self.flatname+appendEncrypt) e = open(self.efilename, "w+") fd = os.open(self.filename, os.O_RDONLY) fstat = os.fstat(fd) fsize = fstat[stat.ST_SIZE] # XXX: bad blocking stuff, move into thread # create a pad at front of file to make it an even multiple of 16 fpad = int(16 - fsize%16); #logger.debug(self.ctx("fsize=%d, padding with %d bytes" # % (fsize, fpad))) paddata = chr(fpad)+(fpad-1)*'\x00' buf = paddata + os.read(fd,16-len(paddata)) e.write(self.eKey.encrypt(buf)); # now write the rest of the file while 1: # XXX: can we do this in larger than 16-byte chunks? buf = os.read(fd,16) if buf == "": break e.write(self.eKey.encrypt(buf)); #e.close() elen = e.tell() e.seek(0,0) os.close(fd) # erasure code the file # XXX: bad blocking stuff, move into thread self.sfiles = fludfilefec.encode_to_files(e, elen, self.encodedir, 'c', code_k, code_m) #logger.debug(self.ctx("coded to: %s" % str(self.sfiles))) # take hashes and rename coded blocks self.segHashesLocal = [] for i in range(len(self.sfiles)): sfile = self.sfiles[i] h = long(hashfile(sfile),16) logger.debug(self.ctx("file block %s hashes to %s", i, fencode(h))) destfile = os.path.join(self.encodedir,fencode(h)) if os.path.exists(destfile): logger.warn(self.ctx("%s exists (%s)", destfile, fencode(h))) self.segHashesLocal.append(h) #logger.debug(self.ctx("moved %s to %s" % (sfile, destfile))) os.rename(sfile, destfile) self.sfiles[i] = destfile mfile = self.mfiles[i] os.rename(mfile, destfile+".m") self.mfiles[i] = destfile+".m" # 4a: query DHT for metadata. d = self.node.client.kFindValue(self.sK) d.addCallback(self._checkForExistingFileMetadata) d.addErrback(self._storeFileErr, "DHT query for metadata failed") self.currentOps[self.eK] = (d, 1) return d # 4b: compare hashlists (locally encrypted vs. DHT -- if available). # for lhash, dhash in zip(segHashesLocal, segHashesDHT): def _checkForExistingFileMetadata(self, storedMetadata): if storedMetadata == None or isinstance(storedMetadata, dict): logger.info(self.ctx( "metadata doesn't yet exist, storing all data")) d = self._storeBlocks(storedMetadata) #d = self._storeBlocksSKIP(storedMetadata) return d else: storedMetadata = fdecode(storedMetadata) logger.info(self.ctx("metadata exists, verifying all data")) if not self._compareMetadata(storedMetadata, self.sfiles): raise ValueError("stored and local metadata do not match") else: logger.info(self.ctx("stored and local metadata match.")) # XXX: need to check for diversity. It could be that data stored # previously to a smaller network (<k+m nodes) and that we should # try to increase diversity and re-store the data. # XXX: also need to make sure we still trust all the nodes in the # metadata list. If not, we should move those blocks elsewhere. d = self._verifyAndStoreBlocks(storedMetadata) return d def _storeBlocksSKIP(self, storedMetadata): # for testing -- skip stores so we can get to storeMeta dlist = [] self.blockMetadata = {'k': code_k, 'n': code_n} for i in range(len(self.segHashesLocal)): hash = self.segHashesLocal[i] sfile = self.sfiles[i] node = random.choice(self.routing.knownExternalNodes()) host = node[0] port = node[1] nID = node[2] nKu = FludRSA.importPublicKey(node[3]) self.blockMetadata[(i, hash)] = long(nKu.id(), 16) return self._storeMetadata(None) # 5a -- store all blocks def _storeBlocks(self, storedMetadata): dlist = [] self.blockMetadata = {'k': code_k, 'n': code_n} for i in range(len(self.segHashesLocal)): hash = self.segHashesLocal[i] sfile = self.sfiles[i] deferred = self._storeBlock(i, hash, sfile, self.mfiles[i]) dlist.append(deferred) logger.debug(self.ctx("_storeBlocksAll")) dl = defer.DeferredList(dlist) dl.addCallback(self._storeMetadata) return dl def _storeBlock(self, i, hash, sfile, mfile, retry=2): if not self.nodeChoices: #self.nodeChoices = self.routing.knownExternalNodes() # XXX: instead of asking for code_k, ask for code_k - still needed self.nodeChoices = self.config.getOrderedNodes(code_k, self.usedNodes.keys()) logger.warn(self.ctx("asked for more nodes, %d nodes found", len(self.nodeChoices))) if not self.nodeChoices: return defer.fail(failure.DefaultException( "cannot store blocks to 0 nodes")) node = random.choice(self.nodeChoices) self.nodeChoices.remove(node) host = node[0] port = node[1] nID = node[2] nKu = FludRSA.importPublicKey(node[3]) location = long(nKu.id(), 16) logger.info(self.ctx("STOREing under %s on %s:%d", fencode(hash), host, port)) logger.debug(self.ctx("mfile is %s", mfile)) deferred = self.node.client.sendStore(sfile, (self.mkey, mfile), host, port, nKu) deferred.addCallback(self._blockStored, nID, i, hash, location) deferred.addErrback(self._retryStoreBlock, i, hash, location, sfile, mfile, nID, host, port, retry) return deferred def _retryStoreBlock(self, error, i, hash, location, sfile, mfile, nID, host, port, retry=None): retry = retry - 1 if retry > 0: logger.warn(self.ctx("STORE to %s (%s:%d) failed, trying again", nID, host, port)) d = self._storeBlock(i, hash, sfile, mfile, retry) d.addCallback(self._blockStored, nID, i, hash, location) # This will fail the entire operation. This is correct # behavior because we've tried on at least N nodes and couldn't # get the block to store -- the caller will have to try the entire # op again. If this proves to be a problem, up the default retry # value in _storeBlock(). d.addErrback(self._storeFileErr, "couldn't store block %s" % fencode(hash), lambda: self.config.modifyReputation(nID, self.config.trustdeltas['PUT_FAIL_PENALTY'])) return d else: self.usedNodes[nID] = True logger.warn(self.ctx("STORE to %s (%s:%d) failed, giving up", nID, host, port)) d = defer.Deferred() d.addErrback(self._storeFileErr, "couldn't store block %s" % fencode(hash), lambda: self.config.modifyReputation(nID, self.config.trustdeltas['PUT_FAIL_PENALTY'])) d.errback() return d def _blockStored(self, result, nID, i, blockhash, location): self.usedNodes[nID] = True logger.debug(self.ctx("_blcokStored %s", fencode(blockhash))) self.config.modifyReputation(location, self.config.trustdeltas['PUT_SUCCEED_REWARD']) self.blockMetadata[(i, blockhash)] = location return fencode(blockhash) def _compareMetadata(self, storedFiles, fileNames): # compares the block names returned from DHT to those in fileNames. # @param storedFiles: dict of longs (hashes) to their locations, # usually obtained from storedMetadata # @param fileNames: local filenames. Only the os.path.basename part # will be used for comparison # @return true if they match up perfectly, false otherwise logger.debug(self.ctx('# remote block names: %d', len(storedFiles))) logger.debug(self.ctx('# local blocks: %d', len(fileNames))) result = True k = storedFiles.pop('k') n = storedFiles.pop('n') for (i, f) in storedFiles: fname = os.path.join(self.encodedir,fencode(f)) if not fname in fileNames: logger.warn(self.ctx("%s not in sfiles", fencode(i))) result = False for i, fname in enumerate(fileNames): hname = os.path.basename(fname) if not storedFiles.has_key((i, fdecode(hname))): logger.warn(self.ctx("%s not in storedMetadata", hname)) result = False if result == False: for i in storedFiles: logger.debug(self.ctx("storedBlock = %s", fencode(i))) for i in fileNames: logger.debug(self.ctx("localBlock = %s", os.path.basename(i))) storedFiles['k'] = k storedFiles['n'] = n return result def _piggybackStoreMetadata(self, piggybackMeta): # piggybackMeta is a (nodeID, {blockID: storingNodeID, }) logger.debug(self.ctx("got piggyBackMeta data")) meta = piggybackMeta[1] sortedKeys = {} k = meta['k'] n = meta['n'] for i in [x for x in meta if x != 'k' and x != 'n']: sortedKeys[i[0]] = i for i in xrange(k+n): self.sfiles.append(fencode(sortedKeys[i][1])) return self._verifyAndStoreBlocks(meta, True) # 5b -- findnode on all stored blocks. def _verifyAndStoreBlocks(self, storedMetadata, noopVerify=False): self.blockMetadata = storedMetadata dlist = [] for i, sfile in enumerate(self.sfiles): # XXX: metadata should be StringIO to begin with f = file(self.mfiles[i]) mfile = StringIO(f.read()) f.close() seg = os.path.basename(sfile) segl = fdecode(seg) nID = self.blockMetadata[(i, segl)] if isinstance(nID, list): logger.info(self.ctx( "multiple location choices, choosing one randomly.")) nID = random.choice(nID) # XXX: for now, this just picks one of the alternatives at # random. If the chosen one fails, should try each of the # others until it works logger.info(self.ctx("looking up %s...", ('%x' % nID)[:8])) deferred = self.node.client.kFindNode(nID) deferred.addCallback(self._verifyBlock, i, sfile, mfile, seg, segl, nID, noopVerify) deferred.addErrback(self._storeFileErr, "couldn't find node %s... for VERIFY" % ('%x' % nID)[:8], self.config.modifyReputation(nID, self.config.trustdeltas['VRFY_FAIL_PENALTY'])) dlist.append(deferred) dl = defer.DeferredList(dlist) #dl.addCallback(self._storeMetadata) # XXX XXX XXX: don't _updateMaster unless we succeed!! dl.addCallback(self._updateMaster, storedMetadata) return dl # 5c -- verify all blocks, store any that fail verify. def _verifyBlock(self, kdata, i, sfile, mfile, seg, segl, nID, noopVerify): # XXX: looks like we occasionally get in here on timed out connections. # Should go to _storeFileErr instead, eh? if isinstance(kdata, str): logger.err(self.ctx("str kdata=%s", kdata)) #if len(kdata['k']) > 1: # #logger.debug(self.ctx("type kdata: %s" % type(kdata))) # #logger.debug(self.ctx("kdata=%s" % kdata)) # #logger.debug(self.ctx("len(kdata['k'])=%d" % len(kdata['k']))) # raise ValueError("couldn't find node %s" % ('%x' % nID)) # #raise ValueError("this shouldn't really be a ValueError..." # # " should be a GotMoreKnodesThanIBargainedForError" # # " (possibly caused when kFindNode fails (timeout) and" # # " we just get our own list of known nodes?): k=%s" # # % kdata['k']) node = kdata['k'][0] host = node[0] port = node[1] id = node[2] if id != nID: logger.debug(self.ctx("couldn't find node %s", ('%x' % nID))) raise ValueError("couldn't find node %s" % ('%x' % nID)) nKu = FludRSA.importPublicKey(node[3]) logger.info(self.ctx("verifying %s on %s:%d", seg, host, port)) if noopVerify: offset = length = 0 verhash = long(hashstring(''), 16) self.sfiles = [] else: fd = os.open(sfile, os.O_RDONLY) fsize = os.fstat(fd)[stat.ST_SIZE] if fsize > 20: # XXX: 20? length = 20 # XXX: 20? offset = random.randrange(fsize-length) else: length = fsize offset = 0 os.lseek(fd, offset, 0) data = os.read(fd, length) os.close(fd) verhash = long(hashstring(data), 16) deferred = self.node.client.sendVerify(seg, offset, length, host, port, nKu, (self.mkey, mfile)) deferred.addCallback(self._checkVerify, nKu, host, port, i, segl, sfile, mfile, verhash) deferred.addErrback(self._checkVerifyErr, nID, i, segl, sfile, mfile, verhash) return deferred def _checkVerify(self, result, nKu, host, port, i, seg, sfile, mfile, hash): if hash != long(result, 16): logger.info(self.ctx( "VERIFY hash didn't match for %s, performing STORE", fencode(seg))) d = self._storeBlock(i, seg, sfile, mfile) return d else: #logger.debug(self.ctx("block passed verify (%s == %s)" # % (hash, long(result,16)))) return fencode(seg) def _checkVerifyErr(self, failure, nID, i, seg, sfile, mfile, hash): self.config.modifyReputations(nID, self.config.trustdeltas['VRFY_FAIL_PENALTY']) logger.debug(self.ctx("Couldn't VERIFY: %s", failure.getErrorMessage())) logger.info(self.ctx("Couldn't VERIFY %s, performing STORE", fencode(seg))) d = self._storeBlock(i, seg, sfile, mfile) return d # 6 - store the metadata. def _storeMetadata(self, dlistresults): # cleanup part of storeMetadata: logger.debug(self.ctx("dlist=%s", str(dlistresults))) # XXX: for any "False" in dlistresults, need to invoke _storeBlocks # again on corresponding entries in sfiles. for i in dlistresults: if i[1] == None: logger.info(self.ctx("failed store/verify")) return False # storeMetadata part of storeMetadata # XXX: should sign metadata to prevent forged entries. #for i in self.blockMetadata: # logger.debug(self.ctx(" %s: %s" # % (fencode(i), fencode(self.blockMetadata[i])))) logger.debug(self.ctx("storing metadata at %s", fencode(self.sK))) logger.debug(self.ctx("len(segMetadata) = %d", len(self.blockMetadata))) d = self.node.client.kStore(self.sK, self.blockMetadata) d.addCallback(self._updateMaster, self.blockMetadata) d.addErrback(self._storeFileErr, "couldn't store file metadata to DHT") return d # 7 - update local master file record (store it to the network later). def _updateMaster(self, res, meta): # clean up locally coded files and encrypted file for sfile in self.sfiles: os.remove(sfile) for mfile in self.mfiles: os.remove(mfile) if self.encodedir: os.rmdir(self.encodedir) if self.efilename: os.remove(self.efilename) key = fencode(self.sK) logger.info(self.ctx("updating local master metadata with %s", key)) # store the filekey locally # update entry for file self.config.updateMasterMeta(self.filename, (self.sK, int(time.time()))) # update entry for parent dirs paths = pathsplit(self.filename) for i in paths: if not self.config.getFromMasterMeta(i): self.config.updateMasterMeta(i, filemetadata(i)) # XXX: not too efficient to write this out for every file. consider # local caching and periodic syncing instead self.config.syncMasterMeta() # cache the metadata locally (optional) fname = os.path.join(self.metadir,key) m = open(fname, 'wb') m.write(fencode(meta)) m.close() # clean up fs metadata file #os.remove(self.mfilename) #return fencode(self.sK) (d, counter) = self.currentOps[self.eK] counter = counter - 1 if counter == 0: logger.debug(self.ctx("counter 0 for currentOps %s", self.eK)) self.currentOps.pop(self.eK) else: logger.debug(self.ctx("setting counter = %d for %s", counter, self.eK)) self.currentOps[self.eK] = (d, counter) return (key, meta) def _storeFileErr(self, failure, message, raiseException=True, functor=None): if functor: functor() (d, counter) = self.currentOps[self.eK] counter = counter - 1 if counter == 0: logger.debug(self.ctx("err counter 0 for currentOps %s", self.eK)) self.currentOps.pop(self.eK) else: logger.debug(self.ctx("err setting counter = %d for %s", counter, self.eK)) self.currentOps[self.eK] = (d, counter) logger.error(self.ctx("%s: %s", message, failure.getErrorMessage())) logger.debug(self.ctx("%s", failure.getTraceback())) if raiseException: raise failure class RetrieveFile: """ Uses the given storage key to retrieve a file. The storage key is used to query the DHT layer for the file metadata record. The file record contains the locations of the file blocks. These are downloaded until the complete file can be regenerated and saved locally. """ def __init__(self, node, key, mkey=True): # 1: Query DHT for sK # 2: Retrieve entries for sK, decoding until efile can be regenerated # 3: Retrieve eK from sK by eK=Kp(eKe). Use eK to decrypt file. Strip # off leading pad. # 4: Save file as filepath=Kp(efilepath). self.node = node self.mkey = mkey try: self.sK = fdecode(key) except Exception, inst: self.deferred = defer.fail(inst) return self.ctx = Ctx(crc32(str(self.sK))).msg self.config = node.config self.Ku = node.config.Ku self.Kr = node.config.Kr self.routing = self.config.routing.knownNodes() self.metadir = self.config.metadir self.parentcodedir = self.config.clientdir self.numBlocksRetrieved = 0 self.blocks = {} self.fsmetas = {} self.deferred = self._retrieveFile() def _retrieveFile(self): # 1: Query DHT for sK logger.debug(self.ctx("querying DHT for %s", self.sK)) d = self.node.client.kFindValue(self.sK) d.addCallback(self._retrieveFileBlocks) d.addErrback(self._findFileErr, "file retrieve failed") return d def _findFileErr(self, failure, message, raiseException=True): logger.error(self.ctx("%s: %s" % (message, failure.getErrorMessage()))) if raiseException: return failure def _retrieveFileBlocks(self, meta): # 2: Retrieve entries for sK, decoding until efile can be regenerated if meta == None: raise LookupError("couldn't recover metadata for %s" % self.sK) self.meta = fdecode(meta) # XXX: need to check for diversity. It could be that data stored # previously to a smaller network (<k+m nodes) and that we should # try to increase diversity and re-store the data. # XXX: also need to make sure we still trust all the nodes in the # metadata list. If not, we should move those blocks elsewhere. if self.meta == None: raise LookupError("couldn't recover metadata for %s" % self.sK) k = self.meta.pop('k') n = self.meta.pop('n') if k != code_k or n != code_n: # XXX: raise ValueError("unsupported coding scheme %d/%d" % (k, n)) logger.info(self.ctx("got metadata %s" % self.meta)) self.decoded = False return self._getSomeBlocks(code_k) def _getSomeBlocks(self, reqs=code_k): tries = 0 if reqs > len(self.meta): reqs = len(self.meta) dlist = [] for i in range(reqs): c = random.choice(self.meta.keys()) block = fencode(c[1]) id = self.meta[c] if isinstance(id, list): logger.info(self.ctx( "multiple location choices, choosing one randomly.")) id = random.choice(id) # XXX: for now, this just picks one of the alternatives at # random. If the chosen one fails, should try each of the # others until it works or exhausts the list #logger.info(self.ctx("retrieving %s from %s" # % (block, fencode(id)))) logger.info(self.ctx("retrieving %s from %s" % (block, id))) # look up nodes by id, then do a retrieve. deferred = self.node.client.kFindNode(id) deferred.addCallback(self._retrieveBlock, block, id) deferred.addErrback(self._findNodeErr, "couldn't find node %s for block %s" % (fencode(id), block), id) dlist.append(deferred) self.meta.pop(c) tries = tries + 1 if tries >= reqs: break; dl = defer.DeferredList(dlist) dl.addCallback(self._retrievedAll) return dl def _findNodeErr(self, failure, msg, id): logger.info(self.ctx("%s: %s" % (message, failure.getErrorMessage()))) self.config.modifyReputation(id, self.config.trustdeltas['FNDN_FAIL_PENALTY']) def _retrieveBlock(self, kdata, block, id): #print type(kdata) #print kdata #if len(kdata['k']) > 1: if kdata['k'][0][2] != id: print "%s != %s" (kdata['k'][0], id) raise ValueError("couldn't find node %s" % fencode(id)) #raise ValueError("this shouldn't really be a ValueError..." # " should be a GotMoreKnodesThanIBargainedForError: k=%s" # % kdata['k']) #else: # print kdata['k'] node = kdata['k'][0] host = node[0] port = node[1] id = node[2] nKu = FludRSA.importPublicKey(node[3]) if not self.decoded: d = self.node.client.sendRetrieve(block, host, port, nKu, self.mkey) d.addCallback(self._retrievedBlock, id, block, self.mkey) d.addErrback(self._retrieveBlockErr, id, "couldn't get block %s from %s" % (block, fencode(id)), host, port, id) return d def _retrievedBlock(self, msg, nID, block, mkey): logger.debug(self.ctx("retrieved block=%s, msg=%s" % (block, msg))) self.config.modifyReputation(nID, self.config.trustdeltas['GET_SUCCEED_REWARD']) blockname = [f for f in msg if f[-len(block):] == block][0] expectedmeta = "%s.%s.meta" % (block, mkey) metanames = [f for f in msg if f[-len(expectedmeta):] == expectedmeta] if not metanames: raise failure.DefaultException("expected metadata was missing") self.blocks[block] = blockname self.fsmetas[block] = metanames[0] self.numBlocksRetrieved += 1 return True def _retrieveBlockErr(self, failure, nID, message, host, port, id): logger.info(self.ctx("%s: %s" % (message, failure.getErrorMessage()))) self.config.modifyReputation(nID, self.config.trustdeltas['GET_FAIL_PENALTY']) # don't propogate the error -- one block doesn't cause the file # retrieve to fail. def _retrievedAll(self, success): logger.info(self.ctx("tried retreiving %d blocks %s" % (len(success), success))) if self.numBlocksRetrieved >= code_k: # XXX: need to make this try again with other blocks if decode # fails return self._decodeData() elif len(self.meta) > 0: tries = 3 # XXX: magic number. Should derive from k & m logger.info(self.ctx("requesting %d more blocks" % tries)) return self._getSomeBlocks(tries) else: logger.info(self.ctx("couldn't decode file after retreiving" " all %d available blocks" % self.numBlocksRetrieved)) #return False raise RuntimeError("couldn't decode file after retreiving all" " %d available blocks" % self.numBlocksRetrieved) def _decodeData(self): logger.debug(self.ctx("_decodeData")) self.fname = os.path.join(self.parentcodedir,fencode(self.sK))+".rec1" d = threads.deferToThread(self.decodeData, self.fname, self.blocks.values(), self.config.clientdir) d.addCallback(self._decodeFsMetadata, self.blocks) d.addErrback(self._decodeError) return d def _decodeFsMetadata(self, decoded, blockname): logger.debug(self.ctx("_decodeFsMetadata")) self.mfname = os.path.join(self.parentcodedir,fencode(self.sK))+".m" d = threads.deferToThread(self.decodeData, self.mfname, self.fsmetas.values(), self.config.clientdir) d.addCallback(self._decodeDone, decoded) d.addErrback(self._decodeError) return d def decodeData(self, outfname, datafnames, datadir=None): logger.info(self.ctx("decoding %s to %s" % (datafnames, outfname))) outf = open(outfname, 'wb') data = [] for fname in datafnames: if datadir: fname = os.path.join(datadir, fname) data.append(open(fname, 'rb')) result = fludfilefec.decode_from_files(outf, data) outf.close() for f in data: f.close() if result: for fname in datafnames: if datadir: fname = os.path.join(datadir, fname) os.remove(os.path.join(datadir, fname)) return result def _decodeError(self, err): logger.warn(self.ctx("could not decode: %s", err)) return err def _decodeDone(self, decoded, metadecoded): if not self.decoded and decoded and metadecoded: logger.info(self.ctx("successfully decoded (retrieved %d blocks --" " all but %d blocks tried)" % (self.numBlocksRetrieved, len(self.meta)))) return self._decryptMeta() else: logger.info(self.ctx("decoded=%s, mdecoded=%s" % (decoded, metadecoded))) raise RuntimeError("decode error after retreiving " " %d blocks: decoded=%s, mdecoded=%" % (self.numBlocksRetrieved, decoded, metadecoded)) def _decryptMeta(self): # XXX: decrypt the metadatafile with Kr to get all the nmeta stuff (eeK # etc.) #mfile = open(os.path.join(self.parentcodedir, fencode(self.sK)+".m")) mfile = open(self.mfname, 'r') meta = mfile.read() mfile.close() logger.info(self.ctx("meta is %s" % str(meta))) self.nmeta = fdecode(meta) os.remove(self.mfname) return self._decryptFile() def _decryptFile(self): # 3: Retrieve eK from sK by eK=Kr(eeK). Use eK to decrypt file. Strip # off leading pad. skey = fencode(self.sK) f1 = open(os.path.join(self.parentcodedir,skey+".rec1"), "r") f2 = open(os.path.join(self.parentcodedir,skey+".rec2"), "w") #logger.info(self.ctx("decoding nmeta eeK for %s" % dir(self))) eeK = fdecode(self.nmeta['eeK']) # d_eK business is to ensure that eK is zero-padded to 32 bytes d_eK = self.Kr.decrypt(eeK) d_eK = '\x00'*(32%len(d_eK))+d_eK # XXX: magic 32, should be keyspace/8 eK = binascii.hexlify(d_eK) eKey = AES.new(binascii.unhexlify(eK)) # XXX: bad blocking stuff, move into thread while 1: buf = f1.read(16) if buf == "": break; f2.write(eKey.decrypt(buf)) f1.close() f2.close() os.remove(os.path.join(self.parentcodedir,skey+".rec1")) f2 = open(os.path.join(self.parentcodedir,skey+".rec2"), "r") f3 = open(os.path.join(self.parentcodedir,skey+".rec3"), "w") padlen = f2.read(1) #print "%s" % repr(padlen) padlen = ord(padlen) padlen -= 1 #print "throwing away %d pad bytes" % padlen pad = f2.read(padlen) # throw away pad. while 1: buf = f2.read(16) if buf == "": break; f3.write(buf) f2.close() f3.close() os.remove(os.path.join(self.parentcodedir,skey+".rec2")) # 4: Move file to its correct path, imbue it with properties from # metadata. # XXX: should we make sure we can read metadata before downloading all # the file data? #print "decoding nmeta meta" efmeta = fdecode(self.nmeta['meta']) fmeta = "" for i in range(0, len(efmeta), 128): fmeta += self.Kr.decrypt(efmeta[i:i+128]) fmeta = fdecode(fmeta) result = [fmeta['path']] if os.path.exists(fmeta['path']): # file is already there -- compare it. If different, save as # path.recovered and keep a list of these (to let the user know # that they'll need to resolve later). Or don't keep a list and # just 'do the right thing' (use the latest version by timestamp, # or always use the backup, or always use the local copy, or # define some other behavior for doing the right thing). logger.info(self.ctx("hash rec=%s" % hashfile(fmeta['path']))) logger.info(self.ctx("hash org=%s" % eK)) if hashfile(fmeta['path']) != eK: # XXX: do something better than log it -- see above comment logger.info(self.ctx( 'different version of file %s already present' % fmeta['path'])) # XXX: should generate '.recovered' extension more carefully, # so as not to overwrite coincidentally named files. fmeta['path'] = fmeta['path']+".recovered" result.insert(0,fmeta['path']) os.rename(os.path.join(self.parentcodedir,skey+".rec3"), fmeta['path']) else: logger.info(self.ctx('same version of file %s already present' % fmeta['path'])) # no need to copy: os.remove(os.path.join(self.parentcodedir,skey+".rec3")) else: # recover parent directories if not present paths = pathsplit(fmeta['path']) for i in paths: if not os.path.exists(i) and i != fmeta['path']: os.mkdir(i) # best effort dir creation, even if missing # directory metadata # XXX: should be using an accessor method on config for # master if i in self.config.master: dirmeta = self.config.getFromMasterMeta(i) os.chmod(i,dirmeta['mode']) os.chown(i,dirmeta['uid'],dirmeta['gid']) # XXX: windows # XXX: atim, mtim, ctim # XXX: should try to make sure we can write to dir, change # perms if necessary. # recover file by renaming to its path os.rename(os.path.join(self.parentcodedir,skey+".rec3"), fmeta['path']) # XXX: chown not supported on Windows os.chown(fmeta['path'], fmeta['uid'], fmeta['gid']) os.utime(fmeta['path'], (fmeta['atim'], fmeta['mtim'])) os.chmod(fmeta['path'], fmeta['mode']) logger.info(self.ctx("successfully restored file metadata")) return tuple(result) class RetrieveFilename: """ Retrieves a File given its local name. Only works if the local master index contains an entry for this filename. """ def __init__(self, node, filename): self.node = node self.filename = filename self.metadir = self.node.config.metadir self.config = self.node.config self.deferred = self._recoverFile() def _recoverFile(self): fmeta = self.config.getFromMasterMeta(self.filename) if fmeta: if isinstance(fmeta, dict): logger.debug("%s is a directory in master metadata", self.filename) # RetrieveFile will restore parent dirs, so we don't need to dlist = [] dirname = self.filename+os.path.sep # XXX: this should be calling a config.getAllFromMasterMeta() for i in [x for x in self.config.master.keys() if dirname == x[:len(dirname)]]: filekey = self.config.getFromMasterMeta[i] metakey = crc32(i) logger.debug("calling RetrieveFile %s" % filekey) d = RetrieveFile(self.node, fencode(filekey), metakey).deferred dlist.append(d) dl = defer.DeferredList(dlist) return dl else: logger.debug("%s is file in master metadata", self.filename) (filekey, backuptime) = self.config.getFromMasterMeta( self.filename) metakey = crc32(self.filename) if filekey != None and filekey != "": logger.debug("calling RetrieveFile %s" % filekey) d = RetrieveFile(self.node, fencode(filekey), metakey).deferred return d return defer.fail(LookupError("bad filekey %s for %s" % (filekey, self.filename))) return defer.fail(LookupError("no record of %s" % self.filename)) class VerifyFile: # XXX: remove me? I don't do anything that StoreFile can't do, plus if # I fail you'd still need to call StoreFile right after... # Or, do we keep me around and rip out all the verify stuff from # StoreFile and put it in here? # -- What we really want to do is look at trading partners for all our # files, and then check each one every day with some random bit of # data from some random file. But we also want eventual coverage of # all stored files. It seems like a good first approach would be to # scan over our local copies of all DHT records and make a list of # storing nodes, which will be of size N. Then, do k VERIFY ops to # each node in N, using a random subset of the files we have stored. # The VerifyFile object's purpose may be orthogonal to that, or # completely unnecessary, as the described scheme can be accomplished # with plain VERIFY ops. def verifyFile(self, filepath): """ Chooses some random blocks from filepath to verify against the store. The algorithm is as follows: sK = H(H(file at filepath)). Look up sK in the local master index. If the record isn't there, return this fact. If the record is there, retrieve its metadata. Verify k blocks as follows: With probibility n/(m+n), code the file locally (to verify coded blocks with a fair probabiliy, i.e., if m=40 and n=20, 33% of the time we will do the coding). Choose k blocks from the resulting blocks and using the file metadata record, do a VERIFY operation using a random offset and random length (if we chose not to do the coding in the previous step, the k blocks must come entirely from the non-coded portion). As we wait for the VERIFYs to return, hash these blocks locally. As each VERIFY returns, compare it with our local hash just computed. Return a list of hosts/nodeids for which the VERIFY failed. """ pass #remove me class RetrieveMasterIndex: def __init__(self, node): self.node = node nodeID = long(self.node.config.nodeID, 16) # 1. CAS = kfindval(nodeID) (CAS for last FLUDHOME/meta/master) logger.info("looking for key %x" % nodeID) self.deferred = self.node.client.kFindValue(nodeID) self.deferred.addCallback(self._foundCAS) self.deferred.addErrback(self._retrieveMasterIndexErr, "couldn't find master metadata") def _foundCAS(self, CAS): # 2. oldmaster = kfindval(CAS) if isinstance(CAS, dict): return defer.fail(ValueError("couldn't find CAS key")) CAS = fdecode(CAS) d = RetrieveFile(self.node, CAS).deferred d.addCallback(self._foundMaster) d.addErrback(self._retrieveMasterIndexErr, "couldn't find Master Index") return d def _foundMaster(self, result): if len(result) == 2: # got two filenames back, must mean we should choose one: the # one from the distributed store os.rename(result[0], result[1]) result = (result[1],) return result def _retrieveMasterIndexErr(self, err, msg): logger.warn(msg) return err class UpdateMasterIndex: def __init__(self, node): self.node = node self.metamaster = os.path.join(self.node.config.metadir, self.node.config.metamaster) # 0.1. oldmaster = RetrieveMasterIndex() self.deferred = RetrieveMasterIndex(node).deferred self.deferred.addCallback(self._removeOldMasterIndex) self.deferred.addErrback(self._storeMasterIndex) def _removeOldMasterIndex(self, res): # 0.2. for i in oldmaster: delete(i) print "removing old master not yet implemented" return self._storeMasterIndex(res) def _storeMasterIndex(self, res_or_err): # 1. store FLUDHOME/meta/master print "going to store %s" % self.metamaster d = StoreFile(self.node, self.metamaster).deferred d.addCallback(self._updateCAS) d.addErrback(self._updateMasterIndexErr, "couldn't store master index") return d def _updateCAS(self, stored): # 2. kstore(nodeID, CAS(FLUDHOME/meta/master)) #print "stored = %s" % str(stored) key, meta = stored logger.info("storing %s at %x" % (key, long(self.node.config.nodeID,16))) d = self.node.client.kStore(long(self.node.config.nodeID,16), key) # XXX: key should be fdecode()ed return d def _updateMasterIndexErr(self, err, msg): logger.warn(msg) return err if __name__ == "__main__": from FludNode import FludNode def successTest(res, fname, whatfor, nextStage=None): logger.info("finished %s" % whatfor) def errTest(failure): logger.info("boom: %s" % failure.getErrorMessage()) raise failure def fileKey(fname): EK = hashfile(fname) return fencode(long(hashstring(EK), 16)) def clearMeta(fname): # delete any metadata that might exist for this file. try: SK = fileKey(fname) os.remove(os.path.join(n.config.kstoredir,SK)) logger.info("test removed %s" % os.path.join(n.config.kstoredir,SK)) except: pass def doStore(fname, msg=None, nextStage=successTest): if msg != None: logger.info("finished %s ------" % msg) logger.info("---------------------------------------------------\n") # do a store logger.info("nextStage is %s" % nextStage) d = StoreFile(n,fname).deferred d.addCallback(nextStage, fname, 'store op', doCorruptSegAndStore) d.addErrback(errTest) def doDelSegAndStore((key, meta), fname, msg=None, nextStage=successTest): # only works if stores are local if msg != None: logger.info("finished %s ------" % msg) logger.info("---------------------------------------------------\n") # delete a block and do a store c = random.choice(meta.keys()) logger.info("removing %s" % fencode(c)) os.remove(os.path.join(n.config.storedir,fencode(c))) logger.info("test removed %s" % os.path.join(n.config.storedir, fencode(c))) d = StoreFile(n,fname).deferred d.addCallback(nextStage, fname, 'lost block op') d.addErrback(errTest) def doCorruptSegAndStore((key, meta), fname, msg=None, nextStage=successTest): # only works if stores are local if msg != None: logger.info("finished %s ------" % msg) logger.info("---------------------------------------------------\n") # corrupt a block and do a store c = random.choice(meta.keys()) logger.info("corrupting %s" % fencode(c)) f = open(os.path.join(n.config.storedir,fencode(c)), 'r') data = f.read() f.close() f = open(os.path.join(n.config.storedir,fencode(c)), 'w') f.write('blah'+data) f.close() d = StoreFile(n,fname).deferred d.addCallback(nextStage, fname, 'corrupted block op') d.addErrback(errTest) def doRetrieve(key, msg=None, nextStage=successTest): if msg != None: logger.info("finished %s ------" % msg) logger.info("---------------------------------------------------\n") d = RetrieveFile(n, key).deferred d.addCallback(nextStage, key, 'retrieve op') d.addErrback(errTest) def doRetrieveName(filename, msg=None, nextStage=successTest): if msg != None: logger.info("finished %s ------" % msg) logger.info("---------------------------------------------------\n") d = RetrieveFilename(n, filename).deferred d.addCallback(nextStage, filename, 'retrieve filename op') d.addErrback(errTest) def runTests(dummy): # test against self -- all stores and queries go to self. fname = "/tmp/nrpy.pdf" #clearMeta(fname) #doStore(fname, None, doDelSegAndStore) # do all stages of testing doStore(fname) # only do one op (for manual testing) #doRetrieve(fileKey(fname)) #doRetrieveName(fname) n = FludNode() n.run() if len(sys.argv) == 3: deferred = n.client.sendkFindNode(sys.argv[1], int(sys.argv[2]), 1) deferred.addCallback(runTests) deferred.addErrback(errTest) else: runTests(None) n.join()
Python
""" Okay, so this isn't a real python module, yet. Wanted to get down a few ideas on versioning. First, the background. Traditional backup systems that provide versioning support allow the user to retrieve the current version of a file, or any of N previous versions that were stored during previous backup operations. Since it was rather trivial to simply keep old versions on the central backup server, this wasn't much of an engineering problem (at worst, disk fills up quickly). With a collaborative backup system, such a scheme is less practical. If fully enabled, it can consume many times the storage space of a simple single-snapshot system. If you want to enforce fairness, you must require that the number of resources you consume are proportional to those that you provide. Encoding already dictates that this ratio is imbalanced towards providing more resources than consuming. But "server-side" versioning, even when using a clever delta-compression technique, really tips the scales. There is good news, however. We can use a single-snapshot system to provide versioning by requiring all versioning to occur locally. That is, the consumer's own hard drive can be used to maintain multiple versions of files, and then the whole thing can be backed up as a single-snapshot to the flud network. Think of a local CVS repository (with many versions contained therein) that is set to be backed up; the backup system doesn't have to worry about versioning -- it just backs up the current data. The local CVS repo is in charge of worrying about versions. To the user, its all the same. The advantages of this scheme are mainly: 1) simplicity 2) storage consumption minimization, opt-in 3) decoupling of versioning layer from backup layer Of the three, #1 is really the most appealing. We just back up the current view. This also greatly simplifies the verification mechanism -- the verifier will always have the complete file from which to do challenge/response queries to the verifiee. We don't have to worry about keeping deltas or partial checksums or anything like that in order to do verification; we just pick a block of bytes at random from within the file, and make sure that the storer can return us the hash of those bytes. #1 also means that we don't have to do anything complicated to figure out what the delta of a delta-compressed version should be (i.e., we don't need to request the old version from the storage system, compare it with our version, then send out a delta) -- in fact, with this scheme we wipe out delta compression altogether, at least from the viewpoint of the storage mechanism (some other local mechanism is welcome to use delta compression to store versions locally, but this mechanism won't need to download lots of data in order to do so, because it will all be local). #2 is nice. It means that if the user isn't interested in versioning, they don't have to do it. This will be the default, in fact. This means that we eliminate a lot of overhead that we would have had if every user was storing versions, even if they didn't need them. It also means that there is an automatic cost for enabling versions, not only for the collaborative system, but for the user's local storage resources. Not to imply that we want to punish the user for enabling versions, but there's no free lunch (and besides, adding local disk is cheap). [as an aside, here, versioning does become necessary quite quickly for things such as email clients that store all mail in a particular folder as one large file, or other applications that use databases in single files -- we don't want the user to have to send the whole file (which can become quite large) every time they get a new email or change the db slightly. The good news is that we can still provide this in its own layer]. Decoupling (#3) is always a good idea, especially when it can be done cleanly. The user (or flud client developer) is free to implement whatever local versioning scheme they want. They can make copies of files and store them in other directories, they could use a version control system such as CVS, they could do their own delta compression and store the deltas in a special directory. They could store as many or as few versions as they want. They can take version snapshots often or seldom, and this is independent of how often they perform backup. And such schemes can be switched out, upgraded, or removed on the fly without anyone really noticing. So, what is this module all about then? Well, for now, nothing but this documentation. If the user wants versioning, they'll have to do it themselves, (keeping in mind that the system *will* store those local versions, and they can rest easy knowing that they can be retrieved). At some future point, we will implement some fancy versioning layer with local delta compression in this module to 'complete the package,' but it is currently a low priority endeavor. Its priority will rise as we get close to providing a 'for your grandma' solution. """
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import fencode, fdecode """ Test code for primitive DHT operations. """ stay_alive = 1 filename = "/tmp/tempstoredata" filekey = os.path.basename(filename) key = 87328673569979667228965797330646992089697345905484734072690869757741450870337L # format of block metadata is # {(i, datakey): storingNodeID, ..., 'n': n, 'm': m} # where i<n+m, datakey is a sha256 of the data stored, and storingNodeID is # either a nodeID or a list of nodeIDs. testval = {(0, 802484L): 465705L, (1, 780638L): 465705L, (2, 169688L): 465705L, (3, 267175L): 465705L, (4, 648636L): 465705L, (5, 838315L): 465705L, (6, 477619L): 465705L, (7, 329906L): 465705L, (8, 610565L): 465705L, (9, 217811L): 465705L, (10, 374124L): 465705L, (11, 357214L): 465705L, (12, 147307L): 465705L, (13, 427751L): 465705L, (14, 927853L): 465705L, (15, 760369L): 465705L, (16, 707029L): 465705L, (17, 479234L): 465705L, (18, 190455L): 465705L, (19, 647489L): 465705L, (20, 620470L): 465705L, (21, 777532L): 465705L, (22, 622383L): 465705L, (23, 573283L): 465705L, (24, 613082L): 465705L, (25, 433593L): 465705L, (26, 584543L): 465705L, (27, 337485L): 465705L, (28, 911014L): 465705L, (29, 594065L): 465705L, (30, 375876L): 465705L, (31, 726818L): 465705L, (32, 835759L): 465705L, (33, 814060L): 465705L, (34, 237176L): 465705L, (35, 538268L): 465705L, (36, 272650L): 465705L, (37, 314058L): 465705L, (38, 257714L): 465705L, (39, 439931L): 465705L, 'k': 20, 'n': 20} logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) #logger.setLevel(logging.DEBUG) logger.setLevel(logging.INFO) def cleanup(_, node): logger.info("waiting %ds to shutdown..." % stay_alive) reactor.callLater(stay_alive, node.stop) def testerror(failure, message, node): """ error handler for test errbacks """ logger.warn("testerror message: %s" % message) logger.warn("testerror: '%s'" % str(failure)) logger.warn("At least 1 test FAILED") def endtests(res, nKu, node, host, port): """ executes after all tests """ try: res = fdecode(res) except ValueError: pass if res != testval: return testerror(None, "retrieved value does not match stored value:" " '%s' != '%s'" % (res, testval), node) logger.info("testkFindVal PASSED") logger.debug("testkFindVal result: %s" % str(res)) logger.info("all tests PASSED") return res def testkFindVal(res, nKu, node, host, port): logger.info("testSendkFindVal PASSED") logger.debug("testSendkFindVal result: %s" % str(res)) logger.info("attempting testkFindValue") deferred = node.client.kFindValue(key) deferred.addCallback(endtests, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkFindValue", node) return deferred def testSendkFindVal(res, nKu, node, host, port): logger.info("testkStore PASSED") logger.debug("testkStore result: %s" % str(res)) logger.info("attempting testSendkFindValue") deferred = node.client.sendkFindValue(host, port, key) deferred.addCallback(testkFindVal, nKu, node, host, port) deferred.addErrback(testerror, "failed at testSendkFindValue", node) return deferred def testkStore(res, nKu, node, host, port): logger.info("testSendkStore PASSED") logger.debug("testSendkStore result: %s" % str(res)) logger.info("attempting testkStore") deferred = node.client.kStore(key, testval) deferred.addCallback(testSendkFindVal, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkStore", node) return deferred def testSendkStore(res, nKu, node, host, port): logger.info("testkFindNode PASSED") logger.debug("testkFindNode result: %s" % str(res)) logger.info("attempting testSendkStore") deferred = node.client.sendkStore(host, port, key, testval) deferred.addCallback(testkStore, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkStore", node) return deferred def testkFindNode(res, nKu, node, host, port): """ executes after testSendkFindNode """ logger.info("testSendkFindNode PASSED") logger.debug("testSendkFindNode result: %s" % str(res)) logger.info("attempting kFindNode") deferred = node.client.kFindNode(key) deferred.addCallback(testSendkStore, nKu, node, host, port) deferred.addErrback(testerror, "failed at kFindNode", node) return deferred def testSendkFindNode(nKu, node, host, port): """ executes after testGetID """ logger.info("testkGetID PASSED") logger.info("attempting sendkFindNode") deferred = node.client.sendkFindNode(host, port, key) deferred.addCallback(testkFindNode, nKu, node, host, port) deferred.addErrback(testerror, "failed at sendkFindNode", node) return deferred def testGetID(node, host, port): """ Tests sendGetID(), and invokes testSendkFindNode on success """ deferred = node.client.sendGetID(host, port) deferred.addCallback(testSendkFindNode, node, host, port) deferred.addErrback(testerror, "failed at testGetID", node) return deferred def runTests(host, port=None, listenport=None): host = getCanonicalIP(host) node = FludNode(port=listenport) if port == None: port = node.config.port logger.info("testing against %s:%s, localport=%s" % (host, port, listenport)) node.run() d = testGetID(node, host, port) d.addBoth(cleanup, node) #testkFindVal("blah", node.config.Ku, node, host, port) node.join() """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) elif len(sys.argv) == 4: runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket, shutil, tempfile from binascii import crc32 from StringIO import StringIO from twisted.python import failure sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.FludCrypto import generateRandom, hashfile from flud.protocol.FludCommUtil import * from flud.fencode import fencode """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ smallfilekey = "" smallfilename = "" smallfilenamebad = "" largefilekey = "" largefilename = "" largefilenamebad = "" metadata = 'aaaa' def testerror(failure, message, node): """ error handler for test errbacks """ print "testerror message: %s" % message print "testerror: %s" % str(failure) print "At least 1 test FAILED" raise failure def testUnexpectedSuccess(res, message, node): print "unexpected success message: %s" % message print "At least 1 test succeeded when it should have failed" raise "bad" def testDELETEBadKeyFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg # the end else: # XXX: here and elsewhere, raise something more descriptive, otherwise # its waay confusing print "the following trace may be misleading..." raise failure def testDELETEBadKey(nKu, node, host, port): print "starting testDELETEBadKey" path = os.path.join("somedir", largefilekey) deferred = node.client.sendDelete(path, crc32(path), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "DELETE with bad key succeeded", node) deferred.addErrback(testDELETEBadKeyFailed, "DELETE with bad key failed as expected", node, nKu, host, port) return deferred def testVERIFYBadKeyFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testDELETEBadKey(nKu, node, host, port) else: raise failure def testVERIFYBadKey(nKu, node, host, port): print "starting testVERIFYBadKey" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize-20 deferred = node.client.sendVerify(smallfilenamebad, offset, 5, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad key succeeded", node) deferred.addErrback(testVERIFYBadKeyFailed, "VERIFY of bad filekey failed as expected", node, nKu, host, port) return deferred def testVERIFYBadLengthFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYBadKey(nKu, node, host, port) else: raise failure def testVERIFYBadLength(nKu, node, host, port): print "starting testVERIFYBadOffset" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize-10 deferred = node.client.sendVerify(smallfilekey, offset, 20, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad length", node) deferred.addErrback(testVERIFYBadLengthFailed, "VERIFY of bad length failed as expected", node, nKu, host, port) return deferred def testVERIFYBadOffsetFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYBadLength(nKu, node, host, port) else: print "VERIFYBadOffset failed as expected, but with wrong failure" raise failure def testVERIFYBadOffset(nKu, node, host, port): print "starting testVERIFYBadOffset" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize+2 deferred = node.client.sendVerify(smallfilekey, offset, 20, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad offset", node) deferred.addErrback(testVERIFYBadOffsetFailed, "VERIFY of bad offset failed as expected", node, nKu, host, port) return deferred def testVERIFYNotFoundFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testVERIFYBadOffset(nKu, node, host, port) else: raise failure def testVERIFYNotFound(nKu, node, host, port): print "starting testVERIFYNotFound" deferred = node.client.sendVerify(largefilekey, 10, 10, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified non-existent file", node) deferred.addErrback(testVERIFYNotFoundFailed, "VERIFY of non-existent file failed as expected", node, nKu, host, port) return deferred def testRETRIEVEIllegalPathFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYNotFound(nKu, node, host, port) else: raise failure def testRETRIEVEIllegalPath(nKu, node, host, port): print "starting testRETRIEVEIllegalPath" deferred = node.client.sendRetrieve(os.path.join("somedir",smallfilekey), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "retrieved file with illegal path", node) deferred.addErrback(testRETRIEVEIllegalPathFailed, "RETRIEVE using illegal path failed as expected", node, nKu, host, port) return deferred def testRETRIEVENotFoundFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testRETRIEVEIllegalPath(nKu, node, host, port) else: raise failure def testRETRIEVENotFound(nKu, node, host, port): print "starting testRETRIEVENotFound" deferred = node.client.sendRetrieve(largefilekey, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "retrieved file that shouldn't exist", node) deferred.addErrback(testRETRIEVENotFoundFailed, "RETRIEVE of non-existent file failed as expected", node, nKu, host, port) return deferred def testSTORELargeFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadCASKeyException'): print "%s" % msg return testRETRIEVENotFound(nKu, node, host, port) else: raise failure def testSTOREBadKeyLarge(nKu, node, host, port): print "starting testSTOREBadKeyLarge" deferred = node.client.sendStore(largefilenamebad, (crc32(largefilenamebad), StringIO(metadata)), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "large file, bad key succeeded", node) deferred.addErrback(testSTORELargeFailed, "large STORE with bad key failed as expected", node, nKu, host, port) return deferred def testSTORESmallFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadCASKeyException'): print "%s" % msg return testSTOREBadKeyLarge(nKu, node, host, port) else: raise failure def testSTOREBadKeySmall(nKu, node, host, port): print "starting testSTOREBadKeySmall" deferred = node.client.sendStore(smallfilenamebad, (crc32(smallfilenamebad), StringIO(metadata)), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "small file, bad key succeeded", node) deferred.addErrback(testSTORESmallFailed, "small STORE with bad key failed as expected", node, nKu, host, port) return deferred def testSTORESuccess(res, nKu, node, host, port): print "testSTORE succeeded: %s" % res return testSTOREBadKeySmall(nKu, node, host, port) def testSTORE(nKu, node, host, port): # store a file successfully for later failure tests (VERIFY, etc) print "starting testSTORE" deferred = node.client.sendStore(smallfilename, (crc32(smallfilename), StringIO(metadata)), host, port, nKu) deferred.addCallback(testSTORESuccess, nKu, node, host, port) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred # XXX: need to test bogus headers for all commands (BAD_REQUEST) # XXX: need to test failures for authentication def testID(node, host, port): """ Tests sendGetID(), and invokes testSTORE on success """ print "starting testID" deferred = node.client.sendGetID(host, port) deferred.addCallback(testSTORE, node, host, port) #deferred.addCallback(testSTOREBadKeySmall, node, host, port) deferred.addErrback(testerror, "failed at testID", node) return deferred def cleanup(err, node): if err: print "cleaning up: %s" % err else: print "cleaning up" os.remove(smallfilename) os.remove(smallfilenamebad) os.remove(largefilename) os.remove(largefilenamebad) reactor.callLater(1, node.stop) def generateTestData(): def generateFiles(minsize): fname = tempfile.mktemp() f = open(fname, 'w') f.write('\0'*minsize) f.write(generateRandom(random.randrange(256)+1)) f.close() filekey = hashfile(fname) filekey = fencode(int(filekey, 16)) filename = os.path.join("/tmp",filekey) os.rename(fname,filename) filenamebad = os.path.join("/tmp/","bad"+filekey[3:]) shutil.copy(filename, filenamebad) return (filekey, filename, filenamebad) global smallfilekey global smallfilename global smallfilenamebad (smallfilekey, smallfilename, smallfilenamebad) = generateFiles(1024) global largefilekey global largefilename global largefilenamebad (largefilekey, largefilename, largefilenamebad) = generateFiles(512000) def runTests(host, port=None, listenport=None): generateTestData() node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testID(node, host, port) d.addBoth(cleanup, node) node.join() """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python """ FludFileOpTest.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. System tests for FludFileOperations """ import sys, os, time, logging, tempfile, shutil from twisted.internet import reactor sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludConfig import FludConfig from flud.FludNode import FludNode from flud.fencode import fencode, fdecode from flud.FludCrypto import generateRandom from flud.FludFileOperations import * import flud.FludDefer as FludDefer from flud.protocol.LocalClient import listMeta logger = logging.getLogger('flud') def testError(failure, message, node): print "testError message: %s" % message print "testError: %s" % str(failure) print "At least 1 test FAILED" return failure def gotSuccess(r, desc): print "%s succeeded" % desc def testConcurrent(r, node, files, desc): #print "r was %s" % r print "testConcurrent %s" % desc dlist = [] for file in files: d = testStoreFile(node, file) dlist.append(d) dl = FludDefer.ErrDeferredList(dlist) dl.addCallback(gotSuccess, desc) dl.addErrback(testError) return dl def checkStoreFile(res, node, fname): master = listMeta(node.config) if fname not in master: return defer.fail(failure.DefaultException("file not stored")) else: print "store on %s verified" % fname return res # <- *VITAL* for concurrent dup ops to succeed. def testStoreFile(node, fname): d = StoreFile(node, fname).deferred d.addCallback(checkStoreFile, node, fname) d.addErrback(testError, fname, node) return d def doTests(node, smallfnames, largefnames, dupsmall, duplarge): d = testStoreFile(node, smallfnames[0]) d.addCallback(testConcurrent, node, smallfnames, "small") d.addCallback(testConcurrent, node, largefnames, "large") d.addCallback(testConcurrent, node, dupsmall, "small duplicates") d.addCallback(testConcurrent, node, duplarge, "large duplicates") #d = testConcurrent(None, node, dupsmall, "small duplicates") #d = testConcurrent(None, node, duplarge, "large duplicates") return d def cleanup(_, node, filenamelist): #print _ for f in filenamelist: try: print "deleting %s" % f os.remove(f) except: print "couldn't remove %s" % f reactor.callLater(1, node.stop) def generateTestFile(minSize): fname = tempfile.mktemp() f = open(fname, 'w') data = generateRandom(minSize/50) for i in range(0, 51+random.randrange(50)): f.write(data) f.close() filename = os.path.join("/tmp",fname) os.rename(fname,filename) return filename def runTests(host, port, listenport=None): f1 = generateTestFile(5120) f2 = generateTestFile(5120) f3 = f2+".dup" shutil.copy(f2, f3) f4 = generateTestFile(513000) f5 = generateTestFile(513000) f6 = f5+".dup" shutil.copy(f5, f6) node = FludNode(port=listenport) if port == None: port = node.config.port node.run() node.connectViaGateway(host, port) d = doTests(node, [f1, f2], [f4, f5], [f2, f3], [f5, f6]) d.addBoth(cleanup, node, [f1, f2, f3, f4, f5, f6]) node.join() if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3])) else: print "must run this test against a flud network (no single node op)" print "usage: %s [<othernodehost othernodeport> |"\ " <othernodehost othernodeport listenport>]" % sys.argv[0]
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket, tempfile from twisted.python import failure from StringIO import StringIO from zlib import crc32 sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient import flud.FludCrypto as FludCrypto from flud.fencode import fencode, fdecode from flud.protocol.FludCommUtil import * from flud.FludDefer import ErrDeferredList """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ # metadatablock: (block#,n,k,blockdata) metadatablock = fencode((1,20,40,'adfdsfdffffffddddddddddddddd')) fake_mkey_offset = 111111 def testerror(failure, message, node): """ error handler for test errbacks """ print "testerror message: %s" % message print "testerror: %s" % str(failure) print "At least 1 test FAILED" return failure def allGood(_, nKu): print "all tests PASSED" return nKu def checkDELETE(res, nKu, fname, fkey, mkey, node, host, port, totalDelete): """ checks to ensure the file was deleted """ # totalDelete = True if this delete op should remove all meta (and data) if totalDelete: # try to retrieve with any metakey, should fail print "expecting failed retrieve, any metakey" return testRETRIEVE(res, nKu, fname, fkey, True, node, host, port, lambda args=(res, nKu): allGood(*args), False) else: # try to retrieve with any metakey, should succeed print "expecting successful retrieve, any metakey" return testRETRIEVE(res, nKu, fname, fkey, True, node, host, port, lambda args=(res, nKu, fname, fkey, mkey+fake_mkey_offset, node, host, port, True): testDELETE(*args)) def testDELETE(res, nKu, fname, fkey, mkey, node, host, port, totalDelete): """ Tests sendDelete, and invokes checkDELETE on success """ print "starting testDELETE %s.%s" % (fname, mkey) #return checkDELETE(None, nKu, fname, fkey, mkey, node, host, port, False) deferred = node.client.sendDelete(fkey, mkey, host, port, nKu) deferred.addCallback(checkDELETE, nKu, fname, fkey, mkey, node, host, port, totalDelete) deferred.addErrback(testerror, "failed at testDELETE", node) return deferred def checkVERIFY(res, nKu, fname, fkey, mkey, node, host, port, hash, newmeta): """ executes after testVERIFY """ if long(hash, 16) != long(res, 16): raise failure.DefaultException("verify didn't match: %s != %s" % (hash, res)) print "checkVERIFY (%s) %s success" % (newmeta, fname) if newmeta: return testDELETE(res, nKu, fname, fkey, mkey, node, host, port, False) else: return testVERIFY(nKu, fname, fkey, mkey, node, host, port, True) def testVERIFY(nKu, fname, fkey, mkey, node, host, port, newmeta): """ Test sendVerify """ # newmeta, if True, will generate new metadata to be stored during verify if newmeta: thismkey = mkey+fake_mkey_offset else: thismkey = mkey print "starting testVERIFY (%s) %s.%s" % (newmeta, fname, thismkey) fd = os.open(fname, os.O_RDONLY) fsize = os.fstat(fd)[stat.ST_SIZE] length = 20 offset = random.randrange(fsize-length) os.lseek(fd, offset, 0) data = os.read(fd, length) os.close(fd) hash = FludCrypto.hashstring(data) deferred = node.client.sendVerify(fkey, offset, length, host, port, nKu, (thismkey, StringIO(metadatablock))) deferred.addCallback(checkVERIFY, nKu, fname, fkey, mkey, node, host, port, hash, newmeta) deferred.addErrback(testerror, "failed at testVERIFY (%s)" % newmeta, node) return deferred def failedRETRIEVE(res, nextCallable): return nextCallable(); def checkRETRIEVE(res, nKu, fname, fkey, mkey, node, host, port, nextCallable): """ Compares the file that was stored with the one that was retrieved """ f1 = open(fname) filename = [f for f in res if f[-len(fkey):] == fkey][0] f2 = open(filename) if (f1.read() != f2.read()): f1.close() f2.close() raise failure.DefaultException( "upload/download (%s, %s) files don't match" % (fname, os.path.join(node.config.clientdir, fkey))) #print "%s (%d) and %s (%d) match" % (fname, os.stat(fname)[stat.ST_SIZE], # filename, os.stat(filename)[stat.ST_SIZE]) f1.close() f2.close() if mkey != True: expectedmeta = "%s.%s.meta" % (fkey, mkey) metanames = [f for f in res if f[-len(expectedmeta):] == expectedmeta] if not metanames: raise failure.DefaultException("expected metadata was missing") f3 = open(metanames[0]) md = f3.read() if md != metadatablock: raise failure.DefaultException("upload/download metadata doesn't" " match (%s != %s)" % (md, metadatablock)) return nextCallable() def testRETRIEVE(res, nKu, fname, fkey, mkey, node, host, port, nextCallable, expectSuccess=True): """ Tests sendRetrieve, and invokes checkRETRIEVE on success """ print "starting testRETRIEVE %s.%s" % (fname, mkey) deferred = node.client.sendRetrieve(fkey, host, port, nKu, mkey) deferred.addCallback(checkRETRIEVE, nKu, fname, fkey, mkey, node, host, port, nextCallable) if expectSuccess: deferred.addErrback(testerror, "failed at testRETRIEVE", node) else: deferred.addErrback(failedRETRIEVE, nextCallable) return deferred def testSTORE2(nKu, fname, fkey, node, host, port): mkey = crc32(fname) mkey2 = mkey+(2*fake_mkey_offset) print "starting testSTORE %s.%s" % (fname, mkey2) deferred = node.client.sendStore(fname, (mkey2, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey2, node, host, port, lambda args=(nKu, fname, fkey, mkey, node, host, port, False): testVERIFY(*args)) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred def testSTORE(nKu, fname, fkey, node, host, port): """ Tests sendStore, and invokes testRETRIEVE on success """ mkey = crc32(fname) print "starting testSTORE %s.%s" % (fname, mkey) deferred = node.client.sendStore(fname, (mkey, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey, node, host, port, lambda args=(nKu, fname, fkey, node, host, port): testSTORE2(*args)) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred def testID(node, host, port): """ Tests sendGetID(), and invokes testSTORE on success """ print "starting testID" deferred = node.client.sendGetID(host, port) deferred.addErrback(testerror, "failed at testID", node) return deferred def testAggSTORE(nKu, aggFiles, node, host, port): print "starting testAggSTORE" dlist = [] for fname, fkey in aggFiles: mkey = crc32(fname) print "testAggSTORE %s (%s)" % (fname, mkey) deferred = node.client.sendStore(fname, (mkey, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey, node, host, port, lambda args=(nKu, fname, fkey, mkey, node, host, port, False): testVERIFY(*args)) deferred.addErrback(testerror, "failed at testAggSTORE", node) dlist.append(deferred) dl = ErrDeferredList(dlist) dl.addCallback(allGood, nKu) dl.addErrback(testerror, "failed at testAggSTORE", node) return dl def cleanup(_, node, filenamelist): for f in filenamelist: try: os.remove(f) except: print "couldn't remove %s" % f reactor.callLater(1, node.stop) def generateTestData(minSize): fname = tempfile.mktemp() f = open(fname, 'w') data = FludCrypto.generateRandom(minSize/50) for i in range(0, 51+random.randrange(50)): f.write(data) f.close() filekey = FludCrypto.hashfile(fname) filekey = fencode(int(filekey, 16)) filename = os.path.join("/tmp",filekey) os.rename(fname,filename) return (filename, filekey) def runTests(host, port=None, listenport=None): (largeFilename, largeFilekey) = generateTestData(512000) (smallFilename, smallFilekey) = generateTestData(5120) aggFiles = [] for i in range(4): aggFiles.append(generateTestData(4096)) node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testID(node, host, port) d.addCallback(testSTORE, largeFilename, largeFilekey, node, host, port) d.addCallback(testSTORE, smallFilename, smallFilekey, node, host, port) d.addCallback(testAggSTORE, aggFiles, node, host, port) d.addBoth(cleanup, node, [i[0] for i in aggFiles] + [largeFilename, smallFilename]) node.join() def main(): localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3])) if __name__ == '__main__': main()
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure from twisted.internet import defer sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import * from flud.FludDefer import ErrDeferredList """ Test code for kprimitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ # XXX: check return from ops to see if they passed (e.g., if STORE fails, we # are notified [currently] by the html page that is returned). # XXX: should make a random file each time this is run... CONCURRENT=50 CONCREPORT=10 node = None # format of block metadata is # {(i, datakey): storingNodeID, ..., 'n': n, 'm': m} # where i<n+m, datakey is a sha256 of the data stored, and storingNodeID is # either a nodeID or a list of nodeIDs testval = {(0, 802484L): 465705L, (1, 780638L): 465705L, (2, 169688L): 465705L, (3, 267175L): 465705L, (4, 648636L): 465705L, (5, 838315L): 465705L, (6, 477619L): 465705L, (7, 329906L): 465705L, (8, 610565L): 465705L, (9, 217811L): 465705L, (10, 374124L): 465705L, (11, 357214L): 465705L, (12, 147307L): 465705L, (13, 427751L): 465705L, (14, 927853L): 465705L, (15, 760369L): 465705L, (16, 707029L): 465705L, (17, 479234L): 465705L, (18, 190455L): 465705L, (19, 647489L): 465705L, (20, 620470L): 465705L, (21, 777532L): 465705L, (22, 622383L): 465705L, (23, 573283L): 465705L, (24, 613082L): 465705L, (25, 433593L): 465705L, (26, 584543L): 465705L, (27, 337485L): 465705L, (28, 911014L): 465705L, (29, 594065L): 465705L, (30, 375876L): 465705L, (31, 726818L): 465705L, (32, 835759L): 465705L, (33, 814060L): 465705L, (34, 237176L): 465705L, (35, 538268L): 465705L, (36, 272650L): 465705L, (37, 314058L): 465705L, (38, 257714L): 465705L, (39, 439931L): 465705L, 'k': 20, 'n': 20} logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) logger.setLevel(logging.DEBUG) def suitesuccess(results): logger.info("all tests in suite passed") #print results return results def suiteerror(failure): logger.info("suite did not complete") logger.info("DEBUG: %s" % failure) return failure def stagesuccess(result, message): logger.info("stage %s succeeded" % message) return result def stageerror(failure, message): logger.info("stage %s failed" % message) #logger.info("DEBUG: %s" % failure) return failure def itersuccess(res, i, message): if i % CONCREPORT == 0: logger.info("itersuccess: %s" % message) return res def itererror(failure, message): logger.info("itererror message: %s" % message) #logger.info("DEBUG: %s" % failure) #logger.info("DEBUG: %s" % dir(failure) failure.printTraceback() return failure def endtests(res, nKu, node, host, port): """ executes after all tests """ try: res = fdecode(res) except ValueError: pass if res != testval: return failure.DefaultException("retrieved value does not match" " stored value: '%s' != '%s'" % (res, testval)) logger.log(logging.INFO,"testkFindVal PASSED: %s\n" % str(res)) logger.log(logging.INFO,"all tests PASSED") return res def testkFindVal(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO,"testSendkFindVal PASSED: %s\n" % str(res)) logger.log(logging.INFO,"attempting testkFindValue") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kFindValue(key) deferred.addErrback(itererror, "kFindValue") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kFindValue") d.addErrback(stageerror, 'kFindValue') d.addCallback(endtests, nKu, node, host, port) return d def testSendkFindVal(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkStore PASSED: %s\n" % str(res)) logger.log(logging.INFO, "attempting testSendkFindValue") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkFindValue(host, port, key) deferred.addErrback(itererror, "sendkFindValue") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkFindValue") d.addErrback(stageerror, 'sendkFindValue') d.addCallback(testkFindVal, nKu, node, host, port, num) return d def testkStore(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testSendkStore PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting testkStore") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kStore(key, testval) deferred.addErrback(itererror, "kStore") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kStore") d.addErrback(stageerror, 'kStore') d.addCallback(testSendkFindVal, nKu, node, host, port, num) return d def testSendkStore(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkFindNode PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting testSendkStore") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkStore(host, port, key, testval) deferred.addErrback(itererror, "sendkStore") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkStore") d.addErrback(stageerror, 'sendkStore') d.addCallback(testkStore, nKu, node, host, port, num) return d def testkFindNode(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testSendkFindNode PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting kFindNode") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kFindNode(key) deferred.addErrback(itererror, "kFindNode") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kFindNode") d.addErrback(stageerror, 'kFindNode') d.addCallback(testSendkStore, nKu, node, host, port, num) return d def testSendkFindNode(nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkGetID PASSED") logger.log(logging.INFO, "attempting sendkFindNode") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkFindNode(host, port, key) deferred.addErrback(itererror, "sendkFindNode") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkFindNode") d.addErrback(stageerror, 'sendkFindNode') d.addCallback(testkFindNode, nKu, node, host, port, num) return d def testGetID(node, host, port, num=CONCURRENT): """ Tests sendGetID(), and invokes testSendkFindNode on success """ deferred = node.client.sendGetID(host, port) deferred.addCallback(testSendkFindNode, node, host, port, num) deferred.addErrback(stageerror, "testGetID") return deferred def runTests(host, port=None, listenport=None): num = CONCURRENT #num = 5 node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testGetID(node, host, port, CONCURRENT) d.addCallback(suitesuccess) d.addErrback(suiteerror) d.addBoth(cleanup, node) node.join() #node.start() # doesn't work, because reactor may not have started # listening by time requests start flying def cleanup(_, node): logger.info("shutting down in 1 seconds...") time.sleep(1) reactor.callLater(1, node.stop) logger.info("done cleaning up") """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: print "Warning: testing against self may result in timeout failures" runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure from twisted.internet import defer sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import * from flud.FludDefer import ErrDeferredList """ Test code for kprimitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ # XXX: check return from ops to see if they passed (e.g., if STORE fails, we # are notified [currently] by the html page that is returned). # XXX: should make a random file each time this is run... CONCURRENT=50 CONCREPORT=10 node = None # format of block metadata is # {(i, datakey): storingNodeID, ..., 'n': n, 'm': m} # where i<n+m, datakey is a sha256 of the data stored, and storingNodeID is # either a nodeID or a list of nodeIDs testval = {(0, 802484L): 465705L, (1, 780638L): 465705L, (2, 169688L): 465705L, (3, 267175L): 465705L, (4, 648636L): 465705L, (5, 838315L): 465705L, (6, 477619L): 465705L, (7, 329906L): 465705L, (8, 610565L): 465705L, (9, 217811L): 465705L, (10, 374124L): 465705L, (11, 357214L): 465705L, (12, 147307L): 465705L, (13, 427751L): 465705L, (14, 927853L): 465705L, (15, 760369L): 465705L, (16, 707029L): 465705L, (17, 479234L): 465705L, (18, 190455L): 465705L, (19, 647489L): 465705L, (20, 620470L): 465705L, (21, 777532L): 465705L, (22, 622383L): 465705L, (23, 573283L): 465705L, (24, 613082L): 465705L, (25, 433593L): 465705L, (26, 584543L): 465705L, (27, 337485L): 465705L, (28, 911014L): 465705L, (29, 594065L): 465705L, (30, 375876L): 465705L, (31, 726818L): 465705L, (32, 835759L): 465705L, (33, 814060L): 465705L, (34, 237176L): 465705L, (35, 538268L): 465705L, (36, 272650L): 465705L, (37, 314058L): 465705L, (38, 257714L): 465705L, (39, 439931L): 465705L, 'k': 20, 'n': 20} logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) logger.setLevel(logging.DEBUG) def suitesuccess(results): logger.info("all tests in suite passed") #print results return results def suiteerror(failure): logger.info("suite did not complete") logger.info("DEBUG: %s" % failure) return failure def stagesuccess(result, message): logger.info("stage %s succeeded" % message) return result def stageerror(failure, message): logger.info("stage %s failed" % message) #logger.info("DEBUG: %s" % failure) return failure def itersuccess(res, i, message): if i % CONCREPORT == 0: logger.info("itersuccess: %s" % message) return res def itererror(failure, message): logger.info("itererror message: %s" % message) #logger.info("DEBUG: %s" % failure) #logger.info("DEBUG: %s" % dir(failure) failure.printTraceback() return failure def endtests(res, nKu, node, host, port): """ executes after all tests """ try: res = fdecode(res) except ValueError: pass if res != testval: return failure.DefaultException("retrieved value does not match" " stored value: '%s' != '%s'" % (res, testval)) logger.log(logging.INFO,"testkFindVal PASSED: %s\n" % str(res)) logger.log(logging.INFO,"all tests PASSED") return res def testkFindVal(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO,"testSendkFindVal PASSED: %s\n" % str(res)) logger.log(logging.INFO,"attempting testkFindValue") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kFindValue(key) deferred.addErrback(itererror, "kFindValue") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kFindValue") d.addErrback(stageerror, 'kFindValue') d.addCallback(endtests, nKu, node, host, port) return d def testSendkFindVal(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkStore PASSED: %s\n" % str(res)) logger.log(logging.INFO, "attempting testSendkFindValue") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkFindValue(host, port, key) deferred.addErrback(itererror, "sendkFindValue") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkFindValue") d.addErrback(stageerror, 'sendkFindValue') d.addCallback(testkFindVal, nKu, node, host, port, num) return d def testkStore(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testSendkStore PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting testkStore") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kStore(key, testval) deferred.addErrback(itererror, "kStore") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kStore") d.addErrback(stageerror, 'kStore') d.addCallback(testSendkFindVal, nKu, node, host, port, num) return d def testSendkStore(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkFindNode PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting testSendkStore") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkStore(host, port, key, testval) deferred.addErrback(itererror, "sendkStore") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkStore") d.addErrback(stageerror, 'sendkStore') d.addCallback(testkStore, nKu, node, host, port, num) return d def testkFindNode(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testSendkFindNode PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting kFindNode") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kFindNode(key) deferred.addErrback(itererror, "kFindNode") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kFindNode") d.addErrback(stageerror, 'kFindNode') d.addCallback(testSendkStore, nKu, node, host, port, num) return d def testSendkFindNode(nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkGetID PASSED") logger.log(logging.INFO, "attempting sendkFindNode") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkFindNode(host, port, key) deferred.addErrback(itererror, "sendkFindNode") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkFindNode") d.addErrback(stageerror, 'sendkFindNode') d.addCallback(testkFindNode, nKu, node, host, port, num) return d def testGetID(node, host, port, num=CONCURRENT): """ Tests sendGetID(), and invokes testSendkFindNode on success """ deferred = node.client.sendGetID(host, port) deferred.addCallback(testSendkFindNode, node, host, port, num) deferred.addErrback(stageerror, "testGetID") return deferred def runTests(host, port=None, listenport=None): num = CONCURRENT #num = 5 node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testGetID(node, host, port, CONCURRENT) d.addCallback(suitesuccess) d.addErrback(suiteerror) d.addBoth(cleanup, node) node.join() #node.start() # doesn't work, because reactor may not have started # listening by time requests start flying def cleanup(_, node): logger.info("shutting down in 1 seconds...") time.sleep(1) reactor.callLater(1, node.stop) logger.info("done cleaning up") """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: print "Warning: testing against self may result in timeout failures" runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket, shutil, tempfile from binascii import crc32 from StringIO import StringIO from twisted.python import failure sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.FludCrypto import generateRandom, hashfile from flud.protocol.FludCommUtil import * from flud.fencode import fencode """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ smallfilekey = "" smallfilename = "" smallfilenamebad = "" largefilekey = "" largefilename = "" largefilenamebad = "" metadata = 'aaaa' def testerror(failure, message, node): """ error handler for test errbacks """ print "testerror message: %s" % message print "testerror: %s" % str(failure) print "At least 1 test FAILED" raise failure def testUnexpectedSuccess(res, message, node): print "unexpected success message: %s" % message print "At least 1 test succeeded when it should have failed" raise "bad" def testDELETEBadKeyFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg # the end else: # XXX: here and elsewhere, raise something more descriptive, otherwise # its waay confusing print "the following trace may be misleading..." raise failure def testDELETEBadKey(nKu, node, host, port): print "starting testDELETEBadKey" path = os.path.join("somedir", largefilekey) deferred = node.client.sendDelete(path, crc32(path), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "DELETE with bad key succeeded", node) deferred.addErrback(testDELETEBadKeyFailed, "DELETE with bad key failed as expected", node, nKu, host, port) return deferred def testVERIFYBadKeyFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testDELETEBadKey(nKu, node, host, port) else: raise failure def testVERIFYBadKey(nKu, node, host, port): print "starting testVERIFYBadKey" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize-20 deferred = node.client.sendVerify(smallfilenamebad, offset, 5, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad key succeeded", node) deferred.addErrback(testVERIFYBadKeyFailed, "VERIFY of bad filekey failed as expected", node, nKu, host, port) return deferred def testVERIFYBadLengthFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYBadKey(nKu, node, host, port) else: raise failure def testVERIFYBadLength(nKu, node, host, port): print "starting testVERIFYBadOffset" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize-10 deferred = node.client.sendVerify(smallfilekey, offset, 20, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad length", node) deferred.addErrback(testVERIFYBadLengthFailed, "VERIFY of bad length failed as expected", node, nKu, host, port) return deferred def testVERIFYBadOffsetFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYBadLength(nKu, node, host, port) else: print "VERIFYBadOffset failed as expected, but with wrong failure" raise failure def testVERIFYBadOffset(nKu, node, host, port): print "starting testVERIFYBadOffset" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize+2 deferred = node.client.sendVerify(smallfilekey, offset, 20, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad offset", node) deferred.addErrback(testVERIFYBadOffsetFailed, "VERIFY of bad offset failed as expected", node, nKu, host, port) return deferred def testVERIFYNotFoundFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testVERIFYBadOffset(nKu, node, host, port) else: raise failure def testVERIFYNotFound(nKu, node, host, port): print "starting testVERIFYNotFound" deferred = node.client.sendVerify(largefilekey, 10, 10, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified non-existent file", node) deferred.addErrback(testVERIFYNotFoundFailed, "VERIFY of non-existent file failed as expected", node, nKu, host, port) return deferred def testRETRIEVEIllegalPathFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYNotFound(nKu, node, host, port) else: raise failure def testRETRIEVEIllegalPath(nKu, node, host, port): print "starting testRETRIEVEIllegalPath" deferred = node.client.sendRetrieve(os.path.join("somedir",smallfilekey), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "retrieved file with illegal path", node) deferred.addErrback(testRETRIEVEIllegalPathFailed, "RETRIEVE using illegal path failed as expected", node, nKu, host, port) return deferred def testRETRIEVENotFoundFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testRETRIEVEIllegalPath(nKu, node, host, port) else: raise failure def testRETRIEVENotFound(nKu, node, host, port): print "starting testRETRIEVENotFound" deferred = node.client.sendRetrieve(largefilekey, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "retrieved file that shouldn't exist", node) deferred.addErrback(testRETRIEVENotFoundFailed, "RETRIEVE of non-existent file failed as expected", node, nKu, host, port) return deferred def testSTORELargeFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadCASKeyException'): print "%s" % msg return testRETRIEVENotFound(nKu, node, host, port) else: raise failure def testSTOREBadKeyLarge(nKu, node, host, port): print "starting testSTOREBadKeyLarge" deferred = node.client.sendStore(largefilenamebad, (crc32(largefilenamebad), StringIO(metadata)), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "large file, bad key succeeded", node) deferred.addErrback(testSTORELargeFailed, "large STORE with bad key failed as expected", node, nKu, host, port) return deferred def testSTORESmallFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadCASKeyException'): print "%s" % msg return testSTOREBadKeyLarge(nKu, node, host, port) else: raise failure def testSTOREBadKeySmall(nKu, node, host, port): print "starting testSTOREBadKeySmall" deferred = node.client.sendStore(smallfilenamebad, (crc32(smallfilenamebad), StringIO(metadata)), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "small file, bad key succeeded", node) deferred.addErrback(testSTORESmallFailed, "small STORE with bad key failed as expected", node, nKu, host, port) return deferred def testSTORESuccess(res, nKu, node, host, port): print "testSTORE succeeded: %s" % res return testSTOREBadKeySmall(nKu, node, host, port) def testSTORE(nKu, node, host, port): # store a file successfully for later failure tests (VERIFY, etc) print "starting testSTORE" deferred = node.client.sendStore(smallfilename, (crc32(smallfilename), StringIO(metadata)), host, port, nKu) deferred.addCallback(testSTORESuccess, nKu, node, host, port) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred # XXX: need to test bogus headers for all commands (BAD_REQUEST) # XXX: need to test failures for authentication def testID(node, host, port): """ Tests sendGetID(), and invokes testSTORE on success """ print "starting testID" deferred = node.client.sendGetID(host, port) deferred.addCallback(testSTORE, node, host, port) #deferred.addCallback(testSTOREBadKeySmall, node, host, port) deferred.addErrback(testerror, "failed at testID", node) return deferred def cleanup(err, node): if err: print "cleaning up: %s" % err else: print "cleaning up" os.remove(smallfilename) os.remove(smallfilenamebad) os.remove(largefilename) os.remove(largefilenamebad) reactor.callLater(1, node.stop) def generateTestData(): def generateFiles(minsize): fname = tempfile.mktemp() f = open(fname, 'w') f.write('\0'*minsize) f.write(generateRandom(random.randrange(256)+1)) f.close() filekey = hashfile(fname) filekey = fencode(int(filekey, 16)) filename = os.path.join("/tmp",filekey) os.rename(fname,filename) filenamebad = os.path.join("/tmp/","bad"+filekey[3:]) shutil.copy(filename, filenamebad) return (filekey, filename, filenamebad) global smallfilekey global smallfilename global smallfilenamebad (smallfilekey, smallfilename, smallfilenamebad) = generateFiles(1024) global largefilekey global largefilename global largefilenamebad (largefilekey, largefilename, largefilenamebad) = generateFiles(512000) def runTests(host, port=None, listenport=None): generateTestData() node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testID(node, host, port) d.addBoth(cleanup, node) node.join() """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python """ FludFileOpTest.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. System tests for FludFileOperations """ import sys, os, time, logging, tempfile, shutil from twisted.internet import reactor sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludConfig import FludConfig from flud.FludNode import FludNode from flud.fencode import fencode, fdecode from flud.FludCrypto import generateRandom from flud.FludFileOperations import * import flud.FludDefer as FludDefer from flud.protocol.LocalClient import listMeta logger = logging.getLogger('flud') def testError(failure, message, node): print "testError message: %s" % message print "testError: %s" % str(failure) print "At least 1 test FAILED" return failure def gotSuccess(r, desc): print "%s succeeded" % desc def testConcurrent(r, node, files, desc): #print "r was %s" % r print "testConcurrent %s" % desc dlist = [] for file in files: d = testStoreFile(node, file) dlist.append(d) dl = FludDefer.ErrDeferredList(dlist) dl.addCallback(gotSuccess, desc) dl.addErrback(testError) return dl def checkStoreFile(res, node, fname): master = listMeta(node.config) if fname not in master: return defer.fail(failure.DefaultException("file not stored")) else: print "store on %s verified" % fname return res # <- *VITAL* for concurrent dup ops to succeed. def testStoreFile(node, fname): d = StoreFile(node, fname).deferred d.addCallback(checkStoreFile, node, fname) d.addErrback(testError, fname, node) return d def doTests(node, smallfnames, largefnames, dupsmall, duplarge): d = testStoreFile(node, smallfnames[0]) d.addCallback(testConcurrent, node, smallfnames, "small") d.addCallback(testConcurrent, node, largefnames, "large") d.addCallback(testConcurrent, node, dupsmall, "small duplicates") d.addCallback(testConcurrent, node, duplarge, "large duplicates") #d = testConcurrent(None, node, dupsmall, "small duplicates") #d = testConcurrent(None, node, duplarge, "large duplicates") return d def cleanup(_, node, filenamelist): #print _ for f in filenamelist: try: print "deleting %s" % f os.remove(f) except: print "couldn't remove %s" % f reactor.callLater(1, node.stop) def generateTestFile(minSize): fname = tempfile.mktemp() f = open(fname, 'w') data = generateRandom(minSize/50) for i in range(0, 51+random.randrange(50)): f.write(data) f.close() filename = os.path.join("/tmp",fname) os.rename(fname,filename) return filename def runTests(host, port, listenport=None): f1 = generateTestFile(5120) f2 = generateTestFile(5120) f3 = f2+".dup" shutil.copy(f2, f3) f4 = generateTestFile(513000) f5 = generateTestFile(513000) f6 = f5+".dup" shutil.copy(f5, f6) node = FludNode(port=listenport) if port == None: port = node.config.port node.run() node.connectViaGateway(host, port) d = doTests(node, [f1, f2], [f4, f5], [f2, f3], [f5, f6]) d.addBoth(cleanup, node, [f1, f2, f3, f4, f5, f6]) node.join() if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3])) else: print "must run this test against a flud network (no single node op)" print "usage: %s [<othernodehost othernodeport> |"\ " <othernodehost othernodeport listenport>]" % sys.argv[0]
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import fencode, fdecode """ Test code for primitive DHT operations. """ stay_alive = 1 filename = "/tmp/tempstoredata" filekey = os.path.basename(filename) key = 87328673569979667228965797330646992089697345905484734072690869757741450870337L # format of block metadata is # {(i, datakey): storingNodeID, ..., 'n': n, 'm': m} # where i<n+m, datakey is a sha256 of the data stored, and storingNodeID is # either a nodeID or a list of nodeIDs. testval = {(0, 802484L): 465705L, (1, 780638L): 465705L, (2, 169688L): 465705L, (3, 267175L): 465705L, (4, 648636L): 465705L, (5, 838315L): 465705L, (6, 477619L): 465705L, (7, 329906L): 465705L, (8, 610565L): 465705L, (9, 217811L): 465705L, (10, 374124L): 465705L, (11, 357214L): 465705L, (12, 147307L): 465705L, (13, 427751L): 465705L, (14, 927853L): 465705L, (15, 760369L): 465705L, (16, 707029L): 465705L, (17, 479234L): 465705L, (18, 190455L): 465705L, (19, 647489L): 465705L, (20, 620470L): 465705L, (21, 777532L): 465705L, (22, 622383L): 465705L, (23, 573283L): 465705L, (24, 613082L): 465705L, (25, 433593L): 465705L, (26, 584543L): 465705L, (27, 337485L): 465705L, (28, 911014L): 465705L, (29, 594065L): 465705L, (30, 375876L): 465705L, (31, 726818L): 465705L, (32, 835759L): 465705L, (33, 814060L): 465705L, (34, 237176L): 465705L, (35, 538268L): 465705L, (36, 272650L): 465705L, (37, 314058L): 465705L, (38, 257714L): 465705L, (39, 439931L): 465705L, 'k': 20, 'n': 20} logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) #logger.setLevel(logging.DEBUG) logger.setLevel(logging.INFO) def cleanup(_, node): logger.info("waiting %ds to shutdown..." % stay_alive) reactor.callLater(stay_alive, node.stop) def testerror(failure, message, node): """ error handler for test errbacks """ logger.warn("testerror message: %s" % message) logger.warn("testerror: '%s'" % str(failure)) logger.warn("At least 1 test FAILED") def endtests(res, nKu, node, host, port): """ executes after all tests """ try: res = fdecode(res) except ValueError: pass if res != testval: return testerror(None, "retrieved value does not match stored value:" " '%s' != '%s'" % (res, testval), node) logger.info("testkFindVal PASSED") logger.debug("testkFindVal result: %s" % str(res)) logger.info("all tests PASSED") return res def testkFindVal(res, nKu, node, host, port): logger.info("testSendkFindVal PASSED") logger.debug("testSendkFindVal result: %s" % str(res)) logger.info("attempting testkFindValue") deferred = node.client.kFindValue(key) deferred.addCallback(endtests, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkFindValue", node) return deferred def testSendkFindVal(res, nKu, node, host, port): logger.info("testkStore PASSED") logger.debug("testkStore result: %s" % str(res)) logger.info("attempting testSendkFindValue") deferred = node.client.sendkFindValue(host, port, key) deferred.addCallback(testkFindVal, nKu, node, host, port) deferred.addErrback(testerror, "failed at testSendkFindValue", node) return deferred def testkStore(res, nKu, node, host, port): logger.info("testSendkStore PASSED") logger.debug("testSendkStore result: %s" % str(res)) logger.info("attempting testkStore") deferred = node.client.kStore(key, testval) deferred.addCallback(testSendkFindVal, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkStore", node) return deferred def testSendkStore(res, nKu, node, host, port): logger.info("testkFindNode PASSED") logger.debug("testkFindNode result: %s" % str(res)) logger.info("attempting testSendkStore") deferred = node.client.sendkStore(host, port, key, testval) deferred.addCallback(testkStore, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkStore", node) return deferred def testkFindNode(res, nKu, node, host, port): """ executes after testSendkFindNode """ logger.info("testSendkFindNode PASSED") logger.debug("testSendkFindNode result: %s" % str(res)) logger.info("attempting kFindNode") deferred = node.client.kFindNode(key) deferred.addCallback(testSendkStore, nKu, node, host, port) deferred.addErrback(testerror, "failed at kFindNode", node) return deferred def testSendkFindNode(nKu, node, host, port): """ executes after testGetID """ logger.info("testkGetID PASSED") logger.info("attempting sendkFindNode") deferred = node.client.sendkFindNode(host, port, key) deferred.addCallback(testkFindNode, nKu, node, host, port) deferred.addErrback(testerror, "failed at sendkFindNode", node) return deferred def testGetID(node, host, port): """ Tests sendGetID(), and invokes testSendkFindNode on success """ deferred = node.client.sendGetID(host, port) deferred.addCallback(testSendkFindNode, node, host, port) deferred.addErrback(testerror, "failed at testGetID", node) return deferred def runTests(host, port=None, listenport=None): host = getCanonicalIP(host) node = FludNode(port=listenport) if port == None: port = node.config.port logger.info("testing against %s:%s, localport=%s" % (host, port, listenport)) node.run() d = testGetID(node, host, port) d.addBoth(cleanup, node) #testkFindVal("blah", node.config.Ku, node, host, port) node.join() """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) elif len(sys.argv) == 4: runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import tarfile, tempfile, random, os, sys import gzip from Crypto.Hash import SHA256 sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.fencode import fencode import flud.TarfileUtils as TarfileUtils def maketarball(numfiles, avgsize, hashnames=False, addmetas=False): tarballname = tempfile.mktemp()+".tar" tarball = tarfile.open(tarballname, 'w') if addmetas: metafname = tempfile.mktemp() metaf = file(metafname, 'w') metaf.write('m'*48) metaf.close() for i in xrange(numfiles): fname = tempfile.mktemp() f = file(fname, 'wb') size = int(avgsize * (random.random()+0.5)) blocksize = 65*1024 if hashnames: sha256 = SHA256.new() for j in range(0, size, blocksize): if j+blocksize > size: block = 'a'*(size-j) else: block = 'a'*blocksize if hashnames: sha256.update(block) f.write(block) f.close() arcname = fname if hashnames: arcname = fencode(int(sha256.hexdigest(),16)) tarball.add(fname, arcname) if addmetas: tarball.add(metafname, arcname+".343434.meta") os.remove(fname) if addmetas: os.remove(metafname) contents = tarball.getnames() tarball.close() return tarballname, contents def gzipTarball(tarball): f = gzip.GzipFile(tarball+".gz", 'wb') f.write(file(tarball, 'rb').read()) f.close() os.remove(tarball) return tarball+".gz" def main(): # test plain TarfileUtils.delete() (tballname, contents) = maketarball(5, 4096) TarfileUtils.delete(tballname, contents[2:4]) tarball = tarfile.open(tballname, 'r') os.remove(tballname) assert(tarball.getnames() == contents[:2]+contents[4:]) tarball.close() # test gzip TarfileUtils.delete() (tballname, contents) = maketarball(5, 4096) tballname = gzipTarball(tballname) TarfileUtils.delete(tballname, contents[2:4]) tarball = tarfile.open(tballname, 'r') os.remove(tballname) assert(tarball.getnames() == contents[:2]+contents[4:]) tarball.close() # test plain TarfileUtils.concatenate() (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(gz, plain) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname1 = gzipTarball(tballname1) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(plain, gz) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname2 = gzipTarball(tballname2) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(gz, gz) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname1 = gzipTarball(tballname1) tballname2 = gzipTarball(tballname2) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.verifyHashes(plain no meta) (tballname, contents) = maketarball(5, 4096, True) assert(TarfileUtils.verifyHashes(tballname, contents[2:4])) os.remove(tballname) # test TarfileUtils.verifyHashes(plain with meta) (tballname, contents) = maketarball(5, 4096, True, True) assert(TarfileUtils.verifyHashes(tballname, contents[2:4]), ".meta") os.remove(tballname) # test TarfileUtils.verifyHashes(gzipped no meta) (tballname, contents) = maketarball(5, 4096, True) tballname = gzipTarball(tballname) assert(TarfileUtils.verifyHashes(tballname, contents[2:4])) os.remove(tballname) # test TarfileUtils.verifyHashes(gzipped with meta) (tballname, contents) = maketarball(5, 4096, True, True) tballname = gzipTarball(tballname) assert(TarfileUtils.verifyHashes(tballname, contents[2:4]), ".meta") os.remove(tballname) print "all tests passed" if __name__ == "__main__": main()
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure from twisted.internet import defer sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) import flud.FludCrypto from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import fencode, fdecode from flud.FludDefer import ErrDeferredList """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ CONCURRENT=300 CONCREPORT=50 node = None files = None logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) logger.setLevel(logging.DEBUG) def suitesuccess(results): logger.info("all tests in suite passed") #print results return results def suiteerror(failure): logger.info("suite did not complete") logger.info("DEBUG: %s" % failure) return failure def stagesuccess(result, message): logger.info("stage %s succeeded" % message) return result def stageerror(failure, message): logger.info("stage %s failed" % message) #logger.info("DEBUG: %s" % failure) return failure def itersuccess(res, i, message): if i % CONCREPORT == 0: logger.info("itersuccess: %s" % message) return res def itererror(failure, message): logger.info("itererror message: %s" % message) #logger.info("DEBUG: %s" % failure) #logger.info("DEBUG: %s" % dir(failure) failure.printTraceback() return failure def checkVERIFY(results, nKu, host, port, hashes, num=CONCURRENT): logger.info(" checking VERIFY results...") for i in range(num): hash = hashes[i] res = results[i][1] if long(hash, 16) != long(res, 16): raise failure.DefaultException("verify didn't match: %s != %s" % (hash, res)) logger.info(" ...VERIFY results good.") return results #True def testVERIFY(res, nKu, host, port, num=CONCURRENT): logger.info("testVERIFY started...") dlist = [] hashes = [] for i in range(num): #if i == 4: # port = 21 fd = os.open(files[i], os.O_RDONLY) fsize = os.fstat(fd)[stat.ST_SIZE] length = 20 offset = random.randrange(fsize-length) os.lseek(fd, offset, 0) data = os.read(fd, length) os.close(fd) hashes.append(flud.FludCrypto.hashstring(data)) filekey = os.path.basename(files[i]) deferred = node.client.sendVerify(filekey, offset, length, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testVERIFY %d" % i) deferred.addErrback(itererror, "failed at testVERIFY %d: %s" % (i, filekey)) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testVERIFY") d.addErrback(stageerror, 'failed at testVERIFY') d.addCallback(checkVERIFY, nKu, host, port, hashes, num) return d def checkRETRIEVE(res, nKu, host, port, num=CONCURRENT): logger.info(" checking RETRIEVE results...") for i in range(num): f1 = open(files[i]) filekey = os.path.basename(files[i]) f2 = open(node.config.clientdir+"/"+filekey) if (f1.read() != f2.read()): f1.close() f2.close() raise failure.DefaultException("upload/download files don't match") f2.close() f1.close() logger.info(" ...RETRIEVE results good.") return testVERIFY(res, nKu, host, port, num) def testRETRIEVE(res, nKu, host, port, num=CONCURRENT): logger.info("testRETRIEVE started...") dlist = [] for i in range(num): #if i == 4: # port = 21 filekey = os.path.basename(files[i]) deferred = node.client.sendRetrieve(filekey, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testRETRIEVE %d" % i) deferred.addErrback(itererror, "failed at testRETRIEVE %d: %s" % (i, filekey)) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testRETRIEVE") d.addErrback(stageerror, 'failed at testRETRIEVE') d.addCallback(checkRETRIEVE, nKu, host, port, num) return d def testSTORE(nKu, host, port, num=CONCURRENT): logger.info("testSTORE started...") dlist = [] for i in range(num): #if i == 4: # port = 21 deferred = node.client.sendStore(files[i], None, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testSTORE %d" % i) deferred.addErrback(itererror, "failed at testSTORE %d" % i) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testSTORE") d.addErrback(stageerror, 'failed at testSTORE') d.addCallback(testRETRIEVE, nKu, host, port, num) #d.addCallback(testVERIFY, nKu, host, port, num) return d def testID(host, port, num=CONCURRENT): logger.info("testID started...") dlist = [] for i in range(num): #if i == 4: # port = 21 deferred = node.client.sendGetID(host, port) deferred.debug = True deferred.addErrback(itererror, "failed at testID %d" % i) dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "testID") d.addErrback(stageerror, 'testID') d.addCallback(testSTORE, host, port, num) return d def runTests(host, port=None, listenport=None): num = CONCURRENT #num = 5 global files, node files = createFakeData() node = FludNode(port=listenport) if port == None: port = node.config.port node.run() if num > len(files): num = len(files) d1 = testID(host, port, num) d1.addCallback(suitesuccess) d1.addErrback(suiteerror) d1.addBoth(cleanup) #nku = FludRSA.importPublicKey({'e': 65537L, 'n': 138646504113696863667807411690225283099791076530135000331764542300161152585426296356409290228001197773401729468267448145387041995053893737880473447042984919037843163552727823101445272608470814297563395471329917904393936481407769396601027233955938405001434483474847834031774504827822809611707032477570548179411L}) #d2 = testSTORE(nku, node, host, port, files, num) #d2.addErrback(suiteerror, 'failed at %s' % d2.testname) node.join() #node.start() # doesn't work, because reactor may not have started # listening by time requests start flying def createFakeData(dir="/tmp", num=CONCURRENT): randsrc = open("/dev/urandom", 'rb') files = [] for i in range(num): randdata = randsrc.read(256) filekey = fencode(int(flud.FludCrypto.hashstring(randdata), 16)) filename = dir+'/'+filekey f = open(filename, 'wb') f.write(randdata) f.close() files.append(filename) randsrc.close() return files def deleteFakeData(files): for f in files: if os.path.exists(f): os.remove(f) else: logger.warn("s already deleted!" % f) def cleanup(dummy=None): logger.info("cleaning up files and shutting down in 1 seconds...") time.sleep(1) deleteFakeData(files) reactor.callLater(1, node.stop) logger.info("done cleaning up") """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: print "Warning: testing against self my result in timeout failures" runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket, tempfile from twisted.python import failure from StringIO import StringIO from zlib import crc32 sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient import flud.FludCrypto as FludCrypto from flud.fencode import fencode, fdecode from flud.protocol.FludCommUtil import * from flud.FludDefer import ErrDeferredList """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ # metadatablock: (block#,n,k,blockdata) metadatablock = fencode((1,20,40,'adfdsfdffffffddddddddddddddd')) fake_mkey_offset = 111111 def testerror(failure, message, node): """ error handler for test errbacks """ print "testerror message: %s" % message print "testerror: %s" % str(failure) print "At least 1 test FAILED" return failure def allGood(_, nKu): print "all tests PASSED" return nKu def checkDELETE(res, nKu, fname, fkey, mkey, node, host, port, totalDelete): """ checks to ensure the file was deleted """ # totalDelete = True if this delete op should remove all meta (and data) if totalDelete: # try to retrieve with any metakey, should fail print "expecting failed retrieve, any metakey" return testRETRIEVE(res, nKu, fname, fkey, True, node, host, port, lambda args=(res, nKu): allGood(*args), False) else: # try to retrieve with any metakey, should succeed print "expecting successful retrieve, any metakey" return testRETRIEVE(res, nKu, fname, fkey, True, node, host, port, lambda args=(res, nKu, fname, fkey, mkey+fake_mkey_offset, node, host, port, True): testDELETE(*args)) def testDELETE(res, nKu, fname, fkey, mkey, node, host, port, totalDelete): """ Tests sendDelete, and invokes checkDELETE on success """ print "starting testDELETE %s.%s" % (fname, mkey) #return checkDELETE(None, nKu, fname, fkey, mkey, node, host, port, False) deferred = node.client.sendDelete(fkey, mkey, host, port, nKu) deferred.addCallback(checkDELETE, nKu, fname, fkey, mkey, node, host, port, totalDelete) deferred.addErrback(testerror, "failed at testDELETE", node) return deferred def checkVERIFY(res, nKu, fname, fkey, mkey, node, host, port, hash, newmeta): """ executes after testVERIFY """ if long(hash, 16) != long(res, 16): raise failure.DefaultException("verify didn't match: %s != %s" % (hash, res)) print "checkVERIFY (%s) %s success" % (newmeta, fname) if newmeta: return testDELETE(res, nKu, fname, fkey, mkey, node, host, port, False) else: return testVERIFY(nKu, fname, fkey, mkey, node, host, port, True) def testVERIFY(nKu, fname, fkey, mkey, node, host, port, newmeta): """ Test sendVerify """ # newmeta, if True, will generate new metadata to be stored during verify if newmeta: thismkey = mkey+fake_mkey_offset else: thismkey = mkey print "starting testVERIFY (%s) %s.%s" % (newmeta, fname, thismkey) fd = os.open(fname, os.O_RDONLY) fsize = os.fstat(fd)[stat.ST_SIZE] length = 20 offset = random.randrange(fsize-length) os.lseek(fd, offset, 0) data = os.read(fd, length) os.close(fd) hash = FludCrypto.hashstring(data) deferred = node.client.sendVerify(fkey, offset, length, host, port, nKu, (thismkey, StringIO(metadatablock))) deferred.addCallback(checkVERIFY, nKu, fname, fkey, mkey, node, host, port, hash, newmeta) deferred.addErrback(testerror, "failed at testVERIFY (%s)" % newmeta, node) return deferred def failedRETRIEVE(res, nextCallable): return nextCallable(); def checkRETRIEVE(res, nKu, fname, fkey, mkey, node, host, port, nextCallable): """ Compares the file that was stored with the one that was retrieved """ f1 = open(fname) filename = [f for f in res if f[-len(fkey):] == fkey][0] f2 = open(filename) if (f1.read() != f2.read()): f1.close() f2.close() raise failure.DefaultException( "upload/download (%s, %s) files don't match" % (fname, os.path.join(node.config.clientdir, fkey))) #print "%s (%d) and %s (%d) match" % (fname, os.stat(fname)[stat.ST_SIZE], # filename, os.stat(filename)[stat.ST_SIZE]) f1.close() f2.close() if mkey != True: expectedmeta = "%s.%s.meta" % (fkey, mkey) metanames = [f for f in res if f[-len(expectedmeta):] == expectedmeta] if not metanames: raise failure.DefaultException("expected metadata was missing") f3 = open(metanames[0]) md = f3.read() if md != metadatablock: raise failure.DefaultException("upload/download metadata doesn't" " match (%s != %s)" % (md, metadatablock)) return nextCallable() def testRETRIEVE(res, nKu, fname, fkey, mkey, node, host, port, nextCallable, expectSuccess=True): """ Tests sendRetrieve, and invokes checkRETRIEVE on success """ print "starting testRETRIEVE %s.%s" % (fname, mkey) deferred = node.client.sendRetrieve(fkey, host, port, nKu, mkey) deferred.addCallback(checkRETRIEVE, nKu, fname, fkey, mkey, node, host, port, nextCallable) if expectSuccess: deferred.addErrback(testerror, "failed at testRETRIEVE", node) else: deferred.addErrback(failedRETRIEVE, nextCallable) return deferred def testSTORE2(nKu, fname, fkey, node, host, port): mkey = crc32(fname) mkey2 = mkey+(2*fake_mkey_offset) print "starting testSTORE %s.%s" % (fname, mkey2) deferred = node.client.sendStore(fname, (mkey2, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey2, node, host, port, lambda args=(nKu, fname, fkey, mkey, node, host, port, False): testVERIFY(*args)) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred def testSTORE(nKu, fname, fkey, node, host, port): """ Tests sendStore, and invokes testRETRIEVE on success """ mkey = crc32(fname) print "starting testSTORE %s.%s" % (fname, mkey) deferred = node.client.sendStore(fname, (mkey, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey, node, host, port, lambda args=(nKu, fname, fkey, node, host, port): testSTORE2(*args)) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred def testID(node, host, port): """ Tests sendGetID(), and invokes testSTORE on success """ print "starting testID" deferred = node.client.sendGetID(host, port) deferred.addErrback(testerror, "failed at testID", node) return deferred def testAggSTORE(nKu, aggFiles, node, host, port): print "starting testAggSTORE" dlist = [] for fname, fkey in aggFiles: mkey = crc32(fname) print "testAggSTORE %s (%s)" % (fname, mkey) deferred = node.client.sendStore(fname, (mkey, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey, node, host, port, lambda args=(nKu, fname, fkey, mkey, node, host, port, False): testVERIFY(*args)) deferred.addErrback(testerror, "failed at testAggSTORE", node) dlist.append(deferred) dl = ErrDeferredList(dlist) dl.addCallback(allGood, nKu) dl.addErrback(testerror, "failed at testAggSTORE", node) return dl def cleanup(_, node, filenamelist): for f in filenamelist: try: os.remove(f) except: print "couldn't remove %s" % f reactor.callLater(1, node.stop) def generateTestData(minSize): fname = tempfile.mktemp() f = open(fname, 'w') data = FludCrypto.generateRandom(minSize/50) for i in range(0, 51+random.randrange(50)): f.write(data) f.close() filekey = FludCrypto.hashfile(fname) filekey = fencode(int(filekey, 16)) filename = os.path.join("/tmp",filekey) os.rename(fname,filename) return (filename, filekey) def runTests(host, port=None, listenport=None): (largeFilename, largeFilekey) = generateTestData(512000) (smallFilename, smallFilekey) = generateTestData(5120) aggFiles = [] for i in range(4): aggFiles.append(generateTestData(4096)) node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testID(node, host, port) d.addCallback(testSTORE, largeFilename, largeFilekey, node, host, port) d.addCallback(testSTORE, smallFilename, smallFilekey, node, host, port) d.addCallback(testAggSTORE, aggFiles, node, host, port) d.addBoth(cleanup, node, [i[0] for i in aggFiles] + [largeFilename, smallFilename]) node.join() def main(): localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3])) if __name__ == '__main__': main()
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure from twisted.internet import defer sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) import flud.FludCrypto from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import fencode, fdecode from flud.FludDefer import ErrDeferredList """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ CONCURRENT=300 CONCREPORT=50 node = None files = None logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) logger.setLevel(logging.DEBUG) def suitesuccess(results): logger.info("all tests in suite passed") #print results return results def suiteerror(failure): logger.info("suite did not complete") logger.info("DEBUG: %s" % failure) return failure def stagesuccess(result, message): logger.info("stage %s succeeded" % message) return result def stageerror(failure, message): logger.info("stage %s failed" % message) #logger.info("DEBUG: %s" % failure) return failure def itersuccess(res, i, message): if i % CONCREPORT == 0: logger.info("itersuccess: %s" % message) return res def itererror(failure, message): logger.info("itererror message: %s" % message) #logger.info("DEBUG: %s" % failure) #logger.info("DEBUG: %s" % dir(failure) failure.printTraceback() return failure def checkVERIFY(results, nKu, host, port, hashes, num=CONCURRENT): logger.info(" checking VERIFY results...") for i in range(num): hash = hashes[i] res = results[i][1] if long(hash, 16) != long(res, 16): raise failure.DefaultException("verify didn't match: %s != %s" % (hash, res)) logger.info(" ...VERIFY results good.") return results #True def testVERIFY(res, nKu, host, port, num=CONCURRENT): logger.info("testVERIFY started...") dlist = [] hashes = [] for i in range(num): #if i == 4: # port = 21 fd = os.open(files[i], os.O_RDONLY) fsize = os.fstat(fd)[stat.ST_SIZE] length = 20 offset = random.randrange(fsize-length) os.lseek(fd, offset, 0) data = os.read(fd, length) os.close(fd) hashes.append(flud.FludCrypto.hashstring(data)) filekey = os.path.basename(files[i]) deferred = node.client.sendVerify(filekey, offset, length, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testVERIFY %d" % i) deferred.addErrback(itererror, "failed at testVERIFY %d: %s" % (i, filekey)) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testVERIFY") d.addErrback(stageerror, 'failed at testVERIFY') d.addCallback(checkVERIFY, nKu, host, port, hashes, num) return d def checkRETRIEVE(res, nKu, host, port, num=CONCURRENT): logger.info(" checking RETRIEVE results...") for i in range(num): f1 = open(files[i]) filekey = os.path.basename(files[i]) f2 = open(node.config.clientdir+"/"+filekey) if (f1.read() != f2.read()): f1.close() f2.close() raise failure.DefaultException("upload/download files don't match") f2.close() f1.close() logger.info(" ...RETRIEVE results good.") return testVERIFY(res, nKu, host, port, num) def testRETRIEVE(res, nKu, host, port, num=CONCURRENT): logger.info("testRETRIEVE started...") dlist = [] for i in range(num): #if i == 4: # port = 21 filekey = os.path.basename(files[i]) deferred = node.client.sendRetrieve(filekey, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testRETRIEVE %d" % i) deferred.addErrback(itererror, "failed at testRETRIEVE %d: %s" % (i, filekey)) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testRETRIEVE") d.addErrback(stageerror, 'failed at testRETRIEVE') d.addCallback(checkRETRIEVE, nKu, host, port, num) return d def testSTORE(nKu, host, port, num=CONCURRENT): logger.info("testSTORE started...") dlist = [] for i in range(num): #if i == 4: # port = 21 deferred = node.client.sendStore(files[i], None, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testSTORE %d" % i) deferred.addErrback(itererror, "failed at testSTORE %d" % i) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testSTORE") d.addErrback(stageerror, 'failed at testSTORE') d.addCallback(testRETRIEVE, nKu, host, port, num) #d.addCallback(testVERIFY, nKu, host, port, num) return d def testID(host, port, num=CONCURRENT): logger.info("testID started...") dlist = [] for i in range(num): #if i == 4: # port = 21 deferred = node.client.sendGetID(host, port) deferred.debug = True deferred.addErrback(itererror, "failed at testID %d" % i) dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "testID") d.addErrback(stageerror, 'testID') d.addCallback(testSTORE, host, port, num) return d def runTests(host, port=None, listenport=None): num = CONCURRENT #num = 5 global files, node files = createFakeData() node = FludNode(port=listenport) if port == None: port = node.config.port node.run() if num > len(files): num = len(files) d1 = testID(host, port, num) d1.addCallback(suitesuccess) d1.addErrback(suiteerror) d1.addBoth(cleanup) #nku = FludRSA.importPublicKey({'e': 65537L, 'n': 138646504113696863667807411690225283099791076530135000331764542300161152585426296356409290228001197773401729468267448145387041995053893737880473447042984919037843163552727823101445272608470814297563395471329917904393936481407769396601027233955938405001434483474847834031774504827822809611707032477570548179411L}) #d2 = testSTORE(nku, node, host, port, files, num) #d2.addErrback(suiteerror, 'failed at %s' % d2.testname) node.join() #node.start() # doesn't work, because reactor may not have started # listening by time requests start flying def createFakeData(dir="/tmp", num=CONCURRENT): randsrc = open("/dev/urandom", 'rb') files = [] for i in range(num): randdata = randsrc.read(256) filekey = fencode(int(flud.FludCrypto.hashstring(randdata), 16)) filename = dir+'/'+filekey f = open(filename, 'wb') f.write(randdata) f.close() files.append(filename) randsrc.close() return files def deleteFakeData(files): for f in files: if os.path.exists(f): os.remove(f) else: logger.warn("s already deleted!" % f) def cleanup(dummy=None): logger.info("cleaning up files and shutting down in 1 seconds...") time.sleep(1) deleteFakeData(files) reactor.callLater(1, node.stop) logger.info("done cleaning up") """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: print "Warning: testing against self my result in timeout failures" runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import tarfile, tempfile, random, os, sys import gzip from Crypto.Hash import SHA256 sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.fencode import fencode import flud.TarfileUtils as TarfileUtils def maketarball(numfiles, avgsize, hashnames=False, addmetas=False): tarballname = tempfile.mktemp()+".tar" tarball = tarfile.open(tarballname, 'w') if addmetas: metafname = tempfile.mktemp() metaf = file(metafname, 'w') metaf.write('m'*48) metaf.close() for i in xrange(numfiles): fname = tempfile.mktemp() f = file(fname, 'wb') size = int(avgsize * (random.random()+0.5)) blocksize = 65*1024 if hashnames: sha256 = SHA256.new() for j in range(0, size, blocksize): if j+blocksize > size: block = 'a'*(size-j) else: block = 'a'*blocksize if hashnames: sha256.update(block) f.write(block) f.close() arcname = fname if hashnames: arcname = fencode(int(sha256.hexdigest(),16)) tarball.add(fname, arcname) if addmetas: tarball.add(metafname, arcname+".343434.meta") os.remove(fname) if addmetas: os.remove(metafname) contents = tarball.getnames() tarball.close() return tarballname, contents def gzipTarball(tarball): f = gzip.GzipFile(tarball+".gz", 'wb') f.write(file(tarball, 'rb').read()) f.close() os.remove(tarball) return tarball+".gz" def main(): # test plain TarfileUtils.delete() (tballname, contents) = maketarball(5, 4096) TarfileUtils.delete(tballname, contents[2:4]) tarball = tarfile.open(tballname, 'r') os.remove(tballname) assert(tarball.getnames() == contents[:2]+contents[4:]) tarball.close() # test gzip TarfileUtils.delete() (tballname, contents) = maketarball(5, 4096) tballname = gzipTarball(tballname) TarfileUtils.delete(tballname, contents[2:4]) tarball = tarfile.open(tballname, 'r') os.remove(tballname) assert(tarball.getnames() == contents[:2]+contents[4:]) tarball.close() # test plain TarfileUtils.concatenate() (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(gz, plain) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname1 = gzipTarball(tballname1) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(plain, gz) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname2 = gzipTarball(tballname2) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(gz, gz) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname1 = gzipTarball(tballname1) tballname2 = gzipTarball(tballname2) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.verifyHashes(plain no meta) (tballname, contents) = maketarball(5, 4096, True) assert(TarfileUtils.verifyHashes(tballname, contents[2:4])) os.remove(tballname) # test TarfileUtils.verifyHashes(plain with meta) (tballname, contents) = maketarball(5, 4096, True, True) assert(TarfileUtils.verifyHashes(tballname, contents[2:4]), ".meta") os.remove(tballname) # test TarfileUtils.verifyHashes(gzipped no meta) (tballname, contents) = maketarball(5, 4096, True) tballname = gzipTarball(tballname) assert(TarfileUtils.verifyHashes(tballname, contents[2:4])) os.remove(tballname) # test TarfileUtils.verifyHashes(gzipped with meta) (tballname, contents) = maketarball(5, 4096, True, True) tballname = gzipTarball(tballname) assert(TarfileUtils.verifyHashes(tballname, contents[2:4]), ".meta") os.remove(tballname) print "all tests passed" if __name__ == "__main__": main()
Python
from distutils.core import setup, Extension import os, sys setup(name="flud", version="0.1.1", description="flud decentralized backup", long_description='a 100% decentralized backup system', author="Alen Peacock", author_email="apeacock@flud.org", url='http://flud.org', license='GPLv3 (c)2004-2007 Alen Peacock', packages=['flud', 'flud.protocol', 'flud.bin', 'flud.test'], package_dir={'flud': 'flud', 'flud.protocol': 'flud/protocol', 'flud.bin': 'flud/bin', 'flud.test': 'flud/test'}, package_data={'flud': ['images/*.png', 'FludNode.tac', 'fludrules.init']}, scripts = ['flud/bin/fludnode', 'flud/bin/tacpath-flud', 'flud/bin/fludscheduler', 'flud/bin/fludclient', 'flud/bin/fludlocalclient', 'flud/bin/flud-mastermetadataViewer', 'flud/bin/flud-metadataViewer', 'flud/bin/start-fludnodes', 'flud/bin/stop-fludnodes', 'flud/bin/gauges-fludnodes', 'flud/bin/clean-fludnodes'] )
Python
""" FludCrypto.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Provides FludRSA (an enhanced RSA.RSAobj), as well as convenience functions for creating hashes, finding hash collisions, etc. """ import binascii import operator import struct import time from Crypto.Hash import SHA256 from Crypto.PublicKey import RSA, pubkey from Crypto.Util.randpool import RandomPool class FludRSA(RSA.RSAobj): """ Subclasses the Crypto.PublicKey.RSAobj object to add access to the privatekey as well as methods for exporting and importing an RSA obj. """ rand = RandomPool() def __init__(self, rsa): self.__setstate__(rsa.__getstate__()) def publickey(self): return FludRSA(RSA.construct((self.n, self.e))) def privatekey(self): return FludRSA(RSA.construct((self.n, self.e, self.d))) def encrypt(self, message): return RSA.RSAobj.encrypt(self, message, "") def exportPublicKey(self): return self.publickey().__getstate__() def exportPrivateKey(self): return self.privatekey().__getstate__() def id(self): """ returns the hashstring of the public key """ #return hashstring(str(self.exportPublicKey())) return hashstring(str(self.exportPublicKey()['n'])) def importPublicKey(key): """ Can take, as key, a dict describing the public key ('e' and 'n'), a string describing n, or a long describing n (in the latter two cases, e is assumed to be 65537L). """ if isinstance(key, str): key = long(key, 16) key = {'e': 65537L, 'n': key} elif isinstance(key, long): key = {'e': 65537L, 'n': key} if isinstance(key, dict): state = key pkey = RSA.construct((0L,0L)) pkey.__setstate__(state) return FludRSA(pkey) else: raise TypeError("type %s not supported by importPublicKey():"\ " try dict with keys of 'e' and 'n', string representing"\ " 'n', or long representing 'n'." % type(key)) importPublicKey = staticmethod(importPublicKey) def importPrivateKey(key): state = key pkey = RSA.construct((0L,0L,0L)) pkey.__setstate__(state) return FludRSA(pkey) importPrivateKey = staticmethod(importPrivateKey) def generate(keylength): return FludRSA(RSA.generate(keylength, FludRSA.rand.get_bytes)) generate = staticmethod(generate) def generateKeys(len=1024): fludkey = FludRSA.generate(len) return fludkey.publickey(), fludkey.privatekey() def hashstring(string): sha256 = SHA256.new() sha256.update(string) return sha256.hexdigest() def hashfile(filename): sha256 = SHA256.new() f = open(filename, "r") while 1: buf = f.read(1048576) # XXX: 1Mb - magic number if buf == "": break sha256.update(buf) f.close() return sha256.hexdigest() def hashstream(file, len): sha256 = SHA256.new() readsize = 1048576 # XXX: 1Mb - magic number while len > 0: if len < readsize: readsize = len buf = file.read(readsize) if buf == "": break sha256.update(buf) len = len - readsize return sha256.hexdigest() def generateRandom(n): rand = RandomPool() # using seperate instance of RandomPool purposely return rand.get_bytes(n) def hashcash(match, len, timestamp=False): """ trys to find a hash collision of len significant bits. Returns the 256-bit string that produced the collision. Uses sha256, so match should be a sha256 hashstring (as a hexstring), and len should be between 0 and 256 (lengths close to 256 are intractable). The timestamp field determines whether the current timestamp should be inserted into the pre-hash result (to stem sybil attacks targetting specific IDs). The result is hex-encoded, so to arrive at the matching hashvalue, you would hashstring(binascii.unhexlify(result)). """ matchint = long(match,16) len = 2**(256-len) if date: gtime = struct.pack("I",int(time.time())) while True: attempt = generateRandom(32) # 32 random bytes = 256 random bits if date: # rewrite the 2 lsBs of attempt with the 2 msBs of gtime (time # granularity is thus 65536 seconds, or just over 18 hours between # intervals -- more than enough for a refresh monthly, weekly, or # even daily value) attempt = attempt[0:30]+gtime[2:4] attempthash = hashstring(attempt) attemptint = long(attempthash,16) distance = operator.xor(matchint, attemptint) if distance < len: break return binascii.hexlify(attempt) # XXX: should move all testing to doctest if __name__ == '__main__': fludkey = FludRSA.generate(1024) print "fludkey (pub) is: "+str(fludkey.exportPublicKey()) print "fludkey (priv) is: "+str(fludkey.exportPrivateKey()) print "" pubkeystring = fludkey.exportPublicKey() pubkeylongn = pubkeystring['n'] pubkeystringn = hex(pubkeystring['n']) privkeystring = fludkey.exportPrivateKey() fludkeyPub = FludRSA.importPublicKey(pubkeystring) print "fludkeyPub is: "+str(fludkeyPub.exportPublicKey()) fludkeyPub2 = FludRSA.importPublicKey(pubkeystringn) print "fludkeyPub2 is: "+str(fludkeyPub2.exportPublicKey()) fludkeyPub3 = FludRSA.importPublicKey(pubkeylongn) print "fludkeyPub3 is: "+str(fludkeyPub3.exportPublicKey()) fludkeyPriv = FludRSA.importPrivateKey(privkeystring) print "fludkeyPriv is: "+str(fludkeyPriv.exportPrivateKey()) plaintext = "test message" print "plaintext is: "+plaintext ciphertext = fludkeyPub.encrypt(plaintext) print "ciphertext is: "+str(ciphertext) plaintext2 = fludkeyPriv.decrypt(ciphertext) print "decrypted plaintext is: "+plaintext2 randstring = str(generateRandom(80)) print "80 bytes of random data: '"+binascii.hexlify(randstring) data1='\x00\x1e4%`K\xef\xf6\xdd\x8a\x0eUP\x7f\xb0G\x1d\xb9\xe4\x82\x11n\n\xff\x1a\xc9\x013\xe9\x8e\x99\xb0]M@y\x86l\xb3l' edata1=fludkeyPub.encrypt(data1)[0] data2=fludkeyPriv.decrypt(edata1) print binascii.hexlify(data1) print binascii.hexlify(data2) print data1 == data2
Python
""" FludServer.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. flud server operations """ import threading, binascii, time, os, stat, httplib, gc, re, sys, logging, sets from twisted.web import server, resource, client from twisted.web.resource import Resource from twisted.internet import reactor, threads, defer from twisted.web import http from twisted.python import threadable, failure from flud.FludCrypto import FludRSA import flud.FludkRouting from ServerPrimitives import * from ServerDHTPrimitives import * from LocalPrimitives import * from FludCommUtil import * threadable.init() class FludServer(threading.Thread): """ This class runs the webserver, responding to all requests. """ def __init__(self, node, port): threading.Thread.__init__(self) self.port = port self.node = node self.clientport = node.config.clientport self.logger = node.logger self.root = ROOT(self) commandmap = node.config.commandmap self.root.putChild(commandmap['ID'], ID(self)) self.root.putChild(commandmap['STORE'], STORE(self)) self.root.putChild(commandmap['RETRIEVE'], RETRIEVE(self)) self.root.putChild(commandmap['VERIFY'], VERIFY(self)) self.root.putChild(commandmap['PROXY'], PROXY(self)) self.root.putChild(commandmap['DELETE'], DELETE(self)) self.root.putChild(commandmap['kFINDNODE'], kFINDNODE(self)) self.root.putChild(commandmap['kFINDVAL'], kFINDVAL(self)) self.root.putChild(commandmap['kSTORE'], kSTORE(self)) self.site = server.Site(self.root) reactor.listenTCP(self.port, self.site) reactor.listenTCP(self.clientport, LocalFactory(node), interface="127.0.0.1") #print "FludServer will listen on port %d, local client on %d"\ # % (self.port, self.clientport) self.logger.log(logging.INFO,\ "FludServer will listen on port %d, local client on %d" % (self.port, self.clientport)) def run(self): self.logger.log(logging.INFO, "FludServer starting") return reactor.run(installSignalHandlers=0) def stop(self): self.logger.log(logging.INFO, "FludServer stopping") reactor.stop()
Python
""" LocalPrimitives.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Protocol for talking to a flud node locally (from client code -- command line, GUI, etc). Each command in the local protocol begins with 4 bytes designating the type of command. These are followed by a status byte, which is either '?'=request, ':'=success response, or '!'=failure response. Arguments to the command follow the status byte. """ import binascii, time, os, stat, httplib, gc, re, sys, logging, sets from twisted.web.resource import Resource from twisted.web import server, resource, client from twisted.internet import protocol, reactor, threads, defer from twisted.protocols import basic from twisted.mail import smtp from twisted.python import failure from Crypto.Cipher import AES from flud.FludCrypto import FludRSA, hashstring, generateRandom import flud.FludkRouting from flud.fencode import fencode, fdecode from flud.FludFileOperations import * from FludCommUtil import * from FludServer import * logger = logging.getLogger("flud.local.server") """ Protocol and Factory for local client/server communication """ MAXCONCURRENT = 300 (CONCURR, MAX, QUEUE) = (0, 1, 2) # indexes into LocalProtocol.commands class LocalProtocol(basic.LineReceiver): authenticated = False commands = {'PUTF': [0, MAXCONCURRENT, []], 'GETF': [0, MAXCONCURRENT, []], 'GETI': [0, MAXCONCURRENT, []], 'FNDN': [0, 1, []], 'STOR': [0, MAXCONCURRENT, []], 'RTRV': [0, MAXCONCURRENT, []], 'VRFY': [0, MAXCONCURRENT, []], 'FNDV': [0, 1, []], 'CRED': [0, 1, []], 'LIST': [0, 1, []], 'GETM': [0, 1, []], 'PUTM': [0, 1, []] } def connectionMade(self): logger.info("client connected") self.authenticated=False def connectionLost(self, reason): self.authenticated=False def doOp(self, command, fname): #print "got command '%s'" % command if command == "PUTF": logger.debug("PUTF %s", fname); return StoreFile(self.factory.node, fname).deferred elif command == "GETI": logger.debug("GETI %s", fname); return RetrieveFile(self.factory.node, fname).deferred elif command == "GETF": logger.debug("GETF %s", fname); return RetrieveFilename(self.factory.node, fname).deferred elif command == "FNDN": logger.debug("FNDN %s" % fname); try: intval = long(fname, 16) except: return defer.fail("fname was not hex") return self.factory.node.client.kFindNode(intval) # The following is for testing aggregation of kFindNode on same key #dl = [] #for i in [1,2,3,4,5]: # d = self.factory.node.client.kFindNode(intval) # dl.append(d) #dlist = defer.DeferredList(dl) #return dlist elif command == "FNDV": logger.debug("FNDV %s", fname); try: intval = long(fname, 16) except: return defer.fail("fname was not hex") return self.factory.node.client.kFindValue(intval) elif command == "CRED": passphrase, email = fdecode(fname) # XXX: allow an optional passphrase hint to be sent in email. passphrase = self.factory.node.config.Kr.decrypt(passphrase) logger.debug("CRED %s to %s", passphrase, email); Kr = self.factory.node.config.Kr.exportPrivateKey() Kr['g'] = self.factory.node.config.groupIDr fKr = fencode(Kr) key = AES.new(binascii.unhexlify(hashstring(passphrase))) fKr = '\x00'*(16-(len(fKr)%16))+fKr efKr = fencode(key.encrypt(fKr)) logger.debug("efKr = %s " % efKr) d = smtp.sendmail('localhost', "your_flud_client@localhost", email, "Subject: Your encrypted flud credentials\n\n" "Hopefully, you'll never need to use this email. Its " "sole purpose is to help you recover your data after a " "catastrophic and complete loss of the original computer " "or hard drive.\n\n" "In that unlucky event, you'll need a copy of your flud " "credentials, which I've included below, sitting between " "the \"---+++---\" markers. These credentials were " "encrypted with a passphrase of your choosing when you " "installed the flud software. I'll only say this " "once:\n\n" "YOU MUST REMEMBER THAT PASSWORD IN ORDER TO RECOVER YOUR " "CREDENTIALS. If you are unable to remember the " "passphrase and your computer fails catastrophically " "(losing its local copy of these credentials), you will " "not be able to recover your data." "\n\n" "Luckily, that's all you should ever need in order to " "recover all your data: your passphrase and these " "credentials." "\n\n" "Please save this email. You may want to print out hard " "copies and store them safely, forward this email to " "other email accounts, etc. Since the credentials are " "encrypted, others won't be able to steal them " "without guessing your passphrase. " "\n\n" "---+++---\n"+efKr+"\n---+++---\n") return d # to decode this email, we search for the '---+++---' markers, make # sure the intervening data is all in one piece (remove any line # breaks \r or \n inserted by email clients) and call this 'cred', # reconstruct the AES key with the H(passphrase) (as above), and # then use the key to .decrypt(fdecode(cred)) and call this dcred, # then fdecode(dcred[dcred.find('d'):]) and call this ddcred, and # finally importPrivateKey(ddcred) and set groupIDr to ddcred['g']. elif command == "LIST": logger.debug("LIST") return defer.succeed(self.factory.config.master) elif command == "GETM": logger.debug("GETM") return RetrieveMasterIndex(self.factory.node).deferred elif command == "PUTM": logger.debug("PUTM") return UpdateMasterIndex(self.factory.node).deferred else: #print "fname is '%s'" % fname host = fname[:fname.find(':')] port = fname[fname.find(':')+1:fname.find(',')] fname = fname[fname.find(',')+1:] print "%s: %s : %s , %s" % (command, host, port, fname) if command == "STOR": logger.debug("STOR"); return self.factory.node.client.sendStore(fname, None, host, int(port)) elif command == "RTRV": logger.debug("RTRV"); return self.factory.node.client.sendRetrieve(fname, host, int(port)) elif command == "VRFY": logger.debug("VRFY"); offset = port[port.find(':')+1:port.find('-')] length = port[port.find('-')+1:] port = port[:port.find(':')] print "%s: %s : %s %s - %s , %s" % (command, host, port, offset, length, fname) return self.factory.node.client.sendVerify(fname, int(offset), int(length), host, int(port)) else: logger.debug("bad op"); return defer.fail("bad op") def serviceQueue(self, command): if len(self.commands[command][QUEUE]) > 0 and \ self.commands[command][CONCURR] <= self.commands[command][MAX]: data = self.commands[command][QUEUE].pop() logger.info("servicing queue['%s'], item %s" % (command, data)) print "taking %s off the queue" % command d = self.doOp(command, data) d.addCallback(self.sendSuccess, command, data) d.addErrback(self.sendFailure, command, data) def sendSuccess(self, resp, command, data, prepend=None): logger.debug("SUCCESS! "+command+":"+data) #logger.debug("response: '%s'" % (resp,)) if prepend: w = "%s:%s %s:%s\r\n" % (prepend, command, fencode(resp), data) else: w = "%s:%s:%s\r\n" % (command, fencode(resp), data) self.transport.write(w) self.commands[command][CONCURR] -= 1 try: self.serviceQueue(command) except: print sys.exec_info() return resp def sendFailure(self, err, command, data, prepend=None): logger.debug("FAILED! %s!%s" % (command, data)) errmsg = err.getErrorMessage() if prepend: w = "%s!%s %s!%s\r\n" % (prepend, command, errmsg, data) else: w = "%s!%s!%s\r\n" % (command, errmsg, data) logger.debug("sending %s" % w) self.transport.write(w) self.commands[command][CONCURR] -= 1 self.serviceQueue(command) return err def lineReceived(self, line): logger.debug("lineReceived: '%s'" % line) # commands: AUTH, PUTF, GETF, VRFY # status: ? = request, : = successful response, ! = failed response command = line[0:4] status = line[4] data = line[5:] #print "data is '%s'" % data if not self.authenticated and command == "AUTH": if status == '?': # asked for AUTH challenge to be sent. send it logger.debug("AUTH challenge requested, sending") echallenge = self.factory.sendChallenge() self.transport.write("AUTH?"+echallenge+"\r\n") elif status == ':' and self.factory.challengeAnswered(data): # sent AUTH response and it passed logger.debug("AUTH challenge successful") self.authenticated = True self.transport.write("AUTH:\r\n") elif status == ':': logger.debug("AUTH challenge failed") self.transport.write("AUTH!\r\n") elif command == "DIAG": if data == "NODE": logger.debug("DIAG NODE") nodetups = self.factory.config.routing.knownExternalNodes() nodes = [] for n in nodetups: node = list(n) if self.factory.config.reputations.has_key(n[2]): node.append(self.factory.config.reputations[n[2]]) else: node.append(0) nodes.append(tuple(node)) self.transport.write("DIAG:NODE%s\r\n" % fencode(nodes)) elif data == "BKTS": logger.debug("DIAG BKTS") bucks = eval("%s" % self.factory.config.routing.kBuckets) self.transport.write("DIAG:BKTS%s\r\n" % fencode(bucks)) else: dcommand = data[:4] ddata = data[5:] logger.debug("DIAG %s %s" % (dcommand, ddata)) self.commands[dcommand][CONCURR] += 1 d = self.doOp(dcommand, ddata) d.addCallback(self.sendSuccess, dcommand, ddata, "DIAG") d.addErrback(self.sendFailure, dcommand, ddata, "DIAG") elif status == '?': # requested an operation to be performed. If we are below our # maximum concurrent ops, do the operation. Otherwise, put it on # the queue to be serviced when current ops finish. Response is # sent back to client when deferreds fire. if self.commands[command][CONCURR] >= self.commands[command][MAX]: #print "putting %s on the queue" % line logger.info("received %s request, enqueuing" % command) self.commands[command][QUEUE].insert(0, data) else: #print "doing %s" % line logger.info("received %s request, executing" % command) print self.commands[command] self.commands[command][CONCURR] += 1 d = self.doOp(command, data) d.addCallback(self.sendSuccess, command, data) d.addErrback(self.sendFailure, command, data) class LocalFactory(protocol.ServerFactory): protocol = LocalProtocol def __init__(self, node): self.node = node self.config = node.config def sendChallenge(self): self.challenge = fencode(generateRandom(challengelength)) echallenge = self.config.Ku.encrypt(self.challenge)[0] echallenge = fencode(echallenge) return echallenge def challengeAnswered(self, resp): return resp == self.challenge
Python
#!/usr/bin/python """ LocalClient.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. LocalClient provides client functions which can be called to send commands to a local FludNode instance. """ import sys, os, time from twisted.internet.protocol import ClientFactory from twisted.protocols.basic import LineReceiver from twisted.internet import reactor from twisted.python import threadable threadable.init() from flud.fencode import fencode, fdecode from LocalPrimitives import * logger = logging.getLogger("flud.local.client") opTimeout = 1200 VALIDOPS = LocalProtocol.commands.keys() + ['AUTH', 'DIAG'] # XXX: print commands should either be raised, or put on factory.msgs class LocalClient(LineReceiver): MAX_LENGTH = 300000 auth=False def connectionMade(self): logger.debug("connection est.") self.auth=False self.sendLine("AUTH?") def lineReceived(self, line): logger.debug("received line '%s'" % line) command = line[0:4] if not command in VALIDOPS: print "error: invalid command op ('%s')-- "\ " are you trying to connect to the wrong port"\ " (local client port is usually external port + 500)?"\ % command return None status = line[4] data = line[5:] if not self.auth: if command == "AUTH" and status == '?': # got challenge, send response logger.debug("got AUTH challenge, sending response") echallenge = data self.sendLine("AUTH:"+self.factory.answerChallenge(echallenge)) return elif command == "AUTH" and status == ':': # response accepted, authenticated logger.debug("AUTH challenge accepted, success") self.auth = True self.factory.clientReady(self) #print "authenticated" else: if command == "AUTH" and status == "!": logger.warn("authentication failed (is FLUDHOME set" " correctly?)") print "authentication failed (is FLUDHOME set correctly?)" else: logger.warn("unknown message received before being" " authenticated:") logger.warn(" %s : %s" % (command, status)) print "unknown message received before being authenticated:" print " %s : %s" % (command, status) self.factory.setDie() elif command == "DIAG": subcommand = data[:4] data = data[4:] if subcommand == "NODE": logger.debug("DIAG NODE: %s" % data) data = fdecode(data) result = "" for i in data: score = i[4] petID = "%064x" % i[2] netID = "%s:%d" % (i[0], i[1]) petID = petID[:(70-len(netID))]+"..." result += "%d %s %s\n" % (score, netID, petID) result += "%d known nodes\n" % len(data) d = self.factory.pending['NODE'].pop('') d.callback(result) return if subcommand == "BKTS": logger.debug("DIAG BKTS") data = fdecode(data) result = "" for i in data: for bucket in i: result += "Bucket %s:\n" % bucket for k in i[bucket]: id = "%064x" % k[2] netID = "%s:%d" % (k[0], k[1]) result += " %s %s...\n" \ % (netID,id[:72-len(netID)]) d = self.factory.pending['BKTS'].pop('') d.callback(result) return elif status == ':': response, data = data.split(status, 1) logger.debug("DIAG %s: success" % subcommand) d = self.factory.pending[subcommand].pop(data) d.callback(fdecode(response)) elif status == "!": response, data = data.split(status, 1) logger.debug("DIAG %s: failure" % subcommand) d = self.factory.pending[subcommand].pop(data) d.errback(failure.DefaultException(response)) elif status == ':': response, data = data.split(status, 1) logger.debug("%s: success" % command) d = self.factory.pending[command].pop(data) d.callback(fdecode(response)) elif status == "!": response, data = data.split(status, 1) logger.debug("%s: failure" % command) if self.factory.pending.has_key(command): if not self.factory.pending[command].has_key(data): print "data key is '%s'" % data print "pending is '%s'" % self.factory.pending[command] if len(self.factory.pending[command]): d = self.factory.pending[command].popitem() d.errback(failure.DefaultException(response)) else: d = self.factory.pending[command].pop(data) d.errback(failure.DefaultException(response)) else: print "failed command '%s' not in pending?" % command print "pending is: %s" % self.factory.pending if command != 'AUTH' and command != 'DIAG' and \ not None in self.factory.pending[command].values(): logger.debug("%s done at %s" % (command, time.ctime())) class LocalClientFactory(ClientFactory): protocol = LocalClient def __init__(self, config): self.config = config self.messageQueue = [] self.client = None self.die = False self.pending = {'PUTF': {}, 'CRED': {}, 'GETI': {}, 'GETF': {}, 'FNDN': {}, 'STOR': {}, 'RTRV': {}, 'VRFY': {}, 'FNDV': {}, 'CRED': {}, 'LIST': {}, 'GETM': {}, 'PUTM': {}, 'NODE': {}, 'BKTS': {}} def clientConnectionFailed(self, connector, reason): #print "connection failed: %s" % reason logger.warn("connection failed: %s" % reason) self.cleanup("connection failed: %s" % reason) def clientConnectionLost(self, connector, reason): #print "connection lost: %s" % reason logger.debug("connection lost: %s" % reason) self.cleanup("connection lost: %s" % reason) def cleanup(self, msg): # override me for cleanup print msg; def clientReady(self, instance): self.client = instance logger.debug("client ready, sending [any] queued msgs") for i in self.messageQueue: self._sendMessage(i) def _sendMessage(self, msg): if self.client: logger.debug("sending msg '%s'" % msg) self.client.sendLine(msg) else: logger.debug("queueing msg '%s'" % msg) self.messageQueue.append(msg) def answerChallenge(self, echallenge): logger.debug("answering challenge") echallenge = (fdecode(echallenge),) challenge = self.config.Kr.decrypt(echallenge) return challenge def expire(self, pending, key): if pending.has_key(fname): logger.debug("timing out operation for %s" % key) #print "timing out operation for %s" % key pending.pop(key) def addFile(self, type, fname): logger.debug("addFile %s %s" % (type, fname)) if not self.pending[type].has_key(fname): d = defer.Deferred() self.pending[type][fname] = d self._sendMessage(type+"?"+fname) return d else: return self.pending[type][fname] def sendPING(self, host, port): logger.debug("sendPING") d = defer.Deferred() d.errback(failure.DefaultException( "ping not yet implemented in FludLocalClient")) return d def sendPUTF(self, fname): logger.debug("sendPUTF %s" % fname) if os.path.isdir(fname): dirlist = os.listdir(fname) dlist = [] for i in dirlist: dlist.append(self.sendPUTF(os.path.join(fname,i))) dl = defer.DeferredList(dlist) return dl elif not self.pending['PUTF'].has_key(fname): d = defer.Deferred() self.pending['PUTF'][fname] = d self._sendMessage("PUTF?"+fname) #reactor.callLater(opTimeout, self.expire, self.pendingPUTF, fname) return d else: return self.pending['PUTF'][fname] def sendCRED(self, passphrase, email): logger.debug("sendCRED") key = fencode((self.config.Ku.encrypt(passphrase)[0], email)) if not self.pending['CRED'].has_key(key): d = defer.Deferred() self.pending['CRED'][key] = d self._sendMessage("CRED?"+key) return d else: return self.pending['CRED'][key] def sendGETI(self, fID): logger.debug("sendGETI") if not self.pending['GETI'].has_key(fID): d = defer.Deferred() self.pending['GETI'][fID] = d self._sendMessage("GETI?"+fID) return d else: return self.pending['GETI'][fID] def sendGETF(self, fname): logger.debug("sendGETF") master = listMeta(self.config) if master.has_key(fname): return self.addFile("GETF",fname) elif fname[-1:] == os.path.sep: dlist = [] for name in master: if fname == name[:len(fname)]: dlist.append(self.addFile("GETF",name)) dl = defer.DeferredList(dlist) return dl def sendFNDN(self, nID): logger.debug("sendFNDN") if not self.pending['FNDN'].has_key(nID): d = defer.Deferred() self.pending['FNDN'][nID] = d self._sendMessage("FNDN?"+nID) return d else: return self.pending['FNDN'][nID] def sendLIST(self): logger.debug("sendLIST") if not self.pending['LIST'].has_key(""): d = defer.Deferred() self.pending['LIST'][''] = d logger.debug("LIST['']=%s" % d) self._sendMessage("LIST?") return d else: return self.pending['LIST'][''] def sendGETM(self): logger.debug("sendGETM") if not self.pending['GETM'].has_key(''): d = defer.Deferred() self.pending['GETM'][''] = d logger.debug("GETM['']=%s" % d) self._sendMessage("GETM?") return d else: return self.pending['GETM'][''] def sendPUTM(self): logger.debug("sendPUTM") if not self.pending['PUTM'].has_key(''): d = defer.Deferred() self.pending['PUTM'][''] = d self._sendMessage("PUTM?") return d else: return self.pending['PUTM'][''] def sendDIAGNODE(self): logger.debug("sendDIAGNODE") if not self.pending['NODE'].has_key(''): d = defer.Deferred() self.pending['NODE'][''] = d self._sendMessage("DIAG?NODE") return d else: return self.pending['NODE'][''] def sendDIAGBKTS(self): logger.debug("sendDIAGBKTS") if not self.pending['BKTS'].has_key(''): d = defer.Deferred() self.pending['BKTS'][''] = d self._sendMessage("DIAG?BKTS") return d else: return self.pending['BKTS'][''] def sendDIAGSTOR(self, command): logger.debug("sendDIAGSTOR") if not self.pending['STOR'].has_key(command): d = defer.Deferred() self.pending['STOR'][command] = d self._sendMessage("DIAG?STOR "+command) return d else: return self.pending['STOR'][command] def sendDIAGRTRV(self, command): logger.debug("sendDIAGRTRV") if not self.pending['RTRV'].has_key(command): d = defer.Deferred() self.pending['RTRV'][command] = d self._sendMessage("DIAG?RTRV "+command) return d else: return self.pending['RTRV'][command] def sendDIAGVRFY(self, command): logger.debug("sendDIAGVRFY") if not self.pending['VRFY'].has_key(command): d = defer.Deferred() self.pending['VRFY'][command] = d self._sendMessage("DIAG?VRFY "+command) return d else: return self.pending['VRFY'][command] def sendDIAGFNDV(self, val): logger.debug("sendDIAGFNDV") if not self.pending['FNDV'].has_key(val): d = defer.Deferred() self.pending['FNDV'][val] = d self._sendMessage("FNDV?"+val) return d else: return self.pending['FNDV'][val] def setDie(self): self.die = True # XXX: this should move into FludNode side of things (LocalClientPrimitives). # anything that calls this should make calls ('LIST', others as necessary) to # get at master metadata, otherwise we could have multiple writer problems. # FludNode should make the file ro while running, too. # And everyone that does anything with the master metadata should do it through # methods of FludConfig, instead of by direct access to the file. def listMeta(config): fmaster = open(os.path.join(config.metadir,config.metamaster), 'r') master = fmaster.read() fmaster.close() if master == "": master = {} else: master = fdecode(master) return master
Python
""" ServerDHTPrimitives.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Primitive server DHT protocol """ import binascii, time, os, stat, httplib, gc, re, sys, logging, random, sets from twisted.web.resource import Resource from twisted.web import server, resource, http, client from twisted.internet import reactor, defer from twisted.python import failure from flud.FludCrypto import FludRSA from flud.fencode import fencode, fdecode from ServerPrimitives import ROOT from FludCommUtil import * logger = logging.getLogger("flud.server.dht") # XXX: move kRouting.insertNode code out of FludConfig. Add a 'addNode' method # to FludProtocol module which calls config.addNode, calls # kRouting.insertNode and if it gets a return value, calls sendGetID with # the callback doing nothing (unless the header comes back in error) and # the errback calling kRouting.replaceNode. Anywhere that we are # currently called config.addNode, call this new method instead. # FUTURE: check flud protocol version for backwards compatibility # XXX: need to make sure we have appropriate timeouts for all comms. # FUTURE: DOS attacks. For now, assume that network hardware can filter these # out (by throttling individual IPs) -- i.e., it isn't our problem. If we # want to defend against this at some point, we need to keep track of who # is generating requests and then ignore them. # XXX: find everywhere we are sending longs and consider sending hex (or our # own base-64) encoded instead # XXX: might want to consider some self-healing for the kademlia layer, as # outlined by this thread: # http://zgp.org/pipermail/p2p-hackers/2003-August/001348.html (should also # consider Zooko's links in the parent to this post) """ The children of ROOT beginning with 'k' are kademlia protocol based. """ # XXX: need to do all the challenge/response jazz in the k classes class kFINDNODE(ROOT): isLeaf = True def render_GET(self, request): """ Return the k closest nodes to the target ID from local k-routing table """ self.setHeaders(request) self.node.DHTtstamp = time.time() try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'key') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by kFINDNODE" logger.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: logger.info("received kFINDNODE request from %s..." % params['nodeID'][:10]) reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) host = getCanonicalIP(request.getClientIP()) #return "{'id': '%s', 'k': %s}"\ # % (self.config.nodeID,\ # self.config.routing.findNode(fdecode(params['key']))) kclosest = self.config.routing.findNode(fdecode(params['key'])) notclose = list(set(self.config.routing.knownExternalNodes()) - set(kclosest)) if len(notclose) > 0 and len(kclosest) > 1: r = random.choice(notclose) #logger.info("**** got some notclose: %s:%d ****" % (r[0],r[1])) kclosest.append(r) #logger.info("returning kFINDNODE response: %s" % kclosest) updateNode(self.node.client, self.config, host, int(params['port']), reqKu, params['nodeID']) return "{'id': '%s', 'k': %s}" % (self.config.nodeID, kclosest) class kSTORE_true(ROOT): # unrestricted kSTORE. Will store any key/value pair, as in generic # kademlia. This should be unregistered in FludServer (can't allow # generic stores). isLeaf = True def render_PUT(self, request): self.setHeaders(request) try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'key', 'val') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by kSTORE_true" logger.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: logger.info("received kSTORE_true request from %s..." % params['nodeID'][:10]) reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) host = getCanonicalIP(request.getClientIP()) updateNode(self.node.client, self.config, host, int(params['port']), reqKu, params['nodeID']) fname = self.config.kstoredir+'/'+params['key'] logger.info("storing dht data to %s" % fname) f = open(fname, "wb") f.write(params['val']) f.close() return "" class kSTORE(ROOT): # XXX: To prevent abuse of the DHT layer, we impose restrictions on its # format. But format alone is not sufficient -- a malicious client # could still format its data in a way that is allowed and gain # arbitrary amounts of freeloading storage space in the DHT. To # prevent this, nodes storing data in the DHT layer must also validate # it. Validation simply requires that the blockIDs described in the # kSTORE actually reside at a significant percentage of the hosts # described in the kSTORE op. In other words, validation requires a # VERIFY op for each block described in the kSTORE op. Validation can # occur randomly sometime after a kSTORE operation, or at the time of # the kSTORE op. The former is better, because it not only allows # purging bad kSTOREs, but prevents them from happening in the first # place (without significant conspiring among all participants). # Since the originator of this request also needs to do a VERIFY, # perhaps we can piggyback these through some means. And, since the # kSTORE is replicated to k other nodes, each of which also should to # a VERIFY, there are several ways to optimize this. One is for the k # nodes to elect a single verifier, and allow the client to learn the # result of the VERIFY op. Another is to allow each k node to do its # own VERIFY, but stagger them in a way such that they can take the # place of the originator's first k VERIFY ops. This could be # coordinated or (perhaps better) allow each k node to randomly pick a # time at which it will VERIFY, distributed over a period for which it # is likely to cover many of the first k VERIFY ops generated by # the originator. The random approach is nice because it is the same # mechanism used by the k nodes to occasionally verify that the DHT # data is valid and should not be purged. isLeaf = True def render_PUT(self, request): self.setHeaders(request) self.node.DHTtstamp = time.time() try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'key', 'val') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by kSTORE" logger.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: logger.info("received kSTORE request from %s..." % params['nodeID'][:10]) reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) host = getCanonicalIP(request.getClientIP()) updateNode(self.node.client, self.config, host, int(params['port']), reqKu, params['nodeID']) fname = self.config.kstoredir+'/'+params['key'] md = fdecode(params['val']) if not self.dataAllowed(params['key'], md, params['nodeID']): msg = "malformed store data" logger.info("bad data was: %s" % md) request.setResponseCode(http.BAD_REQUEST, msg) return msg # XXX: see if there isn't already a 'val' for 'key' present # - if so, compare to val. Metadata can differ. Blocks # shouldn't. However, if blocks do differ, just add the # new values in, up to N (3?) records per key. Flag these # (all N) as ones we want to verify (to storer and storee). # Expunge any blocks that fail verify, and punish storer's # trust. logger.info("storing dht data to %s" % fname) if os.path.exists(fname) and isinstance(md, dict): f = open(fname, "rb") edata = f.read() f.close() md = self.mergeMetadata(md, fdecode(edata)) f = open(fname, "wb") f.write(fencode(md)) f.close() return "" # XXX: return a VERIFY reverse request: segname, offset def dataAllowed(self, key, data, nodeID): # ensures that 'data' is in [one of] the right format[s] (helps prevent # DHT abuse) def validValue(val): if not isinstance(val, long) and not isinstance(val, int): return False # not a valid key/nodeid if val > 2**256 or val < 0: # XXX: magic 2**256, use fludkrouting return False # not a valid key/nodeid return True def validMetadata(blockdata, nodeID): # returns true if the format of data conforms to the standard for # metadata blocks = 0 try: k = blockdata.pop('k') n = blockdata.pop('n') if not isinstance(k, int) or not isinstance(n, int): return False if k != 20 or n != 20: # XXX: magic numbers '20' # XXX: to support other than 20/20, need to constrain an # upper bound and store multiple records with different m/n # under the same key return False m = k+n except: return False for (i, b) in blockdata: if i > m: return False if not validValue(b): #print "%s is invalid key" %i return False location = blockdata[(i,b)] if isinstance(location, list): if len(location) > 5: #print "too many (list) nodeIDs" % j return False for j in location: if not validValue(j): #print "%s is invalid (list) nodeID" % j return False elif not validValue(location): #print "%s is invalid nodeID" % location return False blocks += 1 if blocks != m: return False # not the right number of blocks blockdata['k'] = k blockdata['n'] = n return True def validMasterCAS(key, data, nodeID): # returns true if the data fits the characteristics of a master # metadata CAS key, i.e., if key==nodeID and the data is the right # length. nodeID = fencode(long(nodeID,16)) if key != nodeID: return False # XXX: need to do challange/response on nodeID (just as in the # regular primitives) here, or else imposters can store/replace # this very important data!!! # XXX: do some length stuff - should only be as long as a CAS key return True return (validMetadata(data, nodeID) or validMasterCAS(key, data, nodeID)) def mergeMetadata(self, m1, m2): # merges the data from m1 into m2. After calling, both m1 and m2 # contain the merged data. """ >>> a1 = {'b': {1: (1, 'a', 8), 2: (2, 'b', 8), 5: (1, 'a', 8)}} >>> a2 = {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80)}} >>> mergeit(a1, a2) {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80), (2, 'b', 8)], 10: (10, 't', 80), 5: (1, 'a', 8)}} >>> a1 = {'b': {1: (1, 'a', 8), 2: (2, 'b', 8), 5: (1, 'a', 8)}} >>> a2 = {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80)}} >>> mergeit(a2, a1) {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80), (2, 'b', 8)], 10: (10, 't', 80), 5: (1, 'a', 8)}} >>> a1 = {'b': {1: (1, 'a', 8), 2: [(2, 'b', 8), (7, 'r', 8)], 5: (1, 'a', 8)}} >>> a2 = {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80)}} >>> mergeit(a2, a1) {'b': {1: (1, 'a', 8), 2: [(2, 'b', 8), (7, 'r', 8), (3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80), 5: (1, 'a', 8)}} >>> a1 = {'b': {1: [(1, 'a', 8)], 2: [(2, 'b', 8), (7, 'r', 8)], 5: (1, 'a', 8)}} >>> a2 = {'b': {1: (1, 'a', 8), 2: [(3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80)}} >>> mergeit(a1, a2) {'b': {1: (1, 'a', 8), 2: [(2, 'b', 8), (7, 'r', 8), (3, 'B', 80), (4, 'bb', 80)], 10: (10, 't', 80), 5: (1, 'a', 8)}} """ # first merge blocks ('b' sections) n = {} for i in m2: if m1.has_key(i) and m2[i] != m1[i]: if isinstance(m1[i], list) and len(m1[i]) == 1: m1[i] = m1[i][0] # collapse list of len 1 if isinstance(m2[i], list) and len(m2[i]) == 1: m2[i] = m2[i][0] # collapse list of len 1 # combine if isinstance(m1[i], list) and isinstance(m2[i], list): n[i] = m2[i] n[i].extend(m1[i]) elif isinstance(m2[i], list): n[i] = m2[i] n[i] = n[i].append(m1[i]) elif isinstance(m1[i], list): n[i] = m1[i] n[i] = n[i].append(m2[i]) elif m1[i] == m2[i]: n[i] = m1[i] else: n[i] = [m1[i], m2[i]] else: n[i] = m2[i] for i in m1: if not n.has_key(i): n[i] = m1[i] # now n contains the merged blocks. m1 = m2 = n return m1 class kFINDVAL(ROOT): isLeaf = True def render_GET(self, request): """ Return the value, or if we don't have it, the k closest nodes to the target ID """ self.setHeaders(request) self.node.DHTtstamp = time.time() try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'key') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by kFINDVALUE" logger.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: logger.info("received kFINDVALUE request from %s..." % params['nodeID'][:10]) reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) host = getCanonicalIP(request.getClientIP()) updateNode(self.node.client, self.config, host, int(params['port']), reqKu, params['nodeID']) fname = self.config.kstoredir+'/'+params['key'] if os.path.isfile(fname): f = open(fname, "rb") logger.info("returning data from kFINDVAL") request.setHeader('nodeID',str(self.config.nodeID)) request.setHeader('Content-Type','application/x-flud-data') d = fdecode(f.read()) if isinstance(d, dict) and d.has_key(params['nodeID']): #print d resp = {'b': d['b'], params['nodeID']: d[params['nodeID']]} #resp = {'b': d['b']} #if d.has_key(params['nodeID']): # resp[params['nodeID']] = d[params['nodeID']] else: resp = d request.write(fencode(resp)) f.close() return "" else: # return the following if it isn't there. logger.info("returning nodes from kFINDVAL for %s" % params['key']) request.setHeader('Content-Type','application/x-flud-nodes') return "{'id': '%s', 'k': %s}"\ % (self.config.nodeID,\ self.config.routing.findNode(fdecode(params['key'])))
Python
#!/usr/bin/python """ LocalClient.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. LocalClient provides client functions which can be called to send commands to a local FludNode instance. """ import sys, os, time from twisted.internet.protocol import ClientFactory from twisted.protocols.basic import LineReceiver from twisted.internet import reactor from twisted.python import threadable threadable.init() from flud.fencode import fencode, fdecode from LocalPrimitives import * logger = logging.getLogger("flud.local.client") opTimeout = 1200 VALIDOPS = LocalProtocol.commands.keys() + ['AUTH', 'DIAG'] # XXX: print commands should either be raised, or put on factory.msgs class LocalClient(LineReceiver): MAX_LENGTH = 300000 auth=False def connectionMade(self): logger.debug("connection est.") self.auth=False self.sendLine("AUTH?") def lineReceived(self, line): logger.debug("received line '%s'" % line) command = line[0:4] if not command in VALIDOPS: print "error: invalid command op ('%s')-- "\ " are you trying to connect to the wrong port"\ " (local client port is usually external port + 500)?"\ % command return None status = line[4] data = line[5:] if not self.auth: if command == "AUTH" and status == '?': # got challenge, send response logger.debug("got AUTH challenge, sending response") echallenge = data self.sendLine("AUTH:"+self.factory.answerChallenge(echallenge)) return elif command == "AUTH" and status == ':': # response accepted, authenticated logger.debug("AUTH challenge accepted, success") self.auth = True self.factory.clientReady(self) #print "authenticated" else: if command == "AUTH" and status == "!": logger.warn("authentication failed (is FLUDHOME set" " correctly?)") print "authentication failed (is FLUDHOME set correctly?)" else: logger.warn("unknown message received before being" " authenticated:") logger.warn(" %s : %s" % (command, status)) print "unknown message received before being authenticated:" print " %s : %s" % (command, status) self.factory.setDie() elif command == "DIAG": subcommand = data[:4] data = data[4:] if subcommand == "NODE": logger.debug("DIAG NODE: %s" % data) data = fdecode(data) result = "" for i in data: score = i[4] petID = "%064x" % i[2] netID = "%s:%d" % (i[0], i[1]) petID = petID[:(70-len(netID))]+"..." result += "%d %s %s\n" % (score, netID, petID) result += "%d known nodes\n" % len(data) d = self.factory.pending['NODE'].pop('') d.callback(result) return if subcommand == "BKTS": logger.debug("DIAG BKTS") data = fdecode(data) result = "" for i in data: for bucket in i: result += "Bucket %s:\n" % bucket for k in i[bucket]: id = "%064x" % k[2] netID = "%s:%d" % (k[0], k[1]) result += " %s %s...\n" \ % (netID,id[:72-len(netID)]) d = self.factory.pending['BKTS'].pop('') d.callback(result) return elif status == ':': response, data = data.split(status, 1) logger.debug("DIAG %s: success" % subcommand) d = self.factory.pending[subcommand].pop(data) d.callback(fdecode(response)) elif status == "!": response, data = data.split(status, 1) logger.debug("DIAG %s: failure" % subcommand) d = self.factory.pending[subcommand].pop(data) d.errback(failure.DefaultException(response)) elif status == ':': response, data = data.split(status, 1) logger.debug("%s: success" % command) d = self.factory.pending[command].pop(data) d.callback(fdecode(response)) elif status == "!": response, data = data.split(status, 1) logger.debug("%s: failure" % command) if self.factory.pending.has_key(command): if not self.factory.pending[command].has_key(data): print "data key is '%s'" % data print "pending is '%s'" % self.factory.pending[command] if len(self.factory.pending[command]): d = self.factory.pending[command].popitem() d.errback(failure.DefaultException(response)) else: d = self.factory.pending[command].pop(data) d.errback(failure.DefaultException(response)) else: print "failed command '%s' not in pending?" % command print "pending is: %s" % self.factory.pending if command != 'AUTH' and command != 'DIAG' and \ not None in self.factory.pending[command].values(): logger.debug("%s done at %s" % (command, time.ctime())) class LocalClientFactory(ClientFactory): protocol = LocalClient def __init__(self, config): self.config = config self.messageQueue = [] self.client = None self.die = False self.pending = {'PUTF': {}, 'CRED': {}, 'GETI': {}, 'GETF': {}, 'FNDN': {}, 'STOR': {}, 'RTRV': {}, 'VRFY': {}, 'FNDV': {}, 'CRED': {}, 'LIST': {}, 'GETM': {}, 'PUTM': {}, 'NODE': {}, 'BKTS': {}} def clientConnectionFailed(self, connector, reason): #print "connection failed: %s" % reason logger.warn("connection failed: %s" % reason) self.cleanup("connection failed: %s" % reason) def clientConnectionLost(self, connector, reason): #print "connection lost: %s" % reason logger.debug("connection lost: %s" % reason) self.cleanup("connection lost: %s" % reason) def cleanup(self, msg): # override me for cleanup print msg; def clientReady(self, instance): self.client = instance logger.debug("client ready, sending [any] queued msgs") for i in self.messageQueue: self._sendMessage(i) def _sendMessage(self, msg): if self.client: logger.debug("sending msg '%s'" % msg) self.client.sendLine(msg) else: logger.debug("queueing msg '%s'" % msg) self.messageQueue.append(msg) def answerChallenge(self, echallenge): logger.debug("answering challenge") echallenge = (fdecode(echallenge),) challenge = self.config.Kr.decrypt(echallenge) return challenge def expire(self, pending, key): if pending.has_key(fname): logger.debug("timing out operation for %s" % key) #print "timing out operation for %s" % key pending.pop(key) def addFile(self, type, fname): logger.debug("addFile %s %s" % (type, fname)) if not self.pending[type].has_key(fname): d = defer.Deferred() self.pending[type][fname] = d self._sendMessage(type+"?"+fname) return d else: return self.pending[type][fname] def sendPING(self, host, port): logger.debug("sendPING") d = defer.Deferred() d.errback(failure.DefaultException( "ping not yet implemented in FludLocalClient")) return d def sendPUTF(self, fname): logger.debug("sendPUTF %s" % fname) if os.path.isdir(fname): dirlist = os.listdir(fname) dlist = [] for i in dirlist: dlist.append(self.sendPUTF(os.path.join(fname,i))) dl = defer.DeferredList(dlist) return dl elif not self.pending['PUTF'].has_key(fname): d = defer.Deferred() self.pending['PUTF'][fname] = d self._sendMessage("PUTF?"+fname) #reactor.callLater(opTimeout, self.expire, self.pendingPUTF, fname) return d else: return self.pending['PUTF'][fname] def sendCRED(self, passphrase, email): logger.debug("sendCRED") key = fencode((self.config.Ku.encrypt(passphrase)[0], email)) if not self.pending['CRED'].has_key(key): d = defer.Deferred() self.pending['CRED'][key] = d self._sendMessage("CRED?"+key) return d else: return self.pending['CRED'][key] def sendGETI(self, fID): logger.debug("sendGETI") if not self.pending['GETI'].has_key(fID): d = defer.Deferred() self.pending['GETI'][fID] = d self._sendMessage("GETI?"+fID) return d else: return self.pending['GETI'][fID] def sendGETF(self, fname): logger.debug("sendGETF") master = listMeta(self.config) if master.has_key(fname): return self.addFile("GETF",fname) elif fname[-1:] == os.path.sep: dlist = [] for name in master: if fname == name[:len(fname)]: dlist.append(self.addFile("GETF",name)) dl = defer.DeferredList(dlist) return dl def sendFNDN(self, nID): logger.debug("sendFNDN") if not self.pending['FNDN'].has_key(nID): d = defer.Deferred() self.pending['FNDN'][nID] = d self._sendMessage("FNDN?"+nID) return d else: return self.pending['FNDN'][nID] def sendLIST(self): logger.debug("sendLIST") if not self.pending['LIST'].has_key(""): d = defer.Deferred() self.pending['LIST'][''] = d logger.debug("LIST['']=%s" % d) self._sendMessage("LIST?") return d else: return self.pending['LIST'][''] def sendGETM(self): logger.debug("sendGETM") if not self.pending['GETM'].has_key(''): d = defer.Deferred() self.pending['GETM'][''] = d logger.debug("GETM['']=%s" % d) self._sendMessage("GETM?") return d else: return self.pending['GETM'][''] def sendPUTM(self): logger.debug("sendPUTM") if not self.pending['PUTM'].has_key(''): d = defer.Deferred() self.pending['PUTM'][''] = d self._sendMessage("PUTM?") return d else: return self.pending['PUTM'][''] def sendDIAGNODE(self): logger.debug("sendDIAGNODE") if not self.pending['NODE'].has_key(''): d = defer.Deferred() self.pending['NODE'][''] = d self._sendMessage("DIAG?NODE") return d else: return self.pending['NODE'][''] def sendDIAGBKTS(self): logger.debug("sendDIAGBKTS") if not self.pending['BKTS'].has_key(''): d = defer.Deferred() self.pending['BKTS'][''] = d self._sendMessage("DIAG?BKTS") return d else: return self.pending['BKTS'][''] def sendDIAGSTOR(self, command): logger.debug("sendDIAGSTOR") if not self.pending['STOR'].has_key(command): d = defer.Deferred() self.pending['STOR'][command] = d self._sendMessage("DIAG?STOR "+command) return d else: return self.pending['STOR'][command] def sendDIAGRTRV(self, command): logger.debug("sendDIAGRTRV") if not self.pending['RTRV'].has_key(command): d = defer.Deferred() self.pending['RTRV'][command] = d self._sendMessage("DIAG?RTRV "+command) return d else: return self.pending['RTRV'][command] def sendDIAGVRFY(self, command): logger.debug("sendDIAGVRFY") if not self.pending['VRFY'].has_key(command): d = defer.Deferred() self.pending['VRFY'][command] = d self._sendMessage("DIAG?VRFY "+command) return d else: return self.pending['VRFY'][command] def sendDIAGFNDV(self, val): logger.debug("sendDIAGFNDV") if not self.pending['FNDV'].has_key(val): d = defer.Deferred() self.pending['FNDV'][val] = d self._sendMessage("FNDV?"+val) return d else: return self.pending['FNDV'][val] def setDie(self): self.die = True # XXX: this should move into FludNode side of things (LocalClientPrimitives). # anything that calls this should make calls ('LIST', others as necessary) to # get at master metadata, otherwise we could have multiple writer problems. # FludNode should make the file ro while running, too. # And everyone that does anything with the master metadata should do it through # methods of FludConfig, instead of by direct access to the file. def listMeta(config): fmaster = open(os.path.join(config.metadir,config.metamaster), 'r') master = fmaster.read() fmaster.close() if master == "": master = {} else: master = fdecode(master) return master
Python
""" ConnectionQueue, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. This module manages the connection queue. In order to reduce the probability of the reactor getting tied up servicing requests/responses during periods of extreme busy-ness (and thus 'starving' some ops, causing TimeoutErrors), we throttle the number of outstanding requests that we send to MAXOPS. The rest, we put in the 'waiting' queue, and these are popped off when a spot becomes available. """ import logging MAXOPS = 80 # maximum number of concurrent connections to maintain pending = 0 # number of current connections waiting = [] # queue of waiting connections to make. This queue contains # tuples. The first element of the tuple must have a # startRequest() func that takes the rest of the tuple as # arguments. logger = logging.getLogger("flud.client.connq") def checkWaiting(resp, finishedOne=True): """ This function pops items off the waiting queue and invokes their startRequest() method. It should eventually be called by any process that also calls queueWaiting() (usually as part of callback/errback chain). The 'resp' object passed in will be returned (so that this function can sit transparently in the errback/callback chain). """ numwaiting = len(waiting) logger.debug("in checkWaiting, len(waiting) = %d" % numwaiting) #print "resp = %s..." % fencode(long(resp,16))[:8] #print "resp = %s..." % str(resp) global pending if finishedOne: pending = pending - 1 logger.debug("decremented pending to %s" % pending) if numwaiting > 0 and pending < MAXOPS: saved = waiting.pop(0) Req = saved[0] args = saved[1:] logger.debug("w: %d, p: %d, restoring Request %s(%s)" % (numwaiting, pending, Req.__class__.__name__, str(args))) Req.startRequest(*args) pending += 1 return resp def enqueue(requestTuple): """ Adds a requestTuple to those waiting. The first element of the tuple must have a startRequest() func that takes the rest of the tuple as arguments. This startRequest() function will be called with those arguments when it comes off the queue (via checkWaiting). """ waiting.append(requestTuple) logger.debug("trying to do %s now..." % requestTuple[0].__class__.__name__) checkWaiting(None, finishedOne=False)
Python
""" ClientPrimitives.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Primitive client storage protocol """ from twisted.web import http, client from twisted.internet import reactor, threads, defer, error from twisted.python import failure import time, os, stat, httplib, sys, logging, tarfile, gzip from StringIO import StringIO from flud.FludCrypto import FludRSA from flud.fencode import fencode, fdecode import ConnectionQueue from FludCommUtil import * logger = logging.getLogger("flud.client.op") loggerid = logging.getLogger("flud.client.op.id") loggerstor = logging.getLogger("flud.client.op.stor") loggerstoragg = logging.getLogger("flud.client.op.stor.agg") loggerrtrv = logging.getLogger("flud.client.op.rtrv") loggerdele = logging.getLogger("flud.client.op.dele") loggervrfy = logging.getLogger("flud.client.op.vrfy") loggerauth = logging.getLogger("flud.client.op.auth") MINSTORSIZE = 512000 # anything smaller than this tries to get aggregated TARFILE_TO = 2 # timeout for checking aggregated tar files MAXAUTHRETRY = 4 # number of times to retry auth # FUTURE: check flud protocol version for backwards compatibility # XXX: need to make sure we have appropriate timeouts for all comms. # FUTURE: DOS attacks. For now, assume that network hardware can filter these # out (by throttling individual IPs) -- i.e., it isn't our problem. If we # want to defend against this at some point, we need to keep track of who # is generating requests and then ignore them. # XXX: might want to consider some self-healing for the kademlia layer, as # outlined by this thread: # http://zgp.org/pipermail/p2p-hackers/2003-August/001348.html (should also # consider Zooko's links in the parent to this post) # XXX: disallow requests to self. class REQUEST(object): """ This is a parent class for generating http requests that follow the FludProtocol. """ def __init__(self, host, port, node=None): """ All children should inherit. By convention, subclasses should create a URL and attempt to retrieve it in the constructor. @param node the requestor's node object """ self.host = host self.port = port self.dest = "%s:%d" % (host, port) if node: self.node = node self.config = node.config self.headers = {'Fludprotocol': fludproto_ver, 'User-Agent': 'FludClient 0.1'} class SENDGETID(REQUEST): def __init__(self, node, host, port): """ Send a request to retrive the node's ID. This is a reciprocal request -- must send my own ID in order to get one back. """ host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) Ku = self.node.config.Ku.exportPublicKey() url = "http://"+host+":"+str(port)+"/ID?" url += 'nodeID='+str(self.node.config.nodeID) url += '&port='+str(self.node.config.port) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) #self.nKu = {} self.timeoutcount = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, node, host, port, url)) def startRequest(self, node, host, port, url): loggerid.info("sending SENDGETID to %s" % self.dest) d = self._sendRequest(node, host, port, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) d.addErrback(self._errID, node, host, port, url) def _sendRequest(self, node, host, port, url): factory = getPageFactory(url, timeout=primitive_to, headers=self.headers) d2 = factory.deferred d2.addCallback(self._getID, factory, host, port) d2.addErrback(self._errID, node, host, port, url) return d2 def _getID(self, response, factory, host, port): loggerid.debug( "received ID response: %s" % response) if not hasattr(factory, 'status'): raise failure.DefaultException( "SENDGETID FAILED: no status in factory") if eval(factory.status) != http.OK: raise failure.DefaultException("SENDGETID FAILED to "+self.dest+": " +"server sent status "+factory.status+", '"+response+"'") try: nKu = {} nKu = eval(response) nKu = FludRSA.importPublicKey(nKu) loggerid.info("SENDGETID PASSED to %s" % self.dest) updateNode(self.node.client, self.config, host, port, nKu) return nKu except: raise failure.DefaultException("SENDGETID FAILED to "+self.dest+": " +"received response, but it did not contain valid key") def _errID(self, err, node, host, port, url): if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #print "GETID request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(node, host, port, url) else: #print "not trying again [#%d]" % self.timeoutcount return err # XXX: updateNode #print "_errID: %s" % err #print "_errID: %s" % str(err.stack) return err # XXX: either 1) produce filekey here or 2) send it in as part of API # (similar fixes for SENDRETRIEVE and VERIFY)? Currently filekey is # chosen by caller, and is simply the filename. class SENDSTORE(REQUEST): def __init__(self, nKu, node, host, port, datafile, metadata=None, fsize=0): """ Try to upload a file. """ host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) loggerstor.info("sending STORE request to %s" % self.dest) if not fsize: fsize = os.stat(datafile)[stat.ST_SIZE] Ku = self.node.config.Ku.exportPublicKey() params = [('nodeID', self.node.config.nodeID), ('Ku_e', str(Ku['e'])), ('Ku_n', str(Ku['n'])), ('port', str(self.node.config.port)), ('filekey', os.path.basename(datafile)), ('size', str(fsize))] self.timeoutcount = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, self.headers, nKu, host, port, datafile, metadata, params, True)) #self.deferred = self._sendRequest(self.headers, nKu, host, port, # datafile, params, True) def startRequest(self, headers, nKu, host, port, datafile, metadata, params, skipFile): d = self._sendRequest(headers, nKu, host, port, datafile, metadata, params, skipFile) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, headers, nKu, host, port, datafile, metadata, params, skipfile=False): """ skipfile - set to True if you want to send everything but file data (used to send the unauthorized request before responding to challenge) """ if skipfile: files = [(None, 'filename')] elif metadata: metakey = metadata[0] params.append(('metakey', metakey)) metafile = metadata[1] files = [(datafile, 'filename'), (metafile, 'meta')] else: files = [(datafile, 'filename')] deferred = threads.deferToThread(fileUpload, host, port, '/STORE', files, params, headers=self.headers) deferred.addCallback(self._getSendStore, nKu, host, port, datafile, metadata, params, self.headers) deferred.addErrback(self._errSendStore, "Couldn't upload file %s to %s:%d" % (datafile, host, port), self.headers, nKu, host, port, datafile, metadata, params) return deferred def _getSendStore(self, httpconn, nKu, host, port, datafile, metadata, params, headers): """ Check the response for status. """ deferred2 = threads.deferToThread(httpconn.getresponse) deferred2.addCallback(self._getSendStore2, httpconn, nKu, host, port, datafile, metadata, params, headers) deferred2.addErrback(self._errSendStore, "Couldn't get response", headers, nKu, host, port, datafile, metadata, params, httpconn) return deferred2 def _getSendStore2(self, response, httpconn, nKu, host, port, datafile, metadata, params, headers): httpconn.close() if response.status == http.UNAUTHORIZED: loggerstor.info("SENDSTORE unauthorized, sending credentials") challenge = response.reason d = answerChallengeDeferred(challenge, self.node.config.Kr, self.node.config.groupIDu, nKu.id(), headers) d.addCallback(self._sendRequest, nKu, host, port, datafile, metadata, params) d.addErrback(self._errSendStore, "Couldn't answerChallenge", headers, nKu, host, port, datafile, metadata, params, httpconn) return d elif response.status == http.CONFLICT: result = response.read() # XXX: client should check key before ever sending request raise BadCASKeyException("%s %s" % (response.status, response.reason)) elif response.status != http.OK: result = response.read() raise failure.DefaultException( "received %s in SENDSTORE response: %s" % (response.status, result)) else: result = response.read() updateNode(self.node.client, self.config, host, port, nKu) loggerstor.info("received SENDSTORE response from %s: %s" % (self.dest, str(result))) return result def _errSendStore(self, err, msg, headers, nKu, host, port, datafile, metadata, params, httpconn=None): if err.check('socket.error'): #print "SENDSTORE request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(headers, nKu, host, port, datafile, metadata, params) else: print "Maxtimeouts exceeded: %d" % self.timeoutcount elif err.check(BadCASKeyException): pass else: print "%s: unexpected error in SENDSTORE: %s" % (msg, str(err.getErrorMessage())) # XXX: updateNode if httpconn: httpconn.close() #loggerstor.info(msg+": "+err.getErrorMessage()) return err aggDeferredMap = {} # a map of maps, containing a list of deferreds. The # deferred(s) for file 'x' in tarball 'y' are accessed as # aggDeferredMap['y']['x'] aggTimeoutMap = {} # a map of timout calls for a tarball. The timeout for # tarball 'y' is stored in aggTimeoutMap['y'] class AggregateStore: # XXX: if multiple guys store the same file, we're going to get into bad # cb state (the except clause in errbackTarfiles). Need to catch this # as it happens... (this happens e.g. for small files with the same # filehash, e.g, 0-byte files, file copies etc). Should fix this in # FludClient -- non-agg store has a similar problem (encoded file chunks # get deleted out from under successive STOR ops for the same chunk, i.e. # from two concurrent STORs of the same file contents) def __init__(self, nKu, node, host, port, datafile, metadata): tarfilename = os.path.join(node.config.clientdir,nKu.id())\ +'-'+host+'-'+str(port)+".tar" loggerstoragg.debug("tarfile name is %s" % tarfilename) if not os.path.exists(tarfilename) \ or not aggDeferredMap.has_key(tarfilename): loggerstoragg.debug("creating tarfile %s to append %s" % (tarfilename, datafile)) tar = tarfile.open(tarfilename, "w") tarfileTimeout = reactor.callLater(TARFILE_TO, self.sendTar, tarfilename, nKu, node, host, port) aggDeferredMap[tarfilename] = {} aggTimeoutMap[tarfilename] = tarfileTimeout else: loggerstoragg.debug("opening tarfile %s to append %s" % (tarfilename, datafile)) tar = tarfile.open(tarfilename, "a") if os.path.basename(datafile) not in tar.getnames(): loggerstoragg.info("adding datafile %s to tarball, %s" % (os.path.basename(datafile), tar.getnames())) loggerstoragg.debug("adding data to tarball") tar.add(datafile, os.path.basename(datafile)) else: loggerstoragg.info("skip adding datafile %s to tarball" % datafile) if metadata: metafilename = "%s.%s.meta" % (os.path.basename(datafile), metadata[0]) loggerstoragg.debug("metadata filename is %s" % metafilename) try: if isinstance(metadata[1], StringIO): loggerstoragg.debug("metadata is StringIO") tinfo = tarfile.TarInfo(metafilename) metadata[1].seek(0,2) tinfo.size = metadata[1].tell() metadata[1].seek(0,0) tar.addfile(tinfo, metadata[1]) else: loggerstoragg.debug("metadata is file") tar.add(metadata[1], metafilename) except: import traceback loggerstoragg.debug("exception while adding metadata to" " tarball") print sys.exc_info()[2] traceback.print_exc() tar.close() loggerstoragg.debug("prepping deferred") # XXX: (re)set timeout for tarfilename self.deferred = defer.Deferred() loggerstoragg.debug("adding deferred on %s for %s" % (tarfilename, datafile)) try: aggDeferredMap[tarfilename][os.path.basename(datafile)].append( self.deferred) except KeyError: aggDeferredMap[tarfilename][os.path.basename(datafile)] \ = [self.deferred] self.resetTimeout(aggTimeoutMap[tarfilename], tarfilename) def resetTimeout(self, timeoutFunc, tarball): loggerstoragg.debug("in resetTimeout...") if timeoutFunc.active(): #timeoutFunc.reset(TARFILE_TO) if os.stat(tarball)[stat.ST_SIZE] < MINSTORSIZE: loggerstoragg.debug("...reset") timeoutFunc.reset(TARFILE_TO) return loggerstoragg.debug("...didn't reset") def sendTar(self, tarball, nKu, node, host, port): gtarball = tarball+".gz" loggerstoragg.info( "aggregation op triggered, sending tarfile %s to %s:%d" % (gtarball, host, port)) # XXX: bad blocking io gtar = gzip.GzipFile(gtarball, 'wb') gtar.write(file(tarball, 'r').read()) gtar.close() os.remove(tarball) self.deferred = SENDSTORE(nKu, node, host, port, gtarball).deferred self.deferred.addCallback(self.callbackTarfiles, tarball) self.deferred.addErrback(self.errbackTarfiles, tarball) # XXX: make aggDeferredMap use a non-.tar key, so that we don't have to # keep passing 'tarball' around (since we removed it and are really only # interested in gtarball now, use gtarball at the least) def callbackTarfiles(self, result, tarball): loggerstoragg.debug("callbackTarfiles") gtarball = tarball+".gz" tar = tarfile.open(gtarball, "r:gz") cbs = [] try: for tarinfo in tar: if tarinfo.name[-5:] != '.meta': dlist = aggDeferredMap[tarball].pop(tarinfo.name) loggerstoragg.debug("callingback for %s in %s" " (%d deferreds)" % (tarinfo.name, tarball, len(dlist))) for d in dlist: cbs.append(d) except KeyError: loggerstoragg.warn("aggDeferredMap has keys: %s" % str(aggDeferredMap.keys())) loggerstoragg.warn("aggDeferredMap[%s] has keys: %s" % (tarball, str(aggDeferredMap[tarball].keys()))) tar.close() loggerstoragg.debug("deleting tarball %s" % gtarball) os.remove(gtarball) for cb in cbs: cb.callback(result) def errbackTarfiles(self, failure, tarball): loggerstoragg.debug("errbackTarfiles") gtarball = tarball+".gz" tar = tarfile.open(gtarball, "r:gz") cbs = [] try: for tarinfo in tar: dlist = aggDeferredMap[tarball].pop(tarinfo.name) loggerstoragg.debug("erringback for %s in %s" " (%d deferreds)" % (tarinfo.name, tarball, len(dlist))) for d in dlist: cbs.append(d) except KeyError: loggerstoragg.warn("aggDeferredMap has keys: %s" % str(aggDeferredMap.keys())) loggerstoragg.warn("aggDeferredMap[%s] has keys: %s" % (tarball, str(aggDeferredMap[tarball].keys()))) tar.close() loggerstoragg.debug("NOT deleting tarball %s (for debug)" % gtarball) #os.remove(gtarball) for cb in cbs: cb.errback(failure) class SENDRETRIEVE(REQUEST): def __init__(self, nKu, node, host, port, filekey, metakey=True): """ Try to download a file. """ host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) loggerrtrv.info("sending RETRIEVE request to %s:%s" % (host, str(port))) Ku = self.node.config.Ku.exportPublicKey() url = 'http://'+host+':'+str(port)+'/RETRIEVE/'+filekey+'?' url += 'nodeID='+str(self.node.config.nodeID) url += '&port='+str(self.node.config.port) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) url += "&metakey="+str(metakey) #filename = self.node.config.clientdir+'/'+filekey self.timeoutcount = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, self.headers, nKu, host, port, url)) def startRequest(self, headers, nKu, host, port, url): #print "doing RET: %s" % filename d = self._sendRequest(headers, nKu, host, port, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, headers, nKu, host, port, url): factory = multipartDownloadPageFactory(url, self.node.config.clientdir, headers=headers, timeout=transfer_to) deferred = factory.deferred deferred.addCallback(self._getSendRetrieve, nKu, host, port, factory) deferred.addErrback(self._errSendRetrieve, nKu, host, port, factory, url, headers) return deferred def _getSendRetrieve(self, response, nKu, host, port, factory): if eval(factory.status) == http.OK: # response is None, since it went to file (if a server error # occured, it may be printed in this file) # XXX: need to check that file hashes to key! If we don't do this, # malicious nodes can corrupt entire files without detection! #result = "received SENDRETRIEVE response" loggerrtrv.info(response) updateNode(self.node.client, self.config, host, port, nKu) return response else: raise failure.DefaultException("SENDRETRIEVE FAILED: " +"server sent status "+factory.status+", '"+response+"'") def _errSendRetrieve(self, err, nKu, host, port, factory, url, headers): if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #or \ #err.check('twisted.internet.error.ConnectBindError'): self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "RETR trying again [#%d]..." % self.timeoutcount return self._sendRequest(headers, nKu, host, port, url) else: #print "RETR timeout exceeded: %d" % self.timeoutcount pass elif hasattr(factory, 'status') and \ eval(factory.status) == http.UNAUTHORIZED: loggerrtrv.info("SENDRETRIEVE unauthorized, sending credentials") challenge = err.getErrorMessage()[4:] d = answerChallengeDeferred(challenge, self.node.config.Kr, self.node.config.groupIDu, nKu.id(), headers) d.addCallback(self._sendRequest, nKu, host, port, url) #d.addErrback(self._errSendRetrieve, nKu, host, port, factory, # url, headers) return d #extraheaders = answerChallenge(challenge, self.node.config.Kr, # self.node.config.groupIDu, nKu.id(), self.headers) #return self._sendRequest(nKu, host, port, url, extraheaders) # XXX: these remaining else clauses are really just for debugging... elif hasattr(factory, 'status'): if eval(factory.status) == http.NOT_FOUND: err = NotFoundException(err) elif eval(factory.status) == http.BAD_REQUEST: err = BadRequestException(err) elif err.check('twisted.internet.error.ConnectionRefusedError'): pass # fall through to return err else: print "non-timeout, non-UNAUTH RETR request error: %s" % err # XXX: updateNode loggerrtrv.info("SENDRETRIEVE failed") raise err class SENDDELETE(REQUEST): def __init__(self, nKu, node, host, port, filekey, metakey): """ Try to delete a file. """ host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) loggerdele.info("sending DELETE request to %s:%s" % (host, str(port))) Ku = self.node.config.Ku.exportPublicKey() url = 'http://'+host+':'+str(port)+'/DELETE/'+filekey+'?' url += 'nodeID='+str(self.node.config.nodeID) url += '&port='+str(self.node.config.port) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) url += "&metakey="+str(metakey) self.timeoutcount = 0 self.authRetry = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, self.headers, nKu, host, port, url)) def startRequest(self, headers, nKu, host, port, url): d = self._sendRequest(headers, nKu, host, port, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, headers, nKu, host, port, url): factory = getPageFactory(url, headers=headers, timeout=primitive_to) deferred = factory.deferred deferred.addCallback(self._getSendDelete, nKu, host, port, factory) deferred.addErrback(self._errSendDelete, nKu, host, port, factory, url, headers) return deferred def _getSendDelete(self, response, nKu, host, port, factory): if eval(factory.status) == http.OK: loggerdele.info("received SENDDELETE response") updateNode(self.node.client, self.config, host, port, nKu) return response else: # XXX: updateNode raise failure.DefaultException("SENDDELETE FAILED: " +"server sent status "+factory.status+", '"+response+"'") def _errSendDelete(self, err, nKu, host, port, factory, url, headers): if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #print "DELETE request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(headers, nKu, host, port, url) elif hasattr(factory, 'status') and \ eval(factory.status) == http.UNAUTHORIZED and \ self.authRetry < MAXAUTHRETRY: # XXX: add this authRetry stuff to all the other op classes (so # that we don't DOS ourselves and another node self.authRetry += 1 loggerdele.info("SENDDELETE unauthorized, sending credentials") challenge = err.getErrorMessage()[4:] d = answerChallengeDeferred(challenge, self.node.config.Kr, self.node.config.groupIDu, nKu.id(), headers) d.addCallback(self._sendRequest, nKu, host, port, url) d.addErrback(self._errSendDelete, nKu, host, port, factory, url, headers) return d elif hasattr(factory, 'status'): # XXX: updateNode loggerdele.info("SENDDELETE failed") if eval(factory.status) == http.NOT_FOUND: err = NotFoundException(err) elif eval(factory.status) == http.BAD_REQUEST: err = BadRequestException(err) raise err return err class SENDVERIFY(REQUEST): def __init__(self, nKu, node, host, port, filename, offset, length, meta=None): """ Try to verify a file. If meta is present, it should be a (metakey, filelikeobj) pair. """ host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) filekey = os.path.basename(filename) # XXX: filekey should be hash loggervrfy.info("sending VERIFY request to %s:%s" % (host, str(port))) Ku = self.node.config.Ku.exportPublicKey() url = 'http://'+host+':'+str(port)+'/VERIFY/'+filekey+'?' url += 'nodeID='+str(self.node.config.nodeID) url += '&port='+str(self.node.config.port) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) url += "&offset="+str(offset) url += "&length="+str(length) if meta: url += "&metakey="+str(meta[0]) url += "&meta="+fencode(meta[1].read()) self.timeoutcount = 0 if not isinstance(nKu, FludRSA): raise ValueError("must pass in a FludRSA as nKu to SENDVERIFY") self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, self.headers, nKu, host, port, url)) def startRequest(self, headers, nKu, host, port, url): #loggervrfy.debug("*Doing* VERIFY Request %s" % port) d = self._sendRequest(headers, nKu, host, port, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, headers, nKu, host, port, url): loggervrfy.debug("in VERIFY sendReq %s" % port) factory = getPageFactory(url, headers=headers, timeout=primitive_to) deferred = factory.deferred deferred.addCallback(self._getSendVerify, nKu, host, port, factory) deferred.addErrback(self._errSendVerify, nKu, host, port, factory, url, headers) return deferred def _getSendVerify(self, response, nKu, host, port, factory): loggervrfy.debug("got vrfy response") if eval(factory.status) == http.OK: loggervrfy.info("received SENDVERIFY response") updateNode(self.node.client, self.config, host, port, nKu) return response else: # XXX: updateNode loggervrfy.debug("received non-OK SENDVERIFY response") raise failure.DefaultException("SENDVERIFY FAILED: " +"server sent status "+factory.status+", '"+response+"'") def _errSendVerify(self, err, nKu, host, port, factory, url, headers): loggervrfy.debug("got vrfy err") if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #print "VERIFY request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(headers, nKu, host, port, url) elif hasattr(factory, 'status') and \ eval(factory.status) == http.UNAUTHORIZED: loggervrfy.info("SENDVERIFY unauthorized, sending credentials") challenge = err.getErrorMessage()[4:] d = answerChallengeDeferred(challenge, self.node.config.Kr, self.node.config.groupIDu, nKu.id(), headers) d.addCallback(self._sendRequest, nKu, host, port, url) d.addErrback(self._errVerify, nKu, host, port, factory, url, headers, challenge) #d.addErrback(self._errSendVerify, nKu, host, port, factory, # url, headers) return d elif hasattr(factory, 'status'): # XXX: updateNode loggervrfy.info("SENDVERIFY failed: %s" % err.getErrorMessage()) if eval(factory.status) == http.NOT_FOUND: err = NotFoundException(err) elif eval(factory.status) == http.BAD_REQUEST: err = BadRequestException(err) raise err def _errVerify(self, err, nKu, host, port, factory, url, headers, challenge): # we can get in here after storing the same file as another node when # that data is stored in tarballs under its ID. It was expected that # this would be caught up in _getSendVerify... figure out why it isn't. loggervrfy.debug("factory status=%s" % factory.status) loggervrfy.debug("couldn't answer challenge from %s:%d, WHOOPS: %s" % (host, port, err.getErrorMessage())) loggervrfy.debug("challenge was: '%s'" % challenge) return err def answerChallengeDeferred(challenge, Kr, groupIDu, sID, headers): return threads.deferToThread(answerChallenge, challenge, Kr, groupIDu, sID, headers) def answerChallenge(challenge, Kr, groupIDu, sID, headers={}): loggerauth.debug("got challenge: '%s'" % challenge) sID = binascii.unhexlify(sID) challenge = (fdecode(challenge),) response = fencode(Kr.decrypt(challenge)) # XXX: RSA.decrypt won't restore leading 0's. This causes # some challenges to fail when they shouldn't -- solved for now # on the server side by generating non-0 leading challenges. loggerauth.debug("decrypted challenge to %s" % response) responseID = fdecode(response)[:len(sID)] loggerauth.debug(" response id: %s" % fencode(responseID)) if responseID != sID: # fail the op. # If we don't do this, we may be allowing the server to build a # dictionary useful for attack. The attack is as follows: node A # (server) collects a bunch of un-IDed challenge/response pairs by # issuing challenges to node B (client). Then node A uses those # responses to pose as B to some other server C. This sounds # farfetched, in that such a database would need to be huge, but in # reality, such an attack can happen in real-time, with node A # simultaneously serving requests from B, relaying challenges from C to # B, and then responding with B's responses to C to gain resources # there as an imposter. The ID string prevents this attack. # XXX: trust-- (must go by ip:port, since ID could be innocent) raise ImposterException("node %s is issuing invalid challenges --" " claims to have id=%s" % (fencode(sID), fencode(responseID))) response = fdecode(response)[len(sID):] loggerauth.debug(" challenge response: '%s'" % fencode(response)) response = fencode(response)+":"+groupIDu loggerauth.debug("response:groupIDu=%s" % response) response = binascii.b2a_base64(response) loggerauth.debug("b64(response:groupIDu)=%s" % response) response = "Basic %s" % response headers['Authorization'] = response return headers
Python
""" FludClient.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. flud client ops. """ from twisted.web import client from twisted.internet import error import os, stat, httplib, sys, logging from ClientPrimitives import * from ClientDHTPrimitives import * import FludCommUtil logger = logging.getLogger('flud.client') class FludClient(object): """ This class contains methods which create request objects """ def __init__(self, node): self.node = node self.currentStorOps = {} """ Data storage primitives """ def redoTO(self, f, node, host, port): print "in redoTO: %s" % f #print "in redoTO: %s" % dir(f.getTraceback()) if f.getTraceback().find("error.TimeoutError"): print "retrying........" return self.sendGetID(host, port) else: return f def sendGetID(self, host, port): #return SENDGETID(self.node, host, port).deferred d = SENDGETID(self.node, host, port).deferred #d.addErrback(self.redoTO, self.node, host, port) return d # XXX: we should cache nKu so that we don't do the GETID for all of these # ops every single time def sendStore(self, filename, metadata, host, port, nKu=None): # XXX: need to keep a map of 'filename' to deferreds, in case we are # asked to store the same chunk more than once concurrently (happens # for 0-byte files or from identical copies of the same file, for # example). both SENDSTORE and AggregateStore will choke on this. # if we find a store req in said map, just return that deferred instead # of redoing the op. [note, could mess up node choice... should also do # this on whole-file level in FileOps] # XXX: need to remove from currentStorOps on success or failure key = "%s:%d:%s" % (host, port, filename) if self.currentStorOps.has_key(key): logger.debug("returning saved deferred for %s in sendStore" % filename) return self.currentStorOps[key] def sendStoreWithnKu(nKu, host, port, filename, metadata): return SENDSTORE(nKu, self.node, host, port, filename, metadata).deferred def removeKey(r, key): self.currentStorOps.pop(key) return r if not nKu: # XXX: doesn't do AggregateStore if file is small. Can fix by # moving this AggStore v. SENDSTORE choice into SENDSTORE # proper logger.warn("not doing AggregateStore on small file because" " of missing nKu") print "not doing AggregateStore on small file because" \ " of missing nKu" d = self.sendGetID(host, port) d.addCallback(sendStoreWithnKu, host, port, filename, metadata) self.currentStorOps[key] = d return d fsize = os.stat(filename)[stat.ST_SIZE]; if fsize < MINSTORSIZE: logger.debug("doing AggStore") if metadata: logger.debug("with metadata") d = AggregateStore(nKu, self.node, host, port, filename, metadata).deferred else: logger.debug("SENDSTORE") d = SENDSTORE(nKu, self.node, host, port, filename, metadata).deferred self.currentStorOps[key] = d d.addBoth(removeKey, key) return d # XXX: need a version that takes a metakey, too def sendRetrieve(self, filekey, host, port, nKu=None, metakey=True): def sendRetrieveWithNKu(nKu, host, port, filekey, metakey=True): return SENDRETRIEVE(nKu, self.node, host, port, filekey, metakey).deferred if not nKu: d = self.sendGetID(host, port) d.addCallback(sendRetrieveWithNKu, host, port, filekey, metakey) return d else: return SENDRETRIEVE(nKu, self.node, host, port, filekey, metakey).deferred def sendVerify(self, filekey, offset, length, host, port, nKu=None, meta=None): def sendVerifyWithNKu(nKu, host, port, filekey, offset, length, meta=True): return SENDVERIFY(nKu, self.node, host, port, filekey, offset, length, meta).deferred if not nKu: d = self.sendGetID(host, port) d.addCallback(sendVerifyWithNKu, host, port, filekey, offset, length, meta) return d else: s = SENDVERIFY(nKu, self.node, host, port, filekey, offset, length, meta) return s.deferred def sendDelete(self, filekey, metakey, host, port, nKu=None): def sendDeleteWithNKu(nKu, host, port, filekey, metakey): return SENDDELETE(nKu, self.node, host, port, filekey, metakey).deferred if not nKu: d = self.sendGetID(host, port) d.addCallback(sendDeleteWithNKu, host, port, filekey, metakey) return d else: return SENDDELETE(nKu, self.node, host, port, filekey, metakey).deferred """ DHT single primitives (single call to single peer). These should probably only be called for testing or bootstrapping (sendkFindNode can be used to 'connect' to the flud network via a gateway, for instance). Use the recursive primitives for doing DHT ops. """ def sendkFindNode(self, host, port, key): return SENDkFINDNODE(self.node, host, port, key).deferred def sendkStore(self, host, port, key, val): return SENDkSTORE(self.node, host, port, key, val).deferred def sendkFindValue(self, host, port, key): return SENDkFINDVALUE(self.node, host, port, key).deferred """ DHT recursive primitives (recursive calls to muliple peers) """ def kFindNode(self, key): return kFindNode(self.node, key).deferred def kStore(self, key, val): return kStore(self.node, key, val).deferred def kFindValue(self, key): return kFindValue(self.node, key).deferred
Python
""" ServerPrimitives.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Primitive server storage protocol """ import binascii, time, os, stat, httplib, gc, re, sys, logging, sets import tempfile, tarfile from StringIO import StringIO from twisted.web.resource import Resource from twisted.web import server, resource, client from twisted.internet import reactor, threads, defer from twisted.web import http from twisted.python import failure from flud.FludCrypto import FludRSA, hashstring, hashfile, generateRandom import flud.TarfileUtils as TarfileUtils from flud.fencode import fencode, fdecode import BlockFile from FludCommUtil import * logger = logging.getLogger("flud.server.op") loggerid = logging.getLogger("flud.server.op.id") loggerstor = logging.getLogger("flud.server.op.stor") loggerretr = logging.getLogger("flud.server.op.rtrv") loggervrfy = logging.getLogger("flud.server.op.vrfy") loggerdele = logging.getLogger("flud.server.op.dele") loggerauth = logging.getLogger("flud.server.op.auth") """ These classes represent http requests received by this node, and the actions taken to respond. """ # FUTURE: check flud protocol version for backwards compatibility # XXX: need to refactor challengeResponse stuff so that all share this same # code (put it in REQUEST obj). See if we can do the same for some of # the other bits. # XXX: need to make sure we have appropriate timeouts for all comms. # FUTURE: DOS attacks. For now, assume that network hardware can filter these # out (by throttling individual IPs) -- i.e., it isn't our problem. If we # want to defend against this at some point, we need to keep track of who # is generating requests and then ignore them. # XXX: find everywhere we are sending longs and consider sending hex (or our # own base-64) encoded instead # XXX: might want to consider some self-healing for the kademlia layer, as # outlined by this thread: # http://zgp.org/pipermail/p2p-hackers/2003-August/001348.html (should # also consider Zooko's links in the parent to this post) # XXX: can man-in-the-middle get us with http basic auth, i.e., a node (Bob) # receives a store request from node Alice. Bob 'forwards' this request # to another node, Charlie, posing as Alice. When the challenge comes # back from Charlie, Bob relays it back to Alice, gets Alice's response, # and uses that response to answer Charlie's challenge. Bob then pretends # to store the data, while in reality the data is at Charlie (VERIFYs can # be performed similarly). # [This scenario appears legitimate, but it isn't problematic. Bob can # only store at Charlie if Charlie trusts Bob. And if Bob trust Charlie # so much that he is willing to stake his trust with Alice on Charlie's # reliability, there shouldn't be anything wrong with allowing this. If # Charlie goes down, Bob will go down (partly) with him, as far as trust # is concerned]. # [should disable/remove CHALLENGE and GROUPID so that this only works # for forwarding. Leaving CHALLENGE/GROUPID allows imposters. Limiting # only to forwarding imposters.] # XXX: Auth problem still exists. Http-auth allows servers to know that clients # legitimate, but clients don't know the legitimacy of servers. Proposal: # all client requests are currently responded to with a challenge that is # copied in the header and body. Instead, require that the server sign # the url it was handed by the client, and return this in the body # (challenge remains in the header). Now, the client can verify that the # server is who it says it is. This is still somewhat vulnerable to replay # (if the same url is repeated, which it can be), but this only matters if # the client sends the exact same message again later to an imposter # (imposters can only impersonate the server). Alternatively, could send # a challenge as a param of the request. [I think this is fixed now?] # XXX: disallow requests originating from self. fludproto_ver = '0.1' challengelength = 40 # XXX: is 40 bytes sufficient? class ROOT(Resource): """ Notes on parameters common to most requests: Ku_e - the public key RSA exponent (usually 65537L, but can be other values) Ku_n - the public key RSA modulus. nodeID - the nodeID of the requestor port - the port that the reqestor runs their fludserver on. All REQs will contain a nodeID ('nodeID'), a public key ('Ku_e', 'Ku_n'), and a requestID ('reqID'). [XXX: must be more careful about all this code -- if wrong data is sent, exceptions are raised. Put try blocks around everything.] Client Authentication works as follows: Each client request is responded to with a 401 response in the header, with a challenge in the header message and repeated in the body. The client reissues the response, filling the user 'username' field with the challenge-response and the 'password' field with groupIDu. The client must compute the response to the challenge each time by examining this value. The groupIDu never changes, so the client can simply send this value back with the response. This implies some server-side state; since the server responds with a 401 and expects the client to send a new request with an Authorization header that contains the challenge response, the server has to keep some state for each client. This state expires after a short time, and implies that client must make requests to individual servers serially. """ # FUTURE: might want to switch to nevow's rend.Page as parent... def __init__(self, fludserver): """ All children should inherit. Make sure to call super.__init__ if you override __init__ """ Resource.__init__(self) self.fludserver = fludserver self.node = fludserver.node self.config = fludserver.node.config def getChild(self, name, request): """ should override. """ if name == "": return self return Resource.getChild(self, name, request) def render_GET(self, request): self.setHeaders(request) return "<html>Flud</hrml>" def setHeaders(self, request): request.setHeader('Server','FludServer 0.1') request.setHeader('FludProtocol', fludproto_ver) class ID(ROOT): """ Just received a request to expose my identity. Send public key (from which requestor can determine nodeID). Response codes: 200- OK (default) 204- No Content (returned in case of error or not wanting to divulge ID) """ def render_GET(self, request): self.setHeaders(request) try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port') params = requireParams(request, required) except Exception, inst: msg = "%s in request received by ID" % inst.args[0] loggerid.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: loggerid.info("received ID request from %s..." % params['nodeID'][:10]) loggerid.info("returning ID response") #try: reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) host = getCanonicalIP(request.getClientIP()) updateNode(self.node.client, self.config, host, int(params['port']), reqKu, params['nodeID']) return str(self.config.Ku.exportPublicKey()) #except: # msg = "can't return ID" # loggerid.log(logging.WARN, msg) # request.setResponseCode(http.NO_CONTENT, msg) # return msg class STORE(ROOT): """ A request to store data via http upload. Response codes: 200- OK (default) 400- Bad Request (missing params) 401- Unauthorized (ID hash, CHALLENGE, or GROUPCHALLENGE failed) Each file fragment is stored with its storage key as the file name. The file fragment can be 'touched' (or use last access time if supported) each time it is verified or read, so that we have a way to record age (which also allows a purge strategy). Files are reference-listed (reference count with owners) by the BlockFile object. A preliminary file structure: .flud/ store/ fragmenthash1 fragmenthash2 ... dht/ metadatahash1 metadatahash2 ... meta/ master metadatahash1 metadatahash2 ... dl/ flud.conf fludfile.conf fludrules.init flud.log """ isLeaf = True def render_POST(self, request): self.setHeaders(request) try: required = ('filekey', 'size', 'nodeID', 'Ku_e', 'Ku_n', 'port') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by STORE" loggerstor.info(msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: loggerstor.info("received STORE request from %s..." % params['nodeID'][:10]) host = getCanonicalIP(request.getClientIP()) port = int(params['port']) requestedSize = int(params['size']) reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) #if requestedSize > 10: # msg = "unwilling to enter into storage relationship" # request.setResponseCode(http.PAYMENT_REQUIRED, msg) # return msg return authenticate(request, reqKu, params['nodeID'], host, int(params['port']), self.node.client, self.config, self._storeFile, request, params['filekey'], reqKu, params['nodeID']) def _storeFile(self, request, filekey, reqKu, nodeID): # [XXX: memory management is not happy here. might want to look at # request.registerProducer(). Otherwise, might have to scrap # using the STORE(ROOT(RESOURCE)) deal in favor of # producer/consumer model for STORE ops # (http://itamarst.org/writings/OSCON03/twisted_internet-108.html). # Another option might include subclassing web.resource.Resource # and making this derive from that... Or might be web.Site that # needs to be subclassed... Or maybe web.site.Request - # web.site.Request.process()? Request seems doubly-bad: perhaps a # copy is made somewhere, because memory mushrooms to 2x big # upload, then goes back down to around 1x. # [update: This should be fixable in twisted.web2, but I am informed # that in the current version, there is no workaround] # get the data to a tmp file loggerstor.debug("writing store data to tmpfile") tmpfile = tempfile.mktemp(dir=self.config.storedir) tarball = os.path.join(self.config.storedir,reqKu.id()+".tar") # rename and/or prepend the data appropriately tmpTarMode = None if filekey[-4:] == ".tar": tmpfile = tmpfile+".tar" tmpTarMode = 'r' targetTar = tarball elif filekey[-7:] == ".tar.gz": tmpfile = tmpfile+".tar.gz" tmpTarMode = 'r:gz' targetTar = tarball+".gz" loggerstor.debug("tmpfile is %s" % tmpfile) # XXX: if the server supports both .tar and tar.gz, this is wrong; we'd # need to check *both* for already existing dudes instead of just # choosing one if os.path.exists(tarball+'.gz'): tarball = (tarball+'.gz', 'r:gz') elif os.path.exists(tarball): tarball = (tarball, 'r') else: tarball = None loggerstor.debug("tarball is %s" % str(tarball)) data = request.args.get('filename')[0] # XXX: file in mem! need web2. # XXX: bad blocking stuff here f = open(tmpfile, 'wb') f.write(data) f.close() ftype = os.popen('file %s' % tmpfile) loggerstor.debug("ftype of %s is %s" % (tmpfile, ftype.read())) ftype.close() if tmpTarMode: # client sent a tarball loggerstor.debug("about to chksum %s" % tmpfile) digests = TarfileUtils.verifyHashes(tmpfile, '.meta') loggerstor.debug("chksum returned %s" % digests) ftype = os.popen('file %s' % tmpfile) loggerstor.debug("ftype of %s is %s" % (tmpfile, ftype.read())) ftype.close() if not digests: msg = "Attempted to use non-CAS storage key(s) for" \ " STORE tarball" loggerstor.debug(msg) os.remove(tmpfile) request.setResponseCode(http.CONFLICT, msg) return msg # XXX: add digests to a db of already stored files (for quick # lookup) if tarball: tarname, tarnameMode = tarball loggerstor.debug("concatenating tarfiles %s and %s" % (tarname, tmpfile)) f1 = tarfile.open(tarname, tarnameMode) f2 = tarfile.open(tmpfile, tmpTarMode) f1names = f1.getnames() f2names = f2.getnames() f1.close() f2.close() dupes = [f for f in f1names if f in f2names] TarfileUtils.delete(tmpfile, dupes) ftype = os.popen('file %s' % tarname) loggerstor.debug("ftype of %s is %s" % (tarname, ftype.read())) ftype.close() TarfileUtils.concatenate(tarname, tmpfile) ftype = os.popen('file %s' % tarname) loggerstor.debug("ftype of %s is %s" % (tarname, ftype.read())) ftype.close() else: loggerstor.debug("saving %s as tarfile %s" % (tmpfile, targetTar)) os.rename(tmpfile, targetTar) else: # client sent regular file h = hashfile(tmpfile) if request.args.has_key('meta') and request.args.has_key('metakey'): metakey = request.args.get('metakey')[0] meta = request.args.get('meta')[0] # XXX: file in mem! else: metakey = None meta = None if fencode(long(h, 16)) != filekey: msg = "Attempted to use non-CAS storage key for STORE data " msg += "(%s != %s)" % (filekey, fencode(long(h, 16))) os.remove(tmpfile) request.setResponseCode(http.CONFLICT, msg) return msg fname = os.path.join(self.config.storedir, filekey) if os.path.exists(fname): loggerstor.debug("adding metadata to %s" % fname) f = BlockFile.open(fname,'rb+') if not f.hasNode(nodeID): f.addNode(int(nodeID,16), {metakey: meta}) f.close() os.remove(tmpfile) else: if os.path.exists(nodeID+".tar"): # XXX: need to do something with metadata! print "XXX: need to do something with metadata for tar!" tarball = tarfile.open(tarname, 'r') if fname in tarball.getnames(): loggerstor.debug("%s already stored in tarball" % fname) # if the file is already in the corresponding tarball, # update its timestamp and return success. loggerstor.debug("%s already stored" % filekey) # XXX: update timestamp for filekey in tarball return "Successful STORE" else: loggerstor.debug("tarball for %s, but %s not in tarball" % (nodeID,fname)) if len(data) < 8192 and fname != tarname: #XXX: magic # (blk sz) # If the file is small, move it into the appropriate # tarball. Note that this code is unlikely to ever be # executed if the client is an official flud client, as # they do the tarball aggregation thing already, and all # tarballs will be > 8192. This is, then, really just # defensive coding -- clients aren't required to implement # that tarball aggregation strategy. And it is really only # useful for filesystems with inefficient small file # storage. loggerstor.debug("moving small file '%s' into tarball" % fname) if not os.path.exists(tarname): tarball = tarfile.open(tarname, 'w') else: tarball = tarfile.open(tarname, 'a') # XXX: more bad blocking stuff tarball.add(tmpfile, os.path.basename(fname)) if meta: metafilename = "%s.%s.meta" % (os.path.basename(fname), metakey) loggerstor.debug("adding metadata file to tarball %s" % metafilename) metaio = StringIO(meta) tinfo = tarfile.TarInfo(metafilename) tinfo.size = len(meta) tarball.addfile(tinfo, metaio) tarball.close() os.remove(tmpfile) else: # store the file loggerstor.debug("storing %s" % fname) os.rename(tmpfile, fname) BlockFile.convert(fname, (int(nodeID,16), {metakey: meta})) loggerstor.debug("successful STORE for %s" % filekey) return "Successful STORE" def _storeErr(self, error, request, msg): out = msg+": "+error.getErrorMessage() print "%s" % str(error) loggerstor.info(out) # update trust, routing request.setResponseCode(http.UNAUTHORIZED, "Unauthorized: %s" % msg) # XXX: wrong code return msg def render_GET(self, request): request.setResponseCode(http.BAD_REQUEST, "Bad Request") return "STORE request must be sent using POST" class RETRIEVE(ROOT): """ A request to retrieve data. The file to retrieve is indicated by the URL path, e.g. http://server:port/RETRIEVE/a35cd1339766ef209657a7b Response codes: 200- OK (default) 400- Bad Request (missing params) 401- Unauthorized (ID hash, CHALLENGE, or GROUPCHALLENGE failed) """ isLeaf = True def render_GET(self, request): self.setHeaders(request) try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by RETRIEVE" loggerretr.log(logging.INFO, msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: loggerretr.info("received RETRIEVE request for %s from %s..." % (request.path, params['nodeID'][:10])) host = getCanonicalIP(request.getClientIP()) port = int(params['port']) filekey = re.sub('/RETRIEVE', '', str(request.path)) paths = [p for p in filekey.split(os.path.sep) if p != ''] if len(paths) > 1: msg = "filekey contains illegal path seperator tokens." loggerretr.debug("BAD_REQ: %s" % msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request: %s" % msg) return msg reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) if request.args.has_key('metakey'): returnMeta = request.args['metakey'] if returnMeta == 'True': returnMeta = True else: returnMeta = True return authenticate(request, reqKu, params['nodeID'], host, int(params['port']), self.node.client, self.config, self._sendFile, request, filekey, reqKu, returnMeta) def _sendFile(self, request, filekey, reqKu, returnMeta): fname = self.config.storedir + filekey loggerretr.debug("reading file data from %s" % fname) # XXX: make sure requestor owns the file? tfilekey = filekey[1:] if returnMeta: loggerretr.debug("returnMeta = %s" % returnMeta) request.setHeader('Content-type', 'Multipart/Related') rand_bound = binascii.hexlify(generateRandom(13)) request.setHeader('boundary', rand_bound) if not os.path.exists(fname): # check for tarball for originator tarball = os.path.join(self.config.storedir,reqKu.id()+".tar") tarballs = [] if os.path.exists(tarball+'.gz'): tarballs.append((tarball+'.gz', 'r:gz')) if os.path.exists(tarball): tarballs.append((tarball, 'r')) loggerretr.debug("tarballs = %s" % tarballs) # XXX: does this work? does it close both tarballs if both got # opened? for tarball, openmode in tarballs: tar = tarfile.open(tarball, openmode) try: tinfo = tar.getmember(tfilekey) returnedMeta = False if returnMeta: loggerretr.debug("tar returnMeta %s" % tfilekey) try: metas = [f for f in tar.getnames() if f[:len(tfilekey)] == tfilekey and f[-4:] == 'meta'] loggerretr.debug("tar returnMetas=%s" % metas) for m in metas: minfo = tar.getmember(m) H = [] H.append("--%s" % rand_bound) H.append("Content-Type: " "Application/octet-stream") H.append("Content-ID: %s" % m) H.append("Content-Length: %d" % minfo.size) H.append("") H = '\r\n'.join(H) request.write(H) request.write('\r\n') tarm = tar.extractfile(minfo) loggerretr.debug("successful metadata" " RETRIEVE (from %s)" % tarball) # XXX: bad blocking stuff while 1: buf = tarm.read() if buf == "": break request.write(buf) request.write('\r\n') tarm.close() H = [] H.append("--%s" % rand_bound) H.append("Content-Type: Application/octet-stream") H.append("Content-ID: %s" % tfilekey) H.append("Content-Length: %d" % tinfo.size) H.append("") H = '\r\n'.join(H) request.write(H) request.write('\r\n') returnedMeta = True except: # couldn't find any metadata, just return normal # file loggerretr.debug("no metadata found") pass # XXX: bad blocking stuff tarf = tar.extractfile(tinfo) # XXX: update timestamp on tarf in tarball loggerretr.debug("successful RETRIEVE (from %s)" % tarball) # XXX: bad blocking stuff while 1: buf = tarf.read() if buf == "": break request.write(buf) tarf.close() tar.close() if returnedMeta: T = [] T.append("") T.append("--%s--" % rand_bound) T.append("") T = '\r\n'.join(T) request.write(T) return "" except: tar.close() request.setResponseCode(http.NOT_FOUND, "Not found: %s" % filekey) request.write("Not found: %s" % filekey) else: f = BlockFile.open(fname,"rb") loggerretr.log(logging.INFO, "successful RETRIEVE for %s" % filekey) meta = f.meta(int(reqKu.id(),16)) if returnMeta and meta: loggerretr.debug("returnMeta %s" % filekey) loggerretr.debug("returnMetas=%s" % meta) for m in meta: H = [] H.append("--%s" % rand_bound) H.append("Content-Type: Application/octet-stream") H.append("Content-ID: %s.%s.meta" % (tfilekey, m)) H.append("Content-Length: %d" % len(meta[m])) H.append("") H.append(meta[m]) H = '\r\n'.join(H) request.write(H) request.write('\r\n') H = [] H.append("--%s" % rand_bound) H.append("Content-Type: Application/octet-stream") H.append("Content-ID: %s" % tfilekey) H.append("Content-Length: %d" % f.size()) H.append("") H = '\r\n'.join(H) request.write(H) request.write('\r\n') # XXX: bad blocking stuff while 1: buf = f.read() if buf == "": break request.write(buf) f.close() if returnMeta and meta: T = [] T.append("") T.append("--%s--" % rand_bound) T.append("") request.write('\r\n'.join(T)) return "" def _sendErr(self, error, request, msg): out = msg+": "+error.getErrorMessage() loggerretr.log(logging.INFO, out ) # update trust, routing request.setResponseCode(http.UNAUTHORIZED, "Unauthorized: %s" % msg) request.write(msg) request.finish() class VERIFY(ROOT): """ Just received a storage VERIFY request. Response codes: 200- OK (default) 400- Bad Request (missing params) 401- Unauthorized (ID hash, CHALLENGE, or GROUPCHALLENGE failed) [VERIFY is the most important of the trust-building ops. Issues: 1) how often do we verify. a) each file A) each block of each file 2) proxying hurts trust (see PROXY below) The answer to #1 is, ideally, we check every node who is storing for us every quanta. But doing the accounting for keeping track of who is storing for us is heavy, so instead we just want to check all of our files and hope that gives sufficient coverage, on average, of the nodes we store to. But we have to put some limits on this, and the limits can't be imposed by the sender (since a greedy sender can just modify this), so peer nodes have to do some throttling of these types of requests. But such throttling is tricky, as the requestor must decrease trust when VERIFY ops fail. Also, since we will just randomly select from our files, such a throttling scheme will reduce our accuracy as we store more and more data (we can only verify a smaller percentage). Peers could enforce limits as a ratio of total data stored for a node, but then the peers could act maliciously by artificially lowering this number. In summary, if we don't enforce limits, misbehaving nodes could flood VERIFY requests resulting in effecitve DOS attack. If we do enforce limits, we have to watch out for trust wars where both nodes end up destroying all trust between them. Possible answer: set an agreed-upon threshold a priori. This could be a hardcoded limit, or (better) negotiated between node pairs. If the requestor goes over this limit, he should understand that his trust will be decreased by requestee. If he doesn't understand this, his trust *should* be decreased, and if he decreases his own trust in us as well, we don't care -- he's misbehaving.] """ # XXX: add support for sending multiple verify ops in a single request isLeaf = True def render_GET(self, request): """ A VERIFY contains a file[fragment]id, an offset, and a length. When this message is received, the given file[fragment] should be accessed and its bytes, from offset to offset+length, should be sha256 hashed, and the result returned. [in theory, we respond to all VERIFY requests. In practice, however, we should probably do some throttling of responses to prevent DOS attacks and probing. Or, maybe require that the sender encrypt the challenge with their private key...] [should also enforce some idea of reasonableness on length of bytes to verify] """ self.setHeaders(request) try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'offset', 'length') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by VERIFY" loggervrfy.log(logging.INFO, msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") loggervrfy.debug("BAD REQUEST") return msg else: loggervrfy.log(logging.INFO, "received VERIFY request from %s..." % params['nodeID']) if 'meta' in request.args: params['metakey'] = request.args['metakey'][0] params['meta'] = fdecode(request.args['meta'][0]) loggerretr.info("VERIFY contained meta field with %d chars" % len(params['meta'])) meta = (params['metakey'], params['meta']) else: meta = None host = getCanonicalIP(request.getClientIP()) port = int(params['port']) offset = int(params['offset']) length = int(params['length']) filekey = re.sub('/VERIFY', '', str(request.path)) paths = [p for p in filekey.split(os.path.sep) if p != ''] if len(paths) > 1: msg = "Bad request:"\ " filekey contains illegal path seperator tokens." loggerretr.debug(msg) request.setResponseCode(http.BAD_REQUEST, msg) return msg reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) return authenticate(request, reqKu, params['nodeID'], host, int(params['port']), self.node.client, self.config, self._sendVerify, request, filekey, offset, length, reqKu, params['nodeID'], meta) def _sendVerify(self, request, filekey, offset, length, reqKu, nodeID, meta): fname = self.config.storedir+filekey loggervrfy.debug("request for %s" % fname) if os.path.exists(fname): loggervrfy.debug("looking in regular blockfile for %s" % fname) if meta: f = BlockFile.open(fname, 'rb+') else: f = BlockFile.open(fname, 'rb') else: # check for tarball for originator loggervrfy.debug("checking tarball for %s" % fname) tarballs = [] tarballbase = os.path.join(self.config.storedir, reqKu.id())+".tar" if os.path.exists(tarballbase+".gz"): tarballs.append((tarballbase+".gz", 'r:gz')) if os.path.exists(tarballbase): tarballs.append((tarballbase, 'r')) loggervrfy.debug("tarballs is %s" % tarballs) for tarball, openmode in tarballs: loggervrfy.debug("looking in tarball %s..." % tarball) tar = tarfile.open(tarball, openmode) try: tfilekey = filekey[1:] tarf = tar.extractfile(tfilekey) tari = tar.getmember(tfilekey) # XXX: update timestamp on tarf in tarball fsize = tari.size if offset > fsize or (offset+length) > fsize: # XXX: should limit length loggervrfy.debug("VERIFY response failed (from %s):" " bad offset/length" % tarball) msg = "Bad request: bad offset/length in VERIFY" request.setResponseCode(http.BAD_REQUEST, msg) return msg # XXX: could avoid seek/read if length == 0 tarf.seek(offset) # XXX: bad blocking read data = tarf.read(length) tarf.close() if meta: mfname = "%s.%s.meta" % (tfilekey, meta[0]) loggervrfy.debug("looking for %s" % mfname) if mfname in tar.getnames(): # make sure that the data is the same, if not, # remove it and re-add it tarmf = tar.extractfile(mfname) # XXX: bad blocking read stored_meta = tarmf.read() tarmf.close() if meta[1] != stored_meta: loggervrfy.debug("updating tarball" " metadata for %s.%s" % (tfilekey, meta[0])) tar.close() TarfileUtils.delete(tarball, mfname) if openmode == 'r:gz': tarball = TarfileUtils.gunzipTarball( tarball) tar = tarfile.open(tarball, 'a') metaio = StringIO(meta[1]) tinfo = tarfile.TarInfo(mfname) tinfo.size = len(meta[1]) tar.addfile(tinfo, metaio) tar.close() if openmode == 'r:gz': tarball = TarfileUtils.gzipTarball( tarball) else: loggervrfy.debug("no need to update tarball" " metadata for %s.%s" % (tfilekey, meta[0])) else: # add it loggervrfy.debug("adding tarball metadata" " for %s.%s" % (tfilekey, meta[0])) tar.close() if openmode == 'r:gz': tarball = TarfileUtils.gunzipTarball(tarball) tar = tarfile.open(tarball, 'a') metaio = StringIO(meta[1]) tinfo = tarfile.TarInfo(mfname) tinfo.size = len(meta[1]) tar.addfile(tinfo, metaio) tar.close() if openmode == 'r:gz': tarball = TarfileUtils.gzipTarball( tarball) tar.close() hash = hashstring(data) loggervrfy.info("successful VERIFY (from %s)" % tarball) return hash except: tar.close() loggervrfy.debug("requested file %s doesn't exist" % fname) msg = "Not found: not storing %s" % filekey request.setResponseCode(http.NOT_FOUND, msg) return msg # make sure request is reasonable fsize = os.stat(fname)[stat.ST_SIZE] if offset > fsize or (offset+length) > fsize: # XXX: should limit length loggervrfy.debug("VERIFY response failed (bad offset/length)") msg = "Bad request: bad offset/length in VERIFY" request.setResponseCode(http.BAD_REQUEST, msg) return msg else: # XXX: blocking # XXX: could avoid seek/read if length == 0 (noop for meta update) f.seek(offset) data = f.read(length) if meta: loggervrfy.debug("adding metadata for %s.%s" % (fname, meta[0])) f.addNode(int(nodeID, 16) , {meta[0]: meta[1]}) # XXX: blocking f.close() hash = hashstring(data) loggervrfy.debug("returning VERIFY") return hash def _sendErr(self, error, request, msg): out = "%s:%s" % (msg, error.getErrorMessage()) loggervrfy.info(out) # update trust, routing request.setResponseCode(http.UNAUTHORIZED, "Unauthorized: %s" % msg) request.write(msg) request.finish() class DELETE(ROOT): """ A request to delete data. The file to delete is indicated by the URL path, e.g. http://server:port/DELETE/a35cd1339766ef209657a7b Response codes: 200- OK (default) 400- Bad Request (missing params) 401- Unauthorized (ID hash, CHALLENGE, or GROUPCHALLENGE failed, or nodeID doesn't own this block) """ isLeaf = True def render_GET(self, request): self.setHeaders(request) try: required = ('nodeID', 'Ku_e', 'Ku_n', 'port', 'metakey') params = requireParams(request, required) except Exception, inst: msg = inst.args[0] + " in request received by DELETE" loggerdele.log(logging.INFO, msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request") return msg else: loggerdele.debug("received DELETE request for %s from %s..." % (request.path, params['nodeID'][:10])) host = getCanonicalIP(request.getClientIP()) port = int(params['port']) filekey = re.sub('/DELETE', '', str(request.path)) paths = [p for p in filekey.split(os.path.sep) if p != ''] if len(paths) > 1: msg = "filekey contains illegal path seperator tokens." loggerretr.debug("BAD_REQ: %s" % msg) request.setResponseCode(http.BAD_REQUEST, "Bad Request: %s" % msg) return msg reqKu = {} reqKu['e'] = long(params['Ku_e']) reqKu['n'] = long(params['Ku_n']) reqKu = FludRSA.importPublicKey(reqKu) metakey = params['metakey'] return authenticate(request, reqKu, params['nodeID'], host, int(params['port']), self.node.client, self.config, self._deleteFile, request, filekey, metakey, reqKu, params['nodeID']) def _deleteFile(self, request, filekey, metakey, reqKu, reqID): fname = self.config.storedir + filekey loggerdele.debug("reading file data from %s" % fname) if not os.path.exists(fname): # check for tarball for originator tarballs = [] tarballbase = os.path.join(self.config.storedir, reqKu.id())+".tar" if os.path.exists(tarballbase+".gz"): tarballs.append((tarballbase+".gz", 'r:gz')) if os.path.exists(tarballbase): tarballs.append((tarballbase, 'r')) for tarball, openmode in tarballs: tfilekey = filekey[1:] mfilekey = "%s.%s.meta" % (tfilekey, metakey) loggerdele.debug("opening %s, %s for delete..." % (tarball, openmode)) ftype = os.popen('file %s' % tarball) loggerdele.debug("ftype of %s is %s" % (tarball, ftype.read())) ftype.close() tar = tarfile.open(tarball, openmode) mnames = [n for n in tar.getnames() if n[:len(tfilekey)] == tfilekey] tar.close() if len(mnames) > 2: deleted = TarfileUtils.delete(tarball, mfilekey) else: deleted = TarfileUtils.delete(tarball, [tfilekey, mfilekey]) if deleted: loggerdele.info("DELETED %s (from %s)" % (deleted, tarball)) return "" request.setResponseCode(http.NOT_FOUND, "Not found: %s" % filekey) request.write("Not found: %s" % filekey) else: f = BlockFile.open(fname,"rb+") nID = int(reqID, 16) if f.hasNode(nID): # remove this node/metakey from owning this file block f.delNode(nID, metakey) if f.emptyNodes(): # if this was the only owning node, delete it f.close() os.remove(fname) f.close() loggerdele.debug("returning DELETE response") return "" def _sendErr(self, error, request, msg): out = msg+": "+error.getErrorMessage() loggerdele.log(logging.INFO, out ) # update trust, routing request.setResponseCode(http.UNAUTHORIZED, "Unauthorized: %s" % msg) request.write(msg) request.finish() class PROXY(ROOT): """ This is a special request which wraps another request. If received, the node is required to forward the request and keep state to re-wrap when the response returns. This is useful for several reasons: 1) insert a bit of anonymity into VERIFY ops, so that a malicious node can't know to do an early purge data for a missing node. 2) NAT'd boxes will need to use relays for STORE, VERIFY, and RETRIEVE (since all those ops cause the receive to make a connection back) [There must be an incentive for nodes to offer PROXY (trust), otherwise it is advantageous to turn it off.] [disadvantages of proxy: it hurts trust. How do we know that a bad (trust decreasing) op really was caused by the originator? Couldn't the failure be caused by one of the proxies?] [Should consider using GnuNet's "Excess-Based Economic Model" where each request contains a priority which 'spends' some of the trust at the requestee when resources are scarce. In this model, the proxying node[s] charge a fee on the priority, reducing it by a small amount as they forward the request.] """ isLeaf = True def render_GET(self, request): self.setHeaders(request) result = "NOT YET IMPLEMENTED" return result def authenticate(request, reqKu, reqID, host, port, client, config, callable, *callargs): # 1- make sure that reqKu hashes to reqID # 2- send a challenge/groupchallenge to reqID (encrypt with reqKu) challengeResponse = request.getUser() groupResponse = request.getPassword() if reqKu.id() != reqID: msg = "Ku %s does not hash to nodeID %s, failing request" \ % (reqKu.id(), reqID) loggerauth.info(msg) # XXX: update trust, routing request.setResponseCode(http.UNAUTHORIZED, "Unauthorized: %s" % msg) return msg if not challengeResponse or not groupResponse: loggerauth.info("returning challenge for request from %s:%d" \ % (host, port)) return sendChallenge(request, reqKu, config.nodeID) else: if getChallenge(challengeResponse): expireChallenge(challengeResponse) if groupResponse == hashstring( str(reqKu.exportPublicKey()) +str(config.groupIDr)): updateNode(client, config, host, port, reqKu, reqID) return callable(*callargs) else: err = "Group Challenge Failed" loggerauth.info(err) loggerauth.debug("Group Challenge received was %s" \ % groupResponse) # XXX: update trust, routing request.setResponseCode(http.FORBIDDEN, err); return err else: err = "Challenge Failed" loggerauth.info(err) loggerauth.debug("Challenge received was %s" % challengeResponse) # XXX: update trust, routing request.setResponseCode(http.FORBIDDEN, err); return err def sendChallenge(request, reqKu, id): challenge = generateRandom(challengelength) while challenge[0] == '\x00': # make sure we have at least challengelength bytes challenge = generateRandom(challengelength) addChallenge(challenge) loggerauth.debug("unencrypted challenge is %s" % fencode(binascii.unhexlify(id)+challenge)) echallenge = reqKu.encrypt(binascii.unhexlify(id)+challenge)[0] echallenge = fencode(echallenge) loggerauth.debug("echallenge = %s" % echallenge) # since challenges will result in a new req/resp pair being generated, # these could take much longer than the primitive_to. Expire in # 15*primitive_to reactor.callLater(primitive_to*15, expireChallenge, challenge, True) resp = 'challenge = %s' % echallenge loggerauth.debug("resp = %s" % resp) request.setResponseCode(http.UNAUTHORIZED, echallenge) request.setHeader('Connection', 'close') request.setHeader('WWW-Authenticate', 'Basic realm="%s"' % 'default') request.setHeader('Content-Length', str(len(resp))) request.setHeader('Content-Type', 'text/html') request.setHeader('Pragma','claimreserve=5555') # XXX: this doesn't work return resp outstandingChallenges = {} def addChallenge(challenge): outstandingChallenges[challenge] = True loggerauth.debug("added challenge %s" % fencode(challenge)) def expireChallenge(challenge, expired=False): try: challenge = fdecode(challenge) except: pass if outstandingChallenges.has_key(challenge): del outstandingChallenges[challenge] if expired: # XXX: should put these in an expired challenge list so that we # can send a more useful message on failure (must then do # expirations on expired list -- maybe better to just make # these expirations really liberal). loggerauth.debug("expired challenge %s" % fencode(challenge)) else: loggerauth.debug("deleted challenge %s" % fencode(challenge)) def getChallenge(challenge): try: challenge = fdecode(challenge) except: pass if outstandingChallenges.has_key(challenge): return outstandingChallenges[challenge] else: return None
Python
""" FludCommUtil.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Communications routines used by both client and server code. """ from twisted.web import client from twisted.internet import reactor, defer from twisted.python import failure import binascii, httplib, logging, os, stat, random, socket import inspect from flud.FludExceptions import FludException from flud.FludCrypto import FludRSA, generateRandom from flud.HTTPMultipartDownloader import HTTPMultipartDownloader """ Some constants used by the Flud Protocol classes """ fludproto_ver = '0.1' # XXX: when things timeout, bad news. Unintuitive exceptions spewed. Make this # small and fix all issues. primitive_to = 3800 # default timeout for primitives kprimitive_to = primitive_to/2 # default timeout for kademlia primitives #kprimitive_to = 10 # default timeout for kademlia primitives transfer_to = 3600 # 10-hr limit on file transfers MAXTIMEOUTS = 5 # number of times to retry after connection timeout failure CONNECT_TO = 60 CONNECT_TO_VAR = 5 logger = logging.getLogger('flud.comm') class BadCASKeyException(failure.DefaultException): pass class NotFoundException(failure.DefaultException): pass class BadRequestException(failure.DefaultException): pass i = 0 """ Some utility functions used by both client and server. """ def updateNodes(client, config, nodes): if nodes and not isinstance(nodes, list) and not isinstance(nodes, tuple): raise TypeError("updateNodes must be called with node list, tuple," " or kData dict") logger.debug("updateNodes(%s)" % nodes) for i in nodes: host = i[0] port = i[1] nID = i[2] nKu = FludRSA.importPublicKey(i[3]) updateNode(client, config, host, port, nKu, nID) updateNodePendingGETID = {} def updateNode(client, config, host, port, nKu=None, nID=None): """ Updates this node's view of the given node. This includes updating the known-nodes record, trust, and routing table information """ def updateNodeFail(failure, host, port): logging.getLogger('flud').log(logging.INFO, "couldn't get nodeID from %s:%d: %s" % (host, port, failure)) def callUpdateNode(nKu, client, config, host, port, nID): return updateNode(client, config, host, port, nKu, nID) if isinstance(nID, long): nID = "%064x" % nID if nKu is None: #print "updateNode, no nKu" if nID is None: d = client.sendGetID(host, port) d.addCallback(callUpdateNode, client, config, host, port, nID) d.addErrback(updateNodeFail, host, port) else: #print "updateNode, no nKu but got a nID" if config.nodes.has_key(nID): return updateNode(client, config, host, port, FludRSA.importPublicKey(config.nodes[nID]['Ku']), nID) elif updateNodePendingGETID.has_key(nID): pass else: #print "updateNode, sending GETID" updateNodePendingGETID[nID] = True d = client.sendGetID(host, port) d.addCallback(callUpdateNode, client, config, host, port, nID) d.addErrback(updateNodeFail, host, port) elif isinstance(nKu, FludRSA): #print "updateNode with nKu" if updateNodePendingGETID.has_key(nID): del updateNodePendingGETID[nID] if nID == None: nID = nKu.id() elif nID != nKu.id(): raise ValueError("updateNode: given nID doesn't match given nKu." " '%s' != '%s'" % (nID, nKu.id())) # XXX: looks like an imposter -- instead of raising, mark host:port # pair as bad (trust-- on host:port alone, since we don't know id). if config.nodes.has_key(nID) == False: config.addNode(nID, host, port, nKu) # XXX: trust # routing node = (host, port, long(nID, 16), nKu.exportPublicKey()['n']) replacee = config.routing.updateNode(node) #logger.info("knownnodes now: %s" % config.routing.knownNodes()) #print "knownnodes now: %s" % config.routing.knownNodes() if replacee != None: logging.getLogger('flud').info( "determining if replacement in ktable is needed") s = SENDGETID(replacee[0], replacee[1]) s.addErrback(replaceNode, config.routing, replacee, node) else: #print "updateNode nKu=%s, type=%s" % (nKu, type(nKu)) logging.getLogger('flud').warn( "updateNode can't update without a public key or nodeID") frame = inspect.currentframe() # XXX: try/except here for debugging only try: stack = inspect.stack() for i in stack: print "from %s:%d" % (i[1], i[2]) except: print "couldn't get stack trace" raise ValueError("updateNode needs an nKu of type FludRSA" " (received %s) or an nID of type long or str (received %s)" % (type(nKu), type(nID))) # XXX: should really make it impossible to call without one of these... def replaceNode(error, routing, replacee, replacer): routing.replaceNode(replacee, replacer) print "replaced node in ktable" def requireParams(request, paramNames): # Looks for the named parameters in request. If found, returns # a dict of param/value mappings. If any named parameter is missing, # raises an exception params = {} for i in paramNames: try: params[i] = request.args[i][0] except: raise Exception, "missing parameter '"+i+"'" #XXX: use cust Exc return params def getCanonicalIP(IP): # if IP is 'localhost' or '127.0.0.1', use the canonical local hostname. # (this is mostly useful when multiple clients run on the same host) # XXX: could use gethostbyname to get IP addy instead. if IP == '127.0.0.1' or IP == 'localhost': return socket.getfqdn() else: return socket.getfqdn(IP) def getPageFactory(url, contextFactory=None, *args, **kwargs): def failedConnect(reason, factory): try: i = factory.status return reason except: pass #logger.warn("couldn't connect to %s:%d in getPageFactory: %s" # % (factory.host, factory.port, reason)) #logger.warn("state of factory is %s" % factory) #logger.warn("dir() of factory is %s" % dir(factory)) return reason if len(url) >= 16384: raise ValueError( "Too much data sent: twisted server doesn't appear to" " support urls longer than 16384") scheme, host, port, path = client._parse(url) factory = client.HTTPClientFactory(url, *args, **kwargs) factory.deferred.addErrback(failedConnect, factory) to = CONNECT_TO+random.randrange(2+CONNECT_TO_VAR)-CONNECT_TO_VAR if scheme == 'https': from twisted.internet import ssl if contextFactory is None: contextFactory = ssl.ClientContextFactory() reactor.connectSSL(host, port, factory, contextFactory) else: reactor.connectTCP(host, port, factory, timeout=to) return factory def _dlPageFactory(url, target, factoryClass, contextFactory=None, timeout=None, *args, **kwargs): scheme, host, port, path = client._parse(url) if timeout != None: # XXX: do something like http://twistedmatrix.com/pipermail/twisted-python/2003-August/005504.html pass factory = factoryClass(url, target, *args, **kwargs) to = CONNECT_TO+random.randrange(2+CONNECT_TO_VAR)-CONNECT_TO_VAR if scheme == 'https': from twisted.internet import ssl if contextFactory is None: contextFactory = ssl.ClientContextFactory() reactor.connectSSL(host, port, factory, contextFactory) else: reactor.connectTCP(host, port, factory, timeout=to) return factory def downloadPageFactory(url, file, contextFactory=None, timeout=None, *args, **kwargs): return _dlPageFactory(url, file, client.HTTPDownloader, contextFactory, timeout, *args, **kwargs) def multipartDownloadPageFactory(url, dir, contextFactory=None, timeout=None, *args, **kwargs): return _dlPageFactory(url, dir, HTTPMultipartDownloader, contextFactory, timeout, *args, **kwargs) def fileUpload(host, port, selector, files, form=(), headers={}): """ Performs a file upload via http. host - webserver hostname port - webserver listen port selector - the request (relative URL) files - list of files to upload. list contains tuples, with the first entry as filename/file-like obj and the second as form element name. If the first element is a file-like obj, the element will be used as the filename. If the first element is a filename, the filename's basename will be used as the filename on the form. Type will be "application/octet-stream" form (optional) - a list of pairs of additional name/value form elements (param/values). [hopefully, this method goes away in twisted-web2] """ # XXX: set timeout (based on filesize?) port = int(port) rand_bound = binascii.hexlify(generateRandom(13)) boundary = "---------------------------"+rand_bound CRLF = '\r\n' body_content_type = "application/octet-stream" content_type = "multipart/form-data; boundary="+boundary content_length = 0 H = [] for (param, value) in form: H.append('--' + boundary) H.append('Content-Disposition: form-data; name="%s"' % param) H.append('') H.append('%s' % value) form_data = CRLF.join(H)+CRLF content_length = content_length + len(form_data) fuploads = [] for file, element in files: if file == None: file = "/dev/null" # XXX: not portable if 'read' in dir(file): fname = element file.seek(0,2) file_length = file.tell() file.seek(0,0) else: fname = os.path.basename(file) file_length = os.stat(file)[stat.ST_SIZE] #logger.info("upload file %s len is %d" % (fname, file_length)) H = [] # stuff that goes above file data T = [] # stuff that goes below file data H.append('--' + boundary) H.append('Content-Disposition: form-data; name="%s"; filename="%s"' % (element, fname)) H.append('Content-Type: %s\n' % body_content_type) H.append('') file_headers = CRLF.join(H) content_length = content_length + len(file_headers) + file_length fuploads.append((file_headers, file, file_length)) T.append('--'+boundary+'--') T.append('') T.append('') trailer = CRLF.join(T) content_length = content_length + len(trailer) h = httplib.HTTPConnection(host, port) # XXX: blocking h.putrequest('POST', selector) for pageheader in headers: h.putheader(pageheader, headers[pageheader]) h.putheader('Content-Type', content_type) h.putheader('Content-Length', content_length) h.endheaders() h.send(form_data) for fheader, file, flen in fuploads: if 'read' not in dir(file): file = open(file, 'r') h.send(fheader) h.send(file.read(flen)+CRLF) # XXX: blocking file.close() h.send(trailer) return h class ImposterException(FludException): pass
Python
""" ClientDHTPrimitives.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Primitive client DHT protocol """ import time, os, stat, httplib, sys, random, logging from twisted.web import http, client from twisted.internet import reactor, threads, defer from twisted.python import failure import inspect, pdb from flud.FludCrypto import FludRSA import flud.FludkRouting as FludkRouting from flud.fencode import fencode, fdecode import flud.FludDefer as FludDefer import ConnectionQueue from ClientPrimitives import REQUEST from FludCommUtil import * logger = logging.getLogger("flud.client.dht") # FUTURE: check flud protocol version for backwards compatibility # XXX: need to make sure we have appropriate timeouts for all comms. # FUTURE: DOS attacks. For now, assume that network hardware can filter these # out (by throttling individual IPs) -- i.e., it isn't our problem. If we # want to defend against this at some point, we need to keep track of who # is generating requests and then ignore them. # XXX: might want to consider some self-healing for the kademlia layer, as # outlined by this thread: # http://zgp.org/pipermail/p2p-hackers/2003-August/001348.html (should also # consider Zooko's links in the parent to this post). Basic idea: don't # always take the k-closest -- take x random and k-x of the k-closest. # Can alternate each round (k-closest / x + k-x-closest) for a bit more # diversity (as in "Sybil-resistent DHT routing"). # XXX: right now, calls to updateNode are chained. Might want to think about # doing some of this more asynchronously, so that the recursive parts # aren't waiting for remote GETIDs to return before recursing. """ The first set of classes (those beginning with 'k') perform [multiple] queries given a key or key/value pair. They use the second set of classes (those beginning with 'SEND'), which perform a single query to a given node. """ def serviceWaiting(res, key, pending, waiting): # provides a method for calling multiple callbacks on a saved query. # add serviceWaiting as a callback before returning, and pass in the result, # pending dict and waiting dict. All deferreds in the waiting dict will # be called with the result, the waiting dict will be emptied of those # deferreds, and the pending dict will likewise be emptied. if waiting.has_key(key): for d in waiting[key]: #print "offbacking %s" % key d.callback(res) waiting.pop(key) pending.pop(key) return res pendingkFindNodes = {} waitingkFindNodes = {} class kFindNode: """ Perform a kfindnode lookup. """ def __init__(self, node, key): if pendingkFindNodes.has_key(key): d = defer.Deferred() if not waitingkFindNodes.has_key(key): waitingkFindNodes[key] = [] waitingkFindNodes[key].append(d) logger.debug("piggybacking on previous kfindnode for %s" % key) self.deferred = d return self.node = node self.node.DHTtstamp = time.time() self.key = key self.queried = {} self.outstanding = [] self.pending = [] self.kclosest = [] self.abbrvkey = ("%x" % key)[:8]+"..." self.abbrv = "(%s%s)" % (self.abbrvkey, str(self.node.DHTtstamp)[-7:]) self.debugpath = [] self.deferred = self.startQuery(key) def startQuery(self, key): # query self first kclosest = self.node.config.routing.findNode(key) #logger.debug("local kclosest: %s" % kclosest) localhost = getCanonicalIP('localhost') kd = {'id': self.node.config.nodeID, 'k': kclosest} d = self.updateLists(kd, key, localhost, self.node.config.port, long(self.node.config.nodeID, 16)) d.addErrback(self.errkfindnode, key, localhost, self.node.config.port) pendingkFindNodes[key] = d d.addCallback(serviceWaiting, key, pendingkFindNodes, waitingkFindNodes) return d def sendQuery(self, host, port, id, key): self.outstanding.append((host, port, id)) #d = self.node.client.sendkFindNode(host, port, key) d = SENDkFINDNODE(self.node, host, port, key).deferred return d def updateLists(self, response, key, host, port, closestyet, x=0): logger.info("FN: received kfindnode %s response from %s:%d" % (self.abbrv, host, port)) self.debugpath.append("FN: rec. resp from %s:%d" % (host, port)) if not isinstance(response, dict): # a data value is being returned from findval # XXX: moved this bit into findval and call parent for the rest if response == None: logger.warn("got None from key=%s, %s:%d, x=%d, this usually" " means that the host replied None to a findval query" % (key, host, port, x)) # if we found the fencoded value data, return it return defer.succeed(response) logger.debug("updateLists(%s)" % response) if len(response['k']) == 1 and response['k'][0][2] == key: # if we've found the key, don't keep making queries. logger.debug("FN: %s:%d found key %s" % (host, port, key)) self.debugpath.append("FN: %s:%d found key %s" % (host, port, key)) if response['k'][0] not in self.kclosest: self.kclosest.insert(0,response['k'][0]) self.kclosest = self.kclosest[:FludkRouting.k] return defer.succeed(response) #for i in response['k']: # print " res: %s:%d" % (i[0], i[1]) id = long(response['id'], 16) responder = (host, port, id) if responder in self.outstanding: self.outstanding.remove(responder) self.queried[id] = (host, port) knodes = response['k'] for n in knodes: if not self.queried.has_key(n[2])\ and not n in self.pending and not n in self.outstanding: self.pending.append((n[0], n[1], n[2])) if n not in self.kclosest: k = FludkRouting.k # XXX: remove self it in the list? self.kclosest.append(n) self.kclosest.sort( lambda a, b: FludkRouting.kCompare(a[2], b[2], key)) self.kclosest = self.kclosest[:k] #for n in self.outstanding: # if n in self.pending: # self.pending.remove(n) # remove anyone we've sent queries to... self.pending = list(set(self.pending) - set(self.outstanding)) for i in self.queried: n = (self.queried[i][0], self.queried[i][1], i) if n in self.pending: self.pending.remove(n) # ...and anyone who has responded. self.pending.sort(lambda a, b: FludkRouting.kCompare(a[2], b[2], key)) #print "queried: %s" % str(self.queried) #print "outstanding: %s" % str(self.outstanding) #print "pending: %s" % str(self.pending) return self.decideToContinue(response, key, x) def decideToContinue(self, response, key, x): ##print "x is %s" % str(x) ##for i in self.kclosest: ## print " kclosest %s" % str(i) ##for i in self.queried: ## print " queried %s" % str(self.queried[i]) #if len(filter(lambda x: x not in self.queried, self.kclosest)) <= 0: # print "finishing up at round %d" % x # # XXX: never gets in here... # # XXX: remove anything not in self.kclosest from self.pending # self.pending =\ # filter(lambda x: x not in self.kclosest, self.pending) # #self.pending = self.pending[:FludkRouting.k] #else: # return self.makeQueries(key, x) # this is here so that kFindVal can plug-in by overriding return self.makeQueries(key, x) def makeQueries(self, key, x): #print "doing round %d" % x self.debugpath.append("FN: doing round %d" % x) dlist = [] for n in self.pending[:(FludkRouting.a - len(self.outstanding))]: #print " querying %s:%d" % (n[0], n[1]) self.debugpath.append("FN: querying %s:%d" % (n[0], n[1])) d = self.sendQuery(n[0], n[1], n[2], key) d.addCallback(self.updateLists, key, n[0], n[1], self.kclosest[0][2], x+1) d.addErrback(self.errkfindnode, key, n[0], n[1], raiseException=False) dlist.append(d) dl = defer.DeferredList(dlist) dl.addCallback(self.roundDone, key, x) return dl def roundDone(self, responses, key, x): #print "done %d:" % x #print "roundDone: %s" % responses if len(self.pending) != 0 or len(self.outstanding) != 0: # should only get here for nodes that don't accept connections # XXX: updatenode -- decrease trust for i in self.pending: logger.debug("FN: %s couldn't contact node %s (%s:%d)" % (self.abbrv, fencode(i[2]), i[0], i[1])) self.debugpath.append( "FN: %s couldn't contact node %s (%s:%d)" % (self.abbrv, fencode(i[2]), i[0], i[1])) for n in self.kclosest: if (n[0],n[1],n[2]) == i: self.kclosest.remove(n) logger.info("kFindNode %s terminated successfully after %d queries." % (self.abbrv, len(self.queried))) self.debugpath.append("FN: %s terminated successfully after %d queries." % (self.abbrv, len(self.queried))) self.kclosest.sort( lambda a, b: FludkRouting.kCompare(a[2], b[2], key)) result = {} if FludkRouting.k > len(self.kclosest): k = len(self.kclosest) else: k = FludkRouting.k result['k'] = self.kclosest[:k] #print "result: %s" % result #if len(result['k']) > 1: # # if the results (aggregated from multiple responses) contains the # # exact key, just return the correct answer (successful node # # lookup done). # #print "len(result): %d" % len(result['k']) # #print "result[0][2]: %s %d" % (type(result['k'][0][2]), # # result['k'][0][2]) # #print " key: %s %d" % (type(key), key) # if result['k'][0][2] == key: # #print "key matched!" # result['k'] = (result['k'][0],) return result def errkfindnode(self, failure, key, host, port, raiseException=True): logger.info("kFindNode %s request to %s:%d failed -- %s" % (self.abbrv, host, port, failure.getErrorMessage())) # XXX: updateNode-- if raiseException: return failure class kStore(kFindNode): """ Perform a kStore operation. """ def __init__(self, node, key, val): self.node = node self.node.DHTtstamp = time.time() self.key = key self.val = val d = kFindNode(node,key).deferred d.addCallback(self.store) d.addErrback(self._kStoreErr, None, 0) self.deferred = d def store(self, knodes): knodes = knodes['k'] if len(knodes) < 1: raise RuntimeError("can't complete kStore -- no nodes") dlist = [] for knode in knodes: host = knode[0] port = knode[1] deferred = SENDkSTORE(self.node, host, port, self.key, self.val).deferred deferred.addErrback(self._kStoreErr, host, port) dlist.append(deferred) dl = FludDefer.ErrDeferredList(dlist) dl.addCallback(self._kStoreFinished) dl.addErrback(self._kStoreErr, None, 0) return dl def _kStoreFinished(self, response): #print "_kStoreFinished: %s" % response logger.info("kStore finished") return "" def _kStoreErr(self, failure, host, port): logger.info("couldn't store on %s:%d -- %s" % (host, port, failure.getErrorMessage())) print "_kStoreErr was: %s" % failure # XXX: updateNode-- return failure class kFindValue(kFindNode): """ Perform a kFindValue. """ def __init__(self, node, key): self.done = False kFindNode.__init__(self, node, key) def startQuery(self, key): # query self first. We override kFindNode.startQuery here so that # we don't just return the closest nodeID, but the value itself (if # present) localhost = getCanonicalIP('localhost') d = self.sendQuery(localhost, self.node.config.port, long(self.node.config.nodeID, 16), key) d.addCallback(self.updateLists, key, localhost, self.node.config.port, long(self.node.config.nodeID, 16), 0) d.addErrback(self.errkfindnode, key, localhost, self.node.config.port) return d def sendQuery(self, host, port, id, key): # We override sendQuery here in order to call sendkFindValue and handle # its response self.outstanding.append((host, port, id)) d = SENDkFINDVALUE(self.node, host, port, key).deferred d.addCallback(self._handleFindVal, host, port) d.addErrback(self.errkfindnode, key, host, port) return d def _handleFindVal(self, response, host, port): if not isinstance(response, dict): # stop sending out more queries. self.pending = [] self.done = True #print "%s:%d sent value: %s" % (host, port, str(response)[:50]) #f = {} #f['k'] = [] #f['id'] = "0" #f['val'] = response # pass on returned value #return f else: pass #print "%s:%d sent kData: %s" % (host, port, response) return response def decideToContinue(self, response, key, x): if self.done: #if not response.has_key('val'): # logger.warn("response has no 'val', response is: %s" % response) #return response['val'] return response else: return self.makeQueries(key, x) def roundDone(self, responses, key, x): self.debugpath.append("FV: roundDone %d" % x) if self.done: result = {} # see if everyone's responses agreed... for success, resp in responses: # only look at successful non-kData (dict) responses. if success and resp != None and not isinstance(resp, dict): if result.has_key(resp): result[resp] += 1 else: result[resp] = 1 if len(result) == 0: # ... if no one responded, XXX: do something orther than None? logger.info("couldn't get any results") return None elif len(result) == 1: # ... if they did, return the result return result.keys()[0] else: # ... otherwise, return the result of the majority # (other options include returning all results) logger.info("got conflicting results, determining best...") quorumResult = None bestScore = 0 for r in result: #logger.debug("result %s scored %d" % (r, result[r])) if result[r] > bestScore: bestScore = result[r] quorumResult = r #logger.debug("result %s is new best" % r) logger.info("returning result %s", fdecode(quorumResult)) return quorumResult class SENDkFINDNODE(REQUEST): """ Makes one request to a node for its k-closest nodes closest to key """ def __init__(self, node, host, port, key, commandName="kFINDNODE"): """ """ logger.info("sending %s (findnode) for %s... to %s:%d" % (commandName, ("%x" % key)[:10], host, port)) self.commandName = commandName host = getCanonicalIP(host) REQUEST.__init__(self, host, port, node) Ku = self.node.config.Ku.exportPublicKey() url = 'http://'+host+':'+str(port)+'/'+self.commandName+'?' url += 'nodeID='+str(self.node.config.nodeID) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) url += '&port='+str(self.node.config.port) url += "&key="+str(fencode(key)) self.timeoutcount = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, node, host, port, key, url)) def startRequest(self, node, host, port, key, url): d = self._sendRequest(node, host, port, key, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, node, host, port, key, url): factory = getPageFactory(url, headers=self.headers, timeout=kprimitive_to) factory.deferred.addCallback(self._gotResponse, factory, node, host, port, key) factory.deferred.addErrback(self._errSendk, factory, node, host, port, key, url) return factory.deferred def _gotResponse(self, response, factory, node, host, port, key): logger.debug("kfindnode._gotResponse()") self._checkStatus(factory.status, response, host, port) response = eval(response) nID = long(response['id'], 16) updateNode(node.client, node.config, host, port, None, nID) updateNodes(node.client, node.config, response['k']) return response def _checkStatus(self, status, response, host, port): logger.debug("kfindnode._checkStatus()") if eval(status) != http.OK: raise failure.DefaultException(self.commandName+" FAILED from " +host+":"+str(port)+": received status "+status+", '" +response+"'") def _errSendk(self, err, factory, node, host, port, key, url): if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #print "GETID request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(node, host, port, key, url) else: #print "not trying again [#%d]" % self.timeoutcount return err logger.info("%s to %s failed -- %s" % (self.commandName, self.dest, err.getErrorMessage())) # XXX: updateNode-- return err class SENDkSTORE(REQUEST): """ Sends a single kSTORE to the given host:port, with key=val """ def __init__(self, node, host, port, key, val): logger.info("sending kSTORE to %s:%d" % (host, port)) REQUEST.__init__(self, host, port, node) Ku = node.config.Ku.exportPublicKey() url = 'http://'+host+':'+str(port)+'/kSTORE?' url += 'nodeID='+str(node.config.nodeID) url += "&Ku_e="+str(Ku['e']) url += "&Ku_n="+str(Ku['n']) url += '&port='+str(node.config.port) url += "&key="+str(fencode(key)) url += "&val="+str(fencode(val)) # XXX: instead of a single key/val, protocol will take a series of # vals representing the blocks of the coded file and their # locations (by nodeID). The entire thing will be stored under # the given key. Also may need things like signature[s] from # storing node[s], etc. #print "in SENDkSTORE.__init__, len(val)=%d" % len(str(val)) #print "in SENDkSTORE.__init__, len(enc(val))=%d" % len(fencode(val)) #print "in SENDkSTORE.__init__, len(url)=%d" % len(url) self.timeoutcount = 0 self.deferred = defer.Deferred() ConnectionQueue.enqueue((self, host, port, url)) def startRequest(self, host, port, url): d = self._sendRequest(host, port, url) d.addBoth(ConnectionQueue.checkWaiting) d.addCallback(self.deferred.callback) d.addErrback(self.deferred.errback) def _sendRequest(self, host, port, url): factory = getPageFactory(url,\ headers=self.headers, method='PUT', timeout=kprimitive_to) self.deferred.addCallback(self._kStoreFinished, host, port) self.deferred.addErrback(self._storeErr, host, port, url) return factory.deferred def _kStoreFinished(self, response, host, port): logger.info("kSTORE to %s:%d finished" % (host, port)) return response def _storeErr(self, err, host, port, url): if err.check('twisted.internet.error.TimeoutError') or \ err.check('twisted.internet.error.ConnectionLost'): #print "GETID request error: %s" % err.__class__.__name__ self.timeoutcount += 1 if self.timeoutcount < MAXTIMEOUTS: #print "trying again [#%d]...." % self.timeoutcount return self._sendRequest(host, port, url) else: #print "not trying again [#%d]" % self.timeoutcount return err logger.info("kSTORE to %s failed: %s" % (self.dest, err.getErrorMessage())) # XXX: updateNode-- return err class SENDkFINDVALUE(SENDkFINDNODE): """ Issues a single kFINDVALUE request to host:port for the key. If the value is found at host:port, it is returned, otherwise, a 404 response is received and any errbacks are called. """ def __init__(self, node, host, port, key): SENDkFINDNODE.__init__(self, node, host, port, key, "kFINDVAL") def _gotResponse(self, response, factory, node, host, port, key): self._checkStatus(factory.status, response, host, port) # The following 'if' block is the only thing different from kFINDNODE. # If a node returns the value, content-type will be set to x-flud-data # and we should grab the data instead of continuing the recursive # search. if factory.response_headers.has_key('content-type')\ and factory.response_headers['content-type']\ == ['application/x-flud-data']: logger.info("received SENDkFINDVALUE data.") nID = None if factory.response_headers.has_key('nodeid'): nID = factory.response_headers['nodeid'][0] updateNode(node.client, node.config, host, port, None, nID) return response response = eval(response) nID = long(response['id'], 16) updateNode(node.client, node.config, host, port, None, nID) logger.info("received SENDkFINDVALUE nodes") logger.debug("received SENDkFINDVALUE nodes: %s" % response) updateNodes(node.client, node.config, response['k']) return response
Python
""" FludkRouting.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Implements kademlia-style kbuckets (the routing table for the DHT layer). Although this is not a derivative of Khashmir (written by Andrew Loewenstern, Aaron Swartz, et. al.), we would like to give Khashmir a nod for inspiring portions of the design. Khashmir is distributed under the MIT License and is a very nice piece of work. Take a look at http://khashmir.sourceforge.net/ for more information. """ from bisect import * import logging #k = 5 # This is the max depth of a kBucket k = 12 # This is the max depth of a kBucket and the replication factor # XXX: need to split out k into two vars a = 3 # alpha, the system-wide concurrency parameter idspace = 256 # using sha-256 logger = logging.getLogger("flud.k") class NodeCache: """ An LRU cache for nodes """ def __init__(self, size): self.size = size self.cache = {} self.cacheOrder = [] def insertNode(self, node): """ adds a node to the cache. if this displaces a node, the displaced node is returned """ if node[2] not in self.cache: self.cache[node[2]] = node self.cacheOrder.append(node[2]) if len(self.cacheOrder) > self.size: popped = self.cacheOrder.pop(0) self.cache.pop(popped) return popped def getNode(self, nodeID): """ returns a node from the cache, or None """ if nodeID in self.cache: return self.cache[node] return None def removeNode(self, node): """ removes a node from the cache """ if node[2] in self.cache: self.cache.pop(node[2]) self.cacheOrder.pop(node[2]) def nodes(self): return [self.cache[i] for i in self.cache] def kCompare(a, b, target): """ Uses the XOR metric to compare target to a and b (useful for sorting) @param a an integer (or long) value @param b an integer (or long) value @param target the target ID as an integer (or long) value @return 1 if b > a, -1 if a < b, 0 if a == b >>> l = [1, 2, 23, 14, 5, 4, 5, 3, 20] >>> l.sort(lambda a, b: kCompare(a, b, 5)) >>> l [5, 5, 4, 1, 3, 2, 14, 20, 23] """ x, y = target^a, target^b if x == y: return 0 return int((x - y) / abs(x - y)) class kRouting: """ Contains the kBuckets for this node. Provides methods for inserting, updating, and removing nodes. Most importantly, performs kademlia-style routing by returning the node[s] closest to a particular id. >>> table = kRouting(('1.2.3.4', 34, 123456), 20, 5) >>> table.insertNode(('2.2.3.4', 34, 23456)) >>> table.insertNode(('3.2.3.4', 34, 223456)) >>> table.insertNode(('4.2.3.4', 34, 723456)) >>> table.insertNode(('5.2.3.4', 34, 423456)) >>> table.insertNode(('6.2.3.4', 34, 323456)) >>> table.kBuckets [{'0-80000': [('1.2.3.4', 34, 123456), ('2.2.3.4', 34, 23456), ('3.2.3.4', 34, 223456), ('5.2.3.4', 34, 423456), ('6.2.3.4', 34, 323456)]}, {'80001-100000': [('4.2.3.4', 34, 723456)]}] >>> table.findNode(23456) [('2.2.3.4', 34, 23456), ('1.2.3.4', 34, 123456), ('3.2.3.4', 34, 223456), ('6.2.3.4', 34, 323456), ('5.2.3.4', 34, 423456)] >>> table.findNode(55555) [('2.2.3.4', 34, 23456), ('1.2.3.4', 34, 123456), ('3.2.3.4', 34, 223456), ('6.2.3.4', 34, 323456), ('5.2.3.4', 34, 423456)] >>> table.findNode(722222) [('4.2.3.4', 34, 723456), ('3.2.3.4', 34, 223456), ('1.2.3.4', 34, 123456), ('2.2.3.4', 34, 23456), ('5.2.3.4', 34, 423456)] >>> table.insertNode(('7.2.3.4', 34, 733456)) >>> table.insertNode(('8.2.3.4', 34, 743456)) >>> table.insertNode(('9.2.3.4', 34, 753456)) >>> table.insertNode(('10.2.3.4', 34, 763456)) >>> table.insertNode(('11.2.3.4', 34, 773456)) ('4.2.3.4', 34, 723456) >>> table.replaceNode(('4.2.3.4', 34, 723456), ('11.2.3.4', 34, 773456)) >>> table.kBuckets [{'0-80000': [('1.2.3.4', 34, 123456), ('2.2.3.4', 34, 23456), ('3.2.3.4', 34, 223456), ('5.2.3.4', 34, 423456), ('6.2.3.4', 34, 323456)]}, {'80001-100000': [('7.2.3.4', 34, 733456), ('8.2.3.4', 34, 743456), ('9.2.3.4', 34, 753456), ('10.2.3.4', 34, 763456), ('11.2.3.4', 34, 773456)]}] >>> table.removeNode(('1.2.3.4', 34, 123456)) >>> table.kBuckets [{'0-80000': [('2.2.3.4', 34, 23456), ('3.2.3.4', 34, 223456), ('5.2.3.4', 34, 423456), ('6.2.3.4', 34, 323456)]}, {'80001-100000': [('7.2.3.4', 34, 733456), ('8.2.3.4', 34, 743456), ('9.2.3.4', 34, 753456), ('10.2.3.4', 34, 763456), ('11.2.3.4', 34, 773456)]}] >>> table.knownNodes() [('2.2.3.4', 34, 23456), ('3.2.3.4', 34, 223456), ('5.2.3.4', 34, 423456), ('6.2.3.4', 34, 323456), ('7.2.3.4', 34, 733456), ('8.2.3.4', 34, 743456), ('9.2.3.4', 34, 753456), ('10.2.3.4', 34, 763456), ('11.2.3.4', 34, 773456)] >>> table.knownExternalNodes() [('2.2.3.4', 34, 23456), ('3.2.3.4', 34, 223456), ('5.2.3.4', 34, 423456), ('6.2.3.4', 34, 323456), ('7.2.3.4', 34, 733456), ('8.2.3.4', 34, 743456), ('9.2.3.4', 34, 753456), ('10.2.3.4', 34, 763456), ('11.2.3.4', 34, 773456)] """ def __init__(self, node, bits=idspace, depth=k): """ @param node a (ip, port, id) triple, where id is an int (this is needed to know when to split a bucket). """ self.k = depth self.replacementCache = NodeCache(300) self.kBuckets = [kBucket(0, 2**bits, depth),] #self.kBuckets = [kBucket(0, 1, depth),] #for i in xrange(1,bits): # self.kBuckets.append(kBucket(2**i, 2**(i+1)-1, depth)) self.insertNode(node) self.node = node def insertNode(self, node): """ Inserts a node into the appropriate kBucket. If the node already exists in the appropriate kBucket, it is moved to the tail of the list. If the bucket is full, this method returns the oldest node, which the caller should then ping. If the oldest node is alive, the caller does nothing. Otherwise, the caller should call replaceNode. @param node a (ip, port, id) triple, where id is a long. """ if len(node) < 3: raise ValueError("node must be a triple (ip, port, id)") id = node[2] bucket = self._findBucket(id) try: # XXX: need to transfer key/vals that belong to new node? bucket.updateNode(node) self.replacementCache.removeNode(node) except BucketFullException, e: if (bucket.begin <= self.node[2] < bucket.end): # bucket is full /and/ the local node is in this bucket, # split and try adding it again. self._splitBucket(bucket) self.insertNode(node) logger.debug("split and added %x" % node[2]) return # XXX: need to also split for some other cases, see sections 2.4 # and 4.2. else: # bucket is full but we won't split. Return the oldest node # so that the caller can determine if it should be expunged. # If the old node is not reachable, caller should call # replaceNode() logger.debug("didn't add %x" % node[2]) return bucket.contents[0] logger.debug("didn't add %x" % node[2]) return bucket.contents[0] def removeNode(self, node): """ Invalidates a node. """ bucket = self._findBucket(node[2]) bucket.delNode(node) def replaceNode(self, replacee, replacer): """ Expunges replacee from its bucket, making room to add replacer """ # XXX: constraint checks: replacee & replacer belong to the same bucket, # bucket is currently full, adding replacer doesn't overfill, etc. self.removeNode(replacee) self.insertNode(replacer) def findNode(self, nodeID): """ Returns k closest node triples with which the caller may make additional queries. If nodeID is found, it will be the first result. @param nodeID an int """ nodes = [] bucket = self._findBucket(nodeID) #n = bucket.findNode(nodeID) #if n != None: # nodes.append(n) nodes += bucket.contents if len(nodes) < self.k: nextbucket = self._nextbucket(bucket) prevbucket = self._prevbucket(bucket) while len(nodes) < self.k \ and (nextbucket != None or prevbucket != None): if nextbucket != None: nodes += nextbucket.contents if prevbucket != None: nodes += prevbucket.contents nextbucket = self._nextbucket(nextbucket) prevbucket = self._prevbucket(prevbucket) nodes.sort(lambda a, b, n=nodeID: cmp(n ^ a[2], n ^ b[2])) return nodes[:self.k] def getNode(self, nodeID): """ Attempts to get the given node, returning a <ip, port, id> triple. If the node is not found locally, returns None @param nodeID an int """ bucket = self._findBucket(nodeID) n = bucket.findNode(nodeID) if not n: n = self.replacementCache.getNode(nodeID) if n != None: return n return None def updateNode(self, node): """ Call to update a node, i.e., whenever the node has been recently seen @param node a (ip, port, id) triple, where id is an int. """ self.insertNode(node) def knownExternalNodes(self): result = [] for i in self.kBuckets: for j in i.contents: if j[2] != self.node[2]: result.append(j) result += self.replacementCache.nodes() return result def knownNodes(self): result = [] for i in self.kBuckets: for j in i.contents: result.append(j) result += self.replacementCache.nodes() return result def _nextbucket(self, bucket): if bucket == None: return bucket i = self.kBuckets.index(bucket)+1 if i >= len(self.kBuckets): return None return self.kBuckets[i] def _prevbucket(self, bucket): if bucket == None: return bucket i = self.kBuckets.index(bucket)-1 if i < 0: return None return self.kBuckets[i] def _findBucket(self, i): """ returns the bucket which would contain i. @param i an int """ #print "kBuckets = %s" % str(self.kBuckets) bl = bisect_left(self.kBuckets, i) if bl >= len(self.kBuckets): raise Exception( "tried to find an ID that is larger than ID space: %s" % i) return self.kBuckets[bisect_left(self.kBuckets, i)] def _splitBucket(self, bucket): """ This is called for the special case when the bucket is full and this node is a member of the bucket. When this occurs, the bucket should be split into two new buckets. """ halfpoint = (bucket.end - bucket.begin) / 2 newbucket = kBucket(bucket.end - halfpoint + 1, bucket.end, self.k) self.kBuckets.insert(self.kBuckets.index(bucket.begin) + 1, newbucket) bucket.end -= halfpoint for node in bucket.contents[:]: if node[2] > bucket.end: bucket.delNode(node) newbucket.addNode(node) class kBucket: """ A kBucket is a list of <ip, port, id> triples, ordered according to time last seen (most recent at tail). Every kBucket has a begin and end number, indicating the chunk of the id space that it contains. >>> b = kBucket(0,100,5) >>> b {'0-64': []} >>> n1 = ('1.2.3.4', 45, 'd234a53546e4c23') >>> n2 = ('10.20.30.40', 45, 'abcd234a53546e4') >>> n3 = ('10.20.30.4', 5, 'abcd') >>> b.addNode(n1) >>> b {'0-64': [('1.2.3.4', 45, 'd234a53546e4c23')]} >>> b.addNode(n2) >>> b {'0-64': [('1.2.3.4', 45, 'd234a53546e4c23'), ('10.20.30.40', 45, 'abcd234a53546e4')]} >>> b.addNode(n1) >>> b {'0-64': [('10.20.30.40', 45, 'abcd234a53546e4'), ('1.2.3.4', 45, 'd234a53546e4c23')]} >>> b.delNode(n3) >>> b {'0-64': [('10.20.30.40', 45, 'abcd234a53546e4'), ('1.2.3.4', 45, 'd234a53546e4c23')]} >>> b.addNode(n2) >>> b {'0-64': [('1.2.3.4', 45, 'd234a53546e4c23'), ('10.20.30.40', 45, 'abcd234a53546e4')]} >>> b.updateNode(n1) >>> b {'0-64': [('10.20.30.40', 45, 'abcd234a53546e4'), ('1.2.3.4', 45, 'd234a53546e4c23')]} >>> b.delNode(n2) >>> b {'0-64': [('1.2.3.4', 45, 'd234a53546e4c23')]} >>> b.addNode(n3) >>> f = b.findNode(n3[2]) >>> f == n3 True >>> c = kBucket(101,200,5) >>> d = kBucket(150,250,5) # wouldn't really have overlap in practice >>> e = kBucket(251, 2**256,5) >>> buckets = (b, c, d, e) # if not added insort, must sort for bisect >>> b1 = b >>> b1 == b True >>> b1 != b False >>> b == 50 True >>> b == 0 True >>> b == 100 True >>> b == -1 False >>> b > -1 True >>> b == 101 False >>> b < 101 True >>> b <= 90 True >>> b <= 100 True >>> b <= 101 True >>> b < d True >>> b <= c True >>> b > c False >>> bisect_left(buckets, 98) 0 >>> bisect_left(buckets, 198) 1 >>> bisect_left(buckets, 238) 2 >>> bisect_left(buckets, 298) 3 """ def __init__(self, begin, end, depth=k): self.k = depth self.begin = begin self.end = end self.contents = [] def __repr__(self): return "{'%x-%x': %s}" % (self.begin, self.end, self.contents) #return "{'"+repr(self.begin)+'-'+repr(self.end)+"': "\ # +repr(self.contents)+"}" #return "<kBucket "+repr(self.begin)+'-'+repr(self.end)+": "\ # +repr(self.contents)+">" def addNode(self, node): """ adds the given node to this bucket. If the node is already a member of this bucket, its position is updated to the end of the list. If the bucket is full, raises an exception """ if node in self.contents: self.contents.remove(node) self.contents.append(node) elif len(self.contents) >= self.k: raise BucketFullException() else: ids = [x[2] for x in self.contents] if node[2] in ids: # remove the matching node's old contact info self.contents.pop(ids.index(node[2])) self.contents.append(node) def updateNode(self, node): """ Moves the given node to the tail of the list. If the node isn't present in this bucket, this method attempts to add it by calling addNode (which may throw a BucketFullException if bucket is full) """ self.addNode(node) def delNode(self, node): """ removes the given node, if present, from this bucket """ try: self.contents.remove(node) except: pass def findNode(self, nodeID): for i in self.contents: if i[2] == nodeID: return i return None # The following comparators allow us to use list & bisect on the buckets. # integers, longs, and buckets all may be compared to a bucket. def __eq__(self, i): return i >= self.begin and self.end >= i def __ne__(self, i): return i < self.begin or self.end < i def __lt__(self, i): return self.end < i def __le__(self, i): return self.begin <= i def __gt__(self, i): return self.begin > i def __ge__(self, i): return self.end >= i class BucketFullException(Exception): pass def _test(): import doctest doctest.testmod() if __name__ == '__main__': _test()
Python
import zfec import zfec.easyfec as easyfec import zfec.filefec as filefec from pyutil import fileutil from pyutil.mathutil import pad_size, log_ceil import array, os, re, struct, traceback FORMAT_FORMAT = "%%s.%%0%dd_%%0%dd%%s" RE_FORMAT = "%s.[0-9]+_[0-9]+%s" def encode_to_files(inf, fsize, dirname, prefix, k, m, suffix=".fec", overwrite=False, verbose=False): """ Encode inf, writing the shares to specially named, newly created files. @param fsize: calling read() on inf must yield fsize bytes of data and then raise an EOFError @param dirname: the name of the directory into which the sharefiles will be written """ mlen = len(str(m)) format = FORMAT_FORMAT % (mlen, mlen,) padbytes = pad_size(fsize, k) fns = [] fs = [] try: for shnum in range(m): hdr = filefec._build_header(m, k, padbytes, shnum) fn = os.path.join(dirname, format % (prefix, shnum, m, suffix,)) if verbose: print "Creating share file %r..." % (fn,) if overwrite: f = open(fn, "wb") else: flags = os.O_WRONLY|os.O_CREAT|os.O_EXCL | (hasattr(os, 'O_BINARY') and os.O_BINARY) fd = os.open(fn, flags) f = os.fdopen(fd, "wb") f.write(hdr) fs.append(f) fns.append(fn) sumlen = [0] def cb(blocks, length): assert len(blocks) == len(fs) oldsumlen = sumlen[0] sumlen[0] += length if verbose: if int((float(oldsumlen) / fsize) * 10) \ != int((float(sumlen[0]) / fsize) * 10): print str(int((float(sumlen[0]) / fsize) * 10) * 10) \ + "% ...", if sumlen[0] > fsize: raise IOError("Wrong file size -- possibly the size of the" " file changed during encoding. Original size: %d," " observed size at least: %s" % (fsize, sumlen[0],)) for i in range(len(blocks)): data = blocks[i] fs[i].write(data) length -= len(data) filefec.encode_file_stringy_easyfec(inf, cb, k, m, chunksize=4096) except EnvironmentError, le: print "Cannot complete because of exception: " print le print "Cleaning up..." # clean up while fs: f = fs.pop() f.close() ; del f fn = fns.pop() if verbose: print "Cleaning up: trying to remove %r..." % (fn,) fileutil.remove_if_possible(fn) return None if verbose: print print "Done!" return fns # Note: if you really prefer base-2 and you change this code, then please # denote 2^20 as "MiB" instead of "MB" in order to avoid ambiguity. # Thanks. # http://en.wikipedia.org/wiki/Megabyte MILLION_BYTES=10**6 def decode_from_files(outf, infiles, verbose=False): """ Decode from the first k files in infiles, writing the results to outf. """ assert len(infiles) >= 2 infs = [] shnums = [] m = None k = None padlen = None byteswritten = 0 for f in infiles: (nm, nk, npadlen, shnum,) = filefec._parse_header(f) if not (m is None or m == nm): raise CorruptedShareFilesError("Share files were corrupted --" " share file %r said that m was %s but another share file" " previously said that m was %s" % (f.name, nm, m,)) m = nm if not (k is None or k == nk): raise CorruptedShareFilesError("Share files were corrupted --" " share file %r said that k was %s but another share file" " previously said that k was %s" % (f.name, nk, k,)) if k > len(infiles): raise InsufficientShareFilesError(k, len(infiles)) k = nk if not (padlen is None or padlen == npadlen): raise CorruptedShareFilesError("Share files were corrupted --" " share file %r said that pad length was %s but another" " share file previously said that pad length was %s" % (f.name, npadlen, padlen,)) padlen = npadlen infs.append(f) shnums.append(shnum) if len(infs) == k: break dec = easyfec.Decoder(k, m) while True: chunks = [ inf.read(filefec.CHUNKSIZE) for inf in infs ] if [ch for ch in chunks if len(ch) != len(chunks[-1])]: raise CorruptedShareFilesError("Share files were corrupted --" " all share files are required to be the same length," " but they weren't.") if len(chunks[-1]) == filefec.CHUNKSIZE: # Then this was a full read, so we're still in the sharefiles. resultdata = dec.decode(chunks, shnums, padlen=0) outf.write(resultdata) byteswritten += len(resultdata) if verbose: if ((byteswritten - len(resultdata)) / (10*MILLION_BYTES)) \ != (byteswritten / (10*MILLION_BYTES)): print str(byteswritten / MILLION_BYTES) + " MB ...", else: # Then this was a short read, so we've reached the end of the # sharefiles. resultdata = dec.decode(chunks, shnums, padlen) outf.write(resultdata) return True if verbose: print print "Done!" return True # fludfilefec -- modified zfec filefec for use with flud, based on: # # zfec -- fast forward error correction library with Python interface # # Copyright (C) 2007 Allmydata, Inc. # Author: Zooko Wilcox-O'Hearn # # This file is part of zfec. # # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the Free # Software Foundation; either version 2 of the License, or (at your option) # any later version, with the added permission that, if you become obligated # to release a derived work under this licence (as per section 2.b), you may # delay the fulfillment of this obligation for up to 12 months. See the file # COPYING for details. # # If you would like to inquire about a commercial relationship with Allmydata, # Inc., please contact partnerships@allmydata.com and visit # http://allmydata.com/.
Python
from twisted.python import failure from twisted.internet import defer """ FludDefer.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. """ class ErrDeferredList(defer.DeferredList): """ ErrDeferredList acts just like DeferredList, except that if *any* of the Deferreds in the DeferredList errback(), the NewDeferredList also errback()s. This is different from DeferredList(fireOnOneErrback=True) in that if you use that method, you only know about the first failure, and you won't learn of subsequent failures/success in the list. returnOne indicates whether the full result of the DeferredList should be returned, or just the first result (or first error) """ def __init__(self, list, returnOne=False): defer.DeferredList.__init__(self, list, consumeErrors=True) self.returnOne = returnOne self.addCallback(self.wrapResult) def wrapResult(self, result): #print "DEBUG: result= %s" % result for i in result: if i[0] == False: if self.returnOne: raise failure.DefaultException(i[1]) else: raise failure.DefaultException(result) if self.returnOne: #print "DEBUG: returning %s" % str(result[0][1]) return result[0][1] else: #print "DEBUG: returning %s" % result return result
Python
#!/usr/bin/python """ FludNode.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), verison 3. FludNode is the process that runs to talk with other nodes in the flud backup network. """ from twisted.internet import reactor, defer import threading, signal, sys, time, os, random, logging from flud.FludConfig import FludConfig from flud.protocol.FludServer import FludServer from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import getCanonicalIP PINGTIME=60 SYNCTIME=900 class FludNode(object): """ A node in the flud network. A node is both a client and a server. It listens on a network accessible port for both DHT and Storage layer requests, and it listens on a local port for client interface commands. """ def __init__(self, port=None): self._initLogger() self.config = FludConfig() self.logger.removeHandler(self.screenhandler) self.config.load(serverport=port) self.client = FludClient(self) self.DHTtstamp = time.time()+10 def _initLogger(self): logger = logging.getLogger('flud') self.screenhandler = logging.StreamHandler() self.screenhandler.setLevel(logging.INFO) logger.addHandler(self.screenhandler) self.logger = logger def pingRandom(self, tstamp): return # XXX: see pg. 4, Section 2.2 (short) or 2.3 (long) of the Kademlia # paper -- once an hour, each node should check any buckets that # haven't been refreshed and pick a random id within that space # to findnode(id) on, for all buckets. if tstamp < self.DHTtstamp: #r = random.randrange(2**256) n = self.config.routing.knownExternalNodes() if len(n) > 2: n1 = random.choice(n)[2] n2 = random.choice(n)[2] r = (n1+n2)/2 else: r = random.randrange(2**256) def badNode(error): node.logger.warn("Couldn't ping %s:%s" % (sys.argv[1], sys.argv[2])) d = self.client.kFindNode(r) d.addErrback(badNode) pingtime = random.randrange(PINGTIME/2, PINGTIME) reactor.callLater(pingtime, self.pingRandom, time.time()) def syncConfig(self): self.config.save() reactor.callLater(SYNCTIME, self.syncConfig) def start(self, twistd=False): """ starts the reactor in this thread """ self.webserver = FludServer(self, self.config.port) self.logger.log(logging.INFO, "FludServer starting") reactor.callLater(1, self.pingRandom, time.time()) reactor.callLater(random.randrange(10), self.syncConfig) if not twistd: reactor.run() def run(self): """ starts the reactor in its own thread """ #signal.signal(signal.SIGINT, self.sighandler) signal.signal(signal.SIGINT, signal.SIG_DFL) signal.signal(signal.SIGTERM, signal.SIG_DFL) self.webserver = FludServer(self, self.config.port) self.webserver.start() # XXX: need to do save out current config every X seconds # XXX: need to seperate known_nodes from config, and then update this # every X seconds. only update config when it changes. def stop(self): self.logger.log(logging.INFO, "shutting down FludNode") self.webserver.stop() def join(self): self.webserver.join() def sighandler(self, sig, frame): self.logger.log(logging.INFO, "handling signal %s" % sig) def connectViaGateway(self, host, port): def refresh(knodes): def refreshDone(results): self.logger.info("bucket refreshes finished: %s" % results) print "flud node connected and listening on port %d"\ % self.config.port #print "found knodes %s" % knodes dlist = [] for bucket in self.config.routing.kBuckets: #if True: if bucket.begin <= self.config.routing.node[2] < bucket.end: pass #print "passed on bucket %x-%s" % (bucket.begin, bucket.end) else: refreshID = random.randrange(bucket.begin, bucket.end) #print "refreshing bucket %x-%x by finding %x" \ # % (bucket.begin, bucket.end, refreshID) self.logger.info("refreshing bucket %x-%x by finding %x" % (bucket.begin, bucket.end, refreshID)) deferred = self.client.kFindNode(refreshID) dlist.append(deferred) dl = defer.DeferredList(dlist) dl.addCallback(refreshDone) # XXX: do we need to ping newly discovered known nodes? If not, # we could be vulnerable to a poisoning attack (at first # glance, this attack seems rather impotent...) # XXX: need to call refresh about every 60 minutes. Add a # reactor.callLater here to do it. def badGW(error): self.logger.warn(error) self.logger.warn("Couldn't connect to gateway at %s:%s" % (sys.argv[1], sys.argv[2])) self.logger.debug("connectViaGateway %s%d" % (host, port)) deferred = self.client.sendkFindNode(host, port, self.config.routing.node[2]) deferred.addCallback(refresh) deferred.addErrback(badGW) def getPath(): # this is a hack to be able to get the location of FludNode.tac return os.path.dirname(os.path.abspath(__file__))
Python
""" CheckboxState.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. CheckboxState represents the states which a checkbox in DirCtrl can take """ class CheckboxState: (UNSELECTED, SELECTED, SELECTEDCHILD, SELECTEDPARENT, EXCLUDED, EXCLUDEDCHILD) = range(6) def offset(oldstate, newstate): return newstate - oldstate offset = staticmethod(offset)
Python
class FludException(Exception): pass
Python
#!/usr/bin/python """ FludTestGauges.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Provides gauges for visualizing storage for multiple flud nodes running on the same host. This is really only useful for demos and testing. """ import sys, os, signal, stat, random import wx import wx.lib.buttons as buttons from flud.FludConfig import FludConfig dutotal = 0 def visit(arg, top, files): global dutotal for file in files: dutotal += os.lstat("%s" % (os.path.join(top,file)))[stat.ST_SIZE] arg += dutotal def du(dir): global dutotal dutotal = 0 os.path.walk(dir, visit, dutotal) return dutotal # XXX: too much manual layout. should convert to a managed layout to allow for # resizing, etc. SGAUGEWIDTH = 230 # storage gauge DGAUGEWIDTH = 100 # dht gauge GAUGEHEIGHT = 20 ROWHEIGHT = 30 SEP = 5 LABELWIDTH = 20 POWERWIDTH = 70 RATIOBARHEIGHT = 70 COLWIDTH = SGAUGEWIDTH+DGAUGEWIDTH+LABELWIDTH+POWERWIDTH COLGAPFUDGE = 30 class FludTestGauges(wx.Frame): def __init__(self, parent, title, dirroot, dirs): screenHeight = wx.SystemSettings.GetMetric(wx.SYS_SCREEN_Y)-100 rowheight = ROWHEIGHT+SEP height = len(dirs)*(rowheight)+RATIOBARHEIGHT columns = height / screenHeight + 1 width = COLWIDTH*columns if columns > 1: height = (len(dirs)/columns)*(rowheight)+RATIOBARHEIGHT if (len(dirs) % columns) > 0: height += rowheight wx.Frame.__init__(self, parent, wx.ID_ANY, title, size=(width,height), style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE) self.storebarend = 1024 self.smultiplier = 100.0 / self.storebarend self.sdivisor = 1 self.sbytelabel = "" self.dhtbarend = 512 self.dmultiplier = 100.0 / self.dhtbarend self.ddivisor = 1 self.dbytelabel = "" self.storeheading = wx.StaticText(self, -1, "block storage", (LABELWIDTH, 5)) self.totaldht = wx.StaticText(self, -1, "metadata", (LABELWIDTH+SGAUGEWIDTH+SEP, 5)) self.gauges = [] curCol = 0 curRow = 30 for i in range(len(dirs)): self.gauges.append(wx.Gauge(self, -1, 100, (curCol*COLWIDTH+LABELWIDTH, curRow), (SGAUGEWIDTH, GAUGEHEIGHT))) self.gauges[i].SetBezelFace(3) self.gauges[i].SetShadowWidth(3) self.gauges[i].SetValue(0) self.gauges[i].dir = "%s%s" % (dirroot,dirs[i]) os.environ['FLUDHOME'] = self.gauges[i].dir; conf = FludConfig() conf.load(doLogging = False) print "%s" % conf.nodeID self.gauges[i].label = wx.StaticText(self, -1, "%2s" % dirs[i], (curCol*COLWIDTH, curRow+(rowheight/4)), size=(LABELWIDTH, -1)) self.gauges[i].idlabel = wx.StaticText(self, -1, "%s" % conf.nodeID, (curCol*COLWIDTH+LABELWIDTH, curRow+20)) font = self.gauges[i].idlabel.GetFont() font.SetPointSize(6) self.gauges[i].idlabel.SetFont(font) self.gauges[i].dhtgauge = wx.Gauge(self, -1, 100, (curCol*COLWIDTH+LABELWIDTH+SGAUGEWIDTH+SEP, curRow), (SGAUGEWIDTH/3, GAUGEHEIGHT)) self.gauges[i].power = wx.Button(self, i, "turn OFF", (curCol*COLWIDTH +LABELWIDTH+SGAUGEWIDTH+2*SEP+SGAUGEWIDTH/3, curRow), (POWERWIDTH, ROWHEIGHT)) #self.gauges[i].power = buttons.GenBitmapToggleButton(self, i, # None, # (LABELWIDTH+SGAUGEWIDTH+2*SEP+SGAUGEWIDTH/3, curRow), # (POWERWIDTH, ROWHEIGHT)) #self.gauges[i].button.SetBestSize() self.gauges[i].power.SetToolTipString("power on/off") self.Bind(wx.EVT_BUTTON, self.onClick, self.gauges[i].power) curRow += rowheight if curRow > height-RATIOBARHEIGHT: curCol += 1 curRow = 30 self.totalstore = wx.StaticText(self, -1, "total: 0", (LABELWIDTH, height-40)) self.totaldht = wx.StaticText(self, -1, "total: 0", (LABELWIDTH+SGAUGEWIDTH+SEP, height-40)) self.ratiogauge = wx.Gauge(self, -1, 100, (LABELWIDTH, height-20), (SGAUGEWIDTH+SEP+SGAUGEWIDTH/3, 10)) self.ratiogauge.SetValue(0) self.Bind(wx.EVT_IDLE, self.IdleHandler) self.timer = wx.PyTimer(self.update) self.timer.Start(1000) def onClick(self, event): # XXX: note that under our current startNnodes.sh scheme, the first # node spawned doesn't contact anyone, so if that one is powered off # and then powered back on, it will not be part of the node until # another node pings it # XXX: unix-specific proc management stuff follows idx = event.GetId() home = self.gauges[idx].dir pidfile = os.path.join(home, 'twistd.pid') if os.path.exists(pidfile): print "shutting down %s" % home f = open(pidfile) pid = int(f.read()) f.close() # XXX: ps command no worky on windows, and "-ww" may not worker on # oldskool unixes self.gauges[idx].savedCmd = os.popen( "ps f -wwp %d -o args=" % pid).read() procline = os.popen("ps e -wwp %d" % pid).read() self.gauges[idx].savedEnv = [e for e in procline.split() if e[:4] == 'FLUD'] # XXX: os.kill no worky on windows, need something like: #def windowskill(pid): # import win32api # handle = win32api.OpenProcess(1, 0, pid) # return (0 != win32api.TerminateProcess(handle, 0)) os.kill(pid, signal.SIGTERM) self.gauges[idx].power.SetLabel("turn ON") self.gauges[idx].Hide() self.gauges[idx].dhtgauge.Hide() else: print "powering up %s" % home # XXX: this exec no worky on windows fullcmd = "%s %s" % (' '.join(self.gauges[idx].savedEnv), self.gauges[idx].savedCmd) print fullcmd result = os.popen('%s %s' % (' '.join(self.gauges[idx].savedEnv), self.gauges[idx].savedCmd)).readlines() self.gauges[idx].power.SetLabel("turn OFF") self.gauges[idx].Show() self.gauges[idx].dhtgauge.Show() print result def update(self): def sizeclass(num): divisor = 1 bytelabel = "" if num > 1024: divisor = 1024.0 bytelabel = 'K' if num > 1048576: divisor = 1048576.0 bytelabel = 'M' if num > 1073741824: divisor = 1073741824.0 bytelabel = 'G' return (divisor, bytelabel) storelargest = 0 dhtlargest = 0 storetotal = 0 dhttotal = 0 for i in self.gauges: if os.path.isdir(i.dir): i.storebytes = du(os.path.join(i.dir,'store')) if i.storebytes > storelargest: storelargest = i.storebytes storetotal += i.storebytes i.dhtbytes = du(os.path.join(i.dir,'dht')) if i.dhtbytes > dhtlargest: dhtlargest = i.dhtbytes dhttotal += i.dhtbytes else: i.storebytes = 0 i.dhtbytes = 0 i.Disable() i.power.Disable() while storelargest > self.storebarend: self.storebarend = self.storebarend * 2 self.smultiplier = 100.0 / self.storebarend self.sdivisor, self.sbytelabel = sizeclass(storetotal) while dhtlargest > self.dhtbarend: self.dhtbarend = self.dhtbarend * 2 self.dmultiplier = 100.0 / self.dhtbarend self.ddivisor, self.dbytelabel = sizeclass(dhttotal) #print "-----" for i in self.gauges: i.SetValue(i.storebytes*self.smultiplier) i.dhtgauge.SetValue(i.dhtbytes*self.dmultiplier) #print "%.2f, %.2f" % ((float(i.storebytes)/float(i.dhtbytes)), # (float(i.GetValue())/float(i.dhtgauge.GetValue()))) self.totalstore.SetLabel("total: %.1f%s" % (float(storetotal)/self.sdivisor, self.sbytelabel)) self.totaldht.SetLabel("total: %.1f%s" % (float(dhttotal)/self.ddivisor, self.dbytelabel)) if (dhttotal+storetotal == 0): self.ratiogauge.SetValue(0) else: self.ratiogauge.SetValue((storetotal*100/(dhttotal+storetotal))) def updateGauges(self, update): for index, value in update: self.monitors[index].setValue(value) def IdleHandler(self, event): pass def main(): if len(sys.argv) < 2: print "usage: %s dircommon exts" % sys.argv[0] print " where exts will be appended to dircommon" print " e.g., '%s /home/joe/.flud 1,2,3,4,10,15,20'"\ % sys.argv[0] print " or, '%s /home/joe/.flud 1-10,15,20'"\ % sys.argv[0] sys.exit() root = sys.argv[1] exts = [] dirs = [d.strip() for d in sys.argv[2].split(',')] for i in dirs: if i == "_": exts.append('') # undocumented, means "just dircommon" elif i.find('-') >= 0: start, end = i.split('-') for j in range(int(start),int(end)+1): exts.append(j) else: exts.append(int(i)) app = wx.PySimpleApp() t = FludTestGauges(None, 'Flud Test Gauges', root, exts) t.Show(1) app.MainLoop() if __name__ == '__main__': main()
Python
import os, stat, sys, tarfile, tempfile import gzip from flud.FludCrypto import hashstream from flud.fencode import fencode """ TarfileUtils.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Provides additional tarfile functionality (deletion of a member from a tarball, and concatenation of tarballs). """ def delete(tarball, membernames): """ Deletes a member file[s] from a tarball. Returns the names of deleted members if they are removed, False if the file[s] aren't members. If membernames contains all the members in the tarball, the entire tarball is deleted """ gzipped = False if tarball[-7:] == ".tar.gz": gzipped = True f = tarfile.open(tarball, 'r:gz') else: f = tarfile.open(tarball, 'r') if not isinstance(membernames, list): membernames = [membernames] tarnames = f.getnames() for membername in membernames: if not membername in tarnames: membernames.remove(membername) if len(membernames) < 1: f.close() return False if len(tarnames) == len(membernames): f.close() os.remove(tarball) return True f.close() if gzipped: tarball = gunzipTarball(tarball) f = open(tarball, 'r+') tfile = tempfile.mktemp() if gzipped: f2 = gzip.GzipFile(tfile, 'w') else: f2 = open(tfile, 'w') empty = tarfile.BLOCKSIZE * '\0' done = False removednames = [] while not done: bytes = f.read(tarfile.BLOCKSIZE) if bytes == "": done = True elif bytes == empty: f2.write(bytes) else: name = bytes[0:99] name = name[:name.find(chr(0))] size = int(bytes[124:135], 8) blocks = size / tarfile.BLOCKSIZE if (size % tarfile.BLOCKSIZE) > 0: blocks += 1 if name in membernames: f.seek(blocks*tarfile.BLOCKSIZE + f.tell()) removednames.append(name) else: f2.write(bytes) for i in range(blocks): f2.write(f.read(tarfile.BLOCKSIZE)) f2.close() f.close() if gzipped: os.remove(tarball) tarball = tarball+".gz" os.rename(tfile, tarball) return removednames def concatenate(tarfile1, tarfile2): """ Combines tarfile1 and tarfile2 into tarfile1. tarfile1 is modified in the process, and tarfile2 is deleted. """ gzipped = False if tarfile1[-7:] == ".tar.gz": gzipped = True f1 = gzip.GzipFile(tarfile1, 'r') tarfile1 = tarfile1[:-3] f1unzip = file(tarfile1, 'w') f1unzip.write(f1.read()) f1unzip.close() f1.close() os.remove(tarfile1+".gz") f = open(tarfile1, "r+") done = False e = '\0' empty = tarfile.BLOCKSIZE*e emptyblockcount = 0 while not done: header = f.read(tarfile.BLOCKSIZE) if header == "": print "error: end of archive not found" return elif header == empty: emptyblockcount += 1 if emptyblockcount == 2: done = True else: emptyblockcount = 0 fsize = eval(header[124:135]) skip = int(round(float(fsize) / float(tarfile.BLOCKSIZE) + 0.5)) f.seek(skip*tarfile.BLOCKSIZE, 1) # truncate the file to the spot before the end-of-tar marker trueend = f.tell() - (tarfile.BLOCKSIZE*2) f.seek(trueend) f.truncate() # now write the contents of the second tarfile into this spot if tarfile2[-7:] == ".tar.gz": f2 = gzip.GzipFile(tarfile2, 'r') else: f2 = open(tarfile2, "r") done = False while not done: header = f2.read(tarfile.BLOCKSIZE) if header == "": print "error: end of archive not found" f.seek(trueend) f.write(empty*2) return else: f.write(header) if header == empty: emptyblockcount += 1 if emptyblockcount == 2: done = True else: emptyblockcount = 0 fsize = eval(header[124:135]) bsize = int(round(float(fsize) / float(tarfile.BLOCKSIZE) + 0.5)) # XXX: break this up if large data = f2.read(bsize*tarfile.BLOCKSIZE) f.write(data) f2.close() f.close() if gzipped: f2 = gzip.GzipFile(tarfile1+".gz", 'wb') f = file(tarfile1, 'rb') f2.write(f.read()) f2.close() f.close() os.remove(tarfile1) # and delete the second tarfile os.remove(tarfile2) #print "concatenated %s to %s" % (tarfile2, tarfile1) def verifyHashes(tarball, ignoreExt=None): # return all the names of files in this tarball if hash checksum passes, # otherwise return False digests = [] done = False if tarball[-7:] == ".tar.gz": f = gzip.GzipFile(tarball, 'r:gz') else: f = open(tarball, 'r') empty = tarfile.BLOCKSIZE * '\0' while not done: bytes = f.read(tarfile.BLOCKSIZE) if bytes == "": done = True elif bytes == empty: pass else: if bytes[0] == '\0' and bytes[124] == '\0': print "WARNING: read nulls when expecting file header" break name = bytes[0:99] name = name[:name.find(chr(0))] size = int(bytes[124:135], 8) blocks = size / tarfile.BLOCKSIZE if ignoreExt and name[-len(ignoreExt):] == ignoreExt: # gzip doesn't support f.seek(size, 1) f.seek(f.tell()+size) else: digest = hashstream(f, size) digest = fencode(int(digest,16)) if name == digest: #print "%s == %s" % (name, digest) digests.append(name) else: #print "%s != %s" % (name, digest) f.close() return [] if (size % tarfile.BLOCKSIZE) > 0: blocks += 1 f.seek((blocks * tarfile.BLOCKSIZE) - size + f.tell()) f.close() return digests def gzipTarball(tarball): if tarball[-4:] != '.tar': return None f = gzip.GzipFile(tarball+".gz", 'wb') f.write(file(tarball, 'rb').read()) f.close() os.remove(tarball) return tarball+".gz" def gunzipTarball(tarball): if tarball[-3:] != '.gz': return None f = gzip.GzipFile(tarball, 'rb') file(tarball[:-3], 'wb').write(f.read()) f.close() os.remove(tarball) return tarball[:-3] if __name__ == "__main__": if (len(sys.argv) < 4 or sys.argv[1] != "-d") \ and (len(sys.argv) != 4 or sys.argv[1] != "-c") \ and sys.argv[1] != "-v": print "usage: [-d tarfile tarfilemembers]\n"\ +" [-c tarfile1 tarfile2]\n"\ +" [-v tarfile]\n"\ +" -d deletes tarfilemembers from tarfile,\n"\ +" -c concatenates tarfile1 and tarfile2 into tarfile1\n"\ +" -v verifies that the names of files in tarfile are sha256\n" sys.exit(-1) if sys.argv[1] == "-d": deleted = delete(sys.argv[2], sys.argv[3:]) if deleted == sys.argv[3:]: print "%s successfully deleted from %s" % (deleted, sys.argv[2]) else: faileddeletes = [x for x in sys.argv[3:] if x not in deleted] print "could not delete %s from %s" % (faileddeletes, sys.argv[2]) elif sys.argv[1] == "-c": concatenate(sys.argv[2], sys.argv[3]) print "concatenated %s and %s into %s" % (sys.argv[2], sys.argv[3], sys.argv[2]) elif sys.argv[1] == "-v": digests = verifyHashes(sys.argv[2]) if digests: print "verified tarfile member digests for: %s" % digests else: print "some tarfile members failed digest check"
Python
#!/usr/bin/python """ FludNode.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), verison 3. FludNode is the process that runs to talk with other nodes in the flud backup network. """ from twisted.internet import reactor, defer import threading, signal, sys, time, os, random, logging from flud.FludConfig import FludConfig from flud.protocol.FludServer import FludServer from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import getCanonicalIP PINGTIME=60 SYNCTIME=900 class FludNode(object): """ A node in the flud network. A node is both a client and a server. It listens on a network accessible port for both DHT and Storage layer requests, and it listens on a local port for client interface commands. """ def __init__(self, port=None): self._initLogger() self.config = FludConfig() self.logger.removeHandler(self.screenhandler) self.config.load(serverport=port) self.client = FludClient(self) self.DHTtstamp = time.time()+10 def _initLogger(self): logger = logging.getLogger('flud') self.screenhandler = logging.StreamHandler() self.screenhandler.setLevel(logging.INFO) logger.addHandler(self.screenhandler) self.logger = logger def pingRandom(self, tstamp): return # XXX: see pg. 4, Section 2.2 (short) or 2.3 (long) of the Kademlia # paper -- once an hour, each node should check any buckets that # haven't been refreshed and pick a random id within that space # to findnode(id) on, for all buckets. if tstamp < self.DHTtstamp: #r = random.randrange(2**256) n = self.config.routing.knownExternalNodes() if len(n) > 2: n1 = random.choice(n)[2] n2 = random.choice(n)[2] r = (n1+n2)/2 else: r = random.randrange(2**256) def badNode(error): node.logger.warn("Couldn't ping %s:%s" % (sys.argv[1], sys.argv[2])) d = self.client.kFindNode(r) d.addErrback(badNode) pingtime = random.randrange(PINGTIME/2, PINGTIME) reactor.callLater(pingtime, self.pingRandom, time.time()) def syncConfig(self): self.config.save() reactor.callLater(SYNCTIME, self.syncConfig) def start(self, twistd=False): """ starts the reactor in this thread """ self.webserver = FludServer(self, self.config.port) self.logger.log(logging.INFO, "FludServer starting") reactor.callLater(1, self.pingRandom, time.time()) reactor.callLater(random.randrange(10), self.syncConfig) if not twistd: reactor.run() def run(self): """ starts the reactor in its own thread """ #signal.signal(signal.SIGINT, self.sighandler) signal.signal(signal.SIGINT, signal.SIG_DFL) signal.signal(signal.SIGTERM, signal.SIG_DFL) self.webserver = FludServer(self, self.config.port) self.webserver.start() # XXX: need to do save out current config every X seconds # XXX: need to seperate known_nodes from config, and then update this # every X seconds. only update config when it changes. def stop(self): self.logger.log(logging.INFO, "shutting down FludNode") self.webserver.stop() def join(self): self.webserver.join() def sighandler(self, sig, frame): self.logger.log(logging.INFO, "handling signal %s" % sig) def connectViaGateway(self, host, port): def refresh(knodes): def refreshDone(results): self.logger.info("bucket refreshes finished: %s" % results) print "flud node connected and listening on port %d"\ % self.config.port #print "found knodes %s" % knodes dlist = [] for bucket in self.config.routing.kBuckets: #if True: if bucket.begin <= self.config.routing.node[2] < bucket.end: pass #print "passed on bucket %x-%s" % (bucket.begin, bucket.end) else: refreshID = random.randrange(bucket.begin, bucket.end) #print "refreshing bucket %x-%x by finding %x" \ # % (bucket.begin, bucket.end, refreshID) self.logger.info("refreshing bucket %x-%x by finding %x" % (bucket.begin, bucket.end, refreshID)) deferred = self.client.kFindNode(refreshID) dlist.append(deferred) dl = defer.DeferredList(dlist) dl.addCallback(refreshDone) # XXX: do we need to ping newly discovered known nodes? If not, # we could be vulnerable to a poisoning attack (at first # glance, this attack seems rather impotent...) # XXX: need to call refresh about every 60 minutes. Add a # reactor.callLater here to do it. def badGW(error): self.logger.warn(error) self.logger.warn("Couldn't connect to gateway at %s:%s" % (sys.argv[1], sys.argv[2])) self.logger.debug("connectViaGateway %s%d" % (host, port)) deferred = self.client.sendkFindNode(host, port, self.config.routing.node[2]) deferred.addCallback(refresh) deferred.addErrback(badGW) def getPath(): # this is a hack to be able to get the location of FludNode.tac return os.path.dirname(os.path.abspath(__file__))
Python
#!/usr/bin/python """ FludScheduler.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), verison 3. FludScheduler is the process monitors files for changes, and then tells flud to back them up. """ import sys, os, time, stat from twisted.internet import reactor from flud.FludConfig import FludConfig from flud.protocol.LocalClient import * from flud.CheckboxState import CheckboxState CHECKTIME=5 class FludScheduler: def __init__(self, config, factory): self.config = config self.factory = factory self.fileconfigfile = None self.fileconfigfileMTime = 0 self.fileChangeTime = 0 self.fileconfigSelected = set() self.fileconfigExcluded = set() self.getMasterMetadata() def getMasterMetadata(self): d = self.factory.sendLIST() d.addCallback(self.gotMasterMetadata) d.addErrback(self.errMasterMetadata) return d def gotMasterMetadata(self, master): self.mastermetadata = master def errMasterMetadata(self, err): print err reactor.stop() def readFileConfig(self, mtime=None): print "reading FileConfig" file = open(self.fileconfigfile, 'r') self.fileconfig = eval(file.read()) file.close() if mtime: self.fileconfigfileMTime = mtime else: self.fileconfigfileMTime = os.stat( self.fileconfigfile)[stat.ST_MTIME] self.fileconfigSelected = set([f for f in self.fileconfig if self.fileconfig[f] == CheckboxState.SELECTED or self.fileconfig[f] == CheckboxState.SELECTEDCHILD]) self.fileconfigExcluded = set([f for f in self.fileconfig if self.fileconfig[f] == CheckboxState.EXCLUDED or self.fileconfig[f] == CheckboxState.EXCLUDEDCHILD]) # The file[s]ChangeStat are the worst possible way to detect file changes. # Much more efficient to use inotify/dnotify/fam/gamin/etc., as well as # more correct (no way to detect cp -a or -p, for example, with stat). # But, these are a fallback method when those aren't present, and are fine # for testing. def fileChangedStat(self, file, fileChangeTime=None): if os.path.isfile(file) or os.path.isdir(file): mtime = os.stat(file)[stat.ST_MTIME] if not fileChangeTime: fileChangeTime = self.fileChangeTime if file in self.mastermetadata: fileChangeTime = self.mastermetadata[file][1] else: return True print "mtime = %s, ctime = %s (%s)" % (mtime, fileChangeTime, file) if mtime > fileChangeTime: return True return False def filesChangedStat(self, files, fileChangeTime=None): result = [] for f in files: if self.fileChangedStat(f, fileChangeTime): result.append(f) return result # Change these to point to something other than the xxxStat() methods def fileChanged(self, file, fileChangeTime=None): """ >>> now = time.time() >>> f1 = tmpfile.mktemp() >>> """ return self.fileChangedStat(file, fileChangeTime) def filesChanged(self, files, fileChangeTime=None): return self.filesChangedStat(files, fileChangeTime) def checkFileConfig(self): # check config file to see if it has changed, then reparse it if not self.fileconfigfile: # first time through print "checking fileconfigfile (initial)" if os.environ.has_key('FLUDHOME'): fludhome = os.environ['FLUDHOME'] elif os.environ.has_key('HOME'): fludhome = os.environ['HOME']+"/.flud" else: fludhome = ".flud" # XXX: fludfile.conf should be in config self.fileconfigfile = os.path.join(fludhome, "fludfile.conf") if os.path.isfile(self.fileconfigfile): self.readFileConfig() return True else: print "no fileconfigfile to read" elif os.path.isfile(self.fileconfigfile): if self.fileChanged(self.fileconfigfile, self.fileconfigfileMTime): print "fileconfigfile changed" mtime = time.time() self.readFileConfig(mtime) return True return False def checkFilesystem(self): checkedFiles = set() changedFiles = set() def checkList(list): #print "checkList: %s" % list #print "checkedFiles: %s" % checkedFiles for entry in list: # XXX: if entry is in master metadata, and its mtime is not # earlier than the time used by fileChanged, skip it (add 'and' # clause) if entry not in checkedFiles and \ entry not in self.fileconfigExcluded and \ entry not in self.mastermetadata: print "checkFilesystem for %s" % entry if os.path.isdir(entry): #print "dir %s" % entry dirfiles = [os.path.join(entry, i) for i in os.listdir(entry)] checkedFiles.update([entry,]) checkList(dirfiles) elif self.fileChanged(entry): print "%s changed" % entry if os.path.isfile(entry): changedFiles.update([entry,]) #print "file %s changed" % entry else: print "entry ?? %s ?? changed" % entry checkedFiles.update([entry,]) checkList(self.fileconfigSelected) self.fileChangeTime = time.time() return changedFiles def storefileFailed(self, err, file): print "storing %s failed: %s" % (file, err) err.printTraceback() #print dir(err) def storefileYay(self, r, file): print "storing %s success" % file def storeFiles(self, changedFiles): #print "storing %s" % changedFiles dlist = [] for f in changedFiles: print "storing %s" % f deferred = self.factory.sendPUTF(f) deferred.addCallback(self.storefileYay, f) deferred.addErrback(self.storefileFailed, f) dlist.append(deferred) dl = defer.DeferredList(dlist) return dl #return defer.succeed(True) def restartCheckTimer(self, v): print "restarting timer (%d) to call run()" % CHECKTIME reactor.callLater(CHECKTIME, self.run) def updateMasterMetadata(self, v): return self.getMasterMetadata() def run(self): print "run" self.checkFileConfig() changedFiles = self.checkFilesystem() print "%s changed" % changedFiles d = self.storeFiles(changedFiles) d.addBoth(self.updateMasterMetadata) d.addBoth(self.restartCheckTimer)
Python
#!/usr/bin/python """ FludLocalClient.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. FludLocalClient provides a command-line client for interacting with FludNode. """ import sys, os, time from twisted.internet import reactor from flud.FludConfig import FludConfig from flud.fencode import fencode, fdecode from flud.FludCrypto import hashfile from protocol.LocalClient import * logger = logging.getLogger('flud') class CmdClientFactory(LocalClientFactory): def __init__(self, config): LocalClientFactory.__init__(self, config) self.quit = False self.msgs = [] def callFactory(self, func, commands, msgs): # since we can't call factory methods from the promptUser thread, we # use this as a convenience to put those calls back in the event loop reactor.callFromThread(self.doFactoryMethod, func, commands, msgs) def doFactoryMethod(self, func, commands, msgs): d = func() d.addCallback(self.queueResult, msgs, '%s succeeded' % commands) d.addErrback(self.queueError, msgs, '%s failed' % commands) return d def promptUser(self): helpDict = {} command = raw_input("%s> " % time.ctime()) commands = command.split(' ') # XXX: should tokenize on any whitespace commandkey = commands[0][:4] # core client operations helpDict['exit'] = "exit from the client" helpDict['help'] = "display this help message" helpDict['ping'] = "send a GETID() message: 'ping host port'" helpDict['putf'] = "store a file: 'putf canonicalfilepath'" helpDict['getf'] = "retrieve a file: 'getf canonicalfilepath'" helpDict['geti'] = "retrieve a file by CAS key: 'geti fencodedCASkey'" helpDict['fndn'] = "send a FINDNODE() message: 'fndn hexIDstring'" helpDict['list'] = "list stored files (read from local metadata)" helpDict['putm'] = "store master metadata" helpDict['getm'] = "retrieve master metadata" helpDict['cred'] = "send encrypted private credentials: cred"\ " passphrase emailaddress" helpDict['node'] = "list known nodes" helpDict['buck'] = "print k buckets" helpDict['stat'] = "show pending actions" helpDict['stor'] = "store a block to a given node:"\ " 'stor host:port,fname'" helpDict['rtrv'] = "retrieve a block from a given node:"\ " 'rtrv host:port,fname'" helpDict['vrfy'] = "verify a block on a given node:"\ " 'vrfy host:port:offset-length,fname'" helpDict['fndv'] = "retrieve a value from the DHT: 'fndv hexkey'" helpDict['dlet'] = "delete from the stor: '[XXX]'" if commandkey == 'exit' or commandkey == 'quit': self.quit = True elif commandkey == 'help': self.printHelp(helpDict) elif commandkey == 'ping': # ping a host # format: 'ping host port' func = lambda: self.sendPING(commands[1], commands[2]) self.callFactory(func, commands, self.msgs) elif commandkey == 'putf': # store a file # format: 'putf canonicalfilepath' func = lambda: self.sendPUTF(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'getf': # retrieve a file # format: 'getf canonicalfilepath' func = lambda: self.sendGETF(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'geti': # retrieve a file by CAS ID # format: 'geti fencoded_CAS_ID' func = lambda: self.sendGETI(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'fndn': # find a node (or the k-closest nodes) # format: 'fndn hexIDstring' func = lambda: self.sendFNDN(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'list': # list stored files self.callFactory(self.sendLIST, commands, self.msgs) elif commandkey == 'putm': # store master metadata self.callFactory(self.sendPUTM, commands, self.msgs) elif commandkey == 'getm': # retrieve master metadata self.callFactory(self.sendGETM, commands, self.msgs) elif commandkey == 'cred': # send encrypted private credentials to an email address # format: 'cred passphrase emailaddress' func = lambda: self.sendCRED( command[len(commands[0])+1:-len(commands[-1])-1], commands[-1]) self.callFactory(func, commands, self.msgs) # the following are diagnostic operations, debug-only utility elif commandkey == 'node': # list known nodes self.callFactory(self.sendDIAGNODE, commands, self.msgs) elif commandkey == 'buck': # show k-buckets self.callFactory(self.sendDIAGBKTS, commands, self.msgs) elif commandkey == 'stat': # show pending actions print self.pending elif commandkey == 'stor': # stor a block to a given node. format: 'stor host:port,fname' storcommands = commands[1].split(',') try: fileid = int(storcommands[1], 16) except: linkfile = fencode(long(hashfile(storcommands[1]),16)) if (os.path.islink(linkfile)): os.remove(linkfile) os.symlink(storcommands[1], linkfile) storcommands[1] = linkfile # XXX: delete this file when the command finishes commands[1] = "%s,%s" % (storcommands[0], storcommands[1]) func = lambda: self.sendDIAGSTOR(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'rtrv': # retrive a block from a given node. format: 'rtrv host:port,fname' func = lambda: self.sendDIAGRTRV(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'vrfy': # verify a block on a given node. # format: 'vrfy host:port:offset-length,fname' logger.debug("vrfy(%s)" % commands[1]) func = lambda: self.sendDIAGVRFY(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'dlet': print "not yet implemented" elif commandkey == 'fndv': # try to retrieve a value from the DHT # format: 'fndv key' func = lambda: self.sendDIAGFNDV(commands[1]) self.callFactory(func, commands, self.msgs) elif command != "": reactor.callFromThread(self.queueError, None, self.msgs, "illegal command '%s'" % command) def queueResult(self, r, l, msg): logger.debug("got result %s" % msg) l.append((r, msg)) def queueError(self, r, l, msg): logger.debug("got error %s" % msg) if r: l.append((r.getErrorMessage(), msg)) else: l.append((None, msg)) def printHelp(self, helpDict): helpkeys = helpDict.keys() helpkeys.sort() for i in helpkeys: print "%s:\t %s" % (i, helpDict[i]) def promptLoop(self, r): for c in self.pending: for i in self.pending[c].keys(): if self.pending[c][i] == True: print "%s on %s completed successfully" % (c, i) self.pending[c].pop(i) elif self.pending[c][i] == False: print "%s on %s failed" % (c, i) self.pending[c].pop(i) else: print "%s on %s pending" % (c, i) while len(self.msgs) > 0: # this prints in reverse order, perhaps pop() all into a new list, # reverse, then print (errmsg, m) = self.msgs.pop() if errmsg: print "<- %s:\n%s" % (m, errmsg) else: print "<- %s" % m if self.quit: reactor.stop() else: d = threads.deferToThread(self.promptUser) d.addCallback(self.promptLoopDelayed) d.addErrback(self.err) def promptLoopDelayed(self, r): # give the reactor loop time to fire any quick cbs/ebs reactor.callLater(0.1, self.promptLoop, r) def clientConnectionLost(self, connector, reason): if not self.quit: LocalClientFactory.clientConnectionLost(self, connector, reason) def cleanup(self, msg): self.quit = True self.err(msg) def err(self, r): print "bah!: %s" % r reactor.stop() def main(): config = FludConfig() config.load(doLogging=False) logger.setLevel(logging.DEBUG) handler = logging.FileHandler('/tmp/fludclient.log') formatter = logging.Formatter('%(asctime)s %(filename)s:%(lineno)d' ' %(name)s %(levelname)s: %(message)s', datefmt='%H:%M:%S') handler.setFormatter(formatter) logger.addHandler(handler) factory = CmdClientFactory(config) if len(sys.argv) == 2: config.clientport = int(sys.argv[1]) print "connecting to localhost:%d" % config.clientport reactor.connectTCP('localhost', config.clientport, factory) factory.promptLoop(None) reactor.run() if __name__ == '__main__': main()
Python
#!/usr/bin/python """ FludClient.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. FludClient provides a GUI Client for interacting with FludNode. """ #from twisted.internet import wxreactor #wxreactor.install() import sys, os, string, time, glob import wx import wx.lib.mixins.listctrl as listmix import wx.lib.editor.editor from flud.protocol.LocalClient import * from flud.FludConfig import FludConfig from flud.CheckboxState import CheckboxState FLUSHCHECKTIME = 5*60 # s to wait to flush fludfile.conf datadir = os.path.dirname(os.path.abspath(__file__)) imgdir = os.path.join(datadir,'images') mimeMgr = wx.MimeTypesManager() def getFileIcon(file, il, checkboxes, icondict): ft = mimeMgr.GetFileTypeFromExtension(file[file.rfind('.')+1:]) # XXX: what about from mimetype or magic? if ft == None: return icondict['generic'] else: desc = ft.GetDescription() if icondict.has_key(desc): return icondict[desc] else: icon = ft.GetIcon() if icon == None or not icon.Ok(): #print "couldn't find an icon image for %s" % file icondict[desc] = icondict['generic'] return icondict[desc] bm = wx.BitmapFromIcon(icon) newimages = makeCheckboxBitmaps(bm, checkboxes) #il = self.GetImageList() pos = il.GetImageCount() for i in newimages: il.Add(i) icondict[desc] = pos #print "%s got a %s image" % (file, ft.GetDescription()) return pos def getEmptyBitmapAndDC(width, height): empty = wx.EmptyBitmap(width,height) temp_dc = wx.MemoryDC() temp_dc.SelectObject(empty) temp_dc.Clear() return (empty, temp_dc) def makeCheckboxBitmaps(basebitmap, checkboxes): if basebitmap.GetWidth() != 16 or basebitmap.GetHeight() != 16: img = basebitmap.ConvertToImage() img.Rescale(16, 16) basebitmap = img.ConvertToBitmap() result = [] for i in checkboxes: bm, dc = getEmptyBitmapAndDC(40,16) dc.DrawBitmap(basebitmap, 0, 0, False) dc.DrawBitmap(i, 20, 2, False) result.append(bm) return result def createDefaultImageList(): def getDefaultCheckboxes(): ucbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-unchecked1.png"))) cbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checked1.png"))) ccbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checkedpartial1.png"))) cpbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-parentchecked1.png"))) ebm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excluded1.png"))) ecbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excludedpartial1.png"))) return (ucbm, cbm, ccbm, cpbm, ebm, ecbm) checkboxes = getDefaultCheckboxes() il = wx.ImageList(40, 16) folderimgs = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_FOLDER, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) computer = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_HARDDISK, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) drives = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_HARDDISK, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) cdrom = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_CDROM, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) floppy = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_FLOPPY, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) removable = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_REMOVABLE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) genericfile = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_NORMAL_FILE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) execfile = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_EXECUTABLE_FILE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) j = 0 icondict = {} icondict['folder'] = j for i in folderimgs: il.Add(i) j = j+1 icondict['computer'] = j for i in computer: il.Add(i) j = j+1 icondict['drives'] = j for i in drives: il.Add(i) j = j+1 icondict['cdrom'] = j for i in cdrom: il.Add(i) j = j+1 icondict['floppy'] = j for i in floppy: il.Add(i) j = j+1 icondict['removable'] = j for i in removable: il.Add(i) j = j+1 icondict['generic'] = j for i in genericfile: il.Add(i) j = j+1 icondict['exec'] = j for i in execfile: il.Add(i) j = j+1 return il, checkboxes, icondict class DirCheckboxCtrl(wx.TreeCtrl): def __init__(self, parent, id=-1, dir=None, pos=wx.DefaultPosition, size=wx.Size(300,300), #wx.DefaultSize, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER), validator=wx.DefaultValidator, name=wx.ControlNameStr, allowExclude=True): self.allowExclude = allowExclude wx.TreeCtrl.__init__(self, parent, id, pos, size, style, validator, name) self.listeners = [] self.parent = parent #self.il = self.GetImageList() #self.checkboxes = self.getDefaultCheckboxes() self.initTree(dir) self.Bind(wx.EVT_TREE_ITEM_EXPANDING, self.onExpand, self) self.Bind(wx.EVT_LEFT_UP, self.onClick, self) self.Bind(wx.EVT_TREE_ITEM_GETTOOLTIP, self.onTooltip, self) self.Bind(wx.EVT_CHAR, self.onChar) def initTree(self, dir): self.expandRoot(dir) # XXX: should expandHome() on first run, then load expanded dir state # from saved state on subsequent runs. self.expandHome(dir) def expandRoot(self, dir): if not os.path.isdir(dir): raise ValueError("%s is not a valid directory path") self.defaultImageList, self.checkboxes, self.icondict \ = createDefaultImageList() self.AssignImageList(self.defaultImageList) self.il = self.GetImageList() if dir == None: self.rootID = self.AddRoot(dir, self.icondict['computer'], -1, wx.TreeItemData((dir, True, False, CheckboxState.UNSELECTED))) # XXX: getTopLevelDirs() and add them as children else: self.rootID = self.AddRoot(dir, self.icondict['folder'], -1, wx.TreeItemData((dir, True, False, CheckboxState.UNSELECTED))) self.expandDir(self.rootID) self.Expand(self.rootID) self.stateChangeTime = time.time() self.flushTime = time.time() reactor.callLater(FLUSHCHECKTIME, self.checkFlush) def expandHome(self, dir): home = os.environ['HOME'] if home: traversal = home.split(os.path.sep)[1:] node = self.rootID for d in traversal: (ipath, isdir, expanded, istate) \ = self.GetItemData(node).GetData() self.expandDir(node) children = self.getChildren(node, False) childrennames = [self.GetItemText(x) for x in children] if d in childrennames: p = childrennames.index(d) node = children[p] self.expandDir(node) self.Expand(node) else: print "couldn't traverse to HOME dir on %s" % d break def checkFlush(self): print "checking for flush" if self.stateChangeTime > self.flushTime: self.flushTime = time.time() print "flushing" self.parent.flushFileConfig() reactor.callLater(FLUSHCHECKTIME, self.checkFlush) def expandDir(self, parentID, hideHidden=False, busycursor=True): def isDriveAvailable(path): if len(path) == 2 and path[1] == ':': path = path.lower() if path[0] == 'a' or path[0] == 'b' or diExists(path): return True else: return False return True (path, isDir, expanded, state) = self.GetItemData(parentID).GetData() if expanded: return if not isDriveAvailable(path): return if busycursor: wx.BusyCursor() try: dirlist = os.listdir(path) except: self.SetItemHasChildren(parentID, False) return if len(dirlist) == 0: self.SetItemHasChildren(parentID, False) return dirs = [] files = [] for i in dirlist: if hideHidden and i[0] == '.': # XXX: dotfile format check is *nix specific # XXX: if this is a hidden file, don't add it. pass elif os.path.isdir(os.path.join(path,i)): dirs.append(i) else: files.append(i) dirs.sort() files.sort() for d in dirs: child = self.AppendItem(parentID, d) self.SetPyData(child, (os.path.join(path,d), True, False, 0)) self.SetItemImage(child, self.icondict['folder'], wx.TreeItemIcon_Normal) self.SetItemHasChildren(child) il = self.GetImageList() for f in files: child = self.AppendItem(parentID, f) # XXX: unicode? self.SetPyData(child, (os.path.join(path,f), False, False, 0)) idx = getFileIcon(os.path.join(path,f), il, self.checkboxes, self.icondict) self.SetItemImage(child, idx, wx.TreeItemIcon_Normal) self.SetPyData(parentID, (path, isDir, True, state)) def getStates(self, node=None): if not node: node = self.rootID states = {} (path, isDir, expanded, state) = self.GetItemData(node).GetData() if state in [CheckboxState.SELECTED, CheckboxState.EXCLUDED]: states[path] = state children = self.getChildren(node, False) for child in children: states.update(self.getStates(child)) return states def setStates(self, states): for i in states: found = self.findNode(i) if found: self.setItemState(found, states[i]) def findNode(self, path): if path[0] == '/': path = path[1:] # XXX: unix only traversal = path.split(os.path.sep) if traversal[0] == '': traversal.remove('') node = self.rootID while True: (ipath, isdir, expanded, istate) = self.GetItemData(node).GetData() if len(traversal) == 0: return node self.expandDir(node) children = self.getChildren(node, False) childrennames = [self.GetItemText(x) for x in children] firstpath = traversal[0] if firstpath in childrennames: p = childrennames.index(firstpath) node = children[p] traversal.remove(firstpath) else: #print " the file %s is no longer present!" % path return None return None def onExpand(self, event): self.expandDir(event.GetItem()) self.renderChildren(event.GetItem(), True) def getFullPath(self, node): path = self.tree.GetItemText(node) n = node while True: n = self.tree.GetItemParent(n) if n and n != self.GetRootItem(): path = os.path.join(self.tree.GetItemText(n),path) else: break return path def renderParents(self, item): if item == self.rootID: return n = item (path, isDir, expanded, state) = self.GetItemData(item).GetData() while True: n = self.GetItemParent(n) (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(n).GetData() #print "parent %s" % parentpath if n and n != self.GetRootItem(): newstate = parentstate if parentstate != CheckboxState.UNSELECTED and \ parentstate != CheckboxState.SELECTEDPARENT: # we only care about changing UNSELECT or SELECTEDPARENT # states break else: if state == CheckboxState.SELECTED or \ state == CheckboxState.SELECTEDCHILD or \ state == CheckboxState.SELECTEDPARENT: # if the item (child) is selected in any way, parent # should be too. newstate = CheckboxState.SELECTEDPARENT elif state == CheckboxState.UNSELECTED or \ state == CheckboxState.EXCLUDED: # if the item (child) is unselected or excluded, the # parent should be too, /unless/ there are other # children at the same level who are selected. children = self.getChildren(n, False) newstate = CheckboxState.UNSELECTED for child in children: (cpath, cisdir, cexp, cstate) = self.GetItemData(child).GetData() if cstate == CheckboxState.SELECTED or \ cstate == CheckboxState.SELECTEDCHILD or \ cstate == CheckboxState.SELECTEDPARENT: newstate = parentstate if newstate == parentstate: break imageidx = self.GetItemImage(n) imageidx += CheckboxState.offset(parentstate, newstate) self.SetPyData(n, (parentpath, parentisDir, parentexpanded, newstate)) self.SetItemImage(n, imageidx) else: break def renderChildren(self, parent, recurse=False): (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(parent).GetData() children = self.getChildren(parent, False) for child in children: #path = self.getFullPath(child) (path, isDir, expanded, state) = self.GetItemData(child).GetData() imageidx = self.GetItemImage(child) newstate = state """ Here are the state transitions for children based on current states: ('-' = no state change, 'x' = should never occur, '!' = should be prevented at the parent, '?' = need to consult children) child unsel sel selch selpar excl exclch unsel - ! unsel x - unsel sel selch - - selch - selch par selch selch - - selch - selch selpar x x unsl?selpr x x x excl exlch ! exlch ! - - exclch exlch - exlch ! - - """ #if parentpath == '/data': # print "/data pstate = %d" % parentstate # print " %s = %d" % (path, state) if state == CheckboxState.UNSELECTED: if parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.SELECTEDCHILD: if parentstate == CheckboxState.UNSELECTED: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTEDPARENT: if self.checkChildrenStates(child, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT]): # XXX: did we need to pass in selections to checkChldSt newstate = CheckboxState.SELECTEDPARENT else: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.SELECTEDPARENT: if parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif state == CheckboxState.EXCLUDEDCHILD: if parentstate == CheckboxState.UNSELECTED: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD imageidx += CheckboxState.offset(state, newstate) self.SetPyData(child, (path, isDir, expanded, newstate)) self.SetItemImage(child, imageidx) if recurse: self.renderChildren(child, recurse) # XXX: why do we renderParents here? It hits the same # 'parent's over and over and over again. If we want to do # this, we need to 'collect up' the parents and just call once # -- this kills performance. #print "renderParents(%s)" % path #self.renderParents(child) def getChildren(self, node, recurse=False): result = [] child, cookie = self.GetFirstChild(node) while child: result.append(child) if recurse: result.extend(self.getChildren(child, recurse)) child, cookie = self.GetNextChild(node, cookie) return result def checkChildrenStates(self, node, states, ignorelist=[]): children = self.getChildren(node) for child in children: if child not in ignorelist: (p, d, e, childstate) = self.GetItemData(child).GetData() for state in states: if state == childstate: #print "%s has state %d" % (p, state) return True if self.checkChildrenStates(child, states, ignorelist): # do this even if it is in ignorelist, because it may have # children which are not in the ignorelist return True return False def getTooltip(self, item): text = self.GetItemText(item) (path, isDir, expanded, state) = self.GetItemData(item).GetData() if state == CheckboxState.SELECTED: if isDir: text = "'%s' is SELECTED for backup\n" \ "ALL files within this folder will be backed up\n" \ "(except those explicitly marked for exclusion)" % text else: text = "'%s' is SELECTED for backup" % text elif state == CheckboxState.UNSELECTED: text = "'%s' is NOT selected for backup" % text elif state == CheckboxState.SELECTEDPARENT: text = "some files within '%s' are selected for backup" % text elif state == CheckboxState.SELECTEDCHILD: text = "'%s' will be backed up\n" \ "(one of its parent folders is selected)" % text elif state == CheckboxState.EXCLUDED: if isDir: text = "'%s' is EXCLUDED from backup\n" \ "No files within this folder will be backed up" % text else: text = "'%s' is EXCLUDED from backup" % text elif state == CheckboxState.EXCLUDEDCHILD: text = "'%s' is EXCLUDED from backup\n" \ "(one of its parent folders is EXCLUDED)" % text return text def onTooltip(self, event): item = event.GetItem() text = self.getTooltip(item) if text: event.SetToolTip(text) else: event.StopPropagation() #print dir(event) def onClick(self, event): point = (event.GetX(), event.GetY()) item, flags = self.HitTest(point) if flags & wx.TREE_HITTEST_ONITEMICON: selections = self.GetSelections() self.changeState(item, selections) def onChar(self, event): if event.KeyCode() == ord('F') and event.ShiftDown() \ and event.ControlDown(): self.flushTime = time.time() print "flushing" self.parent.flushFileConfig() event.Skip() def changeState(self, item, selections=[]): self.stateChangeTime = time.time() (path, isDir, expanded, state) = self.GetItemData(item).GetData() if item == self.rootID: parent = None parentstate = CheckboxState.UNSELECTED else: parent = self.GetItemParent(item) (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(parent).GetData() imageidx = self.GetItemImage(item) # determine newstate from existing state, parent state, and state # of children """ Here are the state transitions for the item based on current states and parent states: ('-' = no state change, 'x' = should never occur, '?' = depends on children state) item unsel sel selch selpar excl exclch unsel sel excl sel sel unsel excl sel sel excl?selpar sel x selch excl par selch x excl sel sel selch excl selpar sel excl x sel unsel excl excl x excl x exclch exclch excl exclch x excl x exclch exclch excl """ newstate = state if state == CheckboxState.UNSELECTED: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTEDPARENT: if parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: # XXX: this should be impossible to reach... newstate = CheckboxState.EXCLUDEDCHILD else: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTED: if self.checkChildrenStates(item, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT], selections): newstate = CheckboxState.SELECTEDPARENT elif self.allowExclude: newstate = CheckboxState.EXCLUDED else: if parent in selections or \ (parentstate == CheckboxState.UNSELECTED or \ parentstate == CheckboxState.SELECTEDPARENT): newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif state == CheckboxState.EXCLUDED: if parent in selections or \ (parentstate == CheckboxState.UNSELECTED or \ parentstate == CheckboxState.SELECTEDPARENT): newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD else: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDED if len(selections) > 1: # if we have multiple selections, the idea is to move all the # selections to the newstate defined above, or to valid # unselected or inherited states if the move to newstate would # be invalid. """ Here are the state transitions for the item based on the newstate as determined by the clicked item and the current states: ('-' = no state change, '?' = consult children) item unsel sel selch selpar excl exclch unsel - unsel - - unsel - sel sel - sel sel sel - newstate selch - unsel - - unsel - selpar - unsel - - unsel - excl excl excl?slpr excl excl - excl exclch - unsel - - unsel - """ for i in selections: (mpath, misDir, mexpanded, mstate) = self.GetItemData( i).GetData() mnewstate = mstate if mstate == CheckboxState.UNSELECTED or \ mstate == CheckboxState.SELECTEDCHILD or \ mstate == CheckboxState.SELECTEDPARENT: if newstate == CheckboxState.SELECTED or \ newstate == CheckboxState.EXCLUDED: mnewstate = newstate elif mstate == CheckboxState.SELECTED: if newstate == CheckboxState.UNSELECTED or \ newstate == CheckboxState.SELECTEDCHILD or \ newstate == CheckboxState.SELECTEDPARENT or \ newstate == CheckboxState.EXCLUDEDCHILD: mnewstate = CheckboxState.UNSELECTED elif newstate == CheckboxState.EXCLUDED: if self.checkChildrenStates(i, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT], selections): mnewstate = CheckboxState.SELECTEDPARENT else: mnewstate = newstate elif mstate == CheckboxState.EXCLUDED: if newstate == CheckboxState.UNSELECTED or \ newstate == CheckboxState.SELECTEDCHILD or \ newstate == CheckboxState.SELECTEDPARENT or \ newstate == CheckboxState.EXCLUDEDCHILD: mnewstate = CheckboxState.UNSELECTED elif newstate == CheckboxState.SELECTED: mnewstate = newstate elif mstate == CheckboxState.EXCLUDEDCHILD: if newstate == CheckboxState.EXCLUDED: mnewstate = newstate self.setItemState(i, mnewstate) self.setItemState(item, newstate, (path, isDir, expanded, state, imageidx)) def setItemState(self, item, newstate, oldData=None): if oldData: (path, isDir, expanded, state, imageidx) = oldData else: (path, isDir, expanded, state) = self.GetItemData(item).GetData() imageidx = self.GetItemImage(item) imageidx += CheckboxState.offset(state, newstate) self.SetPyData(item, (path, isDir, expanded, newstate)) self.SetItemImage(item, imageidx) self.renderChildren(item, True) self.renderParents(item) def getTopLevelDrives(self): sys = platform.system() if sys == 'Windows': # XXX: need to test this all out import win32api, string drives = win32api.GetLogicalDriveStrings() driveletters = string.splitfields(drives,'\000') for d in driveletters: type = win32api.GetDriveType("%s:\\" % d) # XXX: set the appropriate icon return driveletters else: # Unix, OSX, etc. return ['/'] def addListener(self, callback): self.listeners.append(callback) def SetPyData(self, item, data): wx.TreeCtrl.SetPyData(self, item, data) for f in self.listeners: f(item, data) """ Tests for DirCheckboxCtrl A number of unit tests must be performed on the DirCheckboxGUI widget when refactoring. Add to this list so that it becomes comprehensive. Basic Tests: 1. Click on a top-level UNSELECTED object in the tree [should become SELECTED]. - Click again [should become EXCLUDED]. - Click again [should become UNSELECTED]. 2. Click on a non-top-level UNSELECTED object in the tree that has no SELECTED children [should become SELECTED, it's parents should become SELECTEDPARENT and its children SELECTEDCHILD]. - Click again [should become EXCLUDED, it's parents who were SELECTEDPARENT should become UNSELECTED, and it's UNSELECTED children should become EXCLUDED]. - Click again [should become UNSELECTED, and it's children should become UNSELECTED]. 3. Change two children to their SELECTED state [parents should be in SELECTEDPARENT state]. - Click one child to become EXCLUDED [parents should stay in SELECTEDPARENT] - Click the same child to become UNSELECTED [parents should stay in SELECTEDPARENT] - Click the other child to become EXCLUDED [parents should become UNSELECTED] 4. Choose a folder and a child item. - Click the child to become SEL [parent should be SELPAR] - Click the parent [parent should become SEL] - Click the parent again [parent should become SELPAR] 5. Choose a folder and a child item. - Click the parent to become SEL [child should become SELCHILD] - Click the child [child should become SEL] - Click the child again [child should become EXCL] - Click the child again [child should become SELCHILD] 6. Pick a node with children at least two-deep. Change two of the at-least-two-deep children to their SELECTED state [parents should be in SELECTEDPARENT state]. - Click parent closest to SELECTED children to SELECTED [two childen remain in SELECTED, all other children become SELECTEDCHILD. Parent[s] of parent remain SELECTEDPARENT] - Click one child twice to become SELECTEDCHILD [child should not be able to be UNSELECTED, parent states should not change] - Click other child twice to become SELECTEDCHILD [child should not be able to be UNSELECTED, parent states should not change] 7. Pick a node with children at least two-deep. - Click deepest parent to SELECTED [Parent[s] of parent become SELECTEDPARENT] - Click same parent again to become EXCLUDED [Parent[s] of parent become UNSELECTED] - Click same parent again to become UNSELECTED [Parent[s] of parent remain UNSELECTED] 8. Pick a node with children at least two-deep. - Click deepest child to become SELECTED [Parent[s] of parent become SELECTEDPARENT] - Click the topmost parent to become SELECTED [children become SELECTEDCHILD] - Click the topmost parent again to become SELECTEDPARENT [middle child should become SELECTEDPARENT] Multi-Selection Tests: 1. Multi-select three items at the same level and in the same state. Toggle between the three main states [SELECTED, EXCLUDED, UNSELECTED] 2. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the SEL item to see that all three items become EXCL. 3. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the EXCL item to see that all three items become UNSEL. 4. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the UNSEL item to see that all three items become SEL. 5. Choose three items that are nested within each other: a parent folder, one of its children folders, and a file/folder in the child folder. Choose one other item from the child folder. - set the top parent to UNSEL - set the child folder to SEL [parent become SELPAR] - set the child item to SEL - set the other item to EXCL - multi-select all four items - 5A. click on the top parent (which was in SELPAR) [All four items should become SEL, all children of any of these items should become SELCHILD]. Toggle twice more [all selected items should toggle to EXCL, then to UNSEL] - 5B. reset as above, click on the child folder [All four items should become EXCL]. Toggle twice more [all selected items should go to UNSEL, then SEL] - 5C. reset as above, click on the child item [All four items should become EXCL]. Toggle twice more [all selected items should go to UNSEL, then SEL] - 5D. reset as above, click on the other item [All four items should become UNSEL]. Toggle twice more [all selected items should go to SEL, then EXCL] 6. Choose a folder, one if its subfolders, a subfolder of the subfolder, and an item in the deepest subfolder, and an item in the first subfolder, e.g.: [] A [] B [] C [] D [] E - change item 'D' to SEL [parents 'A', 'B', and 'C' should go to SELPAR] - change item 'E' to EXCL - multi-select 'A', 'C', and 'E' - toggle 'E' to UNSEL [all other selections should stay in current state] - toggle 'E' to SEL ['A' and 'B' become SEL, their children become SELCHILD] - toggle 'E' back to EXCL [should get our original multi-select setup back] - toggle 'C' to SEL [all selections to SEL, children to SELCHILD] - toggle 'C' to SELPAR ['A' and 'C' to SELPAR, 'E' to UNSEL] - toggle 'E' twice [should get our original mulit-select setup back] """ class CheckFileListCtrlMixin: # for some insane reason, we can't get EVT_LEFT_DOWN (or _UP) to bind in # FileListCtrl itself. But we are sneaky and can do it by lots of clever # hax0ry, like by using this silly mixin. def __init__(self, toCall): self.Bind(wx.EVT_LEFT_UP, toCall) class FileListCtrl(wx.ListCtrl, listmix.ListCtrlAutoWidthMixin, CheckFileListCtrlMixin): """ Implements a file list control, with a peerctrl that contains the filesystem model. Currently, this peerctrl must implement an addListener(), changeState(), GetItemData(), expandDir(), GetSelections(), and GetChildren() API similar to that implemented by DirCheckBoxCtrl. """ def __init__(self, parent, peerctrl, id=-1, pos=wx.DefaultPosition, size=wx.DefaultSize, style=wx.LC_REPORT, validator=wx.DefaultValidator, name=wx.ListCtrlNameStr): wx.ListCtrl.__init__(self, parent, id, pos, size, style, validator, name) CheckFileListCtrlMixin.__init__(self, self.OnClick) listmix.ListCtrlAutoWidthMixin.__init__(self) self.peerctrl = peerctrl self.peerctrl.addListener(self.itemChanged) self.itemdict = {} # a dict with filepath as key, containing tuples of # (index into ListCtrl, reference to peerctrl object) self.stopsearch = False self.il, self.checkboxes, self.icondict = createDefaultImageList() self.AssignImageList(self.il, wx.IMAGE_LIST_SMALL) self.il = self.GetImageList(wx.IMAGE_LIST_SMALL) self.InsertColumn(0, "Filename") self.InsertColumn(1, "Location") #self.InsertColumn(2, "Last Backup") #self.SetColumnWidth(0, wx.LIST_AUTOSIZE) #self.SetColumnWidth(1, -1) #wx.LIST_AUTOSIZE) self.Bind(wx.EVT_MOTION, self.mouseMotion) self.searchSourceItems = [] def itemChanged(self, item, data): (path, isDir, expanded, state) = data if self.itemdict.has_key(path): item = self.itemdict[path][0] image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) def GetAll(self, excludeStates=[]): result = [] start = -1 for i in range(self.GetItemCount()): item = self.GetNextItem(start, wx.LIST_NEXT_ALL) # XXX: only append if not in excludeStates result.append(item) start = item return result def GetSelections(self): result = [] start = -1 for i in range(self.GetSelectedItemCount()): item = self.GetNextItem(start, wx.LIST_NEXT_ALL, wx.LIST_STATE_SELECTED) result.append(item) start = item return result def GetPeerSelections(self, selections): result = [] for item in selections: path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) if self.itemdict.has_key(path): result.append(self.itemdict[path][1]) return result def mouseMotion(self, event): point = event.GetPosition() item, flags = self.HitTest(point) if flags == wx.LIST_HITTEST_ONITEMICON: path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) text = self.peerctrl.getTooltip(self.itemdict[path][1]) tip = wx.ToolTip(text) self.SetToolTip(tip) #tipwin = tip.GetWindow() #tippos = tipwin.GetPosition() #print "%s vs %s" % (tippos, point) #tipwin.SetPosition(point) def OnClick(self, event): point = event.GetPosition() item, flags = self.HitTest(point) if flags == wx.LIST_HITTEST_ONITEMICON: peerselections = self.GetPeerSelections(self.GetSelections()) path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) ditem = self.itemdict[path][1] # raises if not present self.peerctrl.changeState(ditem, peerselections) (path, isDir, expanded, state) \ = self.peerctrl.GetItemData(ditem).GetData() image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) def searchButtonAction(self, event): selections = self.peerctrl.GetSelections() if len(selections) == 0: return ("Please tell me where to search. Select one or more" " folders in the left-hand panel (hold down SHIFT or" " CTRL for multiple selection), then click the 'find!'" " button again.", None) else: self.DeleteAllItems() self.itemdict = {} b = wx.BusyCursor() searchSourceItems = [] for i in selections: self.addResults(i, event.searchstring) searchSourceItems.append(i) self.searchSourceItems = [self.peerctrl.GetItemData(s).GetData()[0] for s in searchSourceItems] print "sources: %s" % self.searchSourceItems return ("Search results will appear as files that match your" " search are found.", None) return (None, None) def addResults(self, ditem, searchstring): (path, isDir, expanded, state) \ = self.peerctrl.GetItemData(ditem).GetData() position = self.GetItemCount() if isDir: if not expanded: self.peerctrl.expandDir(ditem, busycursor=False) children = self.peerctrl.getChildren(ditem) for c in children: self.addResults(c, searchstring) wx.Yield() if self.stopsearch: break else: terms = [x for x in searchstring.split(' ') if x != ''] for term in terms: print path if path.find(term) > 0: image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state dirname, filename = os.path.split(path) index = self.InsertImageStringItem(position, filename, image) self.SetStringItem(index, 1, dirname) self.SetColumnWidth(0, wx.LIST_AUTOSIZE) self.itemdict[path] = (index, ditem) break def setGroup(self, state): items = self.GetAll() item = items[0] peerselections = self.GetPeerSelections(items) path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) ditem = self.itemdict[path][1] # raises if not present while True: # cycle until the items state matches the desired state self.peerctrl.changeState(ditem, peerselections) # can be slow (path, isDir, expanded, nstate) \ = self.peerctrl.GetItemData(ditem).GetData() if nstate == state: break image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) return self.searchSourceItems class GroupSelectionCheckbox(wx.Panel): def __init__(self, parent, id=-1, setGroupState=None): wx.Panel.__init__(self, parent, id) self.setGroupState = setGroupState self.ubm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-unchecked1.png"))) self.cbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checked1.png"))) self.ebm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excluded1.png"))) self.checkboxButton = wx.BitmapButton(self, -1, self.ubm, style=wx.NO_BORDER) self.Bind(wx.EVT_BUTTON, self.onCheckbox, self.checkboxButton) self.description = wx.StaticText(self, -1, "always BACKUP any files that match these search criteria ") self.state = CheckboxState.UNSELECTED self.gbSizer = wx.GridBagSizer(1,2) self.gbSizer.Add(self.checkboxButton, (0,0), flag=wx.ALIGN_CENTER) self.gbSizer.Add(self.description, (0,1), flag=wx.ALIGN_CENTER) self.gbSizer.AddGrowableRow(1) self.SetSizerAndFit(self.gbSizer) def Enable(self, enable=True): self.checkboxButton.Enable(enable) self.description.Enable(enable) def Disable(self): self.Enable(False) def clear(self): self.checkboxButton.SetBitmapLabel(self.ubm) self.state = CheckboxState.UNSELECTED self.description.SetLabel( "always BACKUP any files that match these search criteria") def setState(self, state): self.state = state if self.state == CheckboxState.UNSELECTED: self.checkboxButton.SetBitmapLabel(self.ubm) self.description.SetLabel( "always BACKUP any files that match these search criteria") elif self.state == CheckboxState.SELECTED: self.checkboxButton.SetBitmapLabel(self.cbm) self.description.SetLabel( "always BACKUP any files that match these search criteria") elif self.state == CheckboxState.EXCLUDED: self.checkboxButton.SetBitmapLabel(self.ebm) self.description.SetLabel( "always EXCLUDE any files that match these search criteria") def onCheckbox(self, event): if self.state == CheckboxState.UNSELECTED: self.checkboxButton.SetBitmapLabel(self.cbm) self.state = CheckboxState.SELECTED if self.setGroupState: self.setGroupState(CheckboxState.SELECTED) elif self.state == CheckboxState.SELECTED: self.checkboxButton.SetBitmapLabel(self.ebm) self.description.SetLabel( "always EXCLUDE any files that match these search criteria") self.state = CheckboxState.EXCLUDED if self.setGroupState: self.setGroupState(CheckboxState.EXCLUDED) elif self.state == CheckboxState.EXCLUDED: self.checkboxButton.SetBitmapLabel(self.ubm) self.description.SetLabel( "always BACKUP any files that match these search criteria") self.state = CheckboxState.UNSELECTED if self.setGroupState: self.setGroupState(CheckboxState.UNSELECTED) class SearchPanel(wx.Panel): def __init__(self, parent, dircheckbox, id=-1, searchButtonAction=None): wx.Panel.__init__(self, parent, id) self.dircheckbox = dircheckbox self.searchButtonAction = searchButtonAction self.SetAutoLayout(False) self.rules = {} # should refer to something from fludrules self.searchField = wx.TextCtrl(self, -1, "search for files to backup here", size=wx.Size(-1,-1), style=wx.TE_PROCESS_ENTER) self.searchField.SetToolTipString('find files within directories' ' selected to the left by entering search terms here') self.searchField.Bind(wx.EVT_TEXT_ENTER, self.onSearchClick) self.searchField.Bind(wx.EVT_LEFT_DOWN, self.selectAllText) self.searchField.Bind(wx.EVT_KILL_FOCUS, self.unfocused) self.searchButton = wx.Button(self, -1, 'find!', name='searchButton') self.Bind(wx.EVT_BUTTON, self.onSearchClick, self.searchButton) self.searchResults = FileListCtrl(self, dircheckbox, -1, name='searchResults', style=wx.SUNKEN_BORDER | wx.LC_REPORT) self.searchResults.SetExtraStyle(0) self.searchResults.SetLabel('found files') self.groupSelection = GroupSelectionCheckbox(self, -1, self.setGroup) self.groupSelection.Disable() self.gbSizer = wx.GridBagSizer(3,2) self.gbSizer.Add(self.searchField, (0,0), flag=wx.EXPAND) self.gbSizer.Add(self.searchButton, (0,1)) self.gbSizer.Add(self.searchResults, (1,0), (1,2), flag=wx.EXPAND|wx.TOP, border=5) self.gbSizer.Add(self.groupSelection, (2,0) ) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) def onSearchClick(self, event): event.searchstring = self.searchField.GetValue() if self.searchButton.GetLabel() == 'stop!': self.searchButton.SetLabel('find!') self.searchResults.stopsearch = True return else: self.groupSelection.clear() self.groupSelection.Disable() self.searchButton.SetLabel('stop!') self.searchButton.Update() err, info = self.searchResults.searchButtonAction(event) selections = self.searchResults.searchSourceItems # see if we should set the checkbox button from a previous rule state = None if len(selections) > 0 and self.rules.has_key(selections[0]): rule = self.rules[selections[0]] if self.rules[selections[0]].has_key(event.searchstring): state = self.rules[selections[0]][event.searchstring] for i in selections: if not self.rules.has_key(i) or self.rules[i] != rule: state = None break #for j in self.rules[i]: if state: print "should restore checkbox to %s" % state self.groupSelection.setState(state) self.searchButton.SetLabel('find!') self.searchResults.stopsearch = False if self.searchButtonAction: self.searchButtonAction(event, errmsg=err, infomsg=info) self.groupSelection.Enable() def selectAllText(self, event): if wx.Window.FindFocus() != self.searchField: self.searchField.SetSelection(-1,-1) self.searchField.SetFocus() else: self.searchField.SetSelection(0,0) event.Skip() def unfocused(self, event): self.searchField.SetSelection(0,0) def setGroup(self, state): b = wx.BusyCursor() selections = self.searchResults.setGroup(state) for s in selections: if not self.rules.has_key(s): self.rules[s] = {} if state == CheckboxState.UNSELECTED: try: self.rules.pop(s) except: pass else: self.rules[s][self.searchField.GetValue()] = state print self.rules class FilePanel(wx.SplitterWindow): def __init__(self, parent, searchButtonAction=None): # Use the WANTS_CHARS style so the panel doesn't eat the Return key. wx.SplitterWindow.__init__(self, parent, -1, style=wx.SP_LIVE_UPDATE | wx.CLIP_CHILDREN | wx.WANTS_CHARS) self.Bind(wx.EVT_SIZE, self.OnSize) self.SetNeedUpdating(True) print self.GetSize() self.tree = DirCheckboxCtrl(self, -1, dir="/") # XXX: fludrules.init path should be in config self.fludrules = self.getFludHome()+"/fludrules.init" if not os.path.isfile(self.fludrules): # XXX: do the other first time stuff (email encrypted credentials, # etc.) parent.SetMessage("Welcome. This appears to be the first" " time you've run flud. We've automatically selected some" " files for backup. You can make changes by" " selecting/deselecting files and directories. When you are" " done, simply close this window.") src = open(os.path.join(datadir,'fludrules.init'), 'r') dst = open(self.fludrules, 'w') filerules = src.read() dst.write(filerules) dst.close() src.close() filerules = eval(filerules) rulestates = {} for rule in filerules['baserules']: value = filerules['baserules'][rule] rule = glob.glob(os.path.expandvars(rule)) for r in rule: rulestates[r] = value self.tree.setStates(rulestates) # XXX: fludfile.conf path should be in config self.fludfiles = self.getFludHome()+"/fludfile.conf" print self.fludfiles if os.path.isfile(self.fludfiles): file = open(self.fludfiles, 'r') states = eval(file.read()) self.tree.setStates(states) file.close() self.searchPanel = SearchPanel(self, dircheckbox=self.tree, searchButtonAction=searchButtonAction) self.SetMinimumPaneSize(50) self.SplitVertically(self.tree, self.searchPanel) #, 300) print self.GetSize() def getFludHome(self): if os.environ.has_key('FLUDHOME'): fludhome = os.environ['FLUDHOME'] else: fludhome = os.environ['HOME']+"/.flud" if not os.path.isdir(fludhome): os.mkdir(fludhome, 0700) return fludhome def shutdown(self, event): self.flushFileConfig() event.Skip() def flushFileConfig(self): states = self.tree.getStates() f = open(self.fludfiles, 'w') f.write(str(states)) f.close() for i in states: print "%s %s" % (i, states[i]) def OnSize(self, event): w,h = self.GetClientSizeTuple() if self.tree: self.tree.SetDimensions(0, 0, w, h) event.Skip() class RestoreCheckboxCtrl(DirCheckboxCtrl): # XXX: child/parent selection/deselection isn't quite right still, esp wrt # root node. repro: # -/ # -d1 # -f1 # -d2 # -d3 # -f2 # -f3 # with nothing selected, select d3 and f3, then select root, then deselect # d3 and f3 def __init__(self, parent, id=-1, config=None, pos=wx.DefaultPosition, size=wx.DefaultSize, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER), validator=wx.DefaultValidator, name=wx.ControlNameStr): self.config = config DirCheckboxCtrl.__init__(self, parent, id, config, pos, size, style, validator, name, allowExclude=False) def initTree(self, config): self.expandRoot(config) self.expandUntilMultiple() def expandRoot(self, config): self.defaultImageList, self.checkboxes, self.icondict \ = createDefaultImageList() self.AssignImageList(self.defaultImageList) self.il = self.GetImageList() self.rootID = self.AddRoot("/", self.icondict['computer'], -1, wx.TreeItemData(("", True, False, CheckboxState.UNSELECTED))) self.update() def expandUntilMultiple(self): node = self.rootID while True: (ipath, isdir, expanded, istate) = self.GetItemData(node).GetData() children = self.getChildren(node, False) if len(children) > 1 or len(children) == 0: break; node = children[0] self.Expand(node) def update(self): master = listMeta(self.config) for i in master: if not isinstance(master[i], dict): traversal = i.split(os.path.sep) node = self.rootID path = "/" if traversal[0] == '': traversal.remove('') for n in traversal: path = os.path.join(path, n) children = self.getChildrenDict(node) if n == traversal[-1] and not n in children: child = self.AppendItem(node, n) self.SetPyData(child, (path, False, False, 0)) idx = getFileIcon(i, self.il, self.checkboxes, self.icondict) self.SetItemImage(child, idx, wx.TreeItemIcon_Normal) else: if not n in children: child = self.AppendItem(node, n) self.SetPyData(child, (path, True, False, 0)) self.SetItemImage(child, self.icondict['folder'], wx.TreeItemIcon_Normal) else: child = children[n] node = child self.Expand(self.rootID) def getChildrenDict(self, node): result = {} child, cookie = self.GetFirstChild(node) while child: result[self.GetItemText(child)] = child child, cookie = self.GetNextChild(node, cookie) return result def onExpand(self, event): pass def getSelected(self, startNode=None): if not startNode: startNode = self.rootID children = self.getChildren(startNode) selected = [] for n in children: (path, isDir, expanded, state) = self.GetItemData(n).GetData() if not isDir \ and (state == CheckboxState.SELECTED \ or state == CheckboxState.SELECTEDCHILD): selected.append(n) if isDir and (state == CheckboxState.SELECTED \ or state == CheckboxState.SELECTEDPARENT \ or state == CheckboxState.SELECTEDCHILD): selected += self.getSelected(n) return selected class RestorePanel(wx.Panel): def __init__(self, parent, config, factory): self.config = config self.factory = factory wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) self.tree = RestoreCheckboxCtrl(self, -1, config, #wx.TreeCtrl(self, -1, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER)) self.restoreButton = wx.Button(self, -1, 'restore selected files', name='restoreButton') self.Bind(wx.EVT_BUTTON, self.onRestoreClick, self.restoreButton) self.gbSizer = wx.GridBagSizer(2,1) self.gbSizer.Add(self.tree, (0,0), flag=wx.EXPAND|wx.ALL, border=0) self.gbSizer.Add(self.restoreButton, (1,0), flag=wx.EXPAND|wx.ALL, border=0) self.gbSizer.AddGrowableRow(0) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) def update(self): self.tree.update() def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() def onTooltip(self, event): pass def onRestoreClick(self, event): for n in self.tree.getSelected(): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "restoring %s" % path d = self.factory.sendGETF(path) d.addCallback(self.restored, n) d.addErrback(self.restoreFailed, n) self.tree.UnselectAll() def restored(self, res, n): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "yay, %s" % path self.tree.SetItemTextColour(n, '#005804') self.tree.changeState(n) def restoreFailed(self, err, n): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "boo, %s: %s" % (path, err) self.tree.SetItemTextColour(n, wx.RED) self.tree.changeState(n) class SchedulePanel(wx.Panel): def __init__(self, parent): wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() class FeedbackPanel(wx.Panel): def __init__(self, parent): wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) editor = wx.lib.editor.editor.Editor(parent, -1) def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() class FludNotebook(wx.Notebook): def __init__(self, parent, id=-1, pos=wx.DefaultPosition, size=wx.DefaultSize, style=wx.NB_BOTTOM|wx.NO_BORDER): self.parent = parent self.config = parent.config self.factory = LocalClientFactory(self.config) print "connecting to localhost:%d" % self.config.clientport reactor.connectTCP('localhost', self.config.clientport, self.factory) wx.Notebook.__init__(self, parent, id, pos, style=style) self.filePanel = FilePanel(self, searchButtonAction=parent.searchButtonAction) self.AddPage(self.filePanel, "Backup Files") self.restorePanel = RestorePanel(self, self.config, self.factory) self.AddPage(self.restorePanel, "Restore") self.schedulePanel = SchedulePanel(self) self.AddPage(self.schedulePanel, "Backup Schedule") self.feedbackPanel = FeedbackPanel(self) self.AddPage(self.feedbackPanel, "Feedback") self.Bind(wx.EVT_NOTEBOOK_PAGE_CHANGED, self.changedPage) def shutdown(self, event): self.filePanel.shutdown(event) def changedPage(self, event): page = event.GetSelection() if page == 0: self.SetMessage("Select files and directories for backup" " with the filesystem view on the left, or set up criteria" " for finding files for backup with simple searches," " below right.") elif page == 1: self.SetMessage("Select files/directories to be restored to" " your computer, then click on 'restore!' Files will turn" " green as they arrive.") self.restorePanel.update() elif page == 2: self.SetMessage("Configure how often your computer should backup." "\n (not implemented)") elif page == 3: self.SetMessage("Send feedback to flud programmers. (not" " implemented)") def SetMessage(self, msg): self.parent.SetMessage(msg) class FludLogoPanel(wx.Panel): def __init__(self, parent, id=-1, pos=wx.DefaultPosition, size=wx.Size(10,10), style=wx.TAB_TRAVERSAL, name="logo panel"): wx.Panel.__init__(self, parent, id, pos, size, style, name) self.SetAutoLayout(True) self.SetBackgroundColour(wx.BLACK) self.SetForegroundColour(wx.WHITE) logobmp = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "flud-backup-logo-1-150-nodrop.png"))) pad = 0 self.logowidth = logobmp.GetWidth() self.logoheight = logobmp.GetHeight() self.logo = wx.StaticBitmap(self, -1, logobmp) self.messagePanel = wx.Panel(self, -1) self.messagePanel.SetBackgroundColour(wx.BLACK) self.messagePanel.SetForegroundColour(wx.WHITE) self.message = wx.StaticText(self.messagePanel, -1, "message text area", style=wx.ALIGN_CENTER|wx.ST_NO_AUTORESIZE, size=(-1, self.logoheight-15)) self.message.Bind(wx.EVT_SIZE, self.resizeMessage) self.bsizer = wx.BoxSizer(wx.VERTICAL) self.bsizer.Add(self.message, flag=wx.EXPAND|wx.ALL, border=35) self.bsizer.SetSizeHints(self.messagePanel) self.messagePanel.SetSizer(self.bsizer) self.gbSizer = wx.GridBagSizer(1,2) self.gbSizer.Add(self.logo, (0,0)) self.gbSizer.Add(self.messagePanel, (0,1), flag=wx.EXPAND|wx.ALL) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(1) self.SetSizerAndFit(self.gbSizer) self.SetSize(wx.Size(self.logowidth, self.logoheight)) self.SetSizeHints(self.logowidth, self.logoheight, -1, self.logoheight) def SetMessage(self, msg): (w,h) = self.message.GetSizeTuple() #print "msg area size is %d x %d" % (w,h) self.message.SetLabel(msg) self.message.Wrap(w) #print "msg is '%s'" % self.message.GetLabel() self.message.Center() def resizeMessage(self, evt): # this is mainly to deal with StaticText wonkiness (not calling Wrap() # automatically, not centering properly automatically). It may be # possible to get rid of this with a future wxPython release. (w,h) = self.message.GetSizeTuple() self.message.Wrap(w) m = self.message.GetLabel() m = m.replace('\n',' ') self.message.SetLabel(m) self.message.Wrap(w) self.message.Center() class FludFrame(wx.Frame): def __init__(self, parent, id=wx.ID_ANY, label="flud bakcup client", size=wx.Size(800,600), style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE, config=None): wx.Frame.__init__(self, parent, id, label, size=size, style=style) self.lognull = wx.LogNull() wx.ToolTip.SetDelay(2000) self.clearMessage = False self.logoPanel = FludLogoPanel(self) self.SetMessage('Welcome.') self.config = config self.notebook = FludNotebook(self, size=wx.Size(200,200)) self.operationStatus = wx.StatusBar(name='operationStatus', parent=self, style=0) self.SetStatusBar(self.operationStatus) self.gbSizer = wx.GridBagSizer(2,1) self.gbSizer.Add(self.logoPanel,(0,0), flag=wx.EXPAND) self.gbSizer.Add(self.notebook, (1,0), flag=wx.EXPAND|wx.ALL, border=1) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) self.Bind(wx.EVT_CLOSE, self.shutdown) self.SetSize(size) self.Show(True) def SetMessage(self, message): self.logoPanel.SetMessage(message) def shutdown(self, event): self.notebook.shutdown(event) def searchButtonAction(self, event, errmsg=None, infomsg=None): if errmsg: self.logoPanel.SetMessage(errmsg) self.clearMessage = True elif infomsg: self.logoPanel.SetMessage(infomsg) self.clearMessage = False elif self.clearMessage: self.logoPanel.SetMessage("") #if __name__ == '__main__': # app = wx.PySimpleApp() # # config = FludConfig() # config.load(doLogging=False) # # f = FludFrame(None, wx.ID_ANY, 'flud backup client', size=(795,600), # style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE, # config=config) # # from twisted.internet import reactor # reactor.registerWxApp(app) # reactor.run()
Python
#!/usr/bin/python """ FludTestGauges.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Provides gauges for visualizing storage for multiple flud nodes running on the same host. This is really only useful for demos and testing. """ import sys, os, signal, stat, random import wx import wx.lib.buttons as buttons from flud.FludConfig import FludConfig dutotal = 0 def visit(arg, top, files): global dutotal for file in files: dutotal += os.lstat("%s" % (os.path.join(top,file)))[stat.ST_SIZE] arg += dutotal def du(dir): global dutotal dutotal = 0 os.path.walk(dir, visit, dutotal) return dutotal # XXX: too much manual layout. should convert to a managed layout to allow for # resizing, etc. SGAUGEWIDTH = 230 # storage gauge DGAUGEWIDTH = 100 # dht gauge GAUGEHEIGHT = 20 ROWHEIGHT = 30 SEP = 5 LABELWIDTH = 20 POWERWIDTH = 70 RATIOBARHEIGHT = 70 COLWIDTH = SGAUGEWIDTH+DGAUGEWIDTH+LABELWIDTH+POWERWIDTH COLGAPFUDGE = 30 class FludTestGauges(wx.Frame): def __init__(self, parent, title, dirroot, dirs): screenHeight = wx.SystemSettings.GetMetric(wx.SYS_SCREEN_Y)-100 rowheight = ROWHEIGHT+SEP height = len(dirs)*(rowheight)+RATIOBARHEIGHT columns = height / screenHeight + 1 width = COLWIDTH*columns if columns > 1: height = (len(dirs)/columns)*(rowheight)+RATIOBARHEIGHT if (len(dirs) % columns) > 0: height += rowheight wx.Frame.__init__(self, parent, wx.ID_ANY, title, size=(width,height), style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE) self.storebarend = 1024 self.smultiplier = 100.0 / self.storebarend self.sdivisor = 1 self.sbytelabel = "" self.dhtbarend = 512 self.dmultiplier = 100.0 / self.dhtbarend self.ddivisor = 1 self.dbytelabel = "" self.storeheading = wx.StaticText(self, -1, "block storage", (LABELWIDTH, 5)) self.totaldht = wx.StaticText(self, -1, "metadata", (LABELWIDTH+SGAUGEWIDTH+SEP, 5)) self.gauges = [] curCol = 0 curRow = 30 for i in range(len(dirs)): self.gauges.append(wx.Gauge(self, -1, 100, (curCol*COLWIDTH+LABELWIDTH, curRow), (SGAUGEWIDTH, GAUGEHEIGHT))) self.gauges[i].SetBezelFace(3) self.gauges[i].SetShadowWidth(3) self.gauges[i].SetValue(0) self.gauges[i].dir = "%s%s" % (dirroot,dirs[i]) os.environ['FLUDHOME'] = self.gauges[i].dir; conf = FludConfig() conf.load(doLogging = False) print "%s" % conf.nodeID self.gauges[i].label = wx.StaticText(self, -1, "%2s" % dirs[i], (curCol*COLWIDTH, curRow+(rowheight/4)), size=(LABELWIDTH, -1)) self.gauges[i].idlabel = wx.StaticText(self, -1, "%s" % conf.nodeID, (curCol*COLWIDTH+LABELWIDTH, curRow+20)) font = self.gauges[i].idlabel.GetFont() font.SetPointSize(6) self.gauges[i].idlabel.SetFont(font) self.gauges[i].dhtgauge = wx.Gauge(self, -1, 100, (curCol*COLWIDTH+LABELWIDTH+SGAUGEWIDTH+SEP, curRow), (SGAUGEWIDTH/3, GAUGEHEIGHT)) self.gauges[i].power = wx.Button(self, i, "turn OFF", (curCol*COLWIDTH +LABELWIDTH+SGAUGEWIDTH+2*SEP+SGAUGEWIDTH/3, curRow), (POWERWIDTH, ROWHEIGHT)) #self.gauges[i].power = buttons.GenBitmapToggleButton(self, i, # None, # (LABELWIDTH+SGAUGEWIDTH+2*SEP+SGAUGEWIDTH/3, curRow), # (POWERWIDTH, ROWHEIGHT)) #self.gauges[i].button.SetBestSize() self.gauges[i].power.SetToolTipString("power on/off") self.Bind(wx.EVT_BUTTON, self.onClick, self.gauges[i].power) curRow += rowheight if curRow > height-RATIOBARHEIGHT: curCol += 1 curRow = 30 self.totalstore = wx.StaticText(self, -1, "total: 0", (LABELWIDTH, height-40)) self.totaldht = wx.StaticText(self, -1, "total: 0", (LABELWIDTH+SGAUGEWIDTH+SEP, height-40)) self.ratiogauge = wx.Gauge(self, -1, 100, (LABELWIDTH, height-20), (SGAUGEWIDTH+SEP+SGAUGEWIDTH/3, 10)) self.ratiogauge.SetValue(0) self.Bind(wx.EVT_IDLE, self.IdleHandler) self.timer = wx.PyTimer(self.update) self.timer.Start(1000) def onClick(self, event): # XXX: note that under our current startNnodes.sh scheme, the first # node spawned doesn't contact anyone, so if that one is powered off # and then powered back on, it will not be part of the node until # another node pings it # XXX: unix-specific proc management stuff follows idx = event.GetId() home = self.gauges[idx].dir pidfile = os.path.join(home, 'twistd.pid') if os.path.exists(pidfile): print "shutting down %s" % home f = open(pidfile) pid = int(f.read()) f.close() # XXX: ps command no worky on windows, and "-ww" may not worker on # oldskool unixes self.gauges[idx].savedCmd = os.popen( "ps f -wwp %d -o args=" % pid).read() procline = os.popen("ps e -wwp %d" % pid).read() self.gauges[idx].savedEnv = [e for e in procline.split() if e[:4] == 'FLUD'] # XXX: os.kill no worky on windows, need something like: #def windowskill(pid): # import win32api # handle = win32api.OpenProcess(1, 0, pid) # return (0 != win32api.TerminateProcess(handle, 0)) os.kill(pid, signal.SIGTERM) self.gauges[idx].power.SetLabel("turn ON") self.gauges[idx].Hide() self.gauges[idx].dhtgauge.Hide() else: print "powering up %s" % home # XXX: this exec no worky on windows fullcmd = "%s %s" % (' '.join(self.gauges[idx].savedEnv), self.gauges[idx].savedCmd) print fullcmd result = os.popen('%s %s' % (' '.join(self.gauges[idx].savedEnv), self.gauges[idx].savedCmd)).readlines() self.gauges[idx].power.SetLabel("turn OFF") self.gauges[idx].Show() self.gauges[idx].dhtgauge.Show() print result def update(self): def sizeclass(num): divisor = 1 bytelabel = "" if num > 1024: divisor = 1024.0 bytelabel = 'K' if num > 1048576: divisor = 1048576.0 bytelabel = 'M' if num > 1073741824: divisor = 1073741824.0 bytelabel = 'G' return (divisor, bytelabel) storelargest = 0 dhtlargest = 0 storetotal = 0 dhttotal = 0 for i in self.gauges: if os.path.isdir(i.dir): i.storebytes = du(os.path.join(i.dir,'store')) if i.storebytes > storelargest: storelargest = i.storebytes storetotal += i.storebytes i.dhtbytes = du(os.path.join(i.dir,'dht')) if i.dhtbytes > dhtlargest: dhtlargest = i.dhtbytes dhttotal += i.dhtbytes else: i.storebytes = 0 i.dhtbytes = 0 i.Disable() i.power.Disable() while storelargest > self.storebarend: self.storebarend = self.storebarend * 2 self.smultiplier = 100.0 / self.storebarend self.sdivisor, self.sbytelabel = sizeclass(storetotal) while dhtlargest > self.dhtbarend: self.dhtbarend = self.dhtbarend * 2 self.dmultiplier = 100.0 / self.dhtbarend self.ddivisor, self.dbytelabel = sizeclass(dhttotal) #print "-----" for i in self.gauges: i.SetValue(i.storebytes*self.smultiplier) i.dhtgauge.SetValue(i.dhtbytes*self.dmultiplier) #print "%.2f, %.2f" % ((float(i.storebytes)/float(i.dhtbytes)), # (float(i.GetValue())/float(i.dhtgauge.GetValue()))) self.totalstore.SetLabel("total: %.1f%s" % (float(storetotal)/self.sdivisor, self.sbytelabel)) self.totaldht.SetLabel("total: %.1f%s" % (float(dhttotal)/self.ddivisor, self.dbytelabel)) if (dhttotal+storetotal == 0): self.ratiogauge.SetValue(0) else: self.ratiogauge.SetValue((storetotal*100/(dhttotal+storetotal))) def updateGauges(self, update): for index, value in update: self.monitors[index].setValue(value) def IdleHandler(self, event): pass def main(): if len(sys.argv) < 2: print "usage: %s dircommon exts" % sys.argv[0] print " where exts will be appended to dircommon" print " e.g., '%s /home/joe/.flud 1,2,3,4,10,15,20'"\ % sys.argv[0] print " or, '%s /home/joe/.flud 1-10,15,20'"\ % sys.argv[0] sys.exit() root = sys.argv[1] exts = [] dirs = [d.strip() for d in sys.argv[2].split(',')] for i in dirs: if i == "_": exts.append('') # undocumented, means "just dircommon" elif i.find('-') >= 0: start, end = i.split('-') for j in range(int(start),int(end)+1): exts.append(j) else: exts.append(int(i)) app = wx.PySimpleApp() t = FludTestGauges(None, 'Flud Test Gauges', root, exts) t.Show(1) app.MainLoop() if __name__ == '__main__': main()
Python
#!/usr/bin/python """ FludNode.tac (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL). This is the application file used by twistd to daemonize FludNode. """ import os from twisted.application import service, internet import flud.FludNode from flud.protocol.FludCommUtil import getCanonicalIP port = None gwhost = None gwport = None if 'FLUDPORT' in os.environ: port = int(os.environ['FLUDPORT']) if 'FLUDGWHOST' in os.environ: gwhost = getCanonicalIP(os.environ['FLUDGWHOST']) if 'FLUDGWPORT' in os.environ: gwport = int(os.environ['FLUDGWPORT']) node = flud.FludNode.FludNode(port) if gwhost and gwport: node.connectViaGateway(gwhost, gwport) application = service.Application("flud.FludNode") service = node.start(twistd=True) #service.setServiceParent(application)
Python
import base64 """ fencode.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Provides efficient urlsafe base64 encoding of python types (int, long, string, None, dict, tuple, list) -- in the same vein as BitTorrent's bencode or MNet's mencode. """ class Fencoded: # provides a typed wrapper for previously fencoded data, so that we can # nest fencoded data inside other fencoded structures without bloating it # (note that this could also be used to store strings without b64 bloating, # but that forgoes url safety). See doctests in fencode() for usage # examples. def __init__(self, data): self.data = data def __eq__(self, i): if not isinstance(i, Fencoded): return False return self.data == i.data def fencode(d, lenField=False): """ Takes string data or a number and encodes it to an efficient URL-friendly format. >>> n = None >>> i = 123455566 >>> I = 1233433243434343434343434343434343509669586958695869L >>> s = "hello there, everyone" >>> s2 = "long text ............................................................................... AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAABBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC" >>> d = {'a': 'adfasdfasd', 'aaa': 'rrreeeettt', 'f': 'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa'} >>> d2 = {'a': 123, 'b': 'xyz'} >>> d3 = {'a': 123, 'b': 'xyz', 'c': {'x': 456, 'y': 'abc'}} >>> d3 = {'a': 123, 'b': 'xyz', 'c': d} >>> d4 = {} >>> l = [1,2,3,4,'a','b','cde'] >>> l2 = [i,I,s,d] >>> l3 = [] >>> l4 = [n] >>> t = (1,2,3,4,'a','b','cde') >>> t2 = (i,I,s,d) >>> t3 = () >>> d5 = {t: s, 'n': n, 'a': i, i: "a", 'd': d3, s2: s2} >>> l5 = [1,[],2,[],(),{},3,{t: s, 'n': ()}] >>> fdecode(fencode(n)) == n True >>> fdecode(fencode(i)) == i True >>> fdecode(fencode(I)) == I True >>> fdecode(fencode(-i)) == -i True >>> fdecode(fencode(-I)) == -I True >>> fdecode(fencode(s)) == s True >>> fdecode(fencode(d)) == d True >>> fdecode(fencode(d2)) == d2 True >>> fdecode(fencode(d3)) == d3 True >>> fdecode(fencode(d4)) == d4 True >>> fdecode(fencode(l)) == l True >>> fdecode(fencode(l2)) == l2 True >>> fdecode(fencode(l3)) == l3 True >>> fdecode(fencode(l4)) == l4 True >>> fdecode(fencode(t)) == t True >>> fdecode(fencode(t2)) == t2 True >>> fdecode(fencode(t3)) == t3 True >>> fdecode(fencode(d5)) == d5 True >>> fdecode(fencode(l5)) == l5 True >>> f = Fencoded(fencode(s)) >>> fdecode(fencode(f)) == f True >>> fdecode(fdecode(fencode(f))) == s True >>> fdecode(fencode({i: f})) == {i: f} True >>> fdecode(fdecode(fencode({i: f}))[i]) == s True >>> fdecode(fdecode(fencode({i: f, I: f}))[i]) == s True >>> fdecode(fencode(f), recurse=True) == s True >>> fdecode(fencode(f), recurse=2) == s True >>> f2 = Fencoded(fencode(f)) >>> f3 = Fencoded(fencode(f2)) >>> fdecode(fencode(f3), recurse=True) == s True >>> fdecode(fencode(f3), recurse=3) == f True >>> fdecode(fencode({i: f3, I: f2})) == {i: f3, I: f2} True >>> fdecode(fencode({i: f3, I: f2}), recurse=1) == {i: f3, I: f2} True >>> fdecode(fencode({i: f3, I: f2}), recurse=2) == {i: f2, I: f} True >>> fdecode(fencode({i: f3, I: f2}), recurse=3) == {i: f, I: s} True >>> fdecode(fencode({i: f3, I: f2}), recurse=4) == {i: s, I: s} True >>> fdecode(fencode({i: f3, I: f2}), recurse=True) == {i: s, I: s} True """ def makeLen(i): """ Returns the integer i as a three-byte length value. >>> makeLen(255) '\\x00\\xff' >>> makeLen(65535) '\\xff\\xff' """ if i > 65535 or i < 0: raise ValueError("illegal length for fencoded data" "(0 < x <= 65535)") return fencode(i)[1:-1] if isinstance(d, int) or isinstance(d, long): val = "%x" % d neg = False c = 'i' if isinstance(d, long): c = 'o' if d < 0: neg = True val = val[1:] c = c.upper() if len(val) % 2 != 0: val = "0%s" % val val = val.decode('hex') if len(val) % 2 != 0: val = '\x00' + val val = base64.urlsafe_b64encode(val) if lenField: if len(val) > 65535: raise ValueError("value to large for encode") return c+makeLen(len(val))+val else: return c+val elif isinstance(d, str): # String data may contain characters outside the allowed charset. # urlsafe b64encoding ensures that data can be used inside http urls # (and other plaintext representations). val = base64.urlsafe_b64encode(d) if lenField: if len(val) > 65535: raise ValueError("value to large for encode") return 's'+makeLen(len(val))+val else: return 's'+val elif isinstance(d, dict): result = 'd' contents = "" for i in d: contents = contents + fencode(i,True) + fencode(d[i],True) if lenField: result = result+makeLen(len(contents))+contents else: result = result+contents return result elif isinstance(d, list): result = 'l' contents = '' for i in d: contents = contents + fencode(i,True) if lenField: result = result+makeLen(len(contents))+contents else: result = result+contents return result elif isinstance(d, tuple): result = 't' contents = '' for i in d: contents = contents + fencode(i,True) if lenField: result = result+makeLen(len(contents))+contents else: result = result+contents return result elif isinstance(d, Fencoded): result = 'f' contents = d.data if lenField: result = result+makeLen(len(d.data))+contents else: result = result+contents return result elif d == None: if lenField: return 'n'+makeLen(1)+'0' else: return 'n0' else: raise ValueError("invalid value passed to fencode: %s" % type(d)) def fdecode(d, lenField=False, recurse=1): """ Takes previously fencoded data and decodes it into its python type(s). 'lenField' is used internally, and indicates that the fencoded data has length fields (used for compositiong of tuples, lists, dicts, etc). 'recurse' indicates that fdecode should recursively fdecode Fencoded objects if set to True, or that it should recurse to a depth of 'recurse' when encountering Fencoded objects if it is an integer value. """ def getLen(s): if len(s) != 3 or not isinstance(s, str): raise ValueError("fdecode length strings must be 3 bytes long: '%s'" % s) return fdecode('i'+s+'=', recurse=recurse) def scanval(valstring, lenField=False): """ scans the given valstring and returns a value and the offset where that value ended (as a tuple). If valstring contains more than one value, only the length of the first is returned. Otherwise, the entire length is returned. """ valtype = valstring[0] if lenField: start = 4 end = start+getLen(valstring[1:4]) else: start = 1 end = len(valstring)-1 #print " scanval calling fdecode on val[%d:%d]=%s" % (0, end, valstring) return (fdecode(valstring[0:end], True, recurse=recurse), end) if isinstance(d, Fencoded): return fdecode(d.data, recurse=recurse) if not isinstance(d, str): raise ValueError("decode takes string data or Fencoded object only," " got %s" % type(d)) valtype = d[0] if lenField: length = getLen(d[1:4]) val = d[4:] else: val = d[1:len(d)] if valtype == 'i': val = base64.urlsafe_b64decode(val) val = val.encode('hex') return int(val, 16) elif valtype == 'I': val = base64.urlsafe_b64decode(val) val = val.encode('hex') return -int(val, 16) elif valtype == 'o': val = base64.urlsafe_b64decode(val) val = val.encode('hex') return long(val, 16) elif valtype == 'O': val = base64.urlsafe_b64decode(val) val = val.encode('hex') return -long(val, 16) elif valtype == 's': return base64.urlsafe_b64decode(val) elif valtype == 'd': result = {} while len(val) != 0: #print "string is: %s (len=%d)" % (val, len(val)) (key,l1) = scanval(val, True) #print "got key '%s' of length %d" % (key,l1) (value,l2) = scanval(val[l1:len(val)], True) #print "got value '%s' of length %d" % (value,l2) result[key] = value val = val[l1+l2:] return result elif valtype == 'l': result = [] if lenField: pass while len(val) != 0: (v,l) = scanval(val, True) result.append(v) val = val[l:] return result elif valtype == 't': result = [] if lenField: pass while len(val) != 0: (v,l) = scanval(val, True) result.append(v) val = val[l:] return tuple(result) elif valtype == 'f': if not isinstance(recurse, bool): recurse = recurse-1 if recurse > 0: return fdecode(val, recurse=recurse) return Fencoded(val) elif valtype == 'n': return None else: raise ValueError("invalid value passed to fdecode" " -- cannot fdecode data that wasn't previously fencoded: '%s'" % d[:400]) if __name__ == '__main__': import doctest doctest.testmod()
Python
# XXX: this class goes away class Reputation: """ Each node maintains a list of reputation objects corresponding to reputations of other nodes. Reputations may be self-generated (in which case the originator is this node itself), or may be relayed (in which case some other node is the originator). Self-generated reputations are vastly more reliable than those relayed -- relayed reputations are second-hand information, and are more likely to have false data. """ def __init__(self, ID, originator): """ Variables designated as '%' have values between 0 and 100. """ self.ID = ID self.originator = originator self.confidence = 0 # % originator data stored / nodes the # originator stores to self.verifiability = 0 # % originator data verified success/failed self.availability = 0 # % originator contact attempts success/fail self.bandwidth = 0 # avg bandwidth observed from orig. to ID. self.age = 0 # age of reputation in days def score(self): """ Returns a score for this reputation based in member variables. The reputation must be a local reputation, i.e., the originator must be equal to the global myNodeID. Otherwise, call scoreRelay() >>> myNodeID = "self" >>> rep = Reputation("somenode","self") >>> rep.availability = 50 >>> rep.verifiability = 50 >>> rep.score() 33 >>> rep = Reputation("somenode","someothernode") >>> rep.availability = 30 >>> rep.score() -1 """ # should find a good adjustment of weights (XXX: machine learning?) if self.originator != myNodeID: return -1 return (self.confidence + self.verifiability + self.availability) / 3 # XXX: should also include age and bandwidth def scoreRelay(self): """ Returns a score for this reputation based in member variables. The reputation must be a remote reputation, i.e., the originator must not be equal to the global myNodeID. Otherwise, call score() >>> myNodeID = "self" >>> rep = Reputation("somenode","self") >>> rep.availability = 50 >>> rep.verifiability = 50 >>> rep.scoreRelay() -1 >>> rep = Reputation("somenode","someothernode") >>> rep.availability = 30 >>> rep.scoreRelay() 10 """ if self.originator == myNodeID: return -1 return (self.confidence + self.verifiability + self.availability) / 3 # XXX: should also include age and bandwidth def updateConfidence(self, totalDataStored, totalNodesStoredTo): self.confidence = totalDataStored / totalNodesStoredTo; def _test(self): import doctest doctest.testmod() if __name__ == '__main__': myNodeID = "self" rep = Reputation("other", "self") rep._test()
Python
#!/usr/bin/python """ FludLocalClient.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. FludLocalClient provides a command-line client for interacting with FludNode. """ import sys, os, time from twisted.internet import reactor from flud.FludConfig import FludConfig from flud.fencode import fencode, fdecode from flud.FludCrypto import hashfile from protocol.LocalClient import * logger = logging.getLogger('flud') class CmdClientFactory(LocalClientFactory): def __init__(self, config): LocalClientFactory.__init__(self, config) self.quit = False self.msgs = [] def callFactory(self, func, commands, msgs): # since we can't call factory methods from the promptUser thread, we # use this as a convenience to put those calls back in the event loop reactor.callFromThread(self.doFactoryMethod, func, commands, msgs) def doFactoryMethod(self, func, commands, msgs): d = func() d.addCallback(self.queueResult, msgs, '%s succeeded' % commands) d.addErrback(self.queueError, msgs, '%s failed' % commands) return d def promptUser(self): helpDict = {} command = raw_input("%s> " % time.ctime()) commands = command.split(' ') # XXX: should tokenize on any whitespace commandkey = commands[0][:4] # core client operations helpDict['exit'] = "exit from the client" helpDict['help'] = "display this help message" helpDict['ping'] = "send a GETID() message: 'ping host port'" helpDict['putf'] = "store a file: 'putf canonicalfilepath'" helpDict['getf'] = "retrieve a file: 'getf canonicalfilepath'" helpDict['geti'] = "retrieve a file by CAS key: 'geti fencodedCASkey'" helpDict['fndn'] = "send a FINDNODE() message: 'fndn hexIDstring'" helpDict['list'] = "list stored files (read from local metadata)" helpDict['putm'] = "store master metadata" helpDict['getm'] = "retrieve master metadata" helpDict['cred'] = "send encrypted private credentials: cred"\ " passphrase emailaddress" helpDict['node'] = "list known nodes" helpDict['buck'] = "print k buckets" helpDict['stat'] = "show pending actions" helpDict['stor'] = "store a block to a given node:"\ " 'stor host:port,fname'" helpDict['rtrv'] = "retrieve a block from a given node:"\ " 'rtrv host:port,fname'" helpDict['vrfy'] = "verify a block on a given node:"\ " 'vrfy host:port:offset-length,fname'" helpDict['fndv'] = "retrieve a value from the DHT: 'fndv hexkey'" helpDict['dlet'] = "delete from the stor: '[XXX]'" if commandkey == 'exit' or commandkey == 'quit': self.quit = True elif commandkey == 'help': self.printHelp(helpDict) elif commandkey == 'ping': # ping a host # format: 'ping host port' func = lambda: self.sendPING(commands[1], commands[2]) self.callFactory(func, commands, self.msgs) elif commandkey == 'putf': # store a file # format: 'putf canonicalfilepath' func = lambda: self.sendPUTF(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'getf': # retrieve a file # format: 'getf canonicalfilepath' func = lambda: self.sendGETF(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'geti': # retrieve a file by CAS ID # format: 'geti fencoded_CAS_ID' func = lambda: self.sendGETI(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'fndn': # find a node (or the k-closest nodes) # format: 'fndn hexIDstring' func = lambda: self.sendFNDN(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'list': # list stored files self.callFactory(self.sendLIST, commands, self.msgs) elif commandkey == 'putm': # store master metadata self.callFactory(self.sendPUTM, commands, self.msgs) elif commandkey == 'getm': # retrieve master metadata self.callFactory(self.sendGETM, commands, self.msgs) elif commandkey == 'cred': # send encrypted private credentials to an email address # format: 'cred passphrase emailaddress' func = lambda: self.sendCRED( command[len(commands[0])+1:-len(commands[-1])-1], commands[-1]) self.callFactory(func, commands, self.msgs) # the following are diagnostic operations, debug-only utility elif commandkey == 'node': # list known nodes self.callFactory(self.sendDIAGNODE, commands, self.msgs) elif commandkey == 'buck': # show k-buckets self.callFactory(self.sendDIAGBKTS, commands, self.msgs) elif commandkey == 'stat': # show pending actions print self.pending elif commandkey == 'stor': # stor a block to a given node. format: 'stor host:port,fname' storcommands = commands[1].split(',') try: fileid = int(storcommands[1], 16) except: linkfile = fencode(long(hashfile(storcommands[1]),16)) if (os.path.islink(linkfile)): os.remove(linkfile) os.symlink(storcommands[1], linkfile) storcommands[1] = linkfile # XXX: delete this file when the command finishes commands[1] = "%s,%s" % (storcommands[0], storcommands[1]) func = lambda: self.sendDIAGSTOR(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'rtrv': # retrive a block from a given node. format: 'rtrv host:port,fname' func = lambda: self.sendDIAGRTRV(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'vrfy': # verify a block on a given node. # format: 'vrfy host:port:offset-length,fname' logger.debug("vrfy(%s)" % commands[1]) func = lambda: self.sendDIAGVRFY(commands[1]) self.callFactory(func, commands, self.msgs) elif commandkey == 'dlet': print "not yet implemented" elif commandkey == 'fndv': # try to retrieve a value from the DHT # format: 'fndv key' func = lambda: self.sendDIAGFNDV(commands[1]) self.callFactory(func, commands, self.msgs) elif command != "": reactor.callFromThread(self.queueError, None, self.msgs, "illegal command '%s'" % command) def queueResult(self, r, l, msg): logger.debug("got result %s" % msg) l.append((r, msg)) def queueError(self, r, l, msg): logger.debug("got error %s" % msg) if r: l.append((r.getErrorMessage(), msg)) else: l.append((None, msg)) def printHelp(self, helpDict): helpkeys = helpDict.keys() helpkeys.sort() for i in helpkeys: print "%s:\t %s" % (i, helpDict[i]) def promptLoop(self, r): for c in self.pending: for i in self.pending[c].keys(): if self.pending[c][i] == True: print "%s on %s completed successfully" % (c, i) self.pending[c].pop(i) elif self.pending[c][i] == False: print "%s on %s failed" % (c, i) self.pending[c].pop(i) else: print "%s on %s pending" % (c, i) while len(self.msgs) > 0: # this prints in reverse order, perhaps pop() all into a new list, # reverse, then print (errmsg, m) = self.msgs.pop() if errmsg: print "<- %s:\n%s" % (m, errmsg) else: print "<- %s" % m if self.quit: reactor.stop() else: d = threads.deferToThread(self.promptUser) d.addCallback(self.promptLoopDelayed) d.addErrback(self.err) def promptLoopDelayed(self, r): # give the reactor loop time to fire any quick cbs/ebs reactor.callLater(0.1, self.promptLoop, r) def clientConnectionLost(self, connector, reason): if not self.quit: LocalClientFactory.clientConnectionLost(self, connector, reason) def cleanup(self, msg): self.quit = True self.err(msg) def err(self, r): print "bah!: %s" % r reactor.stop() def main(): config = FludConfig() config.load(doLogging=False) logger.setLevel(logging.DEBUG) handler = logging.FileHandler('/tmp/fludclient.log') formatter = logging.Formatter('%(asctime)s %(filename)s:%(lineno)d' ' %(name)s %(levelname)s: %(message)s', datefmt='%H:%M:%S') handler.setFormatter(formatter) logger.addHandler(handler) factory = CmdClientFactory(config) if len(sys.argv) == 2: config.clientport = int(sys.argv[1]) print "connecting to localhost:%d" % config.clientport reactor.connectTCP('localhost', config.clientport, factory) factory.promptLoop(None) reactor.run() if __name__ == '__main__': main()
Python
import urlparse, os, types from twisted.web import client from twisted.internet import reactor, defer from twisted.python import failure """ HTTPMultipartDownloader.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. HTTPMultipartDownloader will download mulitple files from a multipart/related. Note that it does this by using the Content-ID and Content-Length headers in each multipart, and will fail if those are not present (this could be genericized to operate without those fields without too much effort) This code is modeled after twisted.web.client.HTTPDownloader, which is copyright 2001-2004 Twisted Matrix Laboratories, MIT licensed. """ class HTTPMultipartDownloader(client.HTTPDownloader): """Download multiple files, via multipart/related.""" protocol = client.HTTPPageDownloader value = None def __init__(self, url, dir, method='GET', postdata=None, headers=None, agent="Flud client", supportPartial=0): self.requestedPartial = 0 self.filenames = [] self.dir = dir client.HTTPClientFactory.__init__(self, url, method=method, postdata=postdata, headers=headers, agent=agent) self.deferred = defer.Deferred() self.waiting = 1 def gotHeaders(self, headers): if self.requestedPartial: contentRange = headers.get("content-range", None) if not contentRange: # server doesn't support partial requests, oh well self.requestedPartial = 0 return start, end, realLength = http.parseContentRange(contentRange[0]) if start != self.requestedPartial: # server is acting wierdly self.requestedPartial = 0 def openFile(self, partialContent): if partialContent: file = open(self.filename, 'rb+') file.seek(0, 2) else: file = open(self.filename, 'wb') self.filenames.append(self.filename) return file def pageStart(self, partialContent): """Called on page download start. @param partialContent: tells us if the download is partial download we requested. """ if partialContent and not self.requestedPartial: raise ValueError, "we shouldn't get partial content response if"\ " we didn't want it!" self.partialContent = partialContent if self.waiting: self.waiting = 0 self.inSubHeader = True self.file = None self.boundary = None def getSubHeader(self, data): newboundary = data[:data.find('\r\n')] data = data[len(newboundary)+2:] if not self.boundary: self.boundary = newboundary if self.boundary != newboundary: if self.boundary+"--" == newboundary: # end of multiparts return else: raise ValueError, "found illegal boundary" # XXX: print some of newboundary *safely* #raise ValueError, "found illegal boundary: %s, was %s" \ # % (newboundary[:80], self.boundary) headerEnd = data.find('\r\n\r\n') if headerEnd != -1: self.inSubHeader = False self.subHeaders = {} headers = data[:headerEnd].split('\r\n') for header in headers: k, v = header.split(':',1) self.subHeaders[k.lower()] = v.lstrip(' ') if not self.subHeaders.has_key('content-id'): raise ValueError, "no Content-ID field in multipart,"\ " can't continue" # XXX: need to check for badness (e.g, "../../) in content-id self.filename = os.path.join(self.dir, self.subHeaders['content-id']) self.file = self.openFile(self.partialContent) if not self.subHeaders.has_key('content-length'): raise ValueError, "no Content-Length field in multipart,"\ " can't continue" self.filesizeRemaining = int(self.subHeaders['content-length']) self.pagePart(data[headerEnd+4:]) def pagePart(self, data): if self.inSubHeader: self.getSubHeader(data) else: if not self.file: raise ValueError, "file %s not open for output" % self.filename try: if self.filesizeRemaining > len(data): self.file.write(data) self.filesizeRemaining -= len(data) else: self.file.write(data[:self.filesizeRemaining]) skipto = self.filesizeRemaining self.filesizeRemaining = 0 self.file.close() self.file = None self.inSubHeader = True self.getSubHeader(data[skipto+2:]) except IOError: #raise self.file = None self.deferred.errback(failure.Failure()) def pageEnd(self): if self.file: try: self.file.close() except IOError: self.deferred.errback(failure.Failure()) return self.deferred.callback(self.filenames) def doit(): factory = HTTPMultipartDownloader("/ret", "/tmp/") reactor.connectTCP('localhost', 1080, factory) return factory.deferred def didit(r): print "didit: %s" % str(r) reactor.stop() if __name__ == "__main__": # tries to request http://localhost:1080/ret, which it expects to be # multipart/related with Content-Length headers d = doit() d.addBoth(didit) reactor.run()
Python
#!/usr/bin/python """ FludClient.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. FludClient provides a GUI Client for interacting with FludNode. """ #from twisted.internet import wxreactor #wxreactor.install() import sys, os, string, time, glob import wx import wx.lib.mixins.listctrl as listmix import wx.lib.editor.editor from flud.protocol.LocalClient import * from flud.FludConfig import FludConfig from flud.CheckboxState import CheckboxState FLUSHCHECKTIME = 5*60 # s to wait to flush fludfile.conf datadir = os.path.dirname(os.path.abspath(__file__)) imgdir = os.path.join(datadir,'images') mimeMgr = wx.MimeTypesManager() def getFileIcon(file, il, checkboxes, icondict): ft = mimeMgr.GetFileTypeFromExtension(file[file.rfind('.')+1:]) # XXX: what about from mimetype or magic? if ft == None: return icondict['generic'] else: desc = ft.GetDescription() if icondict.has_key(desc): return icondict[desc] else: icon = ft.GetIcon() if icon == None or not icon.Ok(): #print "couldn't find an icon image for %s" % file icondict[desc] = icondict['generic'] return icondict[desc] bm = wx.BitmapFromIcon(icon) newimages = makeCheckboxBitmaps(bm, checkboxes) #il = self.GetImageList() pos = il.GetImageCount() for i in newimages: il.Add(i) icondict[desc] = pos #print "%s got a %s image" % (file, ft.GetDescription()) return pos def getEmptyBitmapAndDC(width, height): empty = wx.EmptyBitmap(width,height) temp_dc = wx.MemoryDC() temp_dc.SelectObject(empty) temp_dc.Clear() return (empty, temp_dc) def makeCheckboxBitmaps(basebitmap, checkboxes): if basebitmap.GetWidth() != 16 or basebitmap.GetHeight() != 16: img = basebitmap.ConvertToImage() img.Rescale(16, 16) basebitmap = img.ConvertToBitmap() result = [] for i in checkboxes: bm, dc = getEmptyBitmapAndDC(40,16) dc.DrawBitmap(basebitmap, 0, 0, False) dc.DrawBitmap(i, 20, 2, False) result.append(bm) return result def createDefaultImageList(): def getDefaultCheckboxes(): ucbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-unchecked1.png"))) cbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checked1.png"))) ccbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checkedpartial1.png"))) cpbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-parentchecked1.png"))) ebm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excluded1.png"))) ecbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excludedpartial1.png"))) return (ucbm, cbm, ccbm, cpbm, ebm, ecbm) checkboxes = getDefaultCheckboxes() il = wx.ImageList(40, 16) folderimgs = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_FOLDER, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) computer = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_HARDDISK, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) drives = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_HARDDISK, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) cdrom = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_CDROM, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) floppy = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_FLOPPY, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) removable = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_REMOVABLE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) genericfile = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_NORMAL_FILE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) execfile = makeCheckboxBitmaps(wx.ArtProvider_GetBitmap( wx.ART_EXECUTABLE_FILE, wx.ART_CMN_DIALOG, wx.Size(16, 16)), checkboxes) j = 0 icondict = {} icondict['folder'] = j for i in folderimgs: il.Add(i) j = j+1 icondict['computer'] = j for i in computer: il.Add(i) j = j+1 icondict['drives'] = j for i in drives: il.Add(i) j = j+1 icondict['cdrom'] = j for i in cdrom: il.Add(i) j = j+1 icondict['floppy'] = j for i in floppy: il.Add(i) j = j+1 icondict['removable'] = j for i in removable: il.Add(i) j = j+1 icondict['generic'] = j for i in genericfile: il.Add(i) j = j+1 icondict['exec'] = j for i in execfile: il.Add(i) j = j+1 return il, checkboxes, icondict class DirCheckboxCtrl(wx.TreeCtrl): def __init__(self, parent, id=-1, dir=None, pos=wx.DefaultPosition, size=wx.Size(300,300), #wx.DefaultSize, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER), validator=wx.DefaultValidator, name=wx.ControlNameStr, allowExclude=True): self.allowExclude = allowExclude wx.TreeCtrl.__init__(self, parent, id, pos, size, style, validator, name) self.listeners = [] self.parent = parent #self.il = self.GetImageList() #self.checkboxes = self.getDefaultCheckboxes() self.initTree(dir) self.Bind(wx.EVT_TREE_ITEM_EXPANDING, self.onExpand, self) self.Bind(wx.EVT_LEFT_UP, self.onClick, self) self.Bind(wx.EVT_TREE_ITEM_GETTOOLTIP, self.onTooltip, self) self.Bind(wx.EVT_CHAR, self.onChar) def initTree(self, dir): self.expandRoot(dir) # XXX: should expandHome() on first run, then load expanded dir state # from saved state on subsequent runs. self.expandHome(dir) def expandRoot(self, dir): if not os.path.isdir(dir): raise ValueError("%s is not a valid directory path") self.defaultImageList, self.checkboxes, self.icondict \ = createDefaultImageList() self.AssignImageList(self.defaultImageList) self.il = self.GetImageList() if dir == None: self.rootID = self.AddRoot(dir, self.icondict['computer'], -1, wx.TreeItemData((dir, True, False, CheckboxState.UNSELECTED))) # XXX: getTopLevelDirs() and add them as children else: self.rootID = self.AddRoot(dir, self.icondict['folder'], -1, wx.TreeItemData((dir, True, False, CheckboxState.UNSELECTED))) self.expandDir(self.rootID) self.Expand(self.rootID) self.stateChangeTime = time.time() self.flushTime = time.time() reactor.callLater(FLUSHCHECKTIME, self.checkFlush) def expandHome(self, dir): home = os.environ['HOME'] if home: traversal = home.split(os.path.sep)[1:] node = self.rootID for d in traversal: (ipath, isdir, expanded, istate) \ = self.GetItemData(node).GetData() self.expandDir(node) children = self.getChildren(node, False) childrennames = [self.GetItemText(x) for x in children] if d in childrennames: p = childrennames.index(d) node = children[p] self.expandDir(node) self.Expand(node) else: print "couldn't traverse to HOME dir on %s" % d break def checkFlush(self): print "checking for flush" if self.stateChangeTime > self.flushTime: self.flushTime = time.time() print "flushing" self.parent.flushFileConfig() reactor.callLater(FLUSHCHECKTIME, self.checkFlush) def expandDir(self, parentID, hideHidden=False, busycursor=True): def isDriveAvailable(path): if len(path) == 2 and path[1] == ':': path = path.lower() if path[0] == 'a' or path[0] == 'b' or diExists(path): return True else: return False return True (path, isDir, expanded, state) = self.GetItemData(parentID).GetData() if expanded: return if not isDriveAvailable(path): return if busycursor: wx.BusyCursor() try: dirlist = os.listdir(path) except: self.SetItemHasChildren(parentID, False) return if len(dirlist) == 0: self.SetItemHasChildren(parentID, False) return dirs = [] files = [] for i in dirlist: if hideHidden and i[0] == '.': # XXX: dotfile format check is *nix specific # XXX: if this is a hidden file, don't add it. pass elif os.path.isdir(os.path.join(path,i)): dirs.append(i) else: files.append(i) dirs.sort() files.sort() for d in dirs: child = self.AppendItem(parentID, d) self.SetPyData(child, (os.path.join(path,d), True, False, 0)) self.SetItemImage(child, self.icondict['folder'], wx.TreeItemIcon_Normal) self.SetItemHasChildren(child) il = self.GetImageList() for f in files: child = self.AppendItem(parentID, f) # XXX: unicode? self.SetPyData(child, (os.path.join(path,f), False, False, 0)) idx = getFileIcon(os.path.join(path,f), il, self.checkboxes, self.icondict) self.SetItemImage(child, idx, wx.TreeItemIcon_Normal) self.SetPyData(parentID, (path, isDir, True, state)) def getStates(self, node=None): if not node: node = self.rootID states = {} (path, isDir, expanded, state) = self.GetItemData(node).GetData() if state in [CheckboxState.SELECTED, CheckboxState.EXCLUDED]: states[path] = state children = self.getChildren(node, False) for child in children: states.update(self.getStates(child)) return states def setStates(self, states): for i in states: found = self.findNode(i) if found: self.setItemState(found, states[i]) def findNode(self, path): if path[0] == '/': path = path[1:] # XXX: unix only traversal = path.split(os.path.sep) if traversal[0] == '': traversal.remove('') node = self.rootID while True: (ipath, isdir, expanded, istate) = self.GetItemData(node).GetData() if len(traversal) == 0: return node self.expandDir(node) children = self.getChildren(node, False) childrennames = [self.GetItemText(x) for x in children] firstpath = traversal[0] if firstpath in childrennames: p = childrennames.index(firstpath) node = children[p] traversal.remove(firstpath) else: #print " the file %s is no longer present!" % path return None return None def onExpand(self, event): self.expandDir(event.GetItem()) self.renderChildren(event.GetItem(), True) def getFullPath(self, node): path = self.tree.GetItemText(node) n = node while True: n = self.tree.GetItemParent(n) if n and n != self.GetRootItem(): path = os.path.join(self.tree.GetItemText(n),path) else: break return path def renderParents(self, item): if item == self.rootID: return n = item (path, isDir, expanded, state) = self.GetItemData(item).GetData() while True: n = self.GetItemParent(n) (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(n).GetData() #print "parent %s" % parentpath if n and n != self.GetRootItem(): newstate = parentstate if parentstate != CheckboxState.UNSELECTED and \ parentstate != CheckboxState.SELECTEDPARENT: # we only care about changing UNSELECT or SELECTEDPARENT # states break else: if state == CheckboxState.SELECTED or \ state == CheckboxState.SELECTEDCHILD or \ state == CheckboxState.SELECTEDPARENT: # if the item (child) is selected in any way, parent # should be too. newstate = CheckboxState.SELECTEDPARENT elif state == CheckboxState.UNSELECTED or \ state == CheckboxState.EXCLUDED: # if the item (child) is unselected or excluded, the # parent should be too, /unless/ there are other # children at the same level who are selected. children = self.getChildren(n, False) newstate = CheckboxState.UNSELECTED for child in children: (cpath, cisdir, cexp, cstate) = self.GetItemData(child).GetData() if cstate == CheckboxState.SELECTED or \ cstate == CheckboxState.SELECTEDCHILD or \ cstate == CheckboxState.SELECTEDPARENT: newstate = parentstate if newstate == parentstate: break imageidx = self.GetItemImage(n) imageidx += CheckboxState.offset(parentstate, newstate) self.SetPyData(n, (parentpath, parentisDir, parentexpanded, newstate)) self.SetItemImage(n, imageidx) else: break def renderChildren(self, parent, recurse=False): (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(parent).GetData() children = self.getChildren(parent, False) for child in children: #path = self.getFullPath(child) (path, isDir, expanded, state) = self.GetItemData(child).GetData() imageidx = self.GetItemImage(child) newstate = state """ Here are the state transitions for children based on current states: ('-' = no state change, 'x' = should never occur, '!' = should be prevented at the parent, '?' = need to consult children) child unsel sel selch selpar excl exclch unsel - ! unsel x - unsel sel selch - - selch - selch par selch selch - - selch - selch selpar x x unsl?selpr x x x excl exlch ! exlch ! - - exclch exlch - exlch ! - - """ #if parentpath == '/data': # print "/data pstate = %d" % parentstate # print " %s = %d" % (path, state) if state == CheckboxState.UNSELECTED: if parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.SELECTEDCHILD: if parentstate == CheckboxState.UNSELECTED: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTEDPARENT: if self.checkChildrenStates(child, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT]): # XXX: did we need to pass in selections to checkChldSt newstate = CheckboxState.SELECTEDPARENT else: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.SELECTEDPARENT: if parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif state == CheckboxState.EXCLUDEDCHILD: if parentstate == CheckboxState.UNSELECTED: newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD imageidx += CheckboxState.offset(state, newstate) self.SetPyData(child, (path, isDir, expanded, newstate)) self.SetItemImage(child, imageidx) if recurse: self.renderChildren(child, recurse) # XXX: why do we renderParents here? It hits the same # 'parent's over and over and over again. If we want to do # this, we need to 'collect up' the parents and just call once # -- this kills performance. #print "renderParents(%s)" % path #self.renderParents(child) def getChildren(self, node, recurse=False): result = [] child, cookie = self.GetFirstChild(node) while child: result.append(child) if recurse: result.extend(self.getChildren(child, recurse)) child, cookie = self.GetNextChild(node, cookie) return result def checkChildrenStates(self, node, states, ignorelist=[]): children = self.getChildren(node) for child in children: if child not in ignorelist: (p, d, e, childstate) = self.GetItemData(child).GetData() for state in states: if state == childstate: #print "%s has state %d" % (p, state) return True if self.checkChildrenStates(child, states, ignorelist): # do this even if it is in ignorelist, because it may have # children which are not in the ignorelist return True return False def getTooltip(self, item): text = self.GetItemText(item) (path, isDir, expanded, state) = self.GetItemData(item).GetData() if state == CheckboxState.SELECTED: if isDir: text = "'%s' is SELECTED for backup\n" \ "ALL files within this folder will be backed up\n" \ "(except those explicitly marked for exclusion)" % text else: text = "'%s' is SELECTED for backup" % text elif state == CheckboxState.UNSELECTED: text = "'%s' is NOT selected for backup" % text elif state == CheckboxState.SELECTEDPARENT: text = "some files within '%s' are selected for backup" % text elif state == CheckboxState.SELECTEDCHILD: text = "'%s' will be backed up\n" \ "(one of its parent folders is selected)" % text elif state == CheckboxState.EXCLUDED: if isDir: text = "'%s' is EXCLUDED from backup\n" \ "No files within this folder will be backed up" % text else: text = "'%s' is EXCLUDED from backup" % text elif state == CheckboxState.EXCLUDEDCHILD: text = "'%s' is EXCLUDED from backup\n" \ "(one of its parent folders is EXCLUDED)" % text return text def onTooltip(self, event): item = event.GetItem() text = self.getTooltip(item) if text: event.SetToolTip(text) else: event.StopPropagation() #print dir(event) def onClick(self, event): point = (event.GetX(), event.GetY()) item, flags = self.HitTest(point) if flags & wx.TREE_HITTEST_ONITEMICON: selections = self.GetSelections() self.changeState(item, selections) def onChar(self, event): if event.KeyCode() == ord('F') and event.ShiftDown() \ and event.ControlDown(): self.flushTime = time.time() print "flushing" self.parent.flushFileConfig() event.Skip() def changeState(self, item, selections=[]): self.stateChangeTime = time.time() (path, isDir, expanded, state) = self.GetItemData(item).GetData() if item == self.rootID: parent = None parentstate = CheckboxState.UNSELECTED else: parent = self.GetItemParent(item) (parentpath, parentisDir, parentexpanded, parentstate) = self.GetItemData(parent).GetData() imageidx = self.GetItemImage(item) # determine newstate from existing state, parent state, and state # of children """ Here are the state transitions for the item based on current states and parent states: ('-' = no state change, 'x' = should never occur, '?' = depends on children state) item unsel sel selch selpar excl exclch unsel sel excl sel sel unsel excl sel sel excl?selpar sel x selch excl par selch x excl sel sel selch excl selpar sel excl x sel unsel excl excl x excl x exclch exclch excl exclch x excl x exclch exclch excl """ newstate = state if state == CheckboxState.UNSELECTED: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTEDPARENT: if parentstate == CheckboxState.EXCLUDED or \ parentstate == CheckboxState.EXCLUDEDCHILD: # XXX: this should be impossible to reach... newstate = CheckboxState.EXCLUDEDCHILD else: newstate = CheckboxState.SELECTED elif state == CheckboxState.SELECTED: if self.checkChildrenStates(item, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT], selections): newstate = CheckboxState.SELECTEDPARENT elif self.allowExclude: newstate = CheckboxState.EXCLUDED else: if parent in selections or \ (parentstate == CheckboxState.UNSELECTED or \ parentstate == CheckboxState.SELECTEDPARENT): newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD elif state == CheckboxState.EXCLUDED: if parent in selections or \ (parentstate == CheckboxState.UNSELECTED or \ parentstate == CheckboxState.SELECTEDPARENT): newstate = CheckboxState.UNSELECTED elif parentstate == CheckboxState.SELECTED or \ parentstate == CheckboxState.SELECTEDCHILD: newstate = CheckboxState.SELECTEDCHILD else: newstate = CheckboxState.EXCLUDEDCHILD elif state == CheckboxState.EXCLUDEDCHILD: newstate = CheckboxState.EXCLUDED if len(selections) > 1: # if we have multiple selections, the idea is to move all the # selections to the newstate defined above, or to valid # unselected or inherited states if the move to newstate would # be invalid. """ Here are the state transitions for the item based on the newstate as determined by the clicked item and the current states: ('-' = no state change, '?' = consult children) item unsel sel selch selpar excl exclch unsel - unsel - - unsel - sel sel - sel sel sel - newstate selch - unsel - - unsel - selpar - unsel - - unsel - excl excl excl?slpr excl excl - excl exclch - unsel - - unsel - """ for i in selections: (mpath, misDir, mexpanded, mstate) = self.GetItemData( i).GetData() mnewstate = mstate if mstate == CheckboxState.UNSELECTED or \ mstate == CheckboxState.SELECTEDCHILD or \ mstate == CheckboxState.SELECTEDPARENT: if newstate == CheckboxState.SELECTED or \ newstate == CheckboxState.EXCLUDED: mnewstate = newstate elif mstate == CheckboxState.SELECTED: if newstate == CheckboxState.UNSELECTED or \ newstate == CheckboxState.SELECTEDCHILD or \ newstate == CheckboxState.SELECTEDPARENT or \ newstate == CheckboxState.EXCLUDEDCHILD: mnewstate = CheckboxState.UNSELECTED elif newstate == CheckboxState.EXCLUDED: if self.checkChildrenStates(i, [CheckboxState.SELECTED, CheckboxState.SELECTEDPARENT], selections): mnewstate = CheckboxState.SELECTEDPARENT else: mnewstate = newstate elif mstate == CheckboxState.EXCLUDED: if newstate == CheckboxState.UNSELECTED or \ newstate == CheckboxState.SELECTEDCHILD or \ newstate == CheckboxState.SELECTEDPARENT or \ newstate == CheckboxState.EXCLUDEDCHILD: mnewstate = CheckboxState.UNSELECTED elif newstate == CheckboxState.SELECTED: mnewstate = newstate elif mstate == CheckboxState.EXCLUDEDCHILD: if newstate == CheckboxState.EXCLUDED: mnewstate = newstate self.setItemState(i, mnewstate) self.setItemState(item, newstate, (path, isDir, expanded, state, imageidx)) def setItemState(self, item, newstate, oldData=None): if oldData: (path, isDir, expanded, state, imageidx) = oldData else: (path, isDir, expanded, state) = self.GetItemData(item).GetData() imageidx = self.GetItemImage(item) imageidx += CheckboxState.offset(state, newstate) self.SetPyData(item, (path, isDir, expanded, newstate)) self.SetItemImage(item, imageidx) self.renderChildren(item, True) self.renderParents(item) def getTopLevelDrives(self): sys = platform.system() if sys == 'Windows': # XXX: need to test this all out import win32api, string drives = win32api.GetLogicalDriveStrings() driveletters = string.splitfields(drives,'\000') for d in driveletters: type = win32api.GetDriveType("%s:\\" % d) # XXX: set the appropriate icon return driveletters else: # Unix, OSX, etc. return ['/'] def addListener(self, callback): self.listeners.append(callback) def SetPyData(self, item, data): wx.TreeCtrl.SetPyData(self, item, data) for f in self.listeners: f(item, data) """ Tests for DirCheckboxCtrl A number of unit tests must be performed on the DirCheckboxGUI widget when refactoring. Add to this list so that it becomes comprehensive. Basic Tests: 1. Click on a top-level UNSELECTED object in the tree [should become SELECTED]. - Click again [should become EXCLUDED]. - Click again [should become UNSELECTED]. 2. Click on a non-top-level UNSELECTED object in the tree that has no SELECTED children [should become SELECTED, it's parents should become SELECTEDPARENT and its children SELECTEDCHILD]. - Click again [should become EXCLUDED, it's parents who were SELECTEDPARENT should become UNSELECTED, and it's UNSELECTED children should become EXCLUDED]. - Click again [should become UNSELECTED, and it's children should become UNSELECTED]. 3. Change two children to their SELECTED state [parents should be in SELECTEDPARENT state]. - Click one child to become EXCLUDED [parents should stay in SELECTEDPARENT] - Click the same child to become UNSELECTED [parents should stay in SELECTEDPARENT] - Click the other child to become EXCLUDED [parents should become UNSELECTED] 4. Choose a folder and a child item. - Click the child to become SEL [parent should be SELPAR] - Click the parent [parent should become SEL] - Click the parent again [parent should become SELPAR] 5. Choose a folder and a child item. - Click the parent to become SEL [child should become SELCHILD] - Click the child [child should become SEL] - Click the child again [child should become EXCL] - Click the child again [child should become SELCHILD] 6. Pick a node with children at least two-deep. Change two of the at-least-two-deep children to their SELECTED state [parents should be in SELECTEDPARENT state]. - Click parent closest to SELECTED children to SELECTED [two childen remain in SELECTED, all other children become SELECTEDCHILD. Parent[s] of parent remain SELECTEDPARENT] - Click one child twice to become SELECTEDCHILD [child should not be able to be UNSELECTED, parent states should not change] - Click other child twice to become SELECTEDCHILD [child should not be able to be UNSELECTED, parent states should not change] 7. Pick a node with children at least two-deep. - Click deepest parent to SELECTED [Parent[s] of parent become SELECTEDPARENT] - Click same parent again to become EXCLUDED [Parent[s] of parent become UNSELECTED] - Click same parent again to become UNSELECTED [Parent[s] of parent remain UNSELECTED] 8. Pick a node with children at least two-deep. - Click deepest child to become SELECTED [Parent[s] of parent become SELECTEDPARENT] - Click the topmost parent to become SELECTED [children become SELECTEDCHILD] - Click the topmost parent again to become SELECTEDPARENT [middle child should become SELECTEDPARENT] Multi-Selection Tests: 1. Multi-select three items at the same level and in the same state. Toggle between the three main states [SELECTED, EXCLUDED, UNSELECTED] 2. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the SEL item to see that all three items become EXCL. 3. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the EXCL item to see that all three items become UNSEL. 4. Multi-select three items at the same level, one in each of the three states (SEL, EXCL, UNSEL). Toggle the UNSEL item to see that all three items become SEL. 5. Choose three items that are nested within each other: a parent folder, one of its children folders, and a file/folder in the child folder. Choose one other item from the child folder. - set the top parent to UNSEL - set the child folder to SEL [parent become SELPAR] - set the child item to SEL - set the other item to EXCL - multi-select all four items - 5A. click on the top parent (which was in SELPAR) [All four items should become SEL, all children of any of these items should become SELCHILD]. Toggle twice more [all selected items should toggle to EXCL, then to UNSEL] - 5B. reset as above, click on the child folder [All four items should become EXCL]. Toggle twice more [all selected items should go to UNSEL, then SEL] - 5C. reset as above, click on the child item [All four items should become EXCL]. Toggle twice more [all selected items should go to UNSEL, then SEL] - 5D. reset as above, click on the other item [All four items should become UNSEL]. Toggle twice more [all selected items should go to SEL, then EXCL] 6. Choose a folder, one if its subfolders, a subfolder of the subfolder, and an item in the deepest subfolder, and an item in the first subfolder, e.g.: [] A [] B [] C [] D [] E - change item 'D' to SEL [parents 'A', 'B', and 'C' should go to SELPAR] - change item 'E' to EXCL - multi-select 'A', 'C', and 'E' - toggle 'E' to UNSEL [all other selections should stay in current state] - toggle 'E' to SEL ['A' and 'B' become SEL, their children become SELCHILD] - toggle 'E' back to EXCL [should get our original multi-select setup back] - toggle 'C' to SEL [all selections to SEL, children to SELCHILD] - toggle 'C' to SELPAR ['A' and 'C' to SELPAR, 'E' to UNSEL] - toggle 'E' twice [should get our original mulit-select setup back] """ class CheckFileListCtrlMixin: # for some insane reason, we can't get EVT_LEFT_DOWN (or _UP) to bind in # FileListCtrl itself. But we are sneaky and can do it by lots of clever # hax0ry, like by using this silly mixin. def __init__(self, toCall): self.Bind(wx.EVT_LEFT_UP, toCall) class FileListCtrl(wx.ListCtrl, listmix.ListCtrlAutoWidthMixin, CheckFileListCtrlMixin): """ Implements a file list control, with a peerctrl that contains the filesystem model. Currently, this peerctrl must implement an addListener(), changeState(), GetItemData(), expandDir(), GetSelections(), and GetChildren() API similar to that implemented by DirCheckBoxCtrl. """ def __init__(self, parent, peerctrl, id=-1, pos=wx.DefaultPosition, size=wx.DefaultSize, style=wx.LC_REPORT, validator=wx.DefaultValidator, name=wx.ListCtrlNameStr): wx.ListCtrl.__init__(self, parent, id, pos, size, style, validator, name) CheckFileListCtrlMixin.__init__(self, self.OnClick) listmix.ListCtrlAutoWidthMixin.__init__(self) self.peerctrl = peerctrl self.peerctrl.addListener(self.itemChanged) self.itemdict = {} # a dict with filepath as key, containing tuples of # (index into ListCtrl, reference to peerctrl object) self.stopsearch = False self.il, self.checkboxes, self.icondict = createDefaultImageList() self.AssignImageList(self.il, wx.IMAGE_LIST_SMALL) self.il = self.GetImageList(wx.IMAGE_LIST_SMALL) self.InsertColumn(0, "Filename") self.InsertColumn(1, "Location") #self.InsertColumn(2, "Last Backup") #self.SetColumnWidth(0, wx.LIST_AUTOSIZE) #self.SetColumnWidth(1, -1) #wx.LIST_AUTOSIZE) self.Bind(wx.EVT_MOTION, self.mouseMotion) self.searchSourceItems = [] def itemChanged(self, item, data): (path, isDir, expanded, state) = data if self.itemdict.has_key(path): item = self.itemdict[path][0] image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) def GetAll(self, excludeStates=[]): result = [] start = -1 for i in range(self.GetItemCount()): item = self.GetNextItem(start, wx.LIST_NEXT_ALL) # XXX: only append if not in excludeStates result.append(item) start = item return result def GetSelections(self): result = [] start = -1 for i in range(self.GetSelectedItemCount()): item = self.GetNextItem(start, wx.LIST_NEXT_ALL, wx.LIST_STATE_SELECTED) result.append(item) start = item return result def GetPeerSelections(self, selections): result = [] for item in selections: path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) if self.itemdict.has_key(path): result.append(self.itemdict[path][1]) return result def mouseMotion(self, event): point = event.GetPosition() item, flags = self.HitTest(point) if flags == wx.LIST_HITTEST_ONITEMICON: path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) text = self.peerctrl.getTooltip(self.itemdict[path][1]) tip = wx.ToolTip(text) self.SetToolTip(tip) #tipwin = tip.GetWindow() #tippos = tipwin.GetPosition() #print "%s vs %s" % (tippos, point) #tipwin.SetPosition(point) def OnClick(self, event): point = event.GetPosition() item, flags = self.HitTest(point) if flags == wx.LIST_HITTEST_ONITEMICON: peerselections = self.GetPeerSelections(self.GetSelections()) path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) ditem = self.itemdict[path][1] # raises if not present self.peerctrl.changeState(ditem, peerselections) (path, isDir, expanded, state) \ = self.peerctrl.GetItemData(ditem).GetData() image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) def searchButtonAction(self, event): selections = self.peerctrl.GetSelections() if len(selections) == 0: return ("Please tell me where to search. Select one or more" " folders in the left-hand panel (hold down SHIFT or" " CTRL for multiple selection), then click the 'find!'" " button again.", None) else: self.DeleteAllItems() self.itemdict = {} b = wx.BusyCursor() searchSourceItems = [] for i in selections: self.addResults(i, event.searchstring) searchSourceItems.append(i) self.searchSourceItems = [self.peerctrl.GetItemData(s).GetData()[0] for s in searchSourceItems] print "sources: %s" % self.searchSourceItems return ("Search results will appear as files that match your" " search are found.", None) return (None, None) def addResults(self, ditem, searchstring): (path, isDir, expanded, state) \ = self.peerctrl.GetItemData(ditem).GetData() position = self.GetItemCount() if isDir: if not expanded: self.peerctrl.expandDir(ditem, busycursor=False) children = self.peerctrl.getChildren(ditem) for c in children: self.addResults(c, searchstring) wx.Yield() if self.stopsearch: break else: terms = [x for x in searchstring.split(' ') if x != ''] for term in terms: print path if path.find(term) > 0: image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state dirname, filename = os.path.split(path) index = self.InsertImageStringItem(position, filename, image) self.SetStringItem(index, 1, dirname) self.SetColumnWidth(0, wx.LIST_AUTOSIZE) self.itemdict[path] = (index, ditem) break def setGroup(self, state): items = self.GetAll() item = items[0] peerselections = self.GetPeerSelections(items) path = os.path.join(self.GetItem(item,1).GetText(), self.GetItemText(item)) ditem = self.itemdict[path][1] # raises if not present while True: # cycle until the items state matches the desired state self.peerctrl.changeState(ditem, peerselections) # can be slow (path, isDir, expanded, nstate) \ = self.peerctrl.GetItemData(ditem).GetData() if nstate == state: break image = getFileIcon(path, self.il, self.checkboxes, self.icondict) + state self.SetItemImage(item, image) return self.searchSourceItems class GroupSelectionCheckbox(wx.Panel): def __init__(self, parent, id=-1, setGroupState=None): wx.Panel.__init__(self, parent, id) self.setGroupState = setGroupState self.ubm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-unchecked1.png"))) self.cbm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-checked1.png"))) self.ebm = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "checkbox-excluded1.png"))) self.checkboxButton = wx.BitmapButton(self, -1, self.ubm, style=wx.NO_BORDER) self.Bind(wx.EVT_BUTTON, self.onCheckbox, self.checkboxButton) self.description = wx.StaticText(self, -1, "always BACKUP any files that match these search criteria ") self.state = CheckboxState.UNSELECTED self.gbSizer = wx.GridBagSizer(1,2) self.gbSizer.Add(self.checkboxButton, (0,0), flag=wx.ALIGN_CENTER) self.gbSizer.Add(self.description, (0,1), flag=wx.ALIGN_CENTER) self.gbSizer.AddGrowableRow(1) self.SetSizerAndFit(self.gbSizer) def Enable(self, enable=True): self.checkboxButton.Enable(enable) self.description.Enable(enable) def Disable(self): self.Enable(False) def clear(self): self.checkboxButton.SetBitmapLabel(self.ubm) self.state = CheckboxState.UNSELECTED self.description.SetLabel( "always BACKUP any files that match these search criteria") def setState(self, state): self.state = state if self.state == CheckboxState.UNSELECTED: self.checkboxButton.SetBitmapLabel(self.ubm) self.description.SetLabel( "always BACKUP any files that match these search criteria") elif self.state == CheckboxState.SELECTED: self.checkboxButton.SetBitmapLabel(self.cbm) self.description.SetLabel( "always BACKUP any files that match these search criteria") elif self.state == CheckboxState.EXCLUDED: self.checkboxButton.SetBitmapLabel(self.ebm) self.description.SetLabel( "always EXCLUDE any files that match these search criteria") def onCheckbox(self, event): if self.state == CheckboxState.UNSELECTED: self.checkboxButton.SetBitmapLabel(self.cbm) self.state = CheckboxState.SELECTED if self.setGroupState: self.setGroupState(CheckboxState.SELECTED) elif self.state == CheckboxState.SELECTED: self.checkboxButton.SetBitmapLabel(self.ebm) self.description.SetLabel( "always EXCLUDE any files that match these search criteria") self.state = CheckboxState.EXCLUDED if self.setGroupState: self.setGroupState(CheckboxState.EXCLUDED) elif self.state == CheckboxState.EXCLUDED: self.checkboxButton.SetBitmapLabel(self.ubm) self.description.SetLabel( "always BACKUP any files that match these search criteria") self.state = CheckboxState.UNSELECTED if self.setGroupState: self.setGroupState(CheckboxState.UNSELECTED) class SearchPanel(wx.Panel): def __init__(self, parent, dircheckbox, id=-1, searchButtonAction=None): wx.Panel.__init__(self, parent, id) self.dircheckbox = dircheckbox self.searchButtonAction = searchButtonAction self.SetAutoLayout(False) self.rules = {} # should refer to something from fludrules self.searchField = wx.TextCtrl(self, -1, "search for files to backup here", size=wx.Size(-1,-1), style=wx.TE_PROCESS_ENTER) self.searchField.SetToolTipString('find files within directories' ' selected to the left by entering search terms here') self.searchField.Bind(wx.EVT_TEXT_ENTER, self.onSearchClick) self.searchField.Bind(wx.EVT_LEFT_DOWN, self.selectAllText) self.searchField.Bind(wx.EVT_KILL_FOCUS, self.unfocused) self.searchButton = wx.Button(self, -1, 'find!', name='searchButton') self.Bind(wx.EVT_BUTTON, self.onSearchClick, self.searchButton) self.searchResults = FileListCtrl(self, dircheckbox, -1, name='searchResults', style=wx.SUNKEN_BORDER | wx.LC_REPORT) self.searchResults.SetExtraStyle(0) self.searchResults.SetLabel('found files') self.groupSelection = GroupSelectionCheckbox(self, -1, self.setGroup) self.groupSelection.Disable() self.gbSizer = wx.GridBagSizer(3,2) self.gbSizer.Add(self.searchField, (0,0), flag=wx.EXPAND) self.gbSizer.Add(self.searchButton, (0,1)) self.gbSizer.Add(self.searchResults, (1,0), (1,2), flag=wx.EXPAND|wx.TOP, border=5) self.gbSizer.Add(self.groupSelection, (2,0) ) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) def onSearchClick(self, event): event.searchstring = self.searchField.GetValue() if self.searchButton.GetLabel() == 'stop!': self.searchButton.SetLabel('find!') self.searchResults.stopsearch = True return else: self.groupSelection.clear() self.groupSelection.Disable() self.searchButton.SetLabel('stop!') self.searchButton.Update() err, info = self.searchResults.searchButtonAction(event) selections = self.searchResults.searchSourceItems # see if we should set the checkbox button from a previous rule state = None if len(selections) > 0 and self.rules.has_key(selections[0]): rule = self.rules[selections[0]] if self.rules[selections[0]].has_key(event.searchstring): state = self.rules[selections[0]][event.searchstring] for i in selections: if not self.rules.has_key(i) or self.rules[i] != rule: state = None break #for j in self.rules[i]: if state: print "should restore checkbox to %s" % state self.groupSelection.setState(state) self.searchButton.SetLabel('find!') self.searchResults.stopsearch = False if self.searchButtonAction: self.searchButtonAction(event, errmsg=err, infomsg=info) self.groupSelection.Enable() def selectAllText(self, event): if wx.Window.FindFocus() != self.searchField: self.searchField.SetSelection(-1,-1) self.searchField.SetFocus() else: self.searchField.SetSelection(0,0) event.Skip() def unfocused(self, event): self.searchField.SetSelection(0,0) def setGroup(self, state): b = wx.BusyCursor() selections = self.searchResults.setGroup(state) for s in selections: if not self.rules.has_key(s): self.rules[s] = {} if state == CheckboxState.UNSELECTED: try: self.rules.pop(s) except: pass else: self.rules[s][self.searchField.GetValue()] = state print self.rules class FilePanel(wx.SplitterWindow): def __init__(self, parent, searchButtonAction=None): # Use the WANTS_CHARS style so the panel doesn't eat the Return key. wx.SplitterWindow.__init__(self, parent, -1, style=wx.SP_LIVE_UPDATE | wx.CLIP_CHILDREN | wx.WANTS_CHARS) self.Bind(wx.EVT_SIZE, self.OnSize) self.SetNeedUpdating(True) print self.GetSize() self.tree = DirCheckboxCtrl(self, -1, dir="/") # XXX: fludrules.init path should be in config self.fludrules = self.getFludHome()+"/fludrules.init" if not os.path.isfile(self.fludrules): # XXX: do the other first time stuff (email encrypted credentials, # etc.) parent.SetMessage("Welcome. This appears to be the first" " time you've run flud. We've automatically selected some" " files for backup. You can make changes by" " selecting/deselecting files and directories. When you are" " done, simply close this window.") src = open(os.path.join(datadir,'fludrules.init'), 'r') dst = open(self.fludrules, 'w') filerules = src.read() dst.write(filerules) dst.close() src.close() filerules = eval(filerules) rulestates = {} for rule in filerules['baserules']: value = filerules['baserules'][rule] rule = glob.glob(os.path.expandvars(rule)) for r in rule: rulestates[r] = value self.tree.setStates(rulestates) # XXX: fludfile.conf path should be in config self.fludfiles = self.getFludHome()+"/fludfile.conf" print self.fludfiles if os.path.isfile(self.fludfiles): file = open(self.fludfiles, 'r') states = eval(file.read()) self.tree.setStates(states) file.close() self.searchPanel = SearchPanel(self, dircheckbox=self.tree, searchButtonAction=searchButtonAction) self.SetMinimumPaneSize(50) self.SplitVertically(self.tree, self.searchPanel) #, 300) print self.GetSize() def getFludHome(self): if os.environ.has_key('FLUDHOME'): fludhome = os.environ['FLUDHOME'] else: fludhome = os.environ['HOME']+"/.flud" if not os.path.isdir(fludhome): os.mkdir(fludhome, 0700) return fludhome def shutdown(self, event): self.flushFileConfig() event.Skip() def flushFileConfig(self): states = self.tree.getStates() f = open(self.fludfiles, 'w') f.write(str(states)) f.close() for i in states: print "%s %s" % (i, states[i]) def OnSize(self, event): w,h = self.GetClientSizeTuple() if self.tree: self.tree.SetDimensions(0, 0, w, h) event.Skip() class RestoreCheckboxCtrl(DirCheckboxCtrl): # XXX: child/parent selection/deselection isn't quite right still, esp wrt # root node. repro: # -/ # -d1 # -f1 # -d2 # -d3 # -f2 # -f3 # with nothing selected, select d3 and f3, then select root, then deselect # d3 and f3 def __init__(self, parent, id=-1, config=None, pos=wx.DefaultPosition, size=wx.DefaultSize, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER), validator=wx.DefaultValidator, name=wx.ControlNameStr): self.config = config DirCheckboxCtrl.__init__(self, parent, id, config, pos, size, style, validator, name, allowExclude=False) def initTree(self, config): self.expandRoot(config) self.expandUntilMultiple() def expandRoot(self, config): self.defaultImageList, self.checkboxes, self.icondict \ = createDefaultImageList() self.AssignImageList(self.defaultImageList) self.il = self.GetImageList() self.rootID = self.AddRoot("/", self.icondict['computer'], -1, wx.TreeItemData(("", True, False, CheckboxState.UNSELECTED))) self.update() def expandUntilMultiple(self): node = self.rootID while True: (ipath, isdir, expanded, istate) = self.GetItemData(node).GetData() children = self.getChildren(node, False) if len(children) > 1 or len(children) == 0: break; node = children[0] self.Expand(node) def update(self): master = listMeta(self.config) for i in master: if not isinstance(master[i], dict): traversal = i.split(os.path.sep) node = self.rootID path = "/" if traversal[0] == '': traversal.remove('') for n in traversal: path = os.path.join(path, n) children = self.getChildrenDict(node) if n == traversal[-1] and not n in children: child = self.AppendItem(node, n) self.SetPyData(child, (path, False, False, 0)) idx = getFileIcon(i, self.il, self.checkboxes, self.icondict) self.SetItemImage(child, idx, wx.TreeItemIcon_Normal) else: if not n in children: child = self.AppendItem(node, n) self.SetPyData(child, (path, True, False, 0)) self.SetItemImage(child, self.icondict['folder'], wx.TreeItemIcon_Normal) else: child = children[n] node = child self.Expand(self.rootID) def getChildrenDict(self, node): result = {} child, cookie = self.GetFirstChild(node) while child: result[self.GetItemText(child)] = child child, cookie = self.GetNextChild(node, cookie) return result def onExpand(self, event): pass def getSelected(self, startNode=None): if not startNode: startNode = self.rootID children = self.getChildren(startNode) selected = [] for n in children: (path, isDir, expanded, state) = self.GetItemData(n).GetData() if not isDir \ and (state == CheckboxState.SELECTED \ or state == CheckboxState.SELECTEDCHILD): selected.append(n) if isDir and (state == CheckboxState.SELECTED \ or state == CheckboxState.SELECTEDPARENT \ or state == CheckboxState.SELECTEDCHILD): selected += self.getSelected(n) return selected class RestorePanel(wx.Panel): def __init__(self, parent, config, factory): self.config = config self.factory = factory wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) self.tree = RestoreCheckboxCtrl(self, -1, config, #wx.TreeCtrl(self, -1, style=(wx.TR_MULTIPLE | wx.TR_HAS_BUTTONS | wx.TR_TWIST_BUTTONS | wx.TR_NO_LINES | wx.TR_FULL_ROW_HIGHLIGHT | wx.SUNKEN_BORDER)) self.restoreButton = wx.Button(self, -1, 'restore selected files', name='restoreButton') self.Bind(wx.EVT_BUTTON, self.onRestoreClick, self.restoreButton) self.gbSizer = wx.GridBagSizer(2,1) self.gbSizer.Add(self.tree, (0,0), flag=wx.EXPAND|wx.ALL, border=0) self.gbSizer.Add(self.restoreButton, (1,0), flag=wx.EXPAND|wx.ALL, border=0) self.gbSizer.AddGrowableRow(0) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) def update(self): self.tree.update() def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() def onTooltip(self, event): pass def onRestoreClick(self, event): for n in self.tree.getSelected(): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "restoring %s" % path d = self.factory.sendGETF(path) d.addCallback(self.restored, n) d.addErrback(self.restoreFailed, n) self.tree.UnselectAll() def restored(self, res, n): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "yay, %s" % path self.tree.SetItemTextColour(n, '#005804') self.tree.changeState(n) def restoreFailed(self, err, n): (path, isDir, expanded, state) = self.tree.GetItemData(n).GetData() print "boo, %s: %s" % (path, err) self.tree.SetItemTextColour(n, wx.RED) self.tree.changeState(n) class SchedulePanel(wx.Panel): def __init__(self, parent): wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() class FeedbackPanel(wx.Panel): def __init__(self, parent): wx.Panel.__init__(self, parent, -1) self.Bind(wx.EVT_SIZE, self.OnSize) editor = wx.lib.editor.editor.Editor(parent, -1) def OnSize(self, event): w,h = self.GetClientSizeTuple() event.Skip() class FludNotebook(wx.Notebook): def __init__(self, parent, id=-1, pos=wx.DefaultPosition, size=wx.DefaultSize, style=wx.NB_BOTTOM|wx.NO_BORDER): self.parent = parent self.config = parent.config self.factory = LocalClientFactory(self.config) print "connecting to localhost:%d" % self.config.clientport reactor.connectTCP('localhost', self.config.clientport, self.factory) wx.Notebook.__init__(self, parent, id, pos, style=style) self.filePanel = FilePanel(self, searchButtonAction=parent.searchButtonAction) self.AddPage(self.filePanel, "Backup Files") self.restorePanel = RestorePanel(self, self.config, self.factory) self.AddPage(self.restorePanel, "Restore") self.schedulePanel = SchedulePanel(self) self.AddPage(self.schedulePanel, "Backup Schedule") self.feedbackPanel = FeedbackPanel(self) self.AddPage(self.feedbackPanel, "Feedback") self.Bind(wx.EVT_NOTEBOOK_PAGE_CHANGED, self.changedPage) def shutdown(self, event): self.filePanel.shutdown(event) def changedPage(self, event): page = event.GetSelection() if page == 0: self.SetMessage("Select files and directories for backup" " with the filesystem view on the left, or set up criteria" " for finding files for backup with simple searches," " below right.") elif page == 1: self.SetMessage("Select files/directories to be restored to" " your computer, then click on 'restore!' Files will turn" " green as they arrive.") self.restorePanel.update() elif page == 2: self.SetMessage("Configure how often your computer should backup." "\n (not implemented)") elif page == 3: self.SetMessage("Send feedback to flud programmers. (not" " implemented)") def SetMessage(self, msg): self.parent.SetMessage(msg) class FludLogoPanel(wx.Panel): def __init__(self, parent, id=-1, pos=wx.DefaultPosition, size=wx.Size(10,10), style=wx.TAB_TRAVERSAL, name="logo panel"): wx.Panel.__init__(self, parent, id, pos, size, style, name) self.SetAutoLayout(True) self.SetBackgroundColour(wx.BLACK) self.SetForegroundColour(wx.WHITE) logobmp = wx.BitmapFromImage(wx.Image(os.path.join(imgdir, "flud-backup-logo-1-150-nodrop.png"))) pad = 0 self.logowidth = logobmp.GetWidth() self.logoheight = logobmp.GetHeight() self.logo = wx.StaticBitmap(self, -1, logobmp) self.messagePanel = wx.Panel(self, -1) self.messagePanel.SetBackgroundColour(wx.BLACK) self.messagePanel.SetForegroundColour(wx.WHITE) self.message = wx.StaticText(self.messagePanel, -1, "message text area", style=wx.ALIGN_CENTER|wx.ST_NO_AUTORESIZE, size=(-1, self.logoheight-15)) self.message.Bind(wx.EVT_SIZE, self.resizeMessage) self.bsizer = wx.BoxSizer(wx.VERTICAL) self.bsizer.Add(self.message, flag=wx.EXPAND|wx.ALL, border=35) self.bsizer.SetSizeHints(self.messagePanel) self.messagePanel.SetSizer(self.bsizer) self.gbSizer = wx.GridBagSizer(1,2) self.gbSizer.Add(self.logo, (0,0)) self.gbSizer.Add(self.messagePanel, (0,1), flag=wx.EXPAND|wx.ALL) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(1) self.SetSizerAndFit(self.gbSizer) self.SetSize(wx.Size(self.logowidth, self.logoheight)) self.SetSizeHints(self.logowidth, self.logoheight, -1, self.logoheight) def SetMessage(self, msg): (w,h) = self.message.GetSizeTuple() #print "msg area size is %d x %d" % (w,h) self.message.SetLabel(msg) self.message.Wrap(w) #print "msg is '%s'" % self.message.GetLabel() self.message.Center() def resizeMessage(self, evt): # this is mainly to deal with StaticText wonkiness (not calling Wrap() # automatically, not centering properly automatically). It may be # possible to get rid of this with a future wxPython release. (w,h) = self.message.GetSizeTuple() self.message.Wrap(w) m = self.message.GetLabel() m = m.replace('\n',' ') self.message.SetLabel(m) self.message.Wrap(w) self.message.Center() class FludFrame(wx.Frame): def __init__(self, parent, id=wx.ID_ANY, label="flud bakcup client", size=wx.Size(800,600), style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE, config=None): wx.Frame.__init__(self, parent, id, label, size=size, style=style) self.lognull = wx.LogNull() wx.ToolTip.SetDelay(2000) self.clearMessage = False self.logoPanel = FludLogoPanel(self) self.SetMessage('Welcome.') self.config = config self.notebook = FludNotebook(self, size=wx.Size(200,200)) self.operationStatus = wx.StatusBar(name='operationStatus', parent=self, style=0) self.SetStatusBar(self.operationStatus) self.gbSizer = wx.GridBagSizer(2,1) self.gbSizer.Add(self.logoPanel,(0,0), flag=wx.EXPAND) self.gbSizer.Add(self.notebook, (1,0), flag=wx.EXPAND|wx.ALL, border=1) self.gbSizer.AddGrowableRow(1) self.gbSizer.AddGrowableCol(0) self.SetSizerAndFit(self.gbSizer) self.Bind(wx.EVT_CLOSE, self.shutdown) self.SetSize(size) self.Show(True) def SetMessage(self, message): self.logoPanel.SetMessage(message) def shutdown(self, event): self.notebook.shutdown(event) def searchButtonAction(self, event, errmsg=None, infomsg=None): if errmsg: self.logoPanel.SetMessage(errmsg) self.clearMessage = True elif infomsg: self.logoPanel.SetMessage(infomsg) self.clearMessage = False elif self.clearMessage: self.logoPanel.SetMessage("") #if __name__ == '__main__': # app = wx.PySimpleApp() # # config = FludConfig() # config.load(doLogging=False) # # f = FludFrame(None, wx.ID_ANY, 'flud backup client', size=(795,600), # style=wx.DEFAULT_FRAME_STYLE|wx.NO_FULL_REPAINT_ON_RESIZE, # config=config) # # from twisted.internet import reactor # reactor.registerWxApp(app) # reactor.run()
Python
""" FludConfig.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. manages configuration file for flud backup. """ import os, sys, socket, re, logging import ConfigParser import flud.FludCrypto as FludCrypto from flud.FludCrypto import FludRSA from flud.FludkRouting import kRouting from flud.fencode import fencode, fdecode """ default mapping of relative URLs """ def_commandmap = { 'ID': 'ID', 'GROUPID': 'GROUPID', 'STORE': 'STORE', 'RETRIEVE': 'RETRIEVE', 'VERIFY': 'VERIFY', 'PROXY': 'PROXY', 'DELETE': 'DELETE', 'kFINDNODE': 'kFINDNODE', 'kFINDVAL': 'kFINDVAL', 'kSTORE': 'kSTORE' } logger = logging.getLogger('flud') CLIENTPORTOFFSET = 500 def_trustdeltas = { 'NODE_INITIAL_SCORE': 1, 'NODE_POSITIVE_CAP': 100, 'NODE_NEGATIVE_CAP': -100, 'PUT_SUCCEED_REWARD': 2, 'GET_SUCCEED_REWARD': 4, 'VRFY_SUCCEED_REWARD': 4, 'FNDN_FAIL_PENALTY': -1, 'PUT_FAIL_PENALTY': -2, 'GET_FAIL_PENALTY': -10, 'VRFY_FAIL_PENALTY': -10 } class FludDebugLogFilter(logging.Filter): """ Keeps all logging levels defined by loggers, but ups level to DEBUG for loggers whose namespaces match patterns given by wildcards. """ # XXX: doesn't really interact with all logging levels by all loggers, only # with the one defined by the root logger. If children have stricter # loglevels set, this filter won't ever get called on them. def __init__(self, wildcardStrings): self.setWildcards(wildcardStrings) root = logging.getLogger("") if hasattr(root, 'fludDebugLogLevel'): self.effectiveLevel = root.fludDebugLogLevel else: self.effectiveLevel = root.getEffectiveLevel() self.fludDebugLogLevel = root.getEffectiveLevel() root.setLevel(logging.NOTSET) def setWildcards(self, wildcardStrings): self.wildcards = [] if not isinstance(wildcardStrings, list): wildcardStrings = [wildcardStrings] for s in wildcardStrings: self.setWildcard(s) def setWildcard(self, wildcardString): fields = wildcardString.split('.') for i, s in enumerate(fields): #print "%s:%s" % (i, s) if "*" == s: fields[i] = r'[\w.]*' else: try: if s.index(s, '*') > 0: fields[i] = s.replace('*', r'[\w]*') except: pass regex = "^%s$" % r'\.'.join(fields) self.wildcards.append(re.compile(regex)) def filter(self, record): if record.levelno >= self.effectiveLevel: return 1 for w in self.wildcards: m = w.match(record.name) if m: return 1 return 0 # XXX: refactor out the try/except stuff that could be done with has_key() class FludConfig: """ Handles configuration for Flud nodes. Most persistent settings live in this class. Configuration is kept in the directory specified by FLUDHOME if this value is set in the environment, otherwise in HOME/.flud/. If no existing configuration exists, this object will create a configuration with sane default values. """ def __init__(self): self.Kr = 0 self.Ku = 0 self.nodeID = 0 self.groupIDr = 0 self.groupIDu = 0 self.port = -1 self.commandmap = {} self.trustdeltas = {} self.reputations = {} self.nodes = {} try: self.fludhome = os.environ['FLUDHOME'] except: try: home = os.environ['HOME'] self.fludhome = home+"/.flud" except: logger.warn("cannot determine FLUDHOME.") logger.warn("Please set HOME or FLUDHOME environment variable") if os.path.isdir(self.fludhome) == False: os.mkdir(self.fludhome, 0700) self.fludconfig = self.fludhome+"/flud.conf" self.configParser = ConfigParser.ConfigParser() if os.path.isfile(self.fludconfig) == False: conffile = file(self.fludconfig, "w") else: conffile = file(self.fludconfig, "r") self.configParser.readfp(conffile) conffile.close() logger.info('fludhome = %s' % self.fludhome) logger.info('fludconfig = %s' % self.fludconfig) def load(self, serverport=None, doLogging=True): """ If serverport is given, it overrides any value that may be in the configuration file """ self.logfile, self.loglevel = self._getLoggingConf() if doLogging: if os.path.isfile(self.logfile): os.remove(self.logfile) handler = logging.FileHandler(self.logfile) formatter = logging.Formatter('%(asctime)s %(filename)s:%(lineno)d' ' %(name)s %(levelname)s: %(message)s', datefmt='%H:%M:%S') handler.setFormatter(formatter) logger.addHandler(handler) logging.getLogger("").setLevel(self.loglevel) #logger.setLevel(self.loglevel) #logger.setLevel(logging.WARNING) # XXX: overrides user prefs #logger.setLevel(logging.DEBUG) # XXX: overrides user prefs if os.environ.has_key("LOGFILTER"): self.filter = FludDebugLogFilter( os.environ["LOGFILTER"].split(' ')) handler.addFilter(self.filter) # XXX: add a LocalPrimitive that can be called dynamically to # invoke filter.setWildcards() self.Kr, self.Ku, self.nodeID, self.groupIDr, self.groupIDu \ = self._getID() logger.debug('Kr = %s' % self.Kr.exportPrivateKey()) logger.debug('Ku = %s' % self.Ku.exportPublicKey()) logger.debug('nodeID = %s' % self.nodeID) logger.debug('groupIDr = %s' % self.groupIDr) logger.debug('groupIDu = %s' % self.groupIDu) self.port, self.clientport, self.commandmap, self.trustdeltas \ = self._getServerConf() if serverport != None: self.port = serverport self.clientport = serverport + CLIENTPORTOFFSET self.configParser.set("server","port",self.port) self.configParser.set("server","clientport",self.clientport) logger.debug('port = %s' % self.port) logger.debug('clientport = %s' % self.clientport) logger.debug('commandmap = %s' % self.commandmap) self.routing = kRouting((socket.getfqdn(), self.port, long(self.nodeID, 16), self.Ku.exportPublicKey()['n'])) self.storedir, self.generosity, self.minoffer = self._getStoreConf() if os.path.isdir(self.storedir) == False: os.mkdir(self.storedir) os.chmod(self.storedir, 0700) logger.debug('storedir = %s' % self.storedir) self.kstoredir = self._getkStoreConf() if os.path.isdir(self.kstoredir) == False: os.mkdir(self.kstoredir) os.chmod(self.kstoredir, 0700) logger.debug('kstoredir = %s' % self.kstoredir) self.clientdir = self._getClientConf() if os.path.isdir(self.clientdir) == False: os.mkdir(self.clientdir) os.chmod(self.clientdir, 0700) logger.debug('clientdir = %s' % self.clientdir) self.metadir, self.metamaster = self._getMetaConf() if os.path.isdir(self.metadir) == False: os.mkdir(self.metadir) os.chmod(self.metadir, 0700) logger.debug('metadir = %s' % self.metadir) self.reputations = self._getReputations() logger.debug("reputations = %s" % str(self.reputations)) self.nodes = self._getKnownNodes() logger.debug("known nodes = %s" % str(self.nodes)) self.save() os.chmod(self.fludconfig, 0600) self.loadMasterMeta() def save(self): """ saves configuration """ conffile = file(self.fludconfig, "w") self.configParser.write(conffile) conffile.close() def _getLoggingConf(self): """ Returns logging configuration: logfile and loglevel """ if (self.configParser.has_section("logging") == False): self.configParser.add_section("logging") try: logfile = int(self.configParser.get("logging","logfile")) except: logger.debug("no logfile specified, using default") logfile = self.fludhome+'/flud.log' self.configParser.set("logging", "logfile", logfile) try: loglevel = int(self.configParser.get("logging","loglevel")) #loglevel = logging.WARNING # XXX: remove me except: logger.debug("no loglevel specified, using default") loglevel = logging.WARNING self.configParser.set("logging", "loglevel", loglevel) return logfile, loglevel def _getID(self): """ Returns a tuple: private key, public key, nodeID, private group ID, and public group ID from config. If these values don't exist in conf file, they are generated and added. """ # get the keys and IDs from the config file. # If these values don't exist, generate a pub/priv key pair, nodeID, # and groupIDs. if (self.configParser.has_section("identification") == False): self.configParser.add_section("identification") try: privkey = FludRSA.importPrivateKey( eval(self.configParser.get("identification","Kr"))) except: pubkey, privkey = FludCrypto.generateKeys() else: try: pubkey = FludRSA.importPublicKey( eval(self.configParser.get("identification","Ku"))) except: pubkey = privkey.publickey() try: nodeID = self.configParser.get("identification","nodeID") except: #nodeID = FludCrypto.hashstring(str(pubkey.exportPublicKey())) nodeID = pubkey.id() try: privgroupID = self.configParser.get("identification", "groupIDr")[:64] except: privgroupID = 'fludtest' # default groupID hardcoded try: pubgroupID = self.configParser.get("identification","groupIDu") except: pubgroupID = FludCrypto.hashstring(str(pubkey.exportPublicKey()) +privgroupID) # write the settings back out to config object self.configParser.set("identification","Kr",privkey.exportPrivateKey()) self.configParser.set("identification","Ku",pubkey.exportPublicKey()) self.configParser.set("identification","nodeID",nodeID) self.configParser.set("identification","groupIDr",privgroupID) self.configParser.set("identification","groupIDu",pubgroupID) # return the values return privkey, pubkey, nodeID, privgroupID, pubgroupID def _getServerConf(self): """ Returns server configuration: port number """ if (self.configParser.has_section("server") == False): self.configParser.add_section("server") try: port = int(self.configParser.get("server","port")) except: logger.debug("no port specified, using default") port = 8080 # XXX: default should be defined elsewhere. # Should prefer 80. If in use, use 8080+ try: clientport = int(self.configParser.get("server","clientport")) except: logger.debug("no clientport specified, using default") clientport = port+CLIENTPORTOFFSET try: commandmap = eval(self.configParser.get("server","commandmap")) except: logger.debug("no commandmap specified, using default") commandmap = def_commandmap for i in def_commandmap: # ensure that commandmap covers all keys if not commandmap.has_key(i): commandmap[i] = def_commandmap[i] trustdeltas = {} for i in def_trustdeltas: if not trustdeltas.has_key(i): trustdeltas[i] = def_trustdeltas[i] # XXX: could also do a 'parammap' self.configParser.set("server","port",port) self.configParser.set("server","clientport",clientport) self.configParser.set("server","commandmap",commandmap) return port, clientport, commandmap, trustdeltas def _getDirConf(self, configParser, section, default): """ Returns directory configuration """ if (configParser.has_section(section) == False): configParser.add_section(section) try: dir = int(self.configParser.get(section,"dir")) except: logger.debug("no %s directory specified, using default", section) dir = self.fludhome+'/'+default if not os.path.isdir(dir): os.makedirs(dir) self.configParser.set(section,"dir",dir) return dir def _getClientConf(self): """ Returns client configuration: download directory """ return self._getDirConf(self.configParser, "client", "dl") def _getStoreConf(self): """ Returns data store configuration """ storedir = self._getDirConf(self.configParser, "store", "store") try: generosity = self.configParser.get("store", "generosity") except: logger.debug("no generosity specified, using default") generosity = 1.5 try: minoffer = self.configParser.get("store", "minoffer") except: logger.debug("no minoffer specified, using default") minoffer = 1024 return storedir, generosity, minoffer def _getkStoreConf(self): """ Returns dht data store configuration """ return self._getDirConf(self.configParser, "kstore", "dht") def _getMetaConf(self): """ Returns metadata configuration: metadata directory """ metadir = self._getDirConf(self.configParser, "metadata", "meta") try: master = self.configParser.get("meta","master") except: logger.debug("no meta master file specified, using default") master = "master" if not os.path.isfile(metadir+'/'+master): f = open(metadir+'/'+master, 'w') f.close() return (metadir, master) def _getReputations(self): """ Returns dict of reputations known to this node """ # XXX: should probably just throw these in with 'nodes' (for efficiency) return self._getDict(self.configParser, "reputations") def _getKnownNodes(self): """ Returns dict of nodes known to this node """ return {} # XXX: don't read known nodes for now result = self._getDict(self.configParser, "nodes") for i in result: print str(i) self.routing.insertNode( (result[i]['host'], result[i]['port'], long(i, 16), result[i]['nKu'])) return result def _getDict(self, configParser, section): """ creates a dictionary from the list of pairs given by ConfigParser.items(section). Requires that the right-hand side of the config file's "=" operator be a valid python type, as eval() will be invoked on it """ if (configParser.has_section(section) == False): configParser.add_section(section) try: items = configParser.items(section) result = {} for item in items: #print item try: result[str(item[0])] = eval(item[1]) configParser.set(section, item[0], item[1]) except: logger.warn("item '%s' in section '%s'" " of the config file has an unreadable format" % str(item[0]), str(section)) except: logger.warn("Couldn't read %s from config file:" % section) return result def addNode(self, nodeID, host, port, Ku, mygroup=None): """ Convenience method for adding a node to the known. If a node with nodeID already exists, nothing changes. This method /does not/ save the new configuration to file, """ if mygroup == None: mygroup = self.groupIDu if self.nodes.has_key(nodeID) == False: self.nodes[nodeID] = {'host': host, 'port': port, 'Ku': Ku.exportPublicKey(), 'mygroup': mygroup} #logger.log(logging.DEBUG, "nodes: " % str(self.nodes)) # XXX: disabled nodes saving #for k in self.nodes: # self.configParser.set('nodes', k, self.nodes[k]) n = self.routing.insertNode((host, int(port), long(nodeID, 16), Ku.exportPublicKey()['n'])) if n != None: logger.warn("need to ping %s for LRU in routing table!" % str(n)) # XXX: instead of pinging, put it in a replacement cache table # and when one of the nodes needs replaced (future query) # replace it with one of these. Sec 4.1 self.routing.replacementCache.insertNode( (host, int(port), long(nodeID, 16), Ku.exportPublicKey()['n'])) self.reputations[long(nodeID,16)] \ = self.trustdeltas['NODE_INITIAL_SCORE'] # XXX: no management of reputations size: need to manage as a cache def modifyReputation(self, nodeID, delta): """ change reputation of nodeID by delta """ logger.info("modify %s %s" % (nodeID, delta)) if isinstance(nodeID, str): nodeID = long(nodeID,16) if not self.reputations.has_key(nodeID): self.reputations[nodeID] \ = self.trustdeltas['NODE_INITIAL_SCORE'] # XXX: no management of reputations size: need to manage as a cache self.reputations[nodeID] += delta logger.debug("reputation for %d now %d", nodeID, self.reputations[nodeID]) def getOrderedNodes(self, num=None, exclude=None): """ Get nodes ordered by reputation. If num is passed in, return the first 'num' nodes, otherwise all. If exclude list is passed in, try to return nodes not on this list (but do return some excluded if nodes are exhausted, i.e., there aren't num nodes available). """ # XXX: O(n) each time this is called. Better performance if we # maintain sorted list when modified (modifyReputation, addNode), at a # bit higher mem expense. items = self.reputations.items() numitems = len(items) logger.debug("%d items in reps" % numitems) if exclude: items = [(v,k) for (k,v) in items if k not in exclude] if num and len(items) < num and numitems >= num: exitems = [(v,k) for (k,v) in items if k in exclude] items += exitems[num-len(item):] logger.debug("%d items now in reps" % len(items)) else: items = [(v,k) for (k,v) in items] items.sort() items.reverse() items = [(k,v) for (v,k) in items] # need to call routing.getNode() to get node triple and return those if num: logger.debug("returning %d of the %d items" % (num, len(items))) return [self.routing.getNode(f) for (f,v) in items[:num]] else: logger.debug("returning all %d of the items" % len(items)) return [self.routing.getNode(f) for (f,v) in items] # XXX: note that this master metadata all-in-mem scheme doesn't really work # long term; these methods should eventually go to a local db or db-like # something def updateMasterMeta(self, fname, val): """ update fname with val (sK) """ self.master[fname] = val def getFromMasterMeta(self, fname): """ get val (sK) for fname """ try: return self.master[fname] except: return None def deleteFromMasterMeta(self, fname): """ remove fname """ try: self.master.pop(fname) except: pass def loadMasterMeta(self): """ loads fname->sK mappings from file """ fmaster = open(os.path.join(self.metadir, self.metamaster), 'r') master = fmaster.read() fmaster.close() if master == "": master = {} else: master = fdecode(master) self.master = master def syncMasterMeta(self): """ sync in-mem fname->sK mappings to disk """ master = fencode(self.master) fmaster = open(os.path.join(self.metadir, self.metamaster), 'w') fmaster.write(master) fmaster.close() def _test(self): import doctest doctest.testmod() if __name__ == '__main__': fludConfig = FludConfig() fludConfig._test()
Python
#!/usr/bin/python """ FludNode.tac (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL). This is the application file used by twistd to daemonize FludNode. """ import os from twisted.application import service, internet import flud.FludNode from flud.protocol.FludCommUtil import getCanonicalIP port = None gwhost = None gwport = None if 'FLUDPORT' in os.environ: port = int(os.environ['FLUDPORT']) if 'FLUDGWHOST' in os.environ: gwhost = getCanonicalIP(os.environ['FLUDGWHOST']) if 'FLUDGWPORT' in os.environ: gwport = int(os.environ['FLUDGWPORT']) node = flud.FludNode.FludNode(port) if gwhost and gwport: node.connectViaGateway(gwhost, gwport) application = service.Application("flud.FludNode") service = node.start(twistd=True) #service.setServiceParent(application)
Python
#!/usr/bin/python """ FludScheduler.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), verison 3. FludScheduler is the process monitors files for changes, and then tells flud to back them up. """ import sys, os, time, stat from twisted.internet import reactor from flud.FludConfig import FludConfig from flud.protocol.LocalClient import * from flud.CheckboxState import CheckboxState CHECKTIME=5 class FludScheduler: def __init__(self, config, factory): self.config = config self.factory = factory self.fileconfigfile = None self.fileconfigfileMTime = 0 self.fileChangeTime = 0 self.fileconfigSelected = set() self.fileconfigExcluded = set() self.getMasterMetadata() def getMasterMetadata(self): d = self.factory.sendLIST() d.addCallback(self.gotMasterMetadata) d.addErrback(self.errMasterMetadata) return d def gotMasterMetadata(self, master): self.mastermetadata = master def errMasterMetadata(self, err): print err reactor.stop() def readFileConfig(self, mtime=None): print "reading FileConfig" file = open(self.fileconfigfile, 'r') self.fileconfig = eval(file.read()) file.close() if mtime: self.fileconfigfileMTime = mtime else: self.fileconfigfileMTime = os.stat( self.fileconfigfile)[stat.ST_MTIME] self.fileconfigSelected = set([f for f in self.fileconfig if self.fileconfig[f] == CheckboxState.SELECTED or self.fileconfig[f] == CheckboxState.SELECTEDCHILD]) self.fileconfigExcluded = set([f for f in self.fileconfig if self.fileconfig[f] == CheckboxState.EXCLUDED or self.fileconfig[f] == CheckboxState.EXCLUDEDCHILD]) # The file[s]ChangeStat are the worst possible way to detect file changes. # Much more efficient to use inotify/dnotify/fam/gamin/etc., as well as # more correct (no way to detect cp -a or -p, for example, with stat). # But, these are a fallback method when those aren't present, and are fine # for testing. def fileChangedStat(self, file, fileChangeTime=None): if os.path.isfile(file) or os.path.isdir(file): mtime = os.stat(file)[stat.ST_MTIME] if not fileChangeTime: fileChangeTime = self.fileChangeTime if file in self.mastermetadata: fileChangeTime = self.mastermetadata[file][1] else: return True print "mtime = %s, ctime = %s (%s)" % (mtime, fileChangeTime, file) if mtime > fileChangeTime: return True return False def filesChangedStat(self, files, fileChangeTime=None): result = [] for f in files: if self.fileChangedStat(f, fileChangeTime): result.append(f) return result # Change these to point to something other than the xxxStat() methods def fileChanged(self, file, fileChangeTime=None): """ >>> now = time.time() >>> f1 = tmpfile.mktemp() >>> """ return self.fileChangedStat(file, fileChangeTime) def filesChanged(self, files, fileChangeTime=None): return self.filesChangedStat(files, fileChangeTime) def checkFileConfig(self): # check config file to see if it has changed, then reparse it if not self.fileconfigfile: # first time through print "checking fileconfigfile (initial)" if os.environ.has_key('FLUDHOME'): fludhome = os.environ['FLUDHOME'] elif os.environ.has_key('HOME'): fludhome = os.environ['HOME']+"/.flud" else: fludhome = ".flud" # XXX: fludfile.conf should be in config self.fileconfigfile = os.path.join(fludhome, "fludfile.conf") if os.path.isfile(self.fileconfigfile): self.readFileConfig() return True else: print "no fileconfigfile to read" elif os.path.isfile(self.fileconfigfile): if self.fileChanged(self.fileconfigfile, self.fileconfigfileMTime): print "fileconfigfile changed" mtime = time.time() self.readFileConfig(mtime) return True return False def checkFilesystem(self): checkedFiles = set() changedFiles = set() def checkList(list): #print "checkList: %s" % list #print "checkedFiles: %s" % checkedFiles for entry in list: # XXX: if entry is in master metadata, and its mtime is not # earlier than the time used by fileChanged, skip it (add 'and' # clause) if entry not in checkedFiles and \ entry not in self.fileconfigExcluded and \ entry not in self.mastermetadata: print "checkFilesystem for %s" % entry if os.path.isdir(entry): #print "dir %s" % entry dirfiles = [os.path.join(entry, i) for i in os.listdir(entry)] checkedFiles.update([entry,]) checkList(dirfiles) elif self.fileChanged(entry): print "%s changed" % entry if os.path.isfile(entry): changedFiles.update([entry,]) #print "file %s changed" % entry else: print "entry ?? %s ?? changed" % entry checkedFiles.update([entry,]) checkList(self.fileconfigSelected) self.fileChangeTime = time.time() return changedFiles def storefileFailed(self, err, file): print "storing %s failed: %s" % (file, err) err.printTraceback() #print dir(err) def storefileYay(self, r, file): print "storing %s success" % file def storeFiles(self, changedFiles): #print "storing %s" % changedFiles dlist = [] for f in changedFiles: print "storing %s" % f deferred = self.factory.sendPUTF(f) deferred.addCallback(self.storefileYay, f) deferred.addErrback(self.storefileFailed, f) dlist.append(deferred) dl = defer.DeferredList(dlist) return dl #return defer.succeed(True) def restartCheckTimer(self, v): print "restarting timer (%d) to call run()" % CHECKTIME reactor.callLater(CHECKTIME, self.run) def updateMasterMetadata(self, v): return self.getMasterMetadata() def run(self): print "run" self.checkFileConfig() changedFiles = self.checkFilesystem() print "%s changed" % changedFiles d = self.storeFiles(changedFiles) d.addBoth(self.updateMasterMetadata) d.addBoth(self.restartCheckTimer)
Python
""" FludFileOperations.py (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. Implements file storage and retrieval operations using flud primitives. """ import os, stat, sys, logging, binascii, random, time from zlib import crc32 from StringIO import StringIO from twisted.internet import defer, threads from Crypto.Cipher import AES from flud.FludCrypto import FludRSA, hashstring, hashfile from flud.protocol.FludCommUtil import * from flud.fencode import fencode, fdecode import fludfilefec logger = logging.getLogger('flud.fileops') # erasure coding constants code_k = 20 # data blocks code_n = 20 # parity blocks code_m = code_k+code_n # total blocks # temp filenaming defaults appendEncrypt = ".crypt" appendFsMeta = ".nmeta" # XXX: could remove trailing '=' from all stored sha256s (dht keys, storage # keys, etc) and add them back implicitly class Ctx(object): """ Object to encapsulate operation context for logger """ def __init__(self, ctx): self.ctx = ctx self.formatstr = "" self.args = () def msg(self, formatstr, *args): self.formatstr = formatstr self.args = args s = self.formatstr % self.args return self def __repr__(self): return str(self.ctx)+": "+(self.formatstr % self.args) def pathsplit(fname): par, chld = os.path.split(fname) if chld == "": res = [] res.append(par) return res else: res = pathsplit(par) res.append(os.path.join(par,chld)) return res def filemetadata(fname): fstat = os.stat(fname) return {'path' : fname, 'mode' : fstat[stat.ST_MODE], 'uid' : fstat[stat.ST_UID], 'gid' : fstat[stat.ST_GID], 'atim' : fstat[stat.ST_ATIME], 'mtim' : fstat[stat.ST_MTIME], 'ctim' : fstat[stat.ST_CTIME]} class StoreFile: """ Implements the meta operations of storing, retrieving, and verifying files. In addition to using the data primitives laid out in the protocol/ directory, manages the creation of file metadata for local and remote storage. Stores a file to flud by: 1. Create storage and encryption keys for file: Hashes the file once to create an encryption key, eK=H(file), and then again to create the storage key for the file metadata sK=H(eK)=H(H(file)). 2. Create local filesystem file metadata: Encrypts the storage key asymetrically with public key as eeK=e_Ku(eK), and creates local copy of fs metadata with eeK and other file metadata (ownership, name, timestamps). Encrypts this metadata with Ku. (flud file metadata consists of eeK and Ku(fs metadata)). 3. Create data-specific file metadata: Symmetrically encrypt the file with e_file=eK(file). Code e_file into k+m blocks. Perform H(block) on each k+m block. 4. Query DHT for sK. If it exists, grab the metadata record (call it 'storedMetadata') for comparison to one we are generated. Compare data-specific metadata to storedMetadata (if it exists). If it doesn't already exist, this step can be ignored. If it exists and the data doesn't match, either the hash function is broken or a malicious node has poisoned the DHT -- return failure (client can attempt restore under a new key, or challenge the existing stored metadata). 5. Store m+n blocks. If stored metadata exists, can do VERIFYs instead, and only store when VERIFY fails. For each failed VERIFY, must update the data-specific part of the metadata record. Since we don't want to be susceptible to replacement attack, the update is treated as an 'append' to the specific block metadata by the store target. (The store target can monitor its data-specific metadata records and for any that grow beyond a certain size, can shrink all block records to a size of 1 by performing RETRIEVE ops on the block, verfiying hash, and removing any failures. If this fails for all block entries, can prune all but the /first and last/ entries -- keep these to prevent damage by malicious nodes. This strategy is beneficial to the node that performs it, because it prevents storage of arbitrary data in the DHT layer, which is in turn good for the system as a whole) 6. Store file metadata (both node-specific and data-specific) to the DHT. Keep a local copy as well. 7. Update the master file record index for this node, and store it to the DHT layer. """ # XXX: instead of doing all debugging with '"%s ...", self.mkey, ...', make # some props (debug(), warn(), etc) that add the mkey to whatever msg is # passed # XXX: should follow this currentOps model for the other FludFileOps currentOps = {} def __init__(self, node, filename): self.node = node self.filename = filename self.mkey = crc32(self.filename) self.ctx = Ctx(self.mkey).msg self.config = node.config self.Ku = node.config.Ku self.routing = self.config.routing self.metadir = self.config.metadir self.parentcodedir = self.config.clientdir # XXX: clientdir? #self.nodeChoices = self.routing.knownExternalNodes() # ask for code_m + X nodes (we prefer a pool slightly larger than # code_m). XXX: 20 is magic self.nodeChoices = self.config.getOrderedNodes(code_m+20) self.usedNodes = {} self.deferred = self._storeFile() def _storeFile(self): if not os.path.isfile(self.filename): return defer.fail(ValueError("%s is not a file" % self.filename)) # 1: create encryption key (eK) and storage key (sK). Query DHT using # sK self.eK = hashfile(self.filename) logger.debug(self.ctx("_storefile %s (%s)", self.filename, self.eK)) self.sK = long(hashstring(self.eK), 16) self.eeK = self.Ku.encrypt(binascii.unhexlify(self.eK)) self.eKey = AES.new(binascii.unhexlify(self.eK)) #logger.debug(self.ctx("file %s eK:%s, storage key:%d" # % (self.filename, self.eK, self.sK))) # 2: create filesystem metadata locally. sbody = filemetadata(self.filename) sbody = fencode(sbody) self.eNodeFileMetadata = "" for i in range(0, len(sbody), 128): self.eNodeFileMetadata += self.Ku.encrypt(sbody[i:i+128])[0] fsMetadata = fencode({'eeK' : fencode(self.eeK[0]), 'meta' : fencode(self.eNodeFileMetadata)}) # erasure code the metadata # XXX: bad blocking stuff, move into thread self.flatname = fencode(generateRandom(16)) self.mfilename = os.path.join(self.metadir, self.flatname+appendFsMeta) self.encodedir = os.path.join(self.parentcodedir, self.flatname) try: os.mkdir(self.encodedir) except: return defer.fail(failure.DefaultException( "%s already requested" % self.filename)) # XXX: mfiles should be held in mem, as StringIOs (when coder supports # this) self.mfiles = fludfilefec.encode_to_files(StringIO(fsMetadata), len(fsMetadata), self.encodedir, self.mfilename, code_k, code_m) # XXX: piggybacking doesn't work with new metadata scheme, must fix it # to append metadata, or if already in progress, redo via verify ops # if already storing identical file by CAS, piggyback on it if self.currentOps.has_key(self.eK): logger.debug(self.ctx("reusing callback on %s", self.eK)) (d, counter) = self.currentOps[self.eK] self.currentOps[self.eK] = (d, counter+1) # setting sfile, encodedir, and efilename to empty vals is kinda # hokey -- could split _storeMetadata into two funcs instead (the # cleanup part and the store part; see _storeMetadata) self.sfiles = [] self.encodedir = None self.efilename = None d.addCallback(self._piggybackStoreMetadata) return d # 3: encrypt and encode the file locally. self.efilename = os.path.join(self.metadir,self.flatname+appendEncrypt) e = open(self.efilename, "w+") fd = os.open(self.filename, os.O_RDONLY) fstat = os.fstat(fd) fsize = fstat[stat.ST_SIZE] # XXX: bad blocking stuff, move into thread # create a pad at front of file to make it an even multiple of 16 fpad = int(16 - fsize%16); #logger.debug(self.ctx("fsize=%d, padding with %d bytes" # % (fsize, fpad))) paddata = chr(fpad)+(fpad-1)*'\x00' buf = paddata + os.read(fd,16-len(paddata)) e.write(self.eKey.encrypt(buf)); # now write the rest of the file while 1: # XXX: can we do this in larger than 16-byte chunks? buf = os.read(fd,16) if buf == "": break e.write(self.eKey.encrypt(buf)); #e.close() elen = e.tell() e.seek(0,0) os.close(fd) # erasure code the file # XXX: bad blocking stuff, move into thread self.sfiles = fludfilefec.encode_to_files(e, elen, self.encodedir, 'c', code_k, code_m) #logger.debug(self.ctx("coded to: %s" % str(self.sfiles))) # take hashes and rename coded blocks self.segHashesLocal = [] for i in range(len(self.sfiles)): sfile = self.sfiles[i] h = long(hashfile(sfile),16) logger.debug(self.ctx("file block %s hashes to %s", i, fencode(h))) destfile = os.path.join(self.encodedir,fencode(h)) if os.path.exists(destfile): logger.warn(self.ctx("%s exists (%s)", destfile, fencode(h))) self.segHashesLocal.append(h) #logger.debug(self.ctx("moved %s to %s" % (sfile, destfile))) os.rename(sfile, destfile) self.sfiles[i] = destfile mfile = self.mfiles[i] os.rename(mfile, destfile+".m") self.mfiles[i] = destfile+".m" # 4a: query DHT for metadata. d = self.node.client.kFindValue(self.sK) d.addCallback(self._checkForExistingFileMetadata) d.addErrback(self._storeFileErr, "DHT query for metadata failed") self.currentOps[self.eK] = (d, 1) return d # 4b: compare hashlists (locally encrypted vs. DHT -- if available). # for lhash, dhash in zip(segHashesLocal, segHashesDHT): def _checkForExistingFileMetadata(self, storedMetadata): if storedMetadata == None or isinstance(storedMetadata, dict): logger.info(self.ctx( "metadata doesn't yet exist, storing all data")) d = self._storeBlocks(storedMetadata) #d = self._storeBlocksSKIP(storedMetadata) return d else: storedMetadata = fdecode(storedMetadata) logger.info(self.ctx("metadata exists, verifying all data")) if not self._compareMetadata(storedMetadata, self.sfiles): raise ValueError("stored and local metadata do not match") else: logger.info(self.ctx("stored and local metadata match.")) # XXX: need to check for diversity. It could be that data stored # previously to a smaller network (<k+m nodes) and that we should # try to increase diversity and re-store the data. # XXX: also need to make sure we still trust all the nodes in the # metadata list. If not, we should move those blocks elsewhere. d = self._verifyAndStoreBlocks(storedMetadata) return d def _storeBlocksSKIP(self, storedMetadata): # for testing -- skip stores so we can get to storeMeta dlist = [] self.blockMetadata = {'k': code_k, 'n': code_n} for i in range(len(self.segHashesLocal)): hash = self.segHashesLocal[i] sfile = self.sfiles[i] node = random.choice(self.routing.knownExternalNodes()) host = node[0] port = node[1] nID = node[2] nKu = FludRSA.importPublicKey(node[3]) self.blockMetadata[(i, hash)] = long(nKu.id(), 16) return self._storeMetadata(None) # 5a -- store all blocks def _storeBlocks(self, storedMetadata): dlist = [] self.blockMetadata = {'k': code_k, 'n': code_n} for i in range(len(self.segHashesLocal)): hash = self.segHashesLocal[i] sfile = self.sfiles[i] deferred = self._storeBlock(i, hash, sfile, self.mfiles[i]) dlist.append(deferred) logger.debug(self.ctx("_storeBlocksAll")) dl = defer.DeferredList(dlist) dl.addCallback(self._storeMetadata) return dl def _storeBlock(self, i, hash, sfile, mfile, retry=2): if not self.nodeChoices: #self.nodeChoices = self.routing.knownExternalNodes() # XXX: instead of asking for code_k, ask for code_k - still needed self.nodeChoices = self.config.getOrderedNodes(code_k, self.usedNodes.keys()) logger.warn(self.ctx("asked for more nodes, %d nodes found", len(self.nodeChoices))) if not self.nodeChoices: return defer.fail(failure.DefaultException( "cannot store blocks to 0 nodes")) node = random.choice(self.nodeChoices) self.nodeChoices.remove(node) host = node[0] port = node[1] nID = node[2] nKu = FludRSA.importPublicKey(node[3]) location = long(nKu.id(), 16) logger.info(self.ctx("STOREing under %s on %s:%d", fencode(hash), host, port)) logger.debug(self.ctx("mfile is %s", mfile)) deferred = self.node.client.sendStore(sfile, (self.mkey, mfile), host, port, nKu) deferred.addCallback(self._blockStored, nID, i, hash, location) deferred.addErrback(self._retryStoreBlock, i, hash, location, sfile, mfile, nID, host, port, retry) return deferred def _retryStoreBlock(self, error, i, hash, location, sfile, mfile, nID, host, port, retry=None): retry = retry - 1 if retry > 0: logger.warn(self.ctx("STORE to %s (%s:%d) failed, trying again", nID, host, port)) d = self._storeBlock(i, hash, sfile, mfile, retry) d.addCallback(self._blockStored, nID, i, hash, location) # This will fail the entire operation. This is correct # behavior because we've tried on at least N nodes and couldn't # get the block to store -- the caller will have to try the entire # op again. If this proves to be a problem, up the default retry # value in _storeBlock(). d.addErrback(self._storeFileErr, "couldn't store block %s" % fencode(hash), lambda: self.config.modifyReputation(nID, self.config.trustdeltas['PUT_FAIL_PENALTY'])) return d else: self.usedNodes[nID] = True logger.warn(self.ctx("STORE to %s (%s:%d) failed, giving up", nID, host, port)) d = defer.Deferred() d.addErrback(self._storeFileErr, "couldn't store block %s" % fencode(hash), lambda: self.config.modifyReputation(nID, self.config.trustdeltas['PUT_FAIL_PENALTY'])) d.errback() return d def _blockStored(self, result, nID, i, blockhash, location): self.usedNodes[nID] = True logger.debug(self.ctx("_blcokStored %s", fencode(blockhash))) self.config.modifyReputation(location, self.config.trustdeltas['PUT_SUCCEED_REWARD']) self.blockMetadata[(i, blockhash)] = location return fencode(blockhash) def _compareMetadata(self, storedFiles, fileNames): # compares the block names returned from DHT to those in fileNames. # @param storedFiles: dict of longs (hashes) to their locations, # usually obtained from storedMetadata # @param fileNames: local filenames. Only the os.path.basename part # will be used for comparison # @return true if they match up perfectly, false otherwise logger.debug(self.ctx('# remote block names: %d', len(storedFiles))) logger.debug(self.ctx('# local blocks: %d', len(fileNames))) result = True k = storedFiles.pop('k') n = storedFiles.pop('n') for (i, f) in storedFiles: fname = os.path.join(self.encodedir,fencode(f)) if not fname in fileNames: logger.warn(self.ctx("%s not in sfiles", fencode(i))) result = False for i, fname in enumerate(fileNames): hname = os.path.basename(fname) if not storedFiles.has_key((i, fdecode(hname))): logger.warn(self.ctx("%s not in storedMetadata", hname)) result = False if result == False: for i in storedFiles: logger.debug(self.ctx("storedBlock = %s", fencode(i))) for i in fileNames: logger.debug(self.ctx("localBlock = %s", os.path.basename(i))) storedFiles['k'] = k storedFiles['n'] = n return result def _piggybackStoreMetadata(self, piggybackMeta): # piggybackMeta is a (nodeID, {blockID: storingNodeID, }) logger.debug(self.ctx("got piggyBackMeta data")) meta = piggybackMeta[1] sortedKeys = {} k = meta['k'] n = meta['n'] for i in [x for x in meta if x != 'k' and x != 'n']: sortedKeys[i[0]] = i for i in xrange(k+n): self.sfiles.append(fencode(sortedKeys[i][1])) return self._verifyAndStoreBlocks(meta, True) # 5b -- findnode on all stored blocks. def _verifyAndStoreBlocks(self, storedMetadata, noopVerify=False): self.blockMetadata = storedMetadata dlist = [] for i, sfile in enumerate(self.sfiles): # XXX: metadata should be StringIO to begin with f = file(self.mfiles[i]) mfile = StringIO(f.read()) f.close() seg = os.path.basename(sfile) segl = fdecode(seg) nID = self.blockMetadata[(i, segl)] if isinstance(nID, list): logger.info(self.ctx( "multiple location choices, choosing one randomly.")) nID = random.choice(nID) # XXX: for now, this just picks one of the alternatives at # random. If the chosen one fails, should try each of the # others until it works logger.info(self.ctx("looking up %s...", ('%x' % nID)[:8])) deferred = self.node.client.kFindNode(nID) deferred.addCallback(self._verifyBlock, i, sfile, mfile, seg, segl, nID, noopVerify) deferred.addErrback(self._storeFileErr, "couldn't find node %s... for VERIFY" % ('%x' % nID)[:8], self.config.modifyReputation(nID, self.config.trustdeltas['VRFY_FAIL_PENALTY'])) dlist.append(deferred) dl = defer.DeferredList(dlist) #dl.addCallback(self._storeMetadata) # XXX XXX XXX: don't _updateMaster unless we succeed!! dl.addCallback(self._updateMaster, storedMetadata) return dl # 5c -- verify all blocks, store any that fail verify. def _verifyBlock(self, kdata, i, sfile, mfile, seg, segl, nID, noopVerify): # XXX: looks like we occasionally get in here on timed out connections. # Should go to _storeFileErr instead, eh? if isinstance(kdata, str): logger.err(self.ctx("str kdata=%s", kdata)) #if len(kdata['k']) > 1: # #logger.debug(self.ctx("type kdata: %s" % type(kdata))) # #logger.debug(self.ctx("kdata=%s" % kdata)) # #logger.debug(self.ctx("len(kdata['k'])=%d" % len(kdata['k']))) # raise ValueError("couldn't find node %s" % ('%x' % nID)) # #raise ValueError("this shouldn't really be a ValueError..." # # " should be a GotMoreKnodesThanIBargainedForError" # # " (possibly caused when kFindNode fails (timeout) and" # # " we just get our own list of known nodes?): k=%s" # # % kdata['k']) node = kdata['k'][0] host = node[0] port = node[1] id = node[2] if id != nID: logger.debug(self.ctx("couldn't find node %s", ('%x' % nID))) raise ValueError("couldn't find node %s" % ('%x' % nID)) nKu = FludRSA.importPublicKey(node[3]) logger.info(self.ctx("verifying %s on %s:%d", seg, host, port)) if noopVerify: offset = length = 0 verhash = long(hashstring(''), 16) self.sfiles = [] else: fd = os.open(sfile, os.O_RDONLY) fsize = os.fstat(fd)[stat.ST_SIZE] if fsize > 20: # XXX: 20? length = 20 # XXX: 20? offset = random.randrange(fsize-length) else: length = fsize offset = 0 os.lseek(fd, offset, 0) data = os.read(fd, length) os.close(fd) verhash = long(hashstring(data), 16) deferred = self.node.client.sendVerify(seg, offset, length, host, port, nKu, (self.mkey, mfile)) deferred.addCallback(self._checkVerify, nKu, host, port, i, segl, sfile, mfile, verhash) deferred.addErrback(self._checkVerifyErr, nID, i, segl, sfile, mfile, verhash) return deferred def _checkVerify(self, result, nKu, host, port, i, seg, sfile, mfile, hash): if hash != long(result, 16): logger.info(self.ctx( "VERIFY hash didn't match for %s, performing STORE", fencode(seg))) d = self._storeBlock(i, seg, sfile, mfile) return d else: #logger.debug(self.ctx("block passed verify (%s == %s)" # % (hash, long(result,16)))) return fencode(seg) def _checkVerifyErr(self, failure, nID, i, seg, sfile, mfile, hash): self.config.modifyReputations(nID, self.config.trustdeltas['VRFY_FAIL_PENALTY']) logger.debug(self.ctx("Couldn't VERIFY: %s", failure.getErrorMessage())) logger.info(self.ctx("Couldn't VERIFY %s, performing STORE", fencode(seg))) d = self._storeBlock(i, seg, sfile, mfile) return d # 6 - store the metadata. def _storeMetadata(self, dlistresults): # cleanup part of storeMetadata: logger.debug(self.ctx("dlist=%s", str(dlistresults))) # XXX: for any "False" in dlistresults, need to invoke _storeBlocks # again on corresponding entries in sfiles. for i in dlistresults: if i[1] == None: logger.info(self.ctx("failed store/verify")) return False # storeMetadata part of storeMetadata # XXX: should sign metadata to prevent forged entries. #for i in self.blockMetadata: # logger.debug(self.ctx(" %s: %s" # % (fencode(i), fencode(self.blockMetadata[i])))) logger.debug(self.ctx("storing metadata at %s", fencode(self.sK))) logger.debug(self.ctx("len(segMetadata) = %d", len(self.blockMetadata))) d = self.node.client.kStore(self.sK, self.blockMetadata) d.addCallback(self._updateMaster, self.blockMetadata) d.addErrback(self._storeFileErr, "couldn't store file metadata to DHT") return d # 7 - update local master file record (store it to the network later). def _updateMaster(self, res, meta): # clean up locally coded files and encrypted file for sfile in self.sfiles: os.remove(sfile) for mfile in self.mfiles: os.remove(mfile) if self.encodedir: os.rmdir(self.encodedir) if self.efilename: os.remove(self.efilename) key = fencode(self.sK) logger.info(self.ctx("updating local master metadata with %s", key)) # store the filekey locally # update entry for file self.config.updateMasterMeta(self.filename, (self.sK, int(time.time()))) # update entry for parent dirs paths = pathsplit(self.filename) for i in paths: if not self.config.getFromMasterMeta(i): self.config.updateMasterMeta(i, filemetadata(i)) # XXX: not too efficient to write this out for every file. consider # local caching and periodic syncing instead self.config.syncMasterMeta() # cache the metadata locally (optional) fname = os.path.join(self.metadir,key) m = open(fname, 'wb') m.write(fencode(meta)) m.close() # clean up fs metadata file #os.remove(self.mfilename) #return fencode(self.sK) (d, counter) = self.currentOps[self.eK] counter = counter - 1 if counter == 0: logger.debug(self.ctx("counter 0 for currentOps %s", self.eK)) self.currentOps.pop(self.eK) else: logger.debug(self.ctx("setting counter = %d for %s", counter, self.eK)) self.currentOps[self.eK] = (d, counter) return (key, meta) def _storeFileErr(self, failure, message, raiseException=True, functor=None): if functor: functor() (d, counter) = self.currentOps[self.eK] counter = counter - 1 if counter == 0: logger.debug(self.ctx("err counter 0 for currentOps %s", self.eK)) self.currentOps.pop(self.eK) else: logger.debug(self.ctx("err setting counter = %d for %s", counter, self.eK)) self.currentOps[self.eK] = (d, counter) logger.error(self.ctx("%s: %s", message, failure.getErrorMessage())) logger.debug(self.ctx("%s", failure.getTraceback())) if raiseException: raise failure class RetrieveFile: """ Uses the given storage key to retrieve a file. The storage key is used to query the DHT layer for the file metadata record. The file record contains the locations of the file blocks. These are downloaded until the complete file can be regenerated and saved locally. """ def __init__(self, node, key, mkey=True): # 1: Query DHT for sK # 2: Retrieve entries for sK, decoding until efile can be regenerated # 3: Retrieve eK from sK by eK=Kp(eKe). Use eK to decrypt file. Strip # off leading pad. # 4: Save file as filepath=Kp(efilepath). self.node = node self.mkey = mkey try: self.sK = fdecode(key) except Exception, inst: self.deferred = defer.fail(inst) return self.ctx = Ctx(crc32(str(self.sK))).msg self.config = node.config self.Ku = node.config.Ku self.Kr = node.config.Kr self.routing = self.config.routing.knownNodes() self.metadir = self.config.metadir self.parentcodedir = self.config.clientdir self.numBlocksRetrieved = 0 self.blocks = {} self.fsmetas = {} self.deferred = self._retrieveFile() def _retrieveFile(self): # 1: Query DHT for sK logger.debug(self.ctx("querying DHT for %s", self.sK)) d = self.node.client.kFindValue(self.sK) d.addCallback(self._retrieveFileBlocks) d.addErrback(self._findFileErr, "file retrieve failed") return d def _findFileErr(self, failure, message, raiseException=True): logger.error(self.ctx("%s: %s" % (message, failure.getErrorMessage()))) if raiseException: return failure def _retrieveFileBlocks(self, meta): # 2: Retrieve entries for sK, decoding until efile can be regenerated if meta == None: raise LookupError("couldn't recover metadata for %s" % self.sK) self.meta = fdecode(meta) # XXX: need to check for diversity. It could be that data stored # previously to a smaller network (<k+m nodes) and that we should # try to increase diversity and re-store the data. # XXX: also need to make sure we still trust all the nodes in the # metadata list. If not, we should move those blocks elsewhere. if self.meta == None: raise LookupError("couldn't recover metadata for %s" % self.sK) k = self.meta.pop('k') n = self.meta.pop('n') if k != code_k or n != code_n: # XXX: raise ValueError("unsupported coding scheme %d/%d" % (k, n)) logger.info(self.ctx("got metadata %s" % self.meta)) self.decoded = False return self._getSomeBlocks(code_k) def _getSomeBlocks(self, reqs=code_k): tries = 0 if reqs > len(self.meta): reqs = len(self.meta) dlist = [] for i in range(reqs): c = random.choice(self.meta.keys()) block = fencode(c[1]) id = self.meta[c] if isinstance(id, list): logger.info(self.ctx( "multiple location choices, choosing one randomly.")) id = random.choice(id) # XXX: for now, this just picks one of the alternatives at # random. If the chosen one fails, should try each of the # others until it works or exhausts the list #logger.info(self.ctx("retrieving %s from %s" # % (block, fencode(id)))) logger.info(self.ctx("retrieving %s from %s" % (block, id))) # look up nodes by id, then do a retrieve. deferred = self.node.client.kFindNode(id) deferred.addCallback(self._retrieveBlock, block, id) deferred.addErrback(self._findNodeErr, "couldn't find node %s for block %s" % (fencode(id), block), id) dlist.append(deferred) self.meta.pop(c) tries = tries + 1 if tries >= reqs: break; dl = defer.DeferredList(dlist) dl.addCallback(self._retrievedAll) return dl def _findNodeErr(self, failure, msg, id): logger.info(self.ctx("%s: %s" % (message, failure.getErrorMessage()))) self.config.modifyReputation(id, self.config.trustdeltas['FNDN_FAIL_PENALTY']) def _retrieveBlock(self, kdata, block, id): #print type(kdata) #print kdata #if len(kdata['k']) > 1: if kdata['k'][0][2] != id: print "%s != %s" (kdata['k'][0], id) raise ValueError("couldn't find node %s" % fencode(id)) #raise ValueError("this shouldn't really be a ValueError..." # " should be a GotMoreKnodesThanIBargainedForError: k=%s" # % kdata['k']) #else: # print kdata['k'] node = kdata['k'][0] host = node[0] port = node[1] id = node[2] nKu = FludRSA.importPublicKey(node[3]) if not self.decoded: d = self.node.client.sendRetrieve(block, host, port, nKu, self.mkey) d.addCallback(self._retrievedBlock, id, block, self.mkey) d.addErrback(self._retrieveBlockErr, id, "couldn't get block %s from %s" % (block, fencode(id)), host, port, id) return d def _retrievedBlock(self, msg, nID, block, mkey): logger.debug(self.ctx("retrieved block=%s, msg=%s" % (block, msg))) self.config.modifyReputation(nID, self.config.trustdeltas['GET_SUCCEED_REWARD']) blockname = [f for f in msg if f[-len(block):] == block][0] expectedmeta = "%s.%s.meta" % (block, mkey) metanames = [f for f in msg if f[-len(expectedmeta):] == expectedmeta] if not metanames: raise failure.DefaultException("expected metadata was missing") self.blocks[block] = blockname self.fsmetas[block] = metanames[0] self.numBlocksRetrieved += 1 return True def _retrieveBlockErr(self, failure, nID, message, host, port, id): logger.info(self.ctx("%s: %s" % (message, failure.getErrorMessage()))) self.config.modifyReputation(nID, self.config.trustdeltas['GET_FAIL_PENALTY']) # don't propogate the error -- one block doesn't cause the file # retrieve to fail. def _retrievedAll(self, success): logger.info(self.ctx("tried retreiving %d blocks %s" % (len(success), success))) if self.numBlocksRetrieved >= code_k: # XXX: need to make this try again with other blocks if decode # fails return self._decodeData() elif len(self.meta) > 0: tries = 3 # XXX: magic number. Should derive from k & m logger.info(self.ctx("requesting %d more blocks" % tries)) return self._getSomeBlocks(tries) else: logger.info(self.ctx("couldn't decode file after retreiving" " all %d available blocks" % self.numBlocksRetrieved)) #return False raise RuntimeError("couldn't decode file after retreiving all" " %d available blocks" % self.numBlocksRetrieved) def _decodeData(self): logger.debug(self.ctx("_decodeData")) self.fname = os.path.join(self.parentcodedir,fencode(self.sK))+".rec1" d = threads.deferToThread(self.decodeData, self.fname, self.blocks.values(), self.config.clientdir) d.addCallback(self._decodeFsMetadata, self.blocks) d.addErrback(self._decodeError) return d def _decodeFsMetadata(self, decoded, blockname): logger.debug(self.ctx("_decodeFsMetadata")) self.mfname = os.path.join(self.parentcodedir,fencode(self.sK))+".m" d = threads.deferToThread(self.decodeData, self.mfname, self.fsmetas.values(), self.config.clientdir) d.addCallback(self._decodeDone, decoded) d.addErrback(self._decodeError) return d def decodeData(self, outfname, datafnames, datadir=None): logger.info(self.ctx("decoding %s to %s" % (datafnames, outfname))) outf = open(outfname, 'wb') data = [] for fname in datafnames: if datadir: fname = os.path.join(datadir, fname) data.append(open(fname, 'rb')) result = fludfilefec.decode_from_files(outf, data) outf.close() for f in data: f.close() if result: for fname in datafnames: if datadir: fname = os.path.join(datadir, fname) os.remove(os.path.join(datadir, fname)) return result def _decodeError(self, err): logger.warn(self.ctx("could not decode: %s", err)) return err def _decodeDone(self, decoded, metadecoded): if not self.decoded and decoded and metadecoded: logger.info(self.ctx("successfully decoded (retrieved %d blocks --" " all but %d blocks tried)" % (self.numBlocksRetrieved, len(self.meta)))) return self._decryptMeta() else: logger.info(self.ctx("decoded=%s, mdecoded=%s" % (decoded, metadecoded))) raise RuntimeError("decode error after retreiving " " %d blocks: decoded=%s, mdecoded=%" % (self.numBlocksRetrieved, decoded, metadecoded)) def _decryptMeta(self): # XXX: decrypt the metadatafile with Kr to get all the nmeta stuff (eeK # etc.) #mfile = open(os.path.join(self.parentcodedir, fencode(self.sK)+".m")) mfile = open(self.mfname, 'r') meta = mfile.read() mfile.close() logger.info(self.ctx("meta is %s" % str(meta))) self.nmeta = fdecode(meta) os.remove(self.mfname) return self._decryptFile() def _decryptFile(self): # 3: Retrieve eK from sK by eK=Kr(eeK). Use eK to decrypt file. Strip # off leading pad. skey = fencode(self.sK) f1 = open(os.path.join(self.parentcodedir,skey+".rec1"), "r") f2 = open(os.path.join(self.parentcodedir,skey+".rec2"), "w") #logger.info(self.ctx("decoding nmeta eeK for %s" % dir(self))) eeK = fdecode(self.nmeta['eeK']) # d_eK business is to ensure that eK is zero-padded to 32 bytes d_eK = self.Kr.decrypt(eeK) d_eK = '\x00'*(32%len(d_eK))+d_eK # XXX: magic 32, should be keyspace/8 eK = binascii.hexlify(d_eK) eKey = AES.new(binascii.unhexlify(eK)) # XXX: bad blocking stuff, move into thread while 1: buf = f1.read(16) if buf == "": break; f2.write(eKey.decrypt(buf)) f1.close() f2.close() os.remove(os.path.join(self.parentcodedir,skey+".rec1")) f2 = open(os.path.join(self.parentcodedir,skey+".rec2"), "r") f3 = open(os.path.join(self.parentcodedir,skey+".rec3"), "w") padlen = f2.read(1) #print "%s" % repr(padlen) padlen = ord(padlen) padlen -= 1 #print "throwing away %d pad bytes" % padlen pad = f2.read(padlen) # throw away pad. while 1: buf = f2.read(16) if buf == "": break; f3.write(buf) f2.close() f3.close() os.remove(os.path.join(self.parentcodedir,skey+".rec2")) # 4: Move file to its correct path, imbue it with properties from # metadata. # XXX: should we make sure we can read metadata before downloading all # the file data? #print "decoding nmeta meta" efmeta = fdecode(self.nmeta['meta']) fmeta = "" for i in range(0, len(efmeta), 128): fmeta += self.Kr.decrypt(efmeta[i:i+128]) fmeta = fdecode(fmeta) result = [fmeta['path']] if os.path.exists(fmeta['path']): # file is already there -- compare it. If different, save as # path.recovered and keep a list of these (to let the user know # that they'll need to resolve later). Or don't keep a list and # just 'do the right thing' (use the latest version by timestamp, # or always use the backup, or always use the local copy, or # define some other behavior for doing the right thing). logger.info(self.ctx("hash rec=%s" % hashfile(fmeta['path']))) logger.info(self.ctx("hash org=%s" % eK)) if hashfile(fmeta['path']) != eK: # XXX: do something better than log it -- see above comment logger.info(self.ctx( 'different version of file %s already present' % fmeta['path'])) # XXX: should generate '.recovered' extension more carefully, # so as not to overwrite coincidentally named files. fmeta['path'] = fmeta['path']+".recovered" result.insert(0,fmeta['path']) os.rename(os.path.join(self.parentcodedir,skey+".rec3"), fmeta['path']) else: logger.info(self.ctx('same version of file %s already present' % fmeta['path'])) # no need to copy: os.remove(os.path.join(self.parentcodedir,skey+".rec3")) else: # recover parent directories if not present paths = pathsplit(fmeta['path']) for i in paths: if not os.path.exists(i) and i != fmeta['path']: os.mkdir(i) # best effort dir creation, even if missing # directory metadata # XXX: should be using an accessor method on config for # master if i in self.config.master: dirmeta = self.config.getFromMasterMeta(i) os.chmod(i,dirmeta['mode']) os.chown(i,dirmeta['uid'],dirmeta['gid']) # XXX: windows # XXX: atim, mtim, ctim # XXX: should try to make sure we can write to dir, change # perms if necessary. # recover file by renaming to its path os.rename(os.path.join(self.parentcodedir,skey+".rec3"), fmeta['path']) # XXX: chown not supported on Windows os.chown(fmeta['path'], fmeta['uid'], fmeta['gid']) os.utime(fmeta['path'], (fmeta['atim'], fmeta['mtim'])) os.chmod(fmeta['path'], fmeta['mode']) logger.info(self.ctx("successfully restored file metadata")) return tuple(result) class RetrieveFilename: """ Retrieves a File given its local name. Only works if the local master index contains an entry for this filename. """ def __init__(self, node, filename): self.node = node self.filename = filename self.metadir = self.node.config.metadir self.config = self.node.config self.deferred = self._recoverFile() def _recoverFile(self): fmeta = self.config.getFromMasterMeta(self.filename) if fmeta: if isinstance(fmeta, dict): logger.debug("%s is a directory in master metadata", self.filename) # RetrieveFile will restore parent dirs, so we don't need to dlist = [] dirname = self.filename+os.path.sep # XXX: this should be calling a config.getAllFromMasterMeta() for i in [x for x in self.config.master.keys() if dirname == x[:len(dirname)]]: filekey = self.config.getFromMasterMeta[i] metakey = crc32(i) logger.debug("calling RetrieveFile %s" % filekey) d = RetrieveFile(self.node, fencode(filekey), metakey).deferred dlist.append(d) dl = defer.DeferredList(dlist) return dl else: logger.debug("%s is file in master metadata", self.filename) (filekey, backuptime) = self.config.getFromMasterMeta( self.filename) metakey = crc32(self.filename) if filekey != None and filekey != "": logger.debug("calling RetrieveFile %s" % filekey) d = RetrieveFile(self.node, fencode(filekey), metakey).deferred return d return defer.fail(LookupError("bad filekey %s for %s" % (filekey, self.filename))) return defer.fail(LookupError("no record of %s" % self.filename)) class VerifyFile: # XXX: remove me? I don't do anything that StoreFile can't do, plus if # I fail you'd still need to call StoreFile right after... # Or, do we keep me around and rip out all the verify stuff from # StoreFile and put it in here? # -- What we really want to do is look at trading partners for all our # files, and then check each one every day with some random bit of # data from some random file. But we also want eventual coverage of # all stored files. It seems like a good first approach would be to # scan over our local copies of all DHT records and make a list of # storing nodes, which will be of size N. Then, do k VERIFY ops to # each node in N, using a random subset of the files we have stored. # The VerifyFile object's purpose may be orthogonal to that, or # completely unnecessary, as the described scheme can be accomplished # with plain VERIFY ops. def verifyFile(self, filepath): """ Chooses some random blocks from filepath to verify against the store. The algorithm is as follows: sK = H(H(file at filepath)). Look up sK in the local master index. If the record isn't there, return this fact. If the record is there, retrieve its metadata. Verify k blocks as follows: With probibility n/(m+n), code the file locally (to verify coded blocks with a fair probabiliy, i.e., if m=40 and n=20, 33% of the time we will do the coding). Choose k blocks from the resulting blocks and using the file metadata record, do a VERIFY operation using a random offset and random length (if we chose not to do the coding in the previous step, the k blocks must come entirely from the non-coded portion). As we wait for the VERIFYs to return, hash these blocks locally. As each VERIFY returns, compare it with our local hash just computed. Return a list of hosts/nodeids for which the VERIFY failed. """ pass #remove me class RetrieveMasterIndex: def __init__(self, node): self.node = node nodeID = long(self.node.config.nodeID, 16) # 1. CAS = kfindval(nodeID) (CAS for last FLUDHOME/meta/master) logger.info("looking for key %x" % nodeID) self.deferred = self.node.client.kFindValue(nodeID) self.deferred.addCallback(self._foundCAS) self.deferred.addErrback(self._retrieveMasterIndexErr, "couldn't find master metadata") def _foundCAS(self, CAS): # 2. oldmaster = kfindval(CAS) if isinstance(CAS, dict): return defer.fail(ValueError("couldn't find CAS key")) CAS = fdecode(CAS) d = RetrieveFile(self.node, CAS).deferred d.addCallback(self._foundMaster) d.addErrback(self._retrieveMasterIndexErr, "couldn't find Master Index") return d def _foundMaster(self, result): if len(result) == 2: # got two filenames back, must mean we should choose one: the # one from the distributed store os.rename(result[0], result[1]) result = (result[1],) return result def _retrieveMasterIndexErr(self, err, msg): logger.warn(msg) return err class UpdateMasterIndex: def __init__(self, node): self.node = node self.metamaster = os.path.join(self.node.config.metadir, self.node.config.metamaster) # 0.1. oldmaster = RetrieveMasterIndex() self.deferred = RetrieveMasterIndex(node).deferred self.deferred.addCallback(self._removeOldMasterIndex) self.deferred.addErrback(self._storeMasterIndex) def _removeOldMasterIndex(self, res): # 0.2. for i in oldmaster: delete(i) print "removing old master not yet implemented" return self._storeMasterIndex(res) def _storeMasterIndex(self, res_or_err): # 1. store FLUDHOME/meta/master print "going to store %s" % self.metamaster d = StoreFile(self.node, self.metamaster).deferred d.addCallback(self._updateCAS) d.addErrback(self._updateMasterIndexErr, "couldn't store master index") return d def _updateCAS(self, stored): # 2. kstore(nodeID, CAS(FLUDHOME/meta/master)) #print "stored = %s" % str(stored) key, meta = stored logger.info("storing %s at %x" % (key, long(self.node.config.nodeID,16))) d = self.node.client.kStore(long(self.node.config.nodeID,16), key) # XXX: key should be fdecode()ed return d def _updateMasterIndexErr(self, err, msg): logger.warn(msg) return err if __name__ == "__main__": from FludNode import FludNode def successTest(res, fname, whatfor, nextStage=None): logger.info("finished %s" % whatfor) def errTest(failure): logger.info("boom: %s" % failure.getErrorMessage()) raise failure def fileKey(fname): EK = hashfile(fname) return fencode(long(hashstring(EK), 16)) def clearMeta(fname): # delete any metadata that might exist for this file. try: SK = fileKey(fname) os.remove(os.path.join(n.config.kstoredir,SK)) logger.info("test removed %s" % os.path.join(n.config.kstoredir,SK)) except: pass def doStore(fname, msg=None, nextStage=successTest): if msg != None: logger.info("finished %s ------" % msg) logger.info("---------------------------------------------------\n") # do a store logger.info("nextStage is %s" % nextStage) d = StoreFile(n,fname).deferred d.addCallback(nextStage, fname, 'store op', doCorruptSegAndStore) d.addErrback(errTest) def doDelSegAndStore((key, meta), fname, msg=None, nextStage=successTest): # only works if stores are local if msg != None: logger.info("finished %s ------" % msg) logger.info("---------------------------------------------------\n") # delete a block and do a store c = random.choice(meta.keys()) logger.info("removing %s" % fencode(c)) os.remove(os.path.join(n.config.storedir,fencode(c))) logger.info("test removed %s" % os.path.join(n.config.storedir, fencode(c))) d = StoreFile(n,fname).deferred d.addCallback(nextStage, fname, 'lost block op') d.addErrback(errTest) def doCorruptSegAndStore((key, meta), fname, msg=None, nextStage=successTest): # only works if stores are local if msg != None: logger.info("finished %s ------" % msg) logger.info("---------------------------------------------------\n") # corrupt a block and do a store c = random.choice(meta.keys()) logger.info("corrupting %s" % fencode(c)) f = open(os.path.join(n.config.storedir,fencode(c)), 'r') data = f.read() f.close() f = open(os.path.join(n.config.storedir,fencode(c)), 'w') f.write('blah'+data) f.close() d = StoreFile(n,fname).deferred d.addCallback(nextStage, fname, 'corrupted block op') d.addErrback(errTest) def doRetrieve(key, msg=None, nextStage=successTest): if msg != None: logger.info("finished %s ------" % msg) logger.info("---------------------------------------------------\n") d = RetrieveFile(n, key).deferred d.addCallback(nextStage, key, 'retrieve op') d.addErrback(errTest) def doRetrieveName(filename, msg=None, nextStage=successTest): if msg != None: logger.info("finished %s ------" % msg) logger.info("---------------------------------------------------\n") d = RetrieveFilename(n, filename).deferred d.addCallback(nextStage, filename, 'retrieve filename op') d.addErrback(errTest) def runTests(dummy): # test against self -- all stores and queries go to self. fname = "/tmp/nrpy.pdf" #clearMeta(fname) #doStore(fname, None, doDelSegAndStore) # do all stages of testing doStore(fname) # only do one op (for manual testing) #doRetrieve(fileKey(fname)) #doRetrieveName(fname) n = FludNode() n.run() if len(sys.argv) == 3: deferred = n.client.sendkFindNode(sys.argv[1], int(sys.argv[2]), 1) deferred.addCallback(runTests) deferred.addErrback(errTest) else: runTests(None) n.join()
Python
""" Okay, so this isn't a real python module, yet. Wanted to get down a few ideas on versioning. First, the background. Traditional backup systems that provide versioning support allow the user to retrieve the current version of a file, or any of N previous versions that were stored during previous backup operations. Since it was rather trivial to simply keep old versions on the central backup server, this wasn't much of an engineering problem (at worst, disk fills up quickly). With a collaborative backup system, such a scheme is less practical. If fully enabled, it can consume many times the storage space of a simple single-snapshot system. If you want to enforce fairness, you must require that the number of resources you consume are proportional to those that you provide. Encoding already dictates that this ratio is imbalanced towards providing more resources than consuming. But "server-side" versioning, even when using a clever delta-compression technique, really tips the scales. There is good news, however. We can use a single-snapshot system to provide versioning by requiring all versioning to occur locally. That is, the consumer's own hard drive can be used to maintain multiple versions of files, and then the whole thing can be backed up as a single-snapshot to the flud network. Think of a local CVS repository (with many versions contained therein) that is set to be backed up; the backup system doesn't have to worry about versioning -- it just backs up the current data. The local CVS repo is in charge of worrying about versions. To the user, its all the same. The advantages of this scheme are mainly: 1) simplicity 2) storage consumption minimization, opt-in 3) decoupling of versioning layer from backup layer Of the three, #1 is really the most appealing. We just back up the current view. This also greatly simplifies the verification mechanism -- the verifier will always have the complete file from which to do challenge/response queries to the verifiee. We don't have to worry about keeping deltas or partial checksums or anything like that in order to do verification; we just pick a block of bytes at random from within the file, and make sure that the storer can return us the hash of those bytes. #1 also means that we don't have to do anything complicated to figure out what the delta of a delta-compressed version should be (i.e., we don't need to request the old version from the storage system, compare it with our version, then send out a delta) -- in fact, with this scheme we wipe out delta compression altogether, at least from the viewpoint of the storage mechanism (some other local mechanism is welcome to use delta compression to store versions locally, but this mechanism won't need to download lots of data in order to do so, because it will all be local). #2 is nice. It means that if the user isn't interested in versioning, they don't have to do it. This will be the default, in fact. This means that we eliminate a lot of overhead that we would have had if every user was storing versions, even if they didn't need them. It also means that there is an automatic cost for enabling versions, not only for the collaborative system, but for the user's local storage resources. Not to imply that we want to punish the user for enabling versions, but there's no free lunch (and besides, adding local disk is cheap). [as an aside, here, versioning does become necessary quite quickly for things such as email clients that store all mail in a particular folder as one large file, or other applications that use databases in single files -- we don't want the user to have to send the whole file (which can become quite large) every time they get a new email or change the db slightly. The good news is that we can still provide this in its own layer]. Decoupling (#3) is always a good idea, especially when it can be done cleanly. The user (or flud client developer) is free to implement whatever local versioning scheme they want. They can make copies of files and store them in other directories, they could use a version control system such as CVS, they could do their own delta compression and store the deltas in a special directory. They could store as many or as few versions as they want. They can take version snapshots often or seldom, and this is independent of how often they perform backup. And such schemes can be switched out, upgraded, or removed on the fly without anyone really noticing. So, what is this module all about then? Well, for now, nothing but this documentation. If the user wants versioning, they'll have to do it themselves, (keeping in mind that the system *will* store those local versions, and they can rest easy knowing that they can be retrieved). At some future point, we will implement some fancy versioning layer with local delta compression in this module to 'complete the package,' but it is currently a low priority endeavor. Its priority will rise as we get close to providing a 'for your grandma' solution. """
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import fencode, fdecode """ Test code for primitive DHT operations. """ stay_alive = 1 filename = "/tmp/tempstoredata" filekey = os.path.basename(filename) key = 87328673569979667228965797330646992089697345905484734072690869757741450870337L # format of block metadata is # {(i, datakey): storingNodeID, ..., 'n': n, 'm': m} # where i<n+m, datakey is a sha256 of the data stored, and storingNodeID is # either a nodeID or a list of nodeIDs. testval = {(0, 802484L): 465705L, (1, 780638L): 465705L, (2, 169688L): 465705L, (3, 267175L): 465705L, (4, 648636L): 465705L, (5, 838315L): 465705L, (6, 477619L): 465705L, (7, 329906L): 465705L, (8, 610565L): 465705L, (9, 217811L): 465705L, (10, 374124L): 465705L, (11, 357214L): 465705L, (12, 147307L): 465705L, (13, 427751L): 465705L, (14, 927853L): 465705L, (15, 760369L): 465705L, (16, 707029L): 465705L, (17, 479234L): 465705L, (18, 190455L): 465705L, (19, 647489L): 465705L, (20, 620470L): 465705L, (21, 777532L): 465705L, (22, 622383L): 465705L, (23, 573283L): 465705L, (24, 613082L): 465705L, (25, 433593L): 465705L, (26, 584543L): 465705L, (27, 337485L): 465705L, (28, 911014L): 465705L, (29, 594065L): 465705L, (30, 375876L): 465705L, (31, 726818L): 465705L, (32, 835759L): 465705L, (33, 814060L): 465705L, (34, 237176L): 465705L, (35, 538268L): 465705L, (36, 272650L): 465705L, (37, 314058L): 465705L, (38, 257714L): 465705L, (39, 439931L): 465705L, 'k': 20, 'n': 20} logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) #logger.setLevel(logging.DEBUG) logger.setLevel(logging.INFO) def cleanup(_, node): logger.info("waiting %ds to shutdown..." % stay_alive) reactor.callLater(stay_alive, node.stop) def testerror(failure, message, node): """ error handler for test errbacks """ logger.warn("testerror message: %s" % message) logger.warn("testerror: '%s'" % str(failure)) logger.warn("At least 1 test FAILED") def endtests(res, nKu, node, host, port): """ executes after all tests """ try: res = fdecode(res) except ValueError: pass if res != testval: return testerror(None, "retrieved value does not match stored value:" " '%s' != '%s'" % (res, testval), node) logger.info("testkFindVal PASSED") logger.debug("testkFindVal result: %s" % str(res)) logger.info("all tests PASSED") return res def testkFindVal(res, nKu, node, host, port): logger.info("testSendkFindVal PASSED") logger.debug("testSendkFindVal result: %s" % str(res)) logger.info("attempting testkFindValue") deferred = node.client.kFindValue(key) deferred.addCallback(endtests, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkFindValue", node) return deferred def testSendkFindVal(res, nKu, node, host, port): logger.info("testkStore PASSED") logger.debug("testkStore result: %s" % str(res)) logger.info("attempting testSendkFindValue") deferred = node.client.sendkFindValue(host, port, key) deferred.addCallback(testkFindVal, nKu, node, host, port) deferred.addErrback(testerror, "failed at testSendkFindValue", node) return deferred def testkStore(res, nKu, node, host, port): logger.info("testSendkStore PASSED") logger.debug("testSendkStore result: %s" % str(res)) logger.info("attempting testkStore") deferred = node.client.kStore(key, testval) deferred.addCallback(testSendkFindVal, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkStore", node) return deferred def testSendkStore(res, nKu, node, host, port): logger.info("testkFindNode PASSED") logger.debug("testkFindNode result: %s" % str(res)) logger.info("attempting testSendkStore") deferred = node.client.sendkStore(host, port, key, testval) deferred.addCallback(testkStore, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkStore", node) return deferred def testkFindNode(res, nKu, node, host, port): """ executes after testSendkFindNode """ logger.info("testSendkFindNode PASSED") logger.debug("testSendkFindNode result: %s" % str(res)) logger.info("attempting kFindNode") deferred = node.client.kFindNode(key) deferred.addCallback(testSendkStore, nKu, node, host, port) deferred.addErrback(testerror, "failed at kFindNode", node) return deferred def testSendkFindNode(nKu, node, host, port): """ executes after testGetID """ logger.info("testkGetID PASSED") logger.info("attempting sendkFindNode") deferred = node.client.sendkFindNode(host, port, key) deferred.addCallback(testkFindNode, nKu, node, host, port) deferred.addErrback(testerror, "failed at sendkFindNode", node) return deferred def testGetID(node, host, port): """ Tests sendGetID(), and invokes testSendkFindNode on success """ deferred = node.client.sendGetID(host, port) deferred.addCallback(testSendkFindNode, node, host, port) deferred.addErrback(testerror, "failed at testGetID", node) return deferred def runTests(host, port=None, listenport=None): host = getCanonicalIP(host) node = FludNode(port=listenport) if port == None: port = node.config.port logger.info("testing against %s:%s, localport=%s" % (host, port, listenport)) node.run() d = testGetID(node, host, port) d.addBoth(cleanup, node) #testkFindVal("blah", node.config.Ku, node, host, port) node.join() """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) elif len(sys.argv) == 4: runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket, shutil, tempfile from binascii import crc32 from StringIO import StringIO from twisted.python import failure sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.FludCrypto import generateRandom, hashfile from flud.protocol.FludCommUtil import * from flud.fencode import fencode """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ smallfilekey = "" smallfilename = "" smallfilenamebad = "" largefilekey = "" largefilename = "" largefilenamebad = "" metadata = 'aaaa' def testerror(failure, message, node): """ error handler for test errbacks """ print "testerror message: %s" % message print "testerror: %s" % str(failure) print "At least 1 test FAILED" raise failure def testUnexpectedSuccess(res, message, node): print "unexpected success message: %s" % message print "At least 1 test succeeded when it should have failed" raise "bad" def testDELETEBadKeyFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg # the end else: # XXX: here and elsewhere, raise something more descriptive, otherwise # its waay confusing print "the following trace may be misleading..." raise failure def testDELETEBadKey(nKu, node, host, port): print "starting testDELETEBadKey" path = os.path.join("somedir", largefilekey) deferred = node.client.sendDelete(path, crc32(path), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "DELETE with bad key succeeded", node) deferred.addErrback(testDELETEBadKeyFailed, "DELETE with bad key failed as expected", node, nKu, host, port) return deferred def testVERIFYBadKeyFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testDELETEBadKey(nKu, node, host, port) else: raise failure def testVERIFYBadKey(nKu, node, host, port): print "starting testVERIFYBadKey" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize-20 deferred = node.client.sendVerify(smallfilenamebad, offset, 5, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad key succeeded", node) deferred.addErrback(testVERIFYBadKeyFailed, "VERIFY of bad filekey failed as expected", node, nKu, host, port) return deferred def testVERIFYBadLengthFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYBadKey(nKu, node, host, port) else: raise failure def testVERIFYBadLength(nKu, node, host, port): print "starting testVERIFYBadOffset" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize-10 deferred = node.client.sendVerify(smallfilekey, offset, 20, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad length", node) deferred.addErrback(testVERIFYBadLengthFailed, "VERIFY of bad length failed as expected", node, nKu, host, port) return deferred def testVERIFYBadOffsetFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYBadLength(nKu, node, host, port) else: print "VERIFYBadOffset failed as expected, but with wrong failure" raise failure def testVERIFYBadOffset(nKu, node, host, port): print "starting testVERIFYBadOffset" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize+2 deferred = node.client.sendVerify(smallfilekey, offset, 20, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad offset", node) deferred.addErrback(testVERIFYBadOffsetFailed, "VERIFY of bad offset failed as expected", node, nKu, host, port) return deferred def testVERIFYNotFoundFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testVERIFYBadOffset(nKu, node, host, port) else: raise failure def testVERIFYNotFound(nKu, node, host, port): print "starting testVERIFYNotFound" deferred = node.client.sendVerify(largefilekey, 10, 10, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified non-existent file", node) deferred.addErrback(testVERIFYNotFoundFailed, "VERIFY of non-existent file failed as expected", node, nKu, host, port) return deferred def testRETRIEVEIllegalPathFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYNotFound(nKu, node, host, port) else: raise failure def testRETRIEVEIllegalPath(nKu, node, host, port): print "starting testRETRIEVEIllegalPath" deferred = node.client.sendRetrieve(os.path.join("somedir",smallfilekey), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "retrieved file with illegal path", node) deferred.addErrback(testRETRIEVEIllegalPathFailed, "RETRIEVE using illegal path failed as expected", node, nKu, host, port) return deferred def testRETRIEVENotFoundFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testRETRIEVEIllegalPath(nKu, node, host, port) else: raise failure def testRETRIEVENotFound(nKu, node, host, port): print "starting testRETRIEVENotFound" deferred = node.client.sendRetrieve(largefilekey, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "retrieved file that shouldn't exist", node) deferred.addErrback(testRETRIEVENotFoundFailed, "RETRIEVE of non-existent file failed as expected", node, nKu, host, port) return deferred def testSTORELargeFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadCASKeyException'): print "%s" % msg return testRETRIEVENotFound(nKu, node, host, port) else: raise failure def testSTOREBadKeyLarge(nKu, node, host, port): print "starting testSTOREBadKeyLarge" deferred = node.client.sendStore(largefilenamebad, (crc32(largefilenamebad), StringIO(metadata)), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "large file, bad key succeeded", node) deferred.addErrback(testSTORELargeFailed, "large STORE with bad key failed as expected", node, nKu, host, port) return deferred def testSTORESmallFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadCASKeyException'): print "%s" % msg return testSTOREBadKeyLarge(nKu, node, host, port) else: raise failure def testSTOREBadKeySmall(nKu, node, host, port): print "starting testSTOREBadKeySmall" deferred = node.client.sendStore(smallfilenamebad, (crc32(smallfilenamebad), StringIO(metadata)), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "small file, bad key succeeded", node) deferred.addErrback(testSTORESmallFailed, "small STORE with bad key failed as expected", node, nKu, host, port) return deferred def testSTORESuccess(res, nKu, node, host, port): print "testSTORE succeeded: %s" % res return testSTOREBadKeySmall(nKu, node, host, port) def testSTORE(nKu, node, host, port): # store a file successfully for later failure tests (VERIFY, etc) print "starting testSTORE" deferred = node.client.sendStore(smallfilename, (crc32(smallfilename), StringIO(metadata)), host, port, nKu) deferred.addCallback(testSTORESuccess, nKu, node, host, port) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred # XXX: need to test bogus headers for all commands (BAD_REQUEST) # XXX: need to test failures for authentication def testID(node, host, port): """ Tests sendGetID(), and invokes testSTORE on success """ print "starting testID" deferred = node.client.sendGetID(host, port) deferred.addCallback(testSTORE, node, host, port) #deferred.addCallback(testSTOREBadKeySmall, node, host, port) deferred.addErrback(testerror, "failed at testID", node) return deferred def cleanup(err, node): if err: print "cleaning up: %s" % err else: print "cleaning up" os.remove(smallfilename) os.remove(smallfilenamebad) os.remove(largefilename) os.remove(largefilenamebad) reactor.callLater(1, node.stop) def generateTestData(): def generateFiles(minsize): fname = tempfile.mktemp() f = open(fname, 'w') f.write('\0'*minsize) f.write(generateRandom(random.randrange(256)+1)) f.close() filekey = hashfile(fname) filekey = fencode(int(filekey, 16)) filename = os.path.join("/tmp",filekey) os.rename(fname,filename) filenamebad = os.path.join("/tmp/","bad"+filekey[3:]) shutil.copy(filename, filenamebad) return (filekey, filename, filenamebad) global smallfilekey global smallfilename global smallfilenamebad (smallfilekey, smallfilename, smallfilenamebad) = generateFiles(1024) global largefilekey global largefilename global largefilenamebad (largefilekey, largefilename, largefilenamebad) = generateFiles(512000) def runTests(host, port=None, listenport=None): generateTestData() node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testID(node, host, port) d.addBoth(cleanup, node) node.join() """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python """ FludFileOpTest.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. System tests for FludFileOperations """ import sys, os, time, logging, tempfile, shutil from twisted.internet import reactor sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludConfig import FludConfig from flud.FludNode import FludNode from flud.fencode import fencode, fdecode from flud.FludCrypto import generateRandom from flud.FludFileOperations import * import flud.FludDefer as FludDefer from flud.protocol.LocalClient import listMeta logger = logging.getLogger('flud') def testError(failure, message, node): print "testError message: %s" % message print "testError: %s" % str(failure) print "At least 1 test FAILED" return failure def gotSuccess(r, desc): print "%s succeeded" % desc def testConcurrent(r, node, files, desc): #print "r was %s" % r print "testConcurrent %s" % desc dlist = [] for file in files: d = testStoreFile(node, file) dlist.append(d) dl = FludDefer.ErrDeferredList(dlist) dl.addCallback(gotSuccess, desc) dl.addErrback(testError) return dl def checkStoreFile(res, node, fname): master = listMeta(node.config) if fname not in master: return defer.fail(failure.DefaultException("file not stored")) else: print "store on %s verified" % fname return res # <- *VITAL* for concurrent dup ops to succeed. def testStoreFile(node, fname): d = StoreFile(node, fname).deferred d.addCallback(checkStoreFile, node, fname) d.addErrback(testError, fname, node) return d def doTests(node, smallfnames, largefnames, dupsmall, duplarge): d = testStoreFile(node, smallfnames[0]) d.addCallback(testConcurrent, node, smallfnames, "small") d.addCallback(testConcurrent, node, largefnames, "large") d.addCallback(testConcurrent, node, dupsmall, "small duplicates") d.addCallback(testConcurrent, node, duplarge, "large duplicates") #d = testConcurrent(None, node, dupsmall, "small duplicates") #d = testConcurrent(None, node, duplarge, "large duplicates") return d def cleanup(_, node, filenamelist): #print _ for f in filenamelist: try: print "deleting %s" % f os.remove(f) except: print "couldn't remove %s" % f reactor.callLater(1, node.stop) def generateTestFile(minSize): fname = tempfile.mktemp() f = open(fname, 'w') data = generateRandom(minSize/50) for i in range(0, 51+random.randrange(50)): f.write(data) f.close() filename = os.path.join("/tmp",fname) os.rename(fname,filename) return filename def runTests(host, port, listenport=None): f1 = generateTestFile(5120) f2 = generateTestFile(5120) f3 = f2+".dup" shutil.copy(f2, f3) f4 = generateTestFile(513000) f5 = generateTestFile(513000) f6 = f5+".dup" shutil.copy(f5, f6) node = FludNode(port=listenport) if port == None: port = node.config.port node.run() node.connectViaGateway(host, port) d = doTests(node, [f1, f2], [f4, f5], [f2, f3], [f5, f6]) d.addBoth(cleanup, node, [f1, f2, f3, f4, f5, f6]) node.join() if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3])) else: print "must run this test against a flud network (no single node op)" print "usage: %s [<othernodehost othernodeport> |"\ " <othernodehost othernodeport listenport>]" % sys.argv[0]
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket, tempfile from twisted.python import failure from StringIO import StringIO from zlib import crc32 sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient import flud.FludCrypto as FludCrypto from flud.fencode import fencode, fdecode from flud.protocol.FludCommUtil import * from flud.FludDefer import ErrDeferredList """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ # metadatablock: (block#,n,k,blockdata) metadatablock = fencode((1,20,40,'adfdsfdffffffddddddddddddddd')) fake_mkey_offset = 111111 def testerror(failure, message, node): """ error handler for test errbacks """ print "testerror message: %s" % message print "testerror: %s" % str(failure) print "At least 1 test FAILED" return failure def allGood(_, nKu): print "all tests PASSED" return nKu def checkDELETE(res, nKu, fname, fkey, mkey, node, host, port, totalDelete): """ checks to ensure the file was deleted """ # totalDelete = True if this delete op should remove all meta (and data) if totalDelete: # try to retrieve with any metakey, should fail print "expecting failed retrieve, any metakey" return testRETRIEVE(res, nKu, fname, fkey, True, node, host, port, lambda args=(res, nKu): allGood(*args), False) else: # try to retrieve with any metakey, should succeed print "expecting successful retrieve, any metakey" return testRETRIEVE(res, nKu, fname, fkey, True, node, host, port, lambda args=(res, nKu, fname, fkey, mkey+fake_mkey_offset, node, host, port, True): testDELETE(*args)) def testDELETE(res, nKu, fname, fkey, mkey, node, host, port, totalDelete): """ Tests sendDelete, and invokes checkDELETE on success """ print "starting testDELETE %s.%s" % (fname, mkey) #return checkDELETE(None, nKu, fname, fkey, mkey, node, host, port, False) deferred = node.client.sendDelete(fkey, mkey, host, port, nKu) deferred.addCallback(checkDELETE, nKu, fname, fkey, mkey, node, host, port, totalDelete) deferred.addErrback(testerror, "failed at testDELETE", node) return deferred def checkVERIFY(res, nKu, fname, fkey, mkey, node, host, port, hash, newmeta): """ executes after testVERIFY """ if long(hash, 16) != long(res, 16): raise failure.DefaultException("verify didn't match: %s != %s" % (hash, res)) print "checkVERIFY (%s) %s success" % (newmeta, fname) if newmeta: return testDELETE(res, nKu, fname, fkey, mkey, node, host, port, False) else: return testVERIFY(nKu, fname, fkey, mkey, node, host, port, True) def testVERIFY(nKu, fname, fkey, mkey, node, host, port, newmeta): """ Test sendVerify """ # newmeta, if True, will generate new metadata to be stored during verify if newmeta: thismkey = mkey+fake_mkey_offset else: thismkey = mkey print "starting testVERIFY (%s) %s.%s" % (newmeta, fname, thismkey) fd = os.open(fname, os.O_RDONLY) fsize = os.fstat(fd)[stat.ST_SIZE] length = 20 offset = random.randrange(fsize-length) os.lseek(fd, offset, 0) data = os.read(fd, length) os.close(fd) hash = FludCrypto.hashstring(data) deferred = node.client.sendVerify(fkey, offset, length, host, port, nKu, (thismkey, StringIO(metadatablock))) deferred.addCallback(checkVERIFY, nKu, fname, fkey, mkey, node, host, port, hash, newmeta) deferred.addErrback(testerror, "failed at testVERIFY (%s)" % newmeta, node) return deferred def failedRETRIEVE(res, nextCallable): return nextCallable(); def checkRETRIEVE(res, nKu, fname, fkey, mkey, node, host, port, nextCallable): """ Compares the file that was stored with the one that was retrieved """ f1 = open(fname) filename = [f for f in res if f[-len(fkey):] == fkey][0] f2 = open(filename) if (f1.read() != f2.read()): f1.close() f2.close() raise failure.DefaultException( "upload/download (%s, %s) files don't match" % (fname, os.path.join(node.config.clientdir, fkey))) #print "%s (%d) and %s (%d) match" % (fname, os.stat(fname)[stat.ST_SIZE], # filename, os.stat(filename)[stat.ST_SIZE]) f1.close() f2.close() if mkey != True: expectedmeta = "%s.%s.meta" % (fkey, mkey) metanames = [f for f in res if f[-len(expectedmeta):] == expectedmeta] if not metanames: raise failure.DefaultException("expected metadata was missing") f3 = open(metanames[0]) md = f3.read() if md != metadatablock: raise failure.DefaultException("upload/download metadata doesn't" " match (%s != %s)" % (md, metadatablock)) return nextCallable() def testRETRIEVE(res, nKu, fname, fkey, mkey, node, host, port, nextCallable, expectSuccess=True): """ Tests sendRetrieve, and invokes checkRETRIEVE on success """ print "starting testRETRIEVE %s.%s" % (fname, mkey) deferred = node.client.sendRetrieve(fkey, host, port, nKu, mkey) deferred.addCallback(checkRETRIEVE, nKu, fname, fkey, mkey, node, host, port, nextCallable) if expectSuccess: deferred.addErrback(testerror, "failed at testRETRIEVE", node) else: deferred.addErrback(failedRETRIEVE, nextCallable) return deferred def testSTORE2(nKu, fname, fkey, node, host, port): mkey = crc32(fname) mkey2 = mkey+(2*fake_mkey_offset) print "starting testSTORE %s.%s" % (fname, mkey2) deferred = node.client.sendStore(fname, (mkey2, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey2, node, host, port, lambda args=(nKu, fname, fkey, mkey, node, host, port, False): testVERIFY(*args)) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred def testSTORE(nKu, fname, fkey, node, host, port): """ Tests sendStore, and invokes testRETRIEVE on success """ mkey = crc32(fname) print "starting testSTORE %s.%s" % (fname, mkey) deferred = node.client.sendStore(fname, (mkey, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey, node, host, port, lambda args=(nKu, fname, fkey, node, host, port): testSTORE2(*args)) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred def testID(node, host, port): """ Tests sendGetID(), and invokes testSTORE on success """ print "starting testID" deferred = node.client.sendGetID(host, port) deferred.addErrback(testerror, "failed at testID", node) return deferred def testAggSTORE(nKu, aggFiles, node, host, port): print "starting testAggSTORE" dlist = [] for fname, fkey in aggFiles: mkey = crc32(fname) print "testAggSTORE %s (%s)" % (fname, mkey) deferred = node.client.sendStore(fname, (mkey, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey, node, host, port, lambda args=(nKu, fname, fkey, mkey, node, host, port, False): testVERIFY(*args)) deferred.addErrback(testerror, "failed at testAggSTORE", node) dlist.append(deferred) dl = ErrDeferredList(dlist) dl.addCallback(allGood, nKu) dl.addErrback(testerror, "failed at testAggSTORE", node) return dl def cleanup(_, node, filenamelist): for f in filenamelist: try: os.remove(f) except: print "couldn't remove %s" % f reactor.callLater(1, node.stop) def generateTestData(minSize): fname = tempfile.mktemp() f = open(fname, 'w') data = FludCrypto.generateRandom(minSize/50) for i in range(0, 51+random.randrange(50)): f.write(data) f.close() filekey = FludCrypto.hashfile(fname) filekey = fencode(int(filekey, 16)) filename = os.path.join("/tmp",filekey) os.rename(fname,filename) return (filename, filekey) def runTests(host, port=None, listenport=None): (largeFilename, largeFilekey) = generateTestData(512000) (smallFilename, smallFilekey) = generateTestData(5120) aggFiles = [] for i in range(4): aggFiles.append(generateTestData(4096)) node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testID(node, host, port) d.addCallback(testSTORE, largeFilename, largeFilekey, node, host, port) d.addCallback(testSTORE, smallFilename, smallFilekey, node, host, port) d.addCallback(testAggSTORE, aggFiles, node, host, port) d.addBoth(cleanup, node, [i[0] for i in aggFiles] + [largeFilename, smallFilename]) node.join() def main(): localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3])) if __name__ == '__main__': main()
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure from twisted.internet import defer sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import * from flud.FludDefer import ErrDeferredList """ Test code for kprimitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ # XXX: check return from ops to see if they passed (e.g., if STORE fails, we # are notified [currently] by the html page that is returned). # XXX: should make a random file each time this is run... CONCURRENT=50 CONCREPORT=10 node = None # format of block metadata is # {(i, datakey): storingNodeID, ..., 'n': n, 'm': m} # where i<n+m, datakey is a sha256 of the data stored, and storingNodeID is # either a nodeID or a list of nodeIDs testval = {(0, 802484L): 465705L, (1, 780638L): 465705L, (2, 169688L): 465705L, (3, 267175L): 465705L, (4, 648636L): 465705L, (5, 838315L): 465705L, (6, 477619L): 465705L, (7, 329906L): 465705L, (8, 610565L): 465705L, (9, 217811L): 465705L, (10, 374124L): 465705L, (11, 357214L): 465705L, (12, 147307L): 465705L, (13, 427751L): 465705L, (14, 927853L): 465705L, (15, 760369L): 465705L, (16, 707029L): 465705L, (17, 479234L): 465705L, (18, 190455L): 465705L, (19, 647489L): 465705L, (20, 620470L): 465705L, (21, 777532L): 465705L, (22, 622383L): 465705L, (23, 573283L): 465705L, (24, 613082L): 465705L, (25, 433593L): 465705L, (26, 584543L): 465705L, (27, 337485L): 465705L, (28, 911014L): 465705L, (29, 594065L): 465705L, (30, 375876L): 465705L, (31, 726818L): 465705L, (32, 835759L): 465705L, (33, 814060L): 465705L, (34, 237176L): 465705L, (35, 538268L): 465705L, (36, 272650L): 465705L, (37, 314058L): 465705L, (38, 257714L): 465705L, (39, 439931L): 465705L, 'k': 20, 'n': 20} logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) logger.setLevel(logging.DEBUG) def suitesuccess(results): logger.info("all tests in suite passed") #print results return results def suiteerror(failure): logger.info("suite did not complete") logger.info("DEBUG: %s" % failure) return failure def stagesuccess(result, message): logger.info("stage %s succeeded" % message) return result def stageerror(failure, message): logger.info("stage %s failed" % message) #logger.info("DEBUG: %s" % failure) return failure def itersuccess(res, i, message): if i % CONCREPORT == 0: logger.info("itersuccess: %s" % message) return res def itererror(failure, message): logger.info("itererror message: %s" % message) #logger.info("DEBUG: %s" % failure) #logger.info("DEBUG: %s" % dir(failure) failure.printTraceback() return failure def endtests(res, nKu, node, host, port): """ executes after all tests """ try: res = fdecode(res) except ValueError: pass if res != testval: return failure.DefaultException("retrieved value does not match" " stored value: '%s' != '%s'" % (res, testval)) logger.log(logging.INFO,"testkFindVal PASSED: %s\n" % str(res)) logger.log(logging.INFO,"all tests PASSED") return res def testkFindVal(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO,"testSendkFindVal PASSED: %s\n" % str(res)) logger.log(logging.INFO,"attempting testkFindValue") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kFindValue(key) deferred.addErrback(itererror, "kFindValue") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kFindValue") d.addErrback(stageerror, 'kFindValue') d.addCallback(endtests, nKu, node, host, port) return d def testSendkFindVal(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkStore PASSED: %s\n" % str(res)) logger.log(logging.INFO, "attempting testSendkFindValue") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkFindValue(host, port, key) deferred.addErrback(itererror, "sendkFindValue") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkFindValue") d.addErrback(stageerror, 'sendkFindValue') d.addCallback(testkFindVal, nKu, node, host, port, num) return d def testkStore(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testSendkStore PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting testkStore") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kStore(key, testval) deferred.addErrback(itererror, "kStore") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kStore") d.addErrback(stageerror, 'kStore') d.addCallback(testSendkFindVal, nKu, node, host, port, num) return d def testSendkStore(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkFindNode PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting testSendkStore") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkStore(host, port, key, testval) deferred.addErrback(itererror, "sendkStore") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkStore") d.addErrback(stageerror, 'sendkStore') d.addCallback(testkStore, nKu, node, host, port, num) return d def testkFindNode(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testSendkFindNode PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting kFindNode") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kFindNode(key) deferred.addErrback(itererror, "kFindNode") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kFindNode") d.addErrback(stageerror, 'kFindNode') d.addCallback(testSendkStore, nKu, node, host, port, num) return d def testSendkFindNode(nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkGetID PASSED") logger.log(logging.INFO, "attempting sendkFindNode") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkFindNode(host, port, key) deferred.addErrback(itererror, "sendkFindNode") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkFindNode") d.addErrback(stageerror, 'sendkFindNode') d.addCallback(testkFindNode, nKu, node, host, port, num) return d def testGetID(node, host, port, num=CONCURRENT): """ Tests sendGetID(), and invokes testSendkFindNode on success """ deferred = node.client.sendGetID(host, port) deferred.addCallback(testSendkFindNode, node, host, port, num) deferred.addErrback(stageerror, "testGetID") return deferred def runTests(host, port=None, listenport=None): num = CONCURRENT #num = 5 node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testGetID(node, host, port, CONCURRENT) d.addCallback(suitesuccess) d.addErrback(suiteerror) d.addBoth(cleanup, node) node.join() #node.start() # doesn't work, because reactor may not have started # listening by time requests start flying def cleanup(_, node): logger.info("shutting down in 1 seconds...") time.sleep(1) reactor.callLater(1, node.stop) logger.info("done cleaning up") """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: print "Warning: testing against self may result in timeout failures" runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure from twisted.internet import defer sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import * from flud.FludDefer import ErrDeferredList """ Test code for kprimitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ # XXX: check return from ops to see if they passed (e.g., if STORE fails, we # are notified [currently] by the html page that is returned). # XXX: should make a random file each time this is run... CONCURRENT=50 CONCREPORT=10 node = None # format of block metadata is # {(i, datakey): storingNodeID, ..., 'n': n, 'm': m} # where i<n+m, datakey is a sha256 of the data stored, and storingNodeID is # either a nodeID or a list of nodeIDs testval = {(0, 802484L): 465705L, (1, 780638L): 465705L, (2, 169688L): 465705L, (3, 267175L): 465705L, (4, 648636L): 465705L, (5, 838315L): 465705L, (6, 477619L): 465705L, (7, 329906L): 465705L, (8, 610565L): 465705L, (9, 217811L): 465705L, (10, 374124L): 465705L, (11, 357214L): 465705L, (12, 147307L): 465705L, (13, 427751L): 465705L, (14, 927853L): 465705L, (15, 760369L): 465705L, (16, 707029L): 465705L, (17, 479234L): 465705L, (18, 190455L): 465705L, (19, 647489L): 465705L, (20, 620470L): 465705L, (21, 777532L): 465705L, (22, 622383L): 465705L, (23, 573283L): 465705L, (24, 613082L): 465705L, (25, 433593L): 465705L, (26, 584543L): 465705L, (27, 337485L): 465705L, (28, 911014L): 465705L, (29, 594065L): 465705L, (30, 375876L): 465705L, (31, 726818L): 465705L, (32, 835759L): 465705L, (33, 814060L): 465705L, (34, 237176L): 465705L, (35, 538268L): 465705L, (36, 272650L): 465705L, (37, 314058L): 465705L, (38, 257714L): 465705L, (39, 439931L): 465705L, 'k': 20, 'n': 20} logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) logger.setLevel(logging.DEBUG) def suitesuccess(results): logger.info("all tests in suite passed") #print results return results def suiteerror(failure): logger.info("suite did not complete") logger.info("DEBUG: %s" % failure) return failure def stagesuccess(result, message): logger.info("stage %s succeeded" % message) return result def stageerror(failure, message): logger.info("stage %s failed" % message) #logger.info("DEBUG: %s" % failure) return failure def itersuccess(res, i, message): if i % CONCREPORT == 0: logger.info("itersuccess: %s" % message) return res def itererror(failure, message): logger.info("itererror message: %s" % message) #logger.info("DEBUG: %s" % failure) #logger.info("DEBUG: %s" % dir(failure) failure.printTraceback() return failure def endtests(res, nKu, node, host, port): """ executes after all tests """ try: res = fdecode(res) except ValueError: pass if res != testval: return failure.DefaultException("retrieved value does not match" " stored value: '%s' != '%s'" % (res, testval)) logger.log(logging.INFO,"testkFindVal PASSED: %s\n" % str(res)) logger.log(logging.INFO,"all tests PASSED") return res def testkFindVal(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO,"testSendkFindVal PASSED: %s\n" % str(res)) logger.log(logging.INFO,"attempting testkFindValue") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kFindValue(key) deferred.addErrback(itererror, "kFindValue") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kFindValue") d.addErrback(stageerror, 'kFindValue') d.addCallback(endtests, nKu, node, host, port) return d def testSendkFindVal(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkStore PASSED: %s\n" % str(res)) logger.log(logging.INFO, "attempting testSendkFindValue") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkFindValue(host, port, key) deferred.addErrback(itererror, "sendkFindValue") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkFindValue") d.addErrback(stageerror, 'sendkFindValue') d.addCallback(testkFindVal, nKu, node, host, port, num) return d def testkStore(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testSendkStore PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting testkStore") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kStore(key, testval) deferred.addErrback(itererror, "kStore") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kStore") d.addErrback(stageerror, 'kStore') d.addCallback(testSendkFindVal, nKu, node, host, port, num) return d def testSendkStore(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkFindNode PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting testSendkStore") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkStore(host, port, key, testval) deferred.addErrback(itererror, "sendkStore") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkStore") d.addErrback(stageerror, 'sendkStore') d.addCallback(testkStore, nKu, node, host, port, num) return d def testkFindNode(res, nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testSendkFindNode PASSED: %s" % str(res)) logger.log(logging.INFO, "attempting kFindNode") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.kFindNode(key) deferred.addErrback(itererror, "kFindNode") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "kFindNode") d.addErrback(stageerror, 'kFindNode') d.addCallback(testSendkStore, nKu, node, host, port, num) return d def testSendkFindNode(nKu, node, host, port, num=CONCURRENT): logger.log(logging.INFO, "testkGetID PASSED") logger.log(logging.INFO, "attempting sendkFindNode") dlist = [] for i in range(num): key = random.randrange(2**256) deferred = node.client.sendkFindNode(host, port, key) deferred.addErrback(itererror, "sendkFindNode") dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "sendkFindNode") d.addErrback(stageerror, 'sendkFindNode') d.addCallback(testkFindNode, nKu, node, host, port, num) return d def testGetID(node, host, port, num=CONCURRENT): """ Tests sendGetID(), and invokes testSendkFindNode on success """ deferred = node.client.sendGetID(host, port) deferred.addCallback(testSendkFindNode, node, host, port, num) deferred.addErrback(stageerror, "testGetID") return deferred def runTests(host, port=None, listenport=None): num = CONCURRENT #num = 5 node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testGetID(node, host, port, CONCURRENT) d.addCallback(suitesuccess) d.addErrback(suiteerror) d.addBoth(cleanup, node) node.join() #node.start() # doesn't work, because reactor may not have started # listening by time requests start flying def cleanup(_, node): logger.info("shutting down in 1 seconds...") time.sleep(1) reactor.callLater(1, node.stop) logger.info("done cleaning up") """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: print "Warning: testing against self may result in timeout failures" runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket, shutil, tempfile from binascii import crc32 from StringIO import StringIO from twisted.python import failure sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.FludCrypto import generateRandom, hashfile from flud.protocol.FludCommUtil import * from flud.fencode import fencode """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ smallfilekey = "" smallfilename = "" smallfilenamebad = "" largefilekey = "" largefilename = "" largefilenamebad = "" metadata = 'aaaa' def testerror(failure, message, node): """ error handler for test errbacks """ print "testerror message: %s" % message print "testerror: %s" % str(failure) print "At least 1 test FAILED" raise failure def testUnexpectedSuccess(res, message, node): print "unexpected success message: %s" % message print "At least 1 test succeeded when it should have failed" raise "bad" def testDELETEBadKeyFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg # the end else: # XXX: here and elsewhere, raise something more descriptive, otherwise # its waay confusing print "the following trace may be misleading..." raise failure def testDELETEBadKey(nKu, node, host, port): print "starting testDELETEBadKey" path = os.path.join("somedir", largefilekey) deferred = node.client.sendDelete(path, crc32(path), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "DELETE with bad key succeeded", node) deferred.addErrback(testDELETEBadKeyFailed, "DELETE with bad key failed as expected", node, nKu, host, port) return deferred def testVERIFYBadKeyFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testDELETEBadKey(nKu, node, host, port) else: raise failure def testVERIFYBadKey(nKu, node, host, port): print "starting testVERIFYBadKey" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize-20 deferred = node.client.sendVerify(smallfilenamebad, offset, 5, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad key succeeded", node) deferred.addErrback(testVERIFYBadKeyFailed, "VERIFY of bad filekey failed as expected", node, nKu, host, port) return deferred def testVERIFYBadLengthFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYBadKey(nKu, node, host, port) else: raise failure def testVERIFYBadLength(nKu, node, host, port): print "starting testVERIFYBadOffset" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize-10 deferred = node.client.sendVerify(smallfilekey, offset, 20, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad length", node) deferred.addErrback(testVERIFYBadLengthFailed, "VERIFY of bad length failed as expected", node, nKu, host, port) return deferred def testVERIFYBadOffsetFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYBadLength(nKu, node, host, port) else: print "VERIFYBadOffset failed as expected, but with wrong failure" raise failure def testVERIFYBadOffset(nKu, node, host, port): print "starting testVERIFYBadOffset" fsize = os.stat(smallfilename)[stat.ST_SIZE] offset = fsize+2 deferred = node.client.sendVerify(smallfilekey, offset, 20, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified file with bad offset", node) deferred.addErrback(testVERIFYBadOffsetFailed, "VERIFY of bad offset failed as expected", node, nKu, host, port) return deferred def testVERIFYNotFoundFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testVERIFYBadOffset(nKu, node, host, port) else: raise failure def testVERIFYNotFound(nKu, node, host, port): print "starting testVERIFYNotFound" deferred = node.client.sendVerify(largefilekey, 10, 10, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "verified non-existent file", node) deferred.addErrback(testVERIFYNotFoundFailed, "VERIFY of non-existent file failed as expected", node, nKu, host, port) return deferred def testRETRIEVEIllegalPathFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadRequestException'): print "%s" % msg return testVERIFYNotFound(nKu, node, host, port) else: raise failure def testRETRIEVEIllegalPath(nKu, node, host, port): print "starting testRETRIEVEIllegalPath" deferred = node.client.sendRetrieve(os.path.join("somedir",smallfilekey), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "retrieved file with illegal path", node) deferred.addErrback(testRETRIEVEIllegalPathFailed, "RETRIEVE using illegal path failed as expected", node, nKu, host, port) return deferred def testRETRIEVENotFoundFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.NotFoundException'): print "%s" % msg return testRETRIEVEIllegalPath(nKu, node, host, port) else: raise failure def testRETRIEVENotFound(nKu, node, host, port): print "starting testRETRIEVENotFound" deferred = node.client.sendRetrieve(largefilekey, host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "retrieved file that shouldn't exist", node) deferred.addErrback(testRETRIEVENotFoundFailed, "RETRIEVE of non-existent file failed as expected", node, nKu, host, port) return deferred def testSTORELargeFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadCASKeyException'): print "%s" % msg return testRETRIEVENotFound(nKu, node, host, port) else: raise failure def testSTOREBadKeyLarge(nKu, node, host, port): print "starting testSTOREBadKeyLarge" deferred = node.client.sendStore(largefilenamebad, (crc32(largefilenamebad), StringIO(metadata)), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "large file, bad key succeeded", node) deferred.addErrback(testSTORELargeFailed, "large STORE with bad key failed as expected", node, nKu, host, port) return deferred def testSTORESmallFailed(failure, msg, node, nKu, host, port): if failure.check('flud.protocol.FludCommUtil.BadCASKeyException'): print "%s" % msg return testSTOREBadKeyLarge(nKu, node, host, port) else: raise failure def testSTOREBadKeySmall(nKu, node, host, port): print "starting testSTOREBadKeySmall" deferred = node.client.sendStore(smallfilenamebad, (crc32(smallfilenamebad), StringIO(metadata)), host, port, nKu) deferred.addCallback(testUnexpectedSuccess, "small file, bad key succeeded", node) deferred.addErrback(testSTORESmallFailed, "small STORE with bad key failed as expected", node, nKu, host, port) return deferred def testSTORESuccess(res, nKu, node, host, port): print "testSTORE succeeded: %s" % res return testSTOREBadKeySmall(nKu, node, host, port) def testSTORE(nKu, node, host, port): # store a file successfully for later failure tests (VERIFY, etc) print "starting testSTORE" deferred = node.client.sendStore(smallfilename, (crc32(smallfilename), StringIO(metadata)), host, port, nKu) deferred.addCallback(testSTORESuccess, nKu, node, host, port) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred # XXX: need to test bogus headers for all commands (BAD_REQUEST) # XXX: need to test failures for authentication def testID(node, host, port): """ Tests sendGetID(), and invokes testSTORE on success """ print "starting testID" deferred = node.client.sendGetID(host, port) deferred.addCallback(testSTORE, node, host, port) #deferred.addCallback(testSTOREBadKeySmall, node, host, port) deferred.addErrback(testerror, "failed at testID", node) return deferred def cleanup(err, node): if err: print "cleaning up: %s" % err else: print "cleaning up" os.remove(smallfilename) os.remove(smallfilenamebad) os.remove(largefilename) os.remove(largefilenamebad) reactor.callLater(1, node.stop) def generateTestData(): def generateFiles(minsize): fname = tempfile.mktemp() f = open(fname, 'w') f.write('\0'*minsize) f.write(generateRandom(random.randrange(256)+1)) f.close() filekey = hashfile(fname) filekey = fencode(int(filekey, 16)) filename = os.path.join("/tmp",filekey) os.rename(fname,filename) filenamebad = os.path.join("/tmp/","bad"+filekey[3:]) shutil.copy(filename, filenamebad) return (filekey, filename, filenamebad) global smallfilekey global smallfilename global smallfilenamebad (smallfilekey, smallfilename, smallfilenamebad) = generateFiles(1024) global largefilekey global largefilename global largefilenamebad (largefilekey, largefilename, largefilenamebad) = generateFiles(512000) def runTests(host, port=None, listenport=None): generateTestData() node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testID(node, host, port) d.addBoth(cleanup, node) node.join() """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python """ FludFileOpTest.py, (c) 2003-2006 Alen Peacock. This program is distributed under the terms of the GNU General Public License (the GPL), version 3. System tests for FludFileOperations """ import sys, os, time, logging, tempfile, shutil from twisted.internet import reactor sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludConfig import FludConfig from flud.FludNode import FludNode from flud.fencode import fencode, fdecode from flud.FludCrypto import generateRandom from flud.FludFileOperations import * import flud.FludDefer as FludDefer from flud.protocol.LocalClient import listMeta logger = logging.getLogger('flud') def testError(failure, message, node): print "testError message: %s" % message print "testError: %s" % str(failure) print "At least 1 test FAILED" return failure def gotSuccess(r, desc): print "%s succeeded" % desc def testConcurrent(r, node, files, desc): #print "r was %s" % r print "testConcurrent %s" % desc dlist = [] for file in files: d = testStoreFile(node, file) dlist.append(d) dl = FludDefer.ErrDeferredList(dlist) dl.addCallback(gotSuccess, desc) dl.addErrback(testError) return dl def checkStoreFile(res, node, fname): master = listMeta(node.config) if fname not in master: return defer.fail(failure.DefaultException("file not stored")) else: print "store on %s verified" % fname return res # <- *VITAL* for concurrent dup ops to succeed. def testStoreFile(node, fname): d = StoreFile(node, fname).deferred d.addCallback(checkStoreFile, node, fname) d.addErrback(testError, fname, node) return d def doTests(node, smallfnames, largefnames, dupsmall, duplarge): d = testStoreFile(node, smallfnames[0]) d.addCallback(testConcurrent, node, smallfnames, "small") d.addCallback(testConcurrent, node, largefnames, "large") d.addCallback(testConcurrent, node, dupsmall, "small duplicates") d.addCallback(testConcurrent, node, duplarge, "large duplicates") #d = testConcurrent(None, node, dupsmall, "small duplicates") #d = testConcurrent(None, node, duplarge, "large duplicates") return d def cleanup(_, node, filenamelist): #print _ for f in filenamelist: try: print "deleting %s" % f os.remove(f) except: print "couldn't remove %s" % f reactor.callLater(1, node.stop) def generateTestFile(minSize): fname = tempfile.mktemp() f = open(fname, 'w') data = generateRandom(minSize/50) for i in range(0, 51+random.randrange(50)): f.write(data) f.close() filename = os.path.join("/tmp",fname) os.rename(fname,filename) return filename def runTests(host, port, listenport=None): f1 = generateTestFile(5120) f2 = generateTestFile(5120) f3 = f2+".dup" shutil.copy(f2, f3) f4 = generateTestFile(513000) f5 = generateTestFile(513000) f6 = f5+".dup" shutil.copy(f5, f6) node = FludNode(port=listenport) if port == None: port = node.config.port node.run() node.connectViaGateway(host, port) d = doTests(node, [f1, f2], [f4, f5], [f2, f3], [f5, f6]) d.addBoth(cleanup, node, [f1, f2, f3, f4, f5, f6]) node.join() if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3])) else: print "must run this test against a flud network (no single node op)" print "usage: %s [<othernodehost othernodeport> |"\ " <othernodehost othernodeport listenport>]" % sys.argv[0]
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import fencode, fdecode """ Test code for primitive DHT operations. """ stay_alive = 1 filename = "/tmp/tempstoredata" filekey = os.path.basename(filename) key = 87328673569979667228965797330646992089697345905484734072690869757741450870337L # format of block metadata is # {(i, datakey): storingNodeID, ..., 'n': n, 'm': m} # where i<n+m, datakey is a sha256 of the data stored, and storingNodeID is # either a nodeID or a list of nodeIDs. testval = {(0, 802484L): 465705L, (1, 780638L): 465705L, (2, 169688L): 465705L, (3, 267175L): 465705L, (4, 648636L): 465705L, (5, 838315L): 465705L, (6, 477619L): 465705L, (7, 329906L): 465705L, (8, 610565L): 465705L, (9, 217811L): 465705L, (10, 374124L): 465705L, (11, 357214L): 465705L, (12, 147307L): 465705L, (13, 427751L): 465705L, (14, 927853L): 465705L, (15, 760369L): 465705L, (16, 707029L): 465705L, (17, 479234L): 465705L, (18, 190455L): 465705L, (19, 647489L): 465705L, (20, 620470L): 465705L, (21, 777532L): 465705L, (22, 622383L): 465705L, (23, 573283L): 465705L, (24, 613082L): 465705L, (25, 433593L): 465705L, (26, 584543L): 465705L, (27, 337485L): 465705L, (28, 911014L): 465705L, (29, 594065L): 465705L, (30, 375876L): 465705L, (31, 726818L): 465705L, (32, 835759L): 465705L, (33, 814060L): 465705L, (34, 237176L): 465705L, (35, 538268L): 465705L, (36, 272650L): 465705L, (37, 314058L): 465705L, (38, 257714L): 465705L, (39, 439931L): 465705L, 'k': 20, 'n': 20} logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) #logger.setLevel(logging.DEBUG) logger.setLevel(logging.INFO) def cleanup(_, node): logger.info("waiting %ds to shutdown..." % stay_alive) reactor.callLater(stay_alive, node.stop) def testerror(failure, message, node): """ error handler for test errbacks """ logger.warn("testerror message: %s" % message) logger.warn("testerror: '%s'" % str(failure)) logger.warn("At least 1 test FAILED") def endtests(res, nKu, node, host, port): """ executes after all tests """ try: res = fdecode(res) except ValueError: pass if res != testval: return testerror(None, "retrieved value does not match stored value:" " '%s' != '%s'" % (res, testval), node) logger.info("testkFindVal PASSED") logger.debug("testkFindVal result: %s" % str(res)) logger.info("all tests PASSED") return res def testkFindVal(res, nKu, node, host, port): logger.info("testSendkFindVal PASSED") logger.debug("testSendkFindVal result: %s" % str(res)) logger.info("attempting testkFindValue") deferred = node.client.kFindValue(key) deferred.addCallback(endtests, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkFindValue", node) return deferred def testSendkFindVal(res, nKu, node, host, port): logger.info("testkStore PASSED") logger.debug("testkStore result: %s" % str(res)) logger.info("attempting testSendkFindValue") deferred = node.client.sendkFindValue(host, port, key) deferred.addCallback(testkFindVal, nKu, node, host, port) deferred.addErrback(testerror, "failed at testSendkFindValue", node) return deferred def testkStore(res, nKu, node, host, port): logger.info("testSendkStore PASSED") logger.debug("testSendkStore result: %s" % str(res)) logger.info("attempting testkStore") deferred = node.client.kStore(key, testval) deferred.addCallback(testSendkFindVal, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkStore", node) return deferred def testSendkStore(res, nKu, node, host, port): logger.info("testkFindNode PASSED") logger.debug("testkFindNode result: %s" % str(res)) logger.info("attempting testSendkStore") deferred = node.client.sendkStore(host, port, key, testval) deferred.addCallback(testkStore, nKu, node, host, port) deferred.addErrback(testerror, "failed at testkStore", node) return deferred def testkFindNode(res, nKu, node, host, port): """ executes after testSendkFindNode """ logger.info("testSendkFindNode PASSED") logger.debug("testSendkFindNode result: %s" % str(res)) logger.info("attempting kFindNode") deferred = node.client.kFindNode(key) deferred.addCallback(testSendkStore, nKu, node, host, port) deferred.addErrback(testerror, "failed at kFindNode", node) return deferred def testSendkFindNode(nKu, node, host, port): """ executes after testGetID """ logger.info("testkGetID PASSED") logger.info("attempting sendkFindNode") deferred = node.client.sendkFindNode(host, port, key) deferred.addCallback(testkFindNode, nKu, node, host, port) deferred.addErrback(testerror, "failed at sendkFindNode", node) return deferred def testGetID(node, host, port): """ Tests sendGetID(), and invokes testSendkFindNode on success """ deferred = node.client.sendGetID(host, port) deferred.addCallback(testSendkFindNode, node, host, port) deferred.addErrback(testerror, "failed at testGetID", node) return deferred def runTests(host, port=None, listenport=None): host = getCanonicalIP(host) node = FludNode(port=listenport) if port == None: port = node.config.port logger.info("testing against %s:%s, localport=%s" % (host, port, listenport)) node.run() d = testGetID(node, host, port) d.addBoth(cleanup, node) #testkFindVal("blah", node.config.Ku, node, host, port) node.join() """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) elif len(sys.argv) == 4: runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import tarfile, tempfile, random, os, sys import gzip from Crypto.Hash import SHA256 sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.fencode import fencode import flud.TarfileUtils as TarfileUtils def maketarball(numfiles, avgsize, hashnames=False, addmetas=False): tarballname = tempfile.mktemp()+".tar" tarball = tarfile.open(tarballname, 'w') if addmetas: metafname = tempfile.mktemp() metaf = file(metafname, 'w') metaf.write('m'*48) metaf.close() for i in xrange(numfiles): fname = tempfile.mktemp() f = file(fname, 'wb') size = int(avgsize * (random.random()+0.5)) blocksize = 65*1024 if hashnames: sha256 = SHA256.new() for j in range(0, size, blocksize): if j+blocksize > size: block = 'a'*(size-j) else: block = 'a'*blocksize if hashnames: sha256.update(block) f.write(block) f.close() arcname = fname if hashnames: arcname = fencode(int(sha256.hexdigest(),16)) tarball.add(fname, arcname) if addmetas: tarball.add(metafname, arcname+".343434.meta") os.remove(fname) if addmetas: os.remove(metafname) contents = tarball.getnames() tarball.close() return tarballname, contents def gzipTarball(tarball): f = gzip.GzipFile(tarball+".gz", 'wb') f.write(file(tarball, 'rb').read()) f.close() os.remove(tarball) return tarball+".gz" def main(): # test plain TarfileUtils.delete() (tballname, contents) = maketarball(5, 4096) TarfileUtils.delete(tballname, contents[2:4]) tarball = tarfile.open(tballname, 'r') os.remove(tballname) assert(tarball.getnames() == contents[:2]+contents[4:]) tarball.close() # test gzip TarfileUtils.delete() (tballname, contents) = maketarball(5, 4096) tballname = gzipTarball(tballname) TarfileUtils.delete(tballname, contents[2:4]) tarball = tarfile.open(tballname, 'r') os.remove(tballname) assert(tarball.getnames() == contents[:2]+contents[4:]) tarball.close() # test plain TarfileUtils.concatenate() (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(gz, plain) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname1 = gzipTarball(tballname1) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(plain, gz) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname2 = gzipTarball(tballname2) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(gz, gz) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname1 = gzipTarball(tballname1) tballname2 = gzipTarball(tballname2) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.verifyHashes(plain no meta) (tballname, contents) = maketarball(5, 4096, True) assert(TarfileUtils.verifyHashes(tballname, contents[2:4])) os.remove(tballname) # test TarfileUtils.verifyHashes(plain with meta) (tballname, contents) = maketarball(5, 4096, True, True) assert(TarfileUtils.verifyHashes(tballname, contents[2:4]), ".meta") os.remove(tballname) # test TarfileUtils.verifyHashes(gzipped no meta) (tballname, contents) = maketarball(5, 4096, True) tballname = gzipTarball(tballname) assert(TarfileUtils.verifyHashes(tballname, contents[2:4])) os.remove(tballname) # test TarfileUtils.verifyHashes(gzipped with meta) (tballname, contents) = maketarball(5, 4096, True, True) tballname = gzipTarball(tballname) assert(TarfileUtils.verifyHashes(tballname, contents[2:4]), ".meta") os.remove(tballname) print "all tests passed" if __name__ == "__main__": main()
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure from twisted.internet import defer sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) import flud.FludCrypto from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import fencode, fdecode from flud.FludDefer import ErrDeferredList """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ CONCURRENT=300 CONCREPORT=50 node = None files = None logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) logger.setLevel(logging.DEBUG) def suitesuccess(results): logger.info("all tests in suite passed") #print results return results def suiteerror(failure): logger.info("suite did not complete") logger.info("DEBUG: %s" % failure) return failure def stagesuccess(result, message): logger.info("stage %s succeeded" % message) return result def stageerror(failure, message): logger.info("stage %s failed" % message) #logger.info("DEBUG: %s" % failure) return failure def itersuccess(res, i, message): if i % CONCREPORT == 0: logger.info("itersuccess: %s" % message) return res def itererror(failure, message): logger.info("itererror message: %s" % message) #logger.info("DEBUG: %s" % failure) #logger.info("DEBUG: %s" % dir(failure) failure.printTraceback() return failure def checkVERIFY(results, nKu, host, port, hashes, num=CONCURRENT): logger.info(" checking VERIFY results...") for i in range(num): hash = hashes[i] res = results[i][1] if long(hash, 16) != long(res, 16): raise failure.DefaultException("verify didn't match: %s != %s" % (hash, res)) logger.info(" ...VERIFY results good.") return results #True def testVERIFY(res, nKu, host, port, num=CONCURRENT): logger.info("testVERIFY started...") dlist = [] hashes = [] for i in range(num): #if i == 4: # port = 21 fd = os.open(files[i], os.O_RDONLY) fsize = os.fstat(fd)[stat.ST_SIZE] length = 20 offset = random.randrange(fsize-length) os.lseek(fd, offset, 0) data = os.read(fd, length) os.close(fd) hashes.append(flud.FludCrypto.hashstring(data)) filekey = os.path.basename(files[i]) deferred = node.client.sendVerify(filekey, offset, length, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testVERIFY %d" % i) deferred.addErrback(itererror, "failed at testVERIFY %d: %s" % (i, filekey)) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testVERIFY") d.addErrback(stageerror, 'failed at testVERIFY') d.addCallback(checkVERIFY, nKu, host, port, hashes, num) return d def checkRETRIEVE(res, nKu, host, port, num=CONCURRENT): logger.info(" checking RETRIEVE results...") for i in range(num): f1 = open(files[i]) filekey = os.path.basename(files[i]) f2 = open(node.config.clientdir+"/"+filekey) if (f1.read() != f2.read()): f1.close() f2.close() raise failure.DefaultException("upload/download files don't match") f2.close() f1.close() logger.info(" ...RETRIEVE results good.") return testVERIFY(res, nKu, host, port, num) def testRETRIEVE(res, nKu, host, port, num=CONCURRENT): logger.info("testRETRIEVE started...") dlist = [] for i in range(num): #if i == 4: # port = 21 filekey = os.path.basename(files[i]) deferred = node.client.sendRetrieve(filekey, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testRETRIEVE %d" % i) deferred.addErrback(itererror, "failed at testRETRIEVE %d: %s" % (i, filekey)) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testRETRIEVE") d.addErrback(stageerror, 'failed at testRETRIEVE') d.addCallback(checkRETRIEVE, nKu, host, port, num) return d def testSTORE(nKu, host, port, num=CONCURRENT): logger.info("testSTORE started...") dlist = [] for i in range(num): #if i == 4: # port = 21 deferred = node.client.sendStore(files[i], None, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testSTORE %d" % i) deferred.addErrback(itererror, "failed at testSTORE %d" % i) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testSTORE") d.addErrback(stageerror, 'failed at testSTORE') d.addCallback(testRETRIEVE, nKu, host, port, num) #d.addCallback(testVERIFY, nKu, host, port, num) return d def testID(host, port, num=CONCURRENT): logger.info("testID started...") dlist = [] for i in range(num): #if i == 4: # port = 21 deferred = node.client.sendGetID(host, port) deferred.debug = True deferred.addErrback(itererror, "failed at testID %d" % i) dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "testID") d.addErrback(stageerror, 'testID') d.addCallback(testSTORE, host, port, num) return d def runTests(host, port=None, listenport=None): num = CONCURRENT #num = 5 global files, node files = createFakeData() node = FludNode(port=listenport) if port == None: port = node.config.port node.run() if num > len(files): num = len(files) d1 = testID(host, port, num) d1.addCallback(suitesuccess) d1.addErrback(suiteerror) d1.addBoth(cleanup) #nku = FludRSA.importPublicKey({'e': 65537L, 'n': 138646504113696863667807411690225283099791076530135000331764542300161152585426296356409290228001197773401729468267448145387041995053893737880473447042984919037843163552727823101445272608470814297563395471329917904393936481407769396601027233955938405001434483474847834031774504827822809611707032477570548179411L}) #d2 = testSTORE(nku, node, host, port, files, num) #d2.addErrback(suiteerror, 'failed at %s' % d2.testname) node.join() #node.start() # doesn't work, because reactor may not have started # listening by time requests start flying def createFakeData(dir="/tmp", num=CONCURRENT): randsrc = open("/dev/urandom", 'rb') files = [] for i in range(num): randdata = randsrc.read(256) filekey = fencode(int(flud.FludCrypto.hashstring(randdata), 16)) filename = dir+'/'+filekey f = open(filename, 'wb') f.write(randdata) f.close() files.append(filename) randsrc.close() return files def deleteFakeData(files): for f in files: if os.path.exists(f): os.remove(f) else: logger.warn("s already deleted!" % f) def cleanup(dummy=None): logger.info("cleaning up files and shutting down in 1 seconds...") time.sleep(1) deleteFakeData(files) reactor.callLater(1, node.stop) logger.info("done cleaning up") """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: print "Warning: testing against self my result in timeout failures" runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket, tempfile from twisted.python import failure from StringIO import StringIO from zlib import crc32 sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient import flud.FludCrypto as FludCrypto from flud.fencode import fencode, fdecode from flud.protocol.FludCommUtil import * from flud.FludDefer import ErrDeferredList """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ # metadatablock: (block#,n,k,blockdata) metadatablock = fencode((1,20,40,'adfdsfdffffffddddddddddddddd')) fake_mkey_offset = 111111 def testerror(failure, message, node): """ error handler for test errbacks """ print "testerror message: %s" % message print "testerror: %s" % str(failure) print "At least 1 test FAILED" return failure def allGood(_, nKu): print "all tests PASSED" return nKu def checkDELETE(res, nKu, fname, fkey, mkey, node, host, port, totalDelete): """ checks to ensure the file was deleted """ # totalDelete = True if this delete op should remove all meta (and data) if totalDelete: # try to retrieve with any metakey, should fail print "expecting failed retrieve, any metakey" return testRETRIEVE(res, nKu, fname, fkey, True, node, host, port, lambda args=(res, nKu): allGood(*args), False) else: # try to retrieve with any metakey, should succeed print "expecting successful retrieve, any metakey" return testRETRIEVE(res, nKu, fname, fkey, True, node, host, port, lambda args=(res, nKu, fname, fkey, mkey+fake_mkey_offset, node, host, port, True): testDELETE(*args)) def testDELETE(res, nKu, fname, fkey, mkey, node, host, port, totalDelete): """ Tests sendDelete, and invokes checkDELETE on success """ print "starting testDELETE %s.%s" % (fname, mkey) #return checkDELETE(None, nKu, fname, fkey, mkey, node, host, port, False) deferred = node.client.sendDelete(fkey, mkey, host, port, nKu) deferred.addCallback(checkDELETE, nKu, fname, fkey, mkey, node, host, port, totalDelete) deferred.addErrback(testerror, "failed at testDELETE", node) return deferred def checkVERIFY(res, nKu, fname, fkey, mkey, node, host, port, hash, newmeta): """ executes after testVERIFY """ if long(hash, 16) != long(res, 16): raise failure.DefaultException("verify didn't match: %s != %s" % (hash, res)) print "checkVERIFY (%s) %s success" % (newmeta, fname) if newmeta: return testDELETE(res, nKu, fname, fkey, mkey, node, host, port, False) else: return testVERIFY(nKu, fname, fkey, mkey, node, host, port, True) def testVERIFY(nKu, fname, fkey, mkey, node, host, port, newmeta): """ Test sendVerify """ # newmeta, if True, will generate new metadata to be stored during verify if newmeta: thismkey = mkey+fake_mkey_offset else: thismkey = mkey print "starting testVERIFY (%s) %s.%s" % (newmeta, fname, thismkey) fd = os.open(fname, os.O_RDONLY) fsize = os.fstat(fd)[stat.ST_SIZE] length = 20 offset = random.randrange(fsize-length) os.lseek(fd, offset, 0) data = os.read(fd, length) os.close(fd) hash = FludCrypto.hashstring(data) deferred = node.client.sendVerify(fkey, offset, length, host, port, nKu, (thismkey, StringIO(metadatablock))) deferred.addCallback(checkVERIFY, nKu, fname, fkey, mkey, node, host, port, hash, newmeta) deferred.addErrback(testerror, "failed at testVERIFY (%s)" % newmeta, node) return deferred def failedRETRIEVE(res, nextCallable): return nextCallable(); def checkRETRIEVE(res, nKu, fname, fkey, mkey, node, host, port, nextCallable): """ Compares the file that was stored with the one that was retrieved """ f1 = open(fname) filename = [f for f in res if f[-len(fkey):] == fkey][0] f2 = open(filename) if (f1.read() != f2.read()): f1.close() f2.close() raise failure.DefaultException( "upload/download (%s, %s) files don't match" % (fname, os.path.join(node.config.clientdir, fkey))) #print "%s (%d) and %s (%d) match" % (fname, os.stat(fname)[stat.ST_SIZE], # filename, os.stat(filename)[stat.ST_SIZE]) f1.close() f2.close() if mkey != True: expectedmeta = "%s.%s.meta" % (fkey, mkey) metanames = [f for f in res if f[-len(expectedmeta):] == expectedmeta] if not metanames: raise failure.DefaultException("expected metadata was missing") f3 = open(metanames[0]) md = f3.read() if md != metadatablock: raise failure.DefaultException("upload/download metadata doesn't" " match (%s != %s)" % (md, metadatablock)) return nextCallable() def testRETRIEVE(res, nKu, fname, fkey, mkey, node, host, port, nextCallable, expectSuccess=True): """ Tests sendRetrieve, and invokes checkRETRIEVE on success """ print "starting testRETRIEVE %s.%s" % (fname, mkey) deferred = node.client.sendRetrieve(fkey, host, port, nKu, mkey) deferred.addCallback(checkRETRIEVE, nKu, fname, fkey, mkey, node, host, port, nextCallable) if expectSuccess: deferred.addErrback(testerror, "failed at testRETRIEVE", node) else: deferred.addErrback(failedRETRIEVE, nextCallable) return deferred def testSTORE2(nKu, fname, fkey, node, host, port): mkey = crc32(fname) mkey2 = mkey+(2*fake_mkey_offset) print "starting testSTORE %s.%s" % (fname, mkey2) deferred = node.client.sendStore(fname, (mkey2, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey2, node, host, port, lambda args=(nKu, fname, fkey, mkey, node, host, port, False): testVERIFY(*args)) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred def testSTORE(nKu, fname, fkey, node, host, port): """ Tests sendStore, and invokes testRETRIEVE on success """ mkey = crc32(fname) print "starting testSTORE %s.%s" % (fname, mkey) deferred = node.client.sendStore(fname, (mkey, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey, node, host, port, lambda args=(nKu, fname, fkey, node, host, port): testSTORE2(*args)) deferred.addErrback(testerror, "failed at testSTORE", node) return deferred def testID(node, host, port): """ Tests sendGetID(), and invokes testSTORE on success """ print "starting testID" deferred = node.client.sendGetID(host, port) deferred.addErrback(testerror, "failed at testID", node) return deferred def testAggSTORE(nKu, aggFiles, node, host, port): print "starting testAggSTORE" dlist = [] for fname, fkey in aggFiles: mkey = crc32(fname) print "testAggSTORE %s (%s)" % (fname, mkey) deferred = node.client.sendStore(fname, (mkey, StringIO(metadatablock)), host, port, nKu) deferred.addCallback(testRETRIEVE, nKu, fname, fkey, mkey, node, host, port, lambda args=(nKu, fname, fkey, mkey, node, host, port, False): testVERIFY(*args)) deferred.addErrback(testerror, "failed at testAggSTORE", node) dlist.append(deferred) dl = ErrDeferredList(dlist) dl.addCallback(allGood, nKu) dl.addErrback(testerror, "failed at testAggSTORE", node) return dl def cleanup(_, node, filenamelist): for f in filenamelist: try: os.remove(f) except: print "couldn't remove %s" % f reactor.callLater(1, node.stop) def generateTestData(minSize): fname = tempfile.mktemp() f = open(fname, 'w') data = FludCrypto.generateRandom(minSize/50) for i in range(0, 51+random.randrange(50)): f.write(data) f.close() filekey = FludCrypto.hashfile(fname) filekey = fencode(int(filekey, 16)) filename = os.path.join("/tmp",filekey) os.rename(fname,filename) return (filename, filekey) def runTests(host, port=None, listenport=None): (largeFilename, largeFilekey) = generateTestData(512000) (smallFilename, smallFilekey) = generateTestData(5120) aggFiles = [] for i in range(4): aggFiles.append(generateTestData(4096)) node = FludNode(port=listenport) if port == None: port = node.config.port node.run() d = testID(node, host, port) d.addCallback(testSTORE, largeFilename, largeFilekey, node, host, port) d.addCallback(testSTORE, smallFilename, smallFilekey, node, host, port) d.addCallback(testAggSTORE, aggFiles, node, host, port) d.addBoth(cleanup, node, [i[0] for i in aggFiles] + [largeFilename, smallFilename]) node.join() def main(): localhost = socket.getfqdn() if len(sys.argv) == 1: runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3])) if __name__ == '__main__': main()
Python
#!/usr/bin/python import time, os, stat, random, sys, logging, socket from twisted.python import failure from twisted.internet import defer sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) import flud.FludCrypto from flud.FludNode import FludNode from flud.protocol.FludClient import FludClient from flud.protocol.FludCommUtil import * from flud.fencode import fencode, fdecode from flud.FludDefer import ErrDeferredList """ Test code for primitive operations. These ops include all of the descendents of ROOT and REQUEST in FludProtocol. """ CONCURRENT=300 CONCREPORT=50 node = None files = None logger = logging.getLogger('test') screenhandler = logging.StreamHandler() formatter = logging.Formatter('%(asctime)s %(name)s %(levelname)s:' ' %(message)s', datefmt='%H:%M:%S') screenhandler.setFormatter(formatter) logger.addHandler(screenhandler) logger.setLevel(logging.DEBUG) def suitesuccess(results): logger.info("all tests in suite passed") #print results return results def suiteerror(failure): logger.info("suite did not complete") logger.info("DEBUG: %s" % failure) return failure def stagesuccess(result, message): logger.info("stage %s succeeded" % message) return result def stageerror(failure, message): logger.info("stage %s failed" % message) #logger.info("DEBUG: %s" % failure) return failure def itersuccess(res, i, message): if i % CONCREPORT == 0: logger.info("itersuccess: %s" % message) return res def itererror(failure, message): logger.info("itererror message: %s" % message) #logger.info("DEBUG: %s" % failure) #logger.info("DEBUG: %s" % dir(failure) failure.printTraceback() return failure def checkVERIFY(results, nKu, host, port, hashes, num=CONCURRENT): logger.info(" checking VERIFY results...") for i in range(num): hash = hashes[i] res = results[i][1] if long(hash, 16) != long(res, 16): raise failure.DefaultException("verify didn't match: %s != %s" % (hash, res)) logger.info(" ...VERIFY results good.") return results #True def testVERIFY(res, nKu, host, port, num=CONCURRENT): logger.info("testVERIFY started...") dlist = [] hashes = [] for i in range(num): #if i == 4: # port = 21 fd = os.open(files[i], os.O_RDONLY) fsize = os.fstat(fd)[stat.ST_SIZE] length = 20 offset = random.randrange(fsize-length) os.lseek(fd, offset, 0) data = os.read(fd, length) os.close(fd) hashes.append(flud.FludCrypto.hashstring(data)) filekey = os.path.basename(files[i]) deferred = node.client.sendVerify(filekey, offset, length, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testVERIFY %d" % i) deferred.addErrback(itererror, "failed at testVERIFY %d: %s" % (i, filekey)) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testVERIFY") d.addErrback(stageerror, 'failed at testVERIFY') d.addCallback(checkVERIFY, nKu, host, port, hashes, num) return d def checkRETRIEVE(res, nKu, host, port, num=CONCURRENT): logger.info(" checking RETRIEVE results...") for i in range(num): f1 = open(files[i]) filekey = os.path.basename(files[i]) f2 = open(node.config.clientdir+"/"+filekey) if (f1.read() != f2.read()): f1.close() f2.close() raise failure.DefaultException("upload/download files don't match") f2.close() f1.close() logger.info(" ...RETRIEVE results good.") return testVERIFY(res, nKu, host, port, num) def testRETRIEVE(res, nKu, host, port, num=CONCURRENT): logger.info("testRETRIEVE started...") dlist = [] for i in range(num): #if i == 4: # port = 21 filekey = os.path.basename(files[i]) deferred = node.client.sendRetrieve(filekey, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testRETRIEVE %d" % i) deferred.addErrback(itererror, "failed at testRETRIEVE %d: %s" % (i, filekey)) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testRETRIEVE") d.addErrback(stageerror, 'failed at testRETRIEVE') d.addCallback(checkRETRIEVE, nKu, host, port, num) return d def testSTORE(nKu, host, port, num=CONCURRENT): logger.info("testSTORE started...") dlist = [] for i in range(num): #if i == 4: # port = 21 deferred = node.client.sendStore(files[i], None, host, port, nKu) deferred.addCallback(itersuccess, i, "succeeded at testSTORE %d" % i) deferred.addErrback(itererror, "failed at testSTORE %d" % i) dlist.append(deferred) d = ErrDeferredList(dlist) d.addCallback(stagesuccess, "testSTORE") d.addErrback(stageerror, 'failed at testSTORE') d.addCallback(testRETRIEVE, nKu, host, port, num) #d.addCallback(testVERIFY, nKu, host, port, num) return d def testID(host, port, num=CONCURRENT): logger.info("testID started...") dlist = [] for i in range(num): #if i == 4: # port = 21 deferred = node.client.sendGetID(host, port) deferred.debug = True deferred.addErrback(itererror, "failed at testID %d" % i) dlist.append(deferred) d = ErrDeferredList(dlist, returnOne=True) d.addCallback(stagesuccess, "testID") d.addErrback(stageerror, 'testID') d.addCallback(testSTORE, host, port, num) return d def runTests(host, port=None, listenport=None): num = CONCURRENT #num = 5 global files, node files = createFakeData() node = FludNode(port=listenport) if port == None: port = node.config.port node.run() if num > len(files): num = len(files) d1 = testID(host, port, num) d1.addCallback(suitesuccess) d1.addErrback(suiteerror) d1.addBoth(cleanup) #nku = FludRSA.importPublicKey({'e': 65537L, 'n': 138646504113696863667807411690225283099791076530135000331764542300161152585426296356409290228001197773401729468267448145387041995053893737880473447042984919037843163552727823101445272608470814297563395471329917904393936481407769396601027233955938405001434483474847834031774504827822809611707032477570548179411L}) #d2 = testSTORE(nku, node, host, port, files, num) #d2.addErrback(suiteerror, 'failed at %s' % d2.testname) node.join() #node.start() # doesn't work, because reactor may not have started # listening by time requests start flying def createFakeData(dir="/tmp", num=CONCURRENT): randsrc = open("/dev/urandom", 'rb') files = [] for i in range(num): randdata = randsrc.read(256) filekey = fencode(int(flud.FludCrypto.hashstring(randdata), 16)) filename = dir+'/'+filekey f = open(filename, 'wb') f.write(randdata) f.close() files.append(filename) randsrc.close() return files def deleteFakeData(files): for f in files: if os.path.exists(f): os.remove(f) else: logger.warn("s already deleted!" % f) def cleanup(dummy=None): logger.info("cleaning up files and shutting down in 1 seconds...") time.sleep(1) deleteFakeData(files) reactor.callLater(1, node.stop) logger.info("done cleaning up") """ Main currently invokes test code """ if __name__ == '__main__': localhost = socket.getfqdn() if len(sys.argv) == 1: print "Warning: testing against self my result in timeout failures" runTests(localhost) # test by talking to self elif len(sys.argv) == 2: runTests(localhost, eval(sys.argv[1])) # talk to self on port [1] elif len(sys.argv) == 3: runTests(sys.argv[1], eval(sys.argv[2])) # talk to [1] on port [2] elif len(sys.argv) == 4: # talk to [1] on port [2], listen on port [3] runTests(sys.argv[1], eval(sys.argv[2]), eval(sys.argv[3]))
Python
#!/usr/bin/python import tarfile, tempfile, random, os, sys import gzip from Crypto.Hash import SHA256 sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname( os.path.abspath(__file__))))) from flud.fencode import fencode import flud.TarfileUtils as TarfileUtils def maketarball(numfiles, avgsize, hashnames=False, addmetas=False): tarballname = tempfile.mktemp()+".tar" tarball = tarfile.open(tarballname, 'w') if addmetas: metafname = tempfile.mktemp() metaf = file(metafname, 'w') metaf.write('m'*48) metaf.close() for i in xrange(numfiles): fname = tempfile.mktemp() f = file(fname, 'wb') size = int(avgsize * (random.random()+0.5)) blocksize = 65*1024 if hashnames: sha256 = SHA256.new() for j in range(0, size, blocksize): if j+blocksize > size: block = 'a'*(size-j) else: block = 'a'*blocksize if hashnames: sha256.update(block) f.write(block) f.close() arcname = fname if hashnames: arcname = fencode(int(sha256.hexdigest(),16)) tarball.add(fname, arcname) if addmetas: tarball.add(metafname, arcname+".343434.meta") os.remove(fname) if addmetas: os.remove(metafname) contents = tarball.getnames() tarball.close() return tarballname, contents def gzipTarball(tarball): f = gzip.GzipFile(tarball+".gz", 'wb') f.write(file(tarball, 'rb').read()) f.close() os.remove(tarball) return tarball+".gz" def main(): # test plain TarfileUtils.delete() (tballname, contents) = maketarball(5, 4096) TarfileUtils.delete(tballname, contents[2:4]) tarball = tarfile.open(tballname, 'r') os.remove(tballname) assert(tarball.getnames() == contents[:2]+contents[4:]) tarball.close() # test gzip TarfileUtils.delete() (tballname, contents) = maketarball(5, 4096) tballname = gzipTarball(tballname) TarfileUtils.delete(tballname, contents[2:4]) tarball = tarfile.open(tballname, 'r') os.remove(tballname) assert(tarball.getnames() == contents[:2]+contents[4:]) tarball.close() # test plain TarfileUtils.concatenate() (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(gz, plain) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname1 = gzipTarball(tballname1) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(plain, gz) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname2 = gzipTarball(tballname2) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.concatenate(gz, gz) (tballname1, contents1) = maketarball(5, 4096) (tballname2, contents2) = maketarball(5, 4096) tballname1 = gzipTarball(tballname1) tballname2 = gzipTarball(tballname2) TarfileUtils.concatenate(tballname1, tballname2) assert(not os.path.exists(tballname2)) tarball = tarfile.open(tballname1, 'r') os.remove(tballname1) assert(tarball.getnames() == contents1+contents2) # test TarfileUtils.verifyHashes(plain no meta) (tballname, contents) = maketarball(5, 4096, True) assert(TarfileUtils.verifyHashes(tballname, contents[2:4])) os.remove(tballname) # test TarfileUtils.verifyHashes(plain with meta) (tballname, contents) = maketarball(5, 4096, True, True) assert(TarfileUtils.verifyHashes(tballname, contents[2:4]), ".meta") os.remove(tballname) # test TarfileUtils.verifyHashes(gzipped no meta) (tballname, contents) = maketarball(5, 4096, True) tballname = gzipTarball(tballname) assert(TarfileUtils.verifyHashes(tballname, contents[2:4])) os.remove(tballname) # test TarfileUtils.verifyHashes(gzipped with meta) (tballname, contents) = maketarball(5, 4096, True, True) tballname = gzipTarball(tballname) assert(TarfileUtils.verifyHashes(tballname, contents[2:4]), ".meta") os.remove(tballname) print "all tests passed" if __name__ == "__main__": main()
Python