Create proxy_pool.py
Browse files- proxy_pool.py +158 -0
proxy_pool.py
ADDED
|
@@ -0,0 +1,158 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Proxy Pool
|
| 3 |
+
"""
|
| 4 |
+
|
| 5 |
+
import random
|
| 6 |
+
import threading
|
| 7 |
+
import time
|
| 8 |
+
from collections import defaultdict
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
class ProxyEntry:
|
| 12 |
+
__slots__ = (
|
| 13 |
+
'proxy', 'protocol', 'proxy_url', 'latency',
|
| 14 |
+
'proxy_ip', 'verified', 'fail_count', 'total_used',
|
| 15 |
+
)
|
| 16 |
+
|
| 17 |
+
def __init__(self, proxy, protocol, proxy_url,
|
| 18 |
+
latency, proxy_ip, verified):
|
| 19 |
+
self.proxy = proxy
|
| 20 |
+
self.protocol = protocol
|
| 21 |
+
self.proxy_url = proxy_url
|
| 22 |
+
self.latency = latency
|
| 23 |
+
self.proxy_ip = proxy_ip
|
| 24 |
+
self.verified = verified
|
| 25 |
+
self.fail_count = 0
|
| 26 |
+
self.total_used = 0
|
| 27 |
+
|
| 28 |
+
def to_dict(self):
|
| 29 |
+
return {
|
| 30 |
+
"proxy": self.proxy,
|
| 31 |
+
"protocol": self.protocol,
|
| 32 |
+
"proxy_url": self.proxy_url,
|
| 33 |
+
"latency": self.latency,
|
| 34 |
+
"proxy_ip": self.proxy_ip,
|
| 35 |
+
"verified": self.verified,
|
| 36 |
+
"total_used": self.total_used,
|
| 37 |
+
}
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
class ProxyPool:
|
| 41 |
+
def __init__(self):
|
| 42 |
+
self._px = []
|
| 43 |
+
self._um = {}
|
| 44 |
+
self._lk = threading.Lock()
|
| 45 |
+
self._rr = defaultdict(int)
|
| 46 |
+
self._lr = 0.0
|
| 47 |
+
self._rc = 0
|
| 48 |
+
|
| 49 |
+
def refresh(self, new_proxies):
|
| 50 |
+
with self._lk:
|
| 51 |
+
old = dict(self._um)
|
| 52 |
+
fresh, um = [], {}
|
| 53 |
+
for px in new_proxies:
|
| 54 |
+
e = ProxyEntry(
|
| 55 |
+
px["proxy"], px["protocol"],
|
| 56 |
+
px["proxy_url"], px["latency"],
|
| 57 |
+
px.get("proxy_ip", ""),
|
| 58 |
+
px.get("verified", False),
|
| 59 |
+
)
|
| 60 |
+
o = old.get(e.proxy_url)
|
| 61 |
+
if o:
|
| 62 |
+
e.total_used = o.total_used
|
| 63 |
+
e.fail_count = max(0, o.fail_count - 1)
|
| 64 |
+
fresh.append(e)
|
| 65 |
+
um[e.proxy_url] = e
|
| 66 |
+
fresh.sort(key=lambda x: x.latency)
|
| 67 |
+
self._px = fresh
|
| 68 |
+
self._um = um
|
| 69 |
+
self._lr = time.time()
|
| 70 |
+
self._rc += 1
|
| 71 |
+
|
| 72 |
+
def clear(self):
|
| 73 |
+
with self._lk:
|
| 74 |
+
self._px = []
|
| 75 |
+
self._um = {}
|
| 76 |
+
|
| 77 |
+
@property
|
| 78 |
+
def size(self):
|
| 79 |
+
return len(self._px)
|
| 80 |
+
|
| 81 |
+
def _alive(self, protocol=None, verified=False):
|
| 82 |
+
r = [p for p in self._px if p.fail_count < 3]
|
| 83 |
+
if protocol:
|
| 84 |
+
r = [p for p in r if p.protocol == protocol]
|
| 85 |
+
if verified:
|
| 86 |
+
r = [p for p in r if p.verified]
|
| 87 |
+
return r
|
| 88 |
+
|
| 89 |
+
def get_round_robin(self, protocol=None, verified=False):
|
| 90 |
+
with self._lk:
|
| 91 |
+
a = self._alive(protocol, verified)
|
| 92 |
+
if not a:
|
| 93 |
+
return None
|
| 94 |
+
k = f"{protocol}_{verified}"
|
| 95 |
+
i = self._rr[k] % len(a)
|
| 96 |
+
self._rr[k] = i + 1
|
| 97 |
+
a[i].total_used += 1
|
| 98 |
+
return a[i]
|
| 99 |
+
|
| 100 |
+
def get_random(self, protocol=None, verified=False):
|
| 101 |
+
with self._lk:
|
| 102 |
+
a = self._alive(protocol, verified)
|
| 103 |
+
if not a:
|
| 104 |
+
return None
|
| 105 |
+
p = random.choice(a)
|
| 106 |
+
p.total_used += 1
|
| 107 |
+
return p
|
| 108 |
+
|
| 109 |
+
def get_fastest(self, protocol=None, verified=False):
|
| 110 |
+
with self._lk:
|
| 111 |
+
a = self._alive(protocol, verified)
|
| 112 |
+
if not a:
|
| 113 |
+
return None
|
| 114 |
+
a[0].total_used += 1
|
| 115 |
+
return a[0]
|
| 116 |
+
|
| 117 |
+
def get_least_used(self, protocol=None, verified=False):
|
| 118 |
+
with self._lk:
|
| 119 |
+
a = self._alive(protocol, verified)
|
| 120 |
+
if not a:
|
| 121 |
+
return None
|
| 122 |
+
p = min(a, key=lambda x: x.total_used)
|
| 123 |
+
p.total_used += 1
|
| 124 |
+
return p
|
| 125 |
+
|
| 126 |
+
def report_failure(self, url):
|
| 127 |
+
with self._lk:
|
| 128 |
+
p = self._um.get(url)
|
| 129 |
+
if p:
|
| 130 |
+
p.fail_count += 1
|
| 131 |
+
|
| 132 |
+
def report_success(self, url):
|
| 133 |
+
with self._lk:
|
| 134 |
+
p = self._um.get(url)
|
| 135 |
+
if p:
|
| 136 |
+
p.fail_count = 0
|
| 137 |
+
|
| 138 |
+
def get_all(self, protocol=None, verified=False, limit=500):
|
| 139 |
+
with self._lk:
|
| 140 |
+
return [p.to_dict() for p in self._alive(protocol, verified)[:limit]]
|
| 141 |
+
|
| 142 |
+
def get_stats(self):
|
| 143 |
+
with self._lk:
|
| 144 |
+
bp = defaultdict(lambda: {"total": 0, "alive": 0})
|
| 145 |
+
for p in self._px:
|
| 146 |
+
bp[p.protocol]["total"] += 1
|
| 147 |
+
if p.fail_count < 3:
|
| 148 |
+
bp[p.protocol]["alive"] += 1
|
| 149 |
+
alive = sum(d["alive"] for d in bp.values())
|
| 150 |
+
return {
|
| 151 |
+
"total": len(self._px),
|
| 152 |
+
"alive": alive,
|
| 153 |
+
"by_protocol": dict(bp),
|
| 154 |
+
"refresh_count": self._rc,
|
| 155 |
+
"last_refresh_ago": round(
|
| 156 |
+
time.time() - self._lr, 1
|
| 157 |
+
) if self._lr else None,
|
| 158 |
+
}
|