repo_name
stringlengths
6
97
path
stringlengths
3
341
text
stringlengths
8
1.02M
lindemer/pmp
src/test/scala/PmpConfig.scala
<reponame>lindemer/pmp package config import spinal.core._ import spinal.core.SpinalConfig import spinal.core.sim.SimConfig object PmpConfig { val spinalConfig = SpinalConfig( defaultClockDomainFrequency = FixedFrequency(10 MHz), targetDirectory = "rtl" ) def apply() = SimConfig.withWave .withConfig(spinalConfig) .workspacePath("waves") }
ruuda/matasano
src/main/scala/Set1/Challenge7.scala
<filename>src/main/scala/Set1/Challenge7.scala // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import io.Source import com.matasano.Aes import com.matasano.Encoding object Challenge7 { def main(args: Array[String]) = { val file = Source.fromFile("data/challenge7.md") val ciphertext = Encoding.decodeBase64(file.getLines.mkString) val key = Encoding.encodeAscii("YELLOW SUBMARINE") val decrypt = Aes.decrypt(key) val plaintext = ciphertext .grouped(16) .map(decrypt) .map(Encoding.decodeAscii) .mkString println(s"plaintext:\n\n$plaintext") file.close() } }
ruuda/matasano
src/main/scala/Crypto.scala
<filename>src/main/scala/Crypto.scala // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. package com.matasano object Crypto { def xor(a: Vector[Byte], b: Vector[Byte]): Vector[Byte] = a.zip(b).map { case (x, y) => (x ^ y).toByte } def xorSingle(a: Vector[Byte], key: Byte): Vector[Byte] = { val repeated = Vector.fill(a.length) { key } xor(a, repeated) } def xorRepeat(a: Vector[Byte], key: Vector[Byte]): Vector[Byte] = { val repeated = Stream.continually(key).flatten.take(a.length).toVector xor(a, repeated) } def encryptCbc(encrypt: Vector[Byte] => Vector[Byte]) (iv: Vector[Byte], plaintext: Vector[Byte]): Vector[Byte] = { // Assume a block size of 16 for now. require(iv.length == 16) require(plaintext.length % 16 == 0) plaintext.grouped(16).scanLeft(iv) { case (prev, block) => encrypt(xor(prev, block)) } .flatMap(block => block) .toVector } def decryptCbc(decrypt: Vector[Byte] => Vector[Byte]) (iv: Vector[Byte], ciphertext: Vector[Byte]): Vector[Byte] = { // Assume a block size of 16 for now. require(iv.length == 16) require(ciphertext.length % 16 == 0) val blocks = ciphertext.grouped(16).toVector val withIv = Vector(iv) ++ blocks withIv.zip(blocks).flatMap { case (prev, block) => xor(prev, decrypt(block)) } .toVector } }
ruuda/matasano
src/main/scala/Util.scala
// Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. package com.matasano object Util { val bitCount: Int => Int = java.lang.Integer.bitCount def hammingDistance(a: Vector[Byte], b: Vector[Byte]): Int = { // Number of bits that differ. val baseDiff = a.zip(b).map { case (x, y) => bitCount((x ^ y) & 0xff) } .sum // Also count a length mismatch as eight different bits per byte. val extraDiff = math.abs(a.length - b.length) * 8 baseDiff + extraDiff } def padPkcs7(a: Vector[Byte], blockSize: Int): Vector[Byte] = { if (blockSize == 0) return a ++ Vector(1.toByte) val modSz = (a.length % blockSize) val padLen = if (modSz == 0) { blockSize } else { blockSize - modSz } a.padTo(a.length + padLen, padLen.toByte) } def unpadPkcs7(padded: Vector[Byte]): Vector[Byte] = { require(padded.length > 0) val b = padded.last assert(b > 0) val data = padded.take(padded.length - b) val padding = padded.drop(padded.length - b) assert(padding.forall(x => x == b)) data } }
ruuda/matasano
src/test/scala/Aes.scala
<gh_stars>0 // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import org.scalacheck.Gen import org.scalacheck.Properties import org.scalacheck.Prop.forAll import com.matasano.Aes object AesTests extends Properties("Aes") { val blocks = Gen.listOfN(16, Gen.choose(0, 255)).map(_.map(_.toByte).toVector) property("bijectionBlock") = forAll(blocks, blocks) { (block: Vector[Byte], key: Vector[Byte]) => Aes.decrypt(key)(Aes.encrypt(key)(block)) == block } property("bijectionBlock") = forAll(blocks, blocks) { (block: Vector[Byte], key: Vector[Byte]) => Aes.encrypt(key)(Aes.decrypt(key)(block)) == block } }
ruuda/matasano
src/main/scala/Set1/Challenge5.scala
<filename>src/main/scala/Set1/Challenge5.scala // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import com.matasano.Crypto import com.matasano.Encoding object Challenge5 { def main(args: Array[String]) = { val key = Encoding.encodeAscii("ICE") val input = Encoding.encodeAscii("Burning 'em, " + "if you ain't quick and nimble\n" + "I go crazy when I hear a cymbal") val data = Crypto.xorRepeat(input, key) val output = Encoding.encodeHex(data) println(output) } }
ruuda/matasano
src/test/scala/Encoding.scala
<reponame>ruuda/matasano<filename>src/test/scala/Encoding.scala<gh_stars>0 // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import org.scalacheck.Properties import org.scalacheck.Prop.forAll import com.matasano.Encoding object EncodingTests extends Properties("Encoding") { property("bijectionByte") = forAll { (byte: Byte) => Encoding.decodeByte(Encoding.encodeByte(byte)) == byte } property("bijectionTriple") = forAll { (b0: Byte, b1: Byte, b2: Byte) => val triple = Vector(b0, b1, b2) Encoding.decodeTriple(Encoding.encodeTriple(triple)) == triple } property("bijectionTriplePad1") = forAll { (b0: Byte, b1: Byte) => val pair = Vector(b0, b1) Encoding.decodeTriple(Encoding.encodeTriple(pair)) == pair } property("bijectionTriplePad2") = forAll { (b0: Byte) => val byte = Vector(b0) Encoding.decodeTriple(Encoding.encodeTriple(byte)) == byte } property("bijectionHex") = forAll { (data: Vector[Byte]) => Encoding.decodeHex(Encoding.encodeHex(data)) == data } property("bijectionBase64") = forAll { (data: Vector[Byte]) => Encoding.decodeBase64(Encoding.encodeBase64(data)) == data } property("bijectionAscii") = forAll { (data: Vector[Byte]) => Encoding.encodeAscii(Encoding.decodeAscii(data)) == data } }
ruuda/matasano
src/main/scala/Set1/Challenge6.scala
<filename>src/main/scala/Set1/Challenge6.scala // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import io.Source import com.matasano.Crypto import com.matasano.Encoding import com.matasano.TextDetection import com.matasano.Util object Challenge6 { /// Returns a list of key size guesses, from most likely to least likely. def guessKeySize(ciphertext: Vector[Byte], minSize: Int, maxSize: Int): Vector[Int] = { val rankedSizes = (minSize to maxSize).map(keySize => { // Inspect the first ten blocks for this key size. val groups = ciphertext.grouped(keySize).take(10).toVector val distances = for { i <- 0 to groups.length - 1 j <- i + 1 to groups.length - 1 } yield Util.hammingDistance(groups(i), groups(j)) // Compute average number of different bits per byte. val numGroups = groups.length * (groups.length - 1) / 2 var meanDistance = distances.sum.toFloat / numGroups.toFloat var normDistance = meanDistance / keySize.toFloat (normDistance, keySize) }) rankedSizes.sorted.map { case (d, sz) => sz } .toVector } def guessKey(ciphertext: Vector[Byte], keySize: Int): Vector[Byte] = { val rankHist = TextDetection.buildHistogramRanker("data/frequency.md") Vector.tabulate(keySize) { i => { val cipherChars = ciphertext.grouped(keySize) .filter(block => block.length > i) .map(block => block(i)) .toVector val keys = Vector.tabulate(256) { key => { val plaintext = Crypto.xorSingle(cipherChars, key.toByte) val histogram = TextDetection.buildHistogram(plaintext.map(byte => byte.toChar).to) val rank = rankHist(histogram) (key, rank) }} val (bestKey, _) = keys.maxBy { case (key, rank) => rank } bestKey.toByte }} } def main(args: Array[String]) = { val file = Source.fromFile("data/challenge6.md") val ciphertext = Encoding.decodeBase64(file.getLines.mkString) val keySize = guessKeySize(ciphertext, 2, 40).head println(s"key size guess: $keySize") val key = guessKey(ciphertext, keySize) val keyStr = Encoding.decodeAscii(key) println(s"key guess: $keyStr") val plaintext = Encoding.decodeAscii(Crypto.xorRepeat(ciphertext, key)) println(s"plaintext:\n\n$plaintext") file.close() } }
ruuda/matasano
src/test/scala/Set1.scala
// Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import org.scalacheck.Properties import com.matasano.Crypto import com.matasano.Encoding object Set1Spec extends Properties("Set1") { property("challenge1") = { val input = "49276d206b696c6c696e6720796f757220627261696e206c" + "696b65206120706f69736f6e6f7573206d757368726f6f6d" val output = "SSdtIGtpbGxpbmcgeW91ciBicmFpbiBsaWtlIGEgcG9pc29" + "ub3VzIG11c2hyb29t" output == Encoding.encodeBase64(Encoding.decodeHex(input)) } property("challenge2") = { val inputA = Encoding.decodeHex("1c0111001f010100061a024b53535009181c") val inputB = Encoding.decodeHex("686974207468652062756c6c277320657965") val output = Encoding.decodeHex("746865206b696420646f6e277420706c6179") output == Crypto.xor(inputA, inputB) } property("challenge5") = { val input = Encoding.encodeAscii("Burning 'em, " + "if you ain't quick and nimble\n" + "I go crazy when I hear a cymbal") val output = Encoding.decodeHex("0b3637272a2b2e63622c2e69692a23693a2a3" + "<KEY>" + "<KEY>" + "<KEY>") val key = Encoding.encodeAscii("ICE") output == Crypto.xorRepeat(input, key) } }
ruuda/matasano
src/test/scala/Util.scala
// Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import org.scalacheck.Arbitrary import org.scalacheck.Gen import org.scalacheck.Properties import org.scalacheck.Prop.forAll import com.matasano.Encoding import com.matasano.Util object UtilTests extends Properties("Util") { val bytevectors = Arbitrary.arbitrary[Vector[Byte]] val blockLength = Gen.choose(0, 64) property("hammingDistance") = { val a = Encoding.encodeAscii("this is a test") val b = Encoding.encodeAscii("wokka wokka!!!") Util.hammingDistance(a, b) == 37 } property("hammingDistanceBoundedByInput") = forAll { (a: Vector[Byte], b: Vector[Byte]) => val diff = Util.hammingDistance(a, b) val maxDiff = math.max(a.length, b.length) * 8 diff <= maxDiff } property("bijectionPadPkcs7") = forAll(bytevectors, blockLength) { (a: Vector[Byte], b: Int) => a == Util.unpadPkcs7(Util.padPkcs7(a, b)) } }
ruuda/matasano
build.sbt
name := "matasano" libraryDependencies += "org.scalacheck" %% "scalacheck" % "1.13.4" % "test"
ruuda/matasano
src/main/scala/Aes.scala
<filename>src/main/scala/Aes.scala // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. package com.matasano object Aes { import javax.crypto.Cipher import javax.crypto.spec.SecretKeySpec def encrypt(key: Vector[Byte]): Vector[Byte] => Vector[Byte] = { // Use the Java AES implementation only for its block cipher. // The point of this exercise it to learn how to do things myself, // so I will not use the standard block-chaining or padding mechanisms. val cipher = Cipher.getInstance("AES/ECB/NoPadding") val keySpec = new SecretKeySpec(key.toArray, "AES") cipher.init(Cipher.ENCRYPT_MODE, keySpec) block => { require(block.length == 16, "block length must be 16") cipher.doFinal(block.toArray).toVector } } def decrypt(key: Vector[Byte]): Vector[Byte] => Vector[Byte] = { val cipher = Cipher.getInstance("AES/ECB/NoPadding") val keySpec = new SecretKeySpec(key.toArray, "AES") cipher.init(Cipher.DECRYPT_MODE, keySpec) block => { require(block.length == 16, "block length must be 16") cipher.doFinal(block.toArray).toVector } } }
ruuda/matasano
src/main/scala/Set1/Challenge3.scala
<filename>src/main/scala/Set1/Challenge3.scala // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import com.matasano.Crypto import com.matasano.Encoding import com.matasano.TextDetection object Challenge3 { def main(args: Array[String]) = { val input = "1b37373331363f78151b7f2b783431333d" + "78397828372d363c78373e783a393b3736" val ciphertext = Encoding.decodeHex(input) val candidates = (0 to 255).map(key => Crypto.xorSingle(ciphertext, key.toByte)) val strcandidates = candidates.map(bytes => Encoding.decodeAscii(bytes)) val rank = TextDetection.buildRanker("data/frequency.md") val plaintext = strcandidates.maxBy(rank) println(plaintext) } }
ruuda/matasano
src/main/scala/Set1/Challenge1.scala
// Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import com.matasano.Encoding object Challenge1 { def main(args: Array[String]) = { val input = "49276d206b696c6c696e6720796f757220627261696e206c" + "<KEY>" val data = Encoding.decodeHex(input) val output = Encoding.encodeBase64(data) println(output) } }
ruuda/matasano
src/test/scala/Crypto.scala
<reponame>ruuda/matasano // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import org.scalacheck.Gen import org.scalacheck.Properties import org.scalacheck.Prop.forAll import com.matasano.Aes import com.matasano.Crypto object CryptoTests extends Properties("Crypto") { val blocks = Gen.listOfN(16, Gen.choose(0, 255)).map(_.map(_.toByte).toVector) val nBlocks = Gen.listOfN(64, Gen.choose(0, 255)).map(_.map(_.toByte).toVector) property("bijectionEncryptCbc") = forAll(blocks, blocks, nBlocks) { (iv: Vector[Byte], key: Vector[Byte], plaintext: Vector[Byte]) => { val decrypt = Crypto.decryptCbc(Aes.decrypt(key)) _ val encrypt = Crypto.encryptCbc(Aes.encrypt(key)) _ decrypt(iv, encrypt(iv, plaintext)) == plaintext } } }
ruuda/matasano
src/main/scala/TextDetection.scala
<reponame>ruuda/matasano // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. package com.matasano import collection.immutable.SortedMap import io.Source object TextDetection { type Histogram[A] = SortedMap[(Char, Char), A] def buildHistogramPairFromFile(fname: String): Histogram[Int] = { val file = Source.fromFile(fname) var histogram: SortedMap[(Char, Char), Int] = SortedMap() for (line <- file.getLines()) { val lowerLine = line.map(c => c.toLower) for (pair <- lowerLine.zip(lowerLine.drop(1))) { val count = histogram.getOrElse(pair, 0) histogram = histogram.updated(pair, count + 1) } } file.close() histogram } def buildHistogramSingleFromFile(fname: String): SortedMap[Char, Int] = { val file = Source.fromFile(fname) val chars = file.getLines().flatMap(line => line.to) val histogram = buildHistogram(chars) file.close() histogram } def buildHistogram(chars: Iterator[Char]): SortedMap[Char, Int] = { val empty: SortedMap[Char, Int] = SortedMap() chars.map(c => c.toLower).foldLeft(empty) { (hist, c) => val count = hist.getOrElse(c, 0) hist.updated(c, count + 1) } } // Ensures that every value in the histogram is in the interval [0, 2]. def normalizeHistogramMax[Key](counts: SortedMap[Key, Int]): SortedMap[Key, Double] = { val maxCount = counts.foldLeft(0) { case (acc, (k, n)) => math.max(acc, n) } counts.mapValues(n => 2.0 * n / maxCount) } // Ensures that all values in the histogram sum to 1.0. def normalizeHistogram[Key](counts: SortedMap[Key, Int]): SortedMap[Key, Double] = { val sum = counts.foldLeft(0) { case (acc, (k, n)) => acc + n } counts.mapValues(n => n / sum.toDouble) } // Builds a function that ranks strings based on consecutive character // frequencies extracted from the sample file. def buildRanker(fname: String): String => Double = { val histInt = buildHistogramPairFromFile(fname) val histDouble = normalizeHistogramMax(histInt) def rankChars(chars: (Char, Char)): Double = histDouble.getOrElse(chars, 0.05) def rankString(str: String): Double = { val lowerStr = str.map(c => c.toLower) val pairs = lowerStr.zip(lowerStr.drop(1)) pairs.foldLeft(1.0) { (acc, chars) => acc * rankChars(chars) } } rankString } // Builds a function that ranks histograms based on how similar they are to // the character frequencies in the sample file. def buildHistogramRanker(fname: String): SortedMap[Char, Int] => Double = { val histInt = buildHistogramSingleFromFile(fname) val histDouble = normalizeHistogram(histInt) def rankHistogram(sampleHist: SortedMap[Char, Int]): Double = { val sampleDouble = normalizeHistogram(sampleHist) // Consider the histograms as vectors, and compute the squared L2 norm of // their difference. val keys = histDouble.keySet ++ sampleDouble.keySet val diffSqr = keys.foldLeft(0.0) { (acc, c) => { val left = histDouble.getOrElse(c, 0.0) val right = sampleDouble.getOrElse(c, 0.0) acc + (left - right) * (left - right) }} // A higher ranking is better, so negate the difference. -diffSqr } rankHistogram } }
ruuda/matasano
src/main/scala/Set1/Challenge2.scala
// Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import com.matasano.Crypto import com.matasano.Encoding object Challenge2 { def main(args: Array[String]) = { val inputA = Encoding.decodeHex("1c0111001f010100061a024b53535009181c") val inputB = Encoding.decodeHex("686974207468652062756c6c277320657965") val data = Crypto.xor(inputA, inputB) val output = Encoding.encodeHex(data) println(output) } }
ruuda/matasano
src/main/scala/Set1/Challenge8.scala
// Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import io.Source import com.matasano.Encoding import com.matasano.Util object Challenge8 { def repetition(ciphertext: Vector[Byte]): Float = { // Compute the Hamming distance between all pairs of blocks. If the // blocks are encrypted properly then they are random, so their Hamming // distance is 0.5 * 8 * 16 on average. If the blocks were encrypted // individually, similarities will yield a lower distance. val groups = ciphertext.grouped(16).toVector val distances = for { i <- 0 to groups.length - 1 j <- i + 1 to groups.length - 1 } yield { require(groups(i).length == groups(j).length) Util.hammingDistance(groups(i), groups(j)) } val numPairs = groups.length * (groups.length - 1) / 2 var meanDistance = distances.sum.toFloat / numPairs.toFloat // Express the distance relative to the expected distance. meanDistance / 64.0f } def main(args: Array[String]) = { val file = Source.fromFile("data/challenge8.md") val ciphertexts = file.getLines.map(line => Encoding.decodeHex(line)) val ranked = ciphertexts.map(ct => (repetition(ct), ct)) val top = ranked.toVector.sortWith { case ((r1, _), (r2, _)) => r1 < r2 } .take(10) println("Candidates (with confidence, lower is better)") for ((r, ct) <- top) { val hdata = Encoding.encodeHex(ct) println(s"$r $hdata") } file.close() } }
ruuda/matasano
src/test/scala/Set2.scala
// Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import org.scalacheck.Properties import com.matasano.Encoding import com.matasano.Util object Set2Spec extends Properties("Set2") { property("challenge9") = { val input = Encoding.encodeAscii("YELLOW SUBMARINE") val output = Encoding.encodeAscii("YELLOW SUBMARINE\u0004\u0004\u0004\u0004") output == Util.padPkcs7(input, 20) } }
ruuda/matasano
src/main/scala/Set2/Challenge10.scala
<filename>src/main/scala/Set2/Challenge10.scala // Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. import io.Source import com.matasano.Aes import com.matasano.Crypto import com.matasano.Encoding import com.matasano.Util object Challenge10 { def main(args: Array[String]) = { val file = Source.fromFile("data/challenge10.txt") val ciphertext = Encoding.decodeBase64(file.getLines.mkString) val key = Encoding.encodeAscii("YELLOW SUBMARINE") val decrypt = Aes.decrypt(key) val iv: Vector[Byte] = Vector(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) val plaintext = Crypto.decryptCbc(decrypt)(iv, ciphertext) println(Encoding.decodeAscii(Util.unpadPkcs7(plaintext))) file.close() } }
ruuda/matasano
src/main/scala/Encoding.scala
// Copyright 2016 <NAME> // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // A copy of the License is available in the root of the repository. package com.matasano object Encoding { val hexChars = "0123456789abcdef" val base64Chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/" def decodeByte(str: String): Byte = { if (str.length != 2) { throw new Exception("Invalid input, not an integer number of bytes.") } val high = hexChars.indexOf(str(0)) val low = hexChars.indexOf(str(1)) if (high == -1 || low == -1) { throw new Exception(s"Invalid input, '$str' is not hexadecimal.") } else { ((high << 4) | low).toByte } } def encodeByte(byte: Byte): String = { val index0 = (byte & 255) >> 4 val index1 = (byte & 255) val indices = Vector(index0 & 15, index1 & 15) val chars = indices.map(i => hexChars(i)) chars.mkString } def decodeTriple(str: String): Vector[Byte] = { if (str.length != 4) { throw new Exception("Invalid input, not an integer number of bytes.") } // Drop the padding characters. val withoutPadding = str.stripSuffix("==").stripSuffix("=") val numBytes = withoutPadding.length - 1 // Fill up with zeroes to avoid index out of bounds. val nullStr = withoutPadding.padTo(4, base64Chars(0)) val index0 = base64Chars.indexOf(nullStr(0)) val index1 = base64Chars.indexOf(nullStr(1)) val index2 = base64Chars.indexOf(nullStr(2)) val index3 = base64Chars.indexOf(nullStr(3)) if (index0 == -1 || index1 == -1 || index2 == -1 || index3 == -1) { throw new Exception(s"Invalid input, '$str' is not base64.") } val byte0 = (index0 << 2) | (index1 >> 4) val byte1 = (index1 << 4) | (index2 >> 2) val byte2 = (index2 << 6) | (index3) Vector(byte0.toByte, byte1.toByte, byte2.toByte).take(numBytes) } def encodeTriple(triple: Vector[Byte]): String = { // Pad with zeroes if the triple contains less than three bytes. val numBytes = triple.length val padded = triple.padTo(3, 0.toByte) // Note: The & 255 is required here to make the shift behave like a regular // shift of an unsigned integer. val index0 = (padded(0) & 255) >> 2 val index1 = ((padded(0) & 3) << 4) | ((padded(1) & 255) >> 4) val index2 = ((padded(1) & 15) << 2) | ((padded(2) & 255) >> 6) val index3 = (padded(2) & 255) val indices = Vector(index0 & 63, index1 & 63, index2 & 63, index3 & 63) val chars = indices.map(i => base64Chars(i)) // Replace encoded zeroes with padding characters. chars.mkString.take(numBytes + 1).padTo(4, '=') } def decodeHex(str: String): Vector[Byte] = str.grouped(2).map(b => decodeByte(b)).toVector def encodeHex(data: Vector[Byte]): String = data.map(b => encodeByte(b)).mkString def decodeBase64(str: String): Vector[Byte] = str.grouped(4).flatMap(t => decodeTriple(t)).toVector def encodeBase64(data: Vector[Byte]): String = data.grouped(3).map(t => encodeTriple(t)).mkString def decodeAscii(data: Vector[Byte]): String = // Actually use Latin 1 to ensure that the upper 128 bytes // can be mapped as well. Do not use UTF-8, because not every // byte sequence is valid UTF-8. new String(data.toArray, "ISO8859_1") def encodeAscii(str: String): Vector[Byte] = str.getBytes("ISO8859_1").toVector }
kmbnw/multitreat
scala/build.sbt
name := "multitreat" version := "0.1a" organization := "net.kmbnw" scalaVersion := "2.10.5" libraryDependencies += "org.apache.spark" %% "spark-core" % "1.6.2" libraryDependencies += "org.apache.spark" %% "spark-sql" % "1.6.2" libraryDependencies += "org.apache.spark" %% "spark-hive" % "1.6.2" libraryDependencies += "com.holdenkarau" %% "spark-testing-base" % "1.6.1_0.3.3" resolvers += Resolver.mavenLocal javaOptions ++= Seq("-Xms512M", "-Xmx2048M", "-XX:MaxPermSize=2048M", "-XX:+CMSClassUnloadingEnabled") parallelExecution in Test := false
kmbnw/multitreat
scala/src/test/scala/net/kmbnw/multitreat/TestMultitreat.scala
<gh_stars>1-10 /* * Copyright 2016 <NAME> * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package net.kmbnw.multitreat import org.apache.spark.sql.functions._ import org.apache.spark.sql.DataFrame import com.holdenkarau.spark.testing._ class test extends DataFrameSuiteBase { private val tol = 0.001 test("designNumericSingleGroup") { val sqlCtx = sqlContext import sqlCtx.implicits._ val expected = sqlContext.read.json("src/test/resources/salary_one_group.json") val df = expected.drop("title_catN") val treatmentPlan = new Multitreat("amount", List("title")) val treatments = treatmentPlan.designNumeric(df) val treated = treatmentPlan.applyTreatments(df, treatments) // apparently these have to be explicitly ordered or the equality check can fail assertDataFrameApproximateEquals( expected.select("amount", "employer", "title", "title_catN"), treated.select("amount", "employer", "title", "title_catN"), tol) /*val input2 = sc.parallelize(List(4, 5, 6)).toDF intercept[org.scalatest.exceptions.TestFailedException] { assertDataFrameEquals(input1, input2) // not equal }*/ } test("designNumericMultiGroup") { val sqlCtx = sqlContext import sqlCtx.implicits._ val expected = sqlContext.read.json("src/test/resources/salary_multi_group.json") val df = expected.drop("title_catN").drop("employer_catN") val treatmentPlan = new Multitreat("amount", List("title", "employer")) val treatments = treatmentPlan.designNumeric(df) val treated = treatmentPlan.applyTreatments(df, treatments) // apparently these have to be explicitly ordered or the equality check can fail assertDataFrameApproximateEquals( expected.select("amount", "employer", "title", "title_catN", "employer_catN"), treated.select("amount", "employer", "title", "title_catN", "employer_catN"), tol) } // single row means standard deviation of zero; ensure that is OK test("designNumericOneRow") { val sqlCtx = sqlContext import sqlCtx.implicits._ val expected = sqlContext.read.json("src/test/resources/salary_one_row.json") val df = expected.drop("title_catN") val treatmentPlan = new Multitreat("amount", List("title")) val treatments = treatmentPlan.designNumeric(df) val treated = treatmentPlan.applyTreatments(df, treatments) // apparently these have to be explicitly ordered or the equality check can fail assertDataFrameApproximateEquals( expected.select("amount", "employer", "title", "title_catN"), treated.select("amount", "employer", "title", "title_catN"), tol) } }
kmbnw/multitreat
scala/src/main/scala/net/kmbnw/multitreat/TreatmentPlan.scala
<filename>scala/src/main/scala/net/kmbnw/multitreat/TreatmentPlan.scala /* * Copyright 2016 <NAME> * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package net.kmbnw.multitreat import org.apache.spark.sql.functions._ import org.apache.spark.sql.DataFrame // e.g. val treatmentPlan = new TreatmentPlan("salary", List("job_title", "employer")) // val treatments = treatmentPlan.designNumeric(df) // treatments.show() // val treated = treatmentPlan.applyTreatments(df, treatments) class TreatmentPlan( val targetCol: String, val treatmentCols: Seq[String]) { def designNumeric(df: DataFrame): Map[String, DataFrame] = { return treatmentCols.zip( treatmentCols.map(x => designNumericOnCol(df, x)) ).toMap } def applyTreatments( df: DataFrame, treatments: Map[String, DataFrame]): DataFrame = { var treated: DataFrame = df // TODO this can probably be done with fold or reduce for ((colname, treatment) <- treatments) { treated = treated.join(treatment, colname) } return treated } private def designNumericOnCol(df: DataFrame, groupCol: String): DataFrame = { // overall dataframe mean and standard deviation val dfNoNA = df.select(targetCol, groupCol).na.drop() val naFill = 1e-6 val sampleMean = dfNoNA.select(avg(targetCol)).first().getDouble(0) val sampleSd = dfNoNA.select(stddev(targetCol)).first().getDouble(0) // using the simple version of lambda from the paper: lambda = n / (m + n) // where m = group_sd / sample_sd // there is a fill-in for when only one sample exists of 1e-6 // TODO I would like to make lambda user-settable val dfMeans = dfNoNA.groupBy(groupCol).agg( avg(targetCol).as("mean"), count(targetCol).divide( stddev(targetCol).divide(sampleSd) .plus(count(targetCol))).as("lambda") ).na.fill(naFill, Array("lambda")) // this is the Bayesian formula: // lambda * group_mean + (1 - lambda) * sample_mean val lambda = dfMeans.col("lambda") val groupMean = dfMeans.col("mean") val treatedCol = groupCol + "_catN" val ret = dfMeans.withColumn( treatedCol, lambda.multiply(groupMean).plus( lambda.minus(1).multiply(-1 * sampleMean)) ) return ret.select(groupCol, treatedCol) } }
phymbert/spark-search
sql/src/test/scala/org/apache/spark/search/sql/TestData.scala
<reponame>phymbert/spark-search<filename>sql/src/test/scala/org/apache/spark/search/sql/TestData.scala /* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.sql import java.io.{File, Reader} import java.util.function.Consumer import org.apache.lucene.analysis.Analyzer.TokenStreamComponents import org.apache.lucene.analysis.shingle.ShingleFilter import org.apache.lucene.analysis.standard.StandardTokenizer import org.apache.lucene.analysis.synonym.{SynonymGraphFilter, SynonymMap} import org.apache.lucene.analysis.{Analyzer, LowerCaseFilter, TokenStream} import org.apache.lucene.util.CharsRef import org.apache.spark.sql.{Dataset, SparkSession} object TestData { case class Company(name: String) lazy val companies1: String = new File(this.getClass.getResource("/companies-1.csv").toURI).getAbsolutePath def companies1DS(spark: SparkSession): Dataset[Company] = { import spark.implicits._ spark.read .option("header", "true") .option("inferSchema", "true") .csv(companies1) .as[Company] } lazy val companies2: String = new File(this.getClass.getResource("/companies-2.csv").toURI).getAbsolutePath def companies2DS(spark: SparkSession): Dataset[Company] = { import spark.implicits._ spark.read .option("header", "true") .option("inferSchema", "true") .csv(companies2) .as[Company] } class TestCompanyNameAnalyzer extends Analyzer { override def createComponents(fieldName: String): TokenStreamComponents = { val src = new StandardTokenizer() var tok: TokenStream = new LowerCaseFilter(src) val builder = new SynonymMap.Builder(true) builder.add(new CharsRef("ltd"), new CharsRef("l.t.d"), true) builder.add(new CharsRef("ltd"), new CharsRef("limited"), true) builder.add(new CharsRef("inc"), new CharsRef("corporation"), true) builder.add(new CharsRef("inc"), new CharsRef("corp"), true) tok = new SynonymGraphFilter(tok, builder.build(), false) val shingle = new ShingleFilter(tok, 2, 2) shingle.setOutputUnigrams(true) shingle.setOutputUnigramsIfNoShingles(true) tok = shingle new TokenStreamComponents(new Consumer[Reader] { override def accept(r: Reader): Unit = src.setReader(r) }, tok) } } }
phymbert/spark-search
core/src/main/scala/org/apache/spark/search/rdd/SearchRDD.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import org.apache.lucene.search.Query import org.apache.spark.SparkContext import org.apache.spark.rdd.RDD import org.apache.spark.search._ import scala.reflect.ClassTag /** * A search RDD Spark Search brings * advanced full text search features to your RDD. * * {@link org.apache.spark.search.rdd.SearchRDDLucene} * * @tparam S Doc type to index * @author <NAME> */ trait SearchRDD[S] { /** * Default top K. */ val defaultTopK = 1000000 /** * Return the number of indexed elements in the RDD. */ def count(): Long /** * Count how many documents match the given text query. * * @param query Matching query * @return Matched doc count */ def count(query: String): Long = count(parseQueryString(query, options)) /** * Count how many documents match the given lucene query. * * @param query Query to match * @return Matched doc count */ def count(query: StaticQueryProvider): Long /** * Searches the top k hits for this query string. * * @param query Lucene query syntax * @param topK topK to return * @param minScore minimum score of matching documents * @return topK hits collected to the driver as an array * @note this method should only be used if the topK is expected to be small, as * all the data is loaded into the driver's memory. */ def searchList(query: String, topK: Int = defaultTopK, minScore: Double = 0): Array[SearchRecord[S]] = searchListQuery(parseQueryString(query, options), topK, minScore) /** * Searches the top topK hits for this Lucene query. * * @param query Lucene query syntax * @param topK topK to return * @param minScore minimum score of matching documents * @return topK hits * @note this method should only be used if the topK is expected to be small, as * all the data is loaded into the driver's memory. */ def searchListQuery(query: StaticQueryProvider, topK: Int = defaultTopK, minScore: Double = 0): Array[SearchRecord[S]] /** * Searches for the top K hits * per partition for this query string * and returns an RDD with all hits sorted by score in descendent order. * * @param query Lucene query syntax * @param topKByPartition topK to return per partition * @param minScore minimum score of matching documents * @return topK per partition hits RDD sorted by score in descendent order */ def search(query: String, topKByPartition: Int = defaultTopK, minScore: Double = 0): RDD[SearchRecord[S]] = searchQuery(parseQueryString(query, options), topKByPartition, minScore) /** * Searches for the top K hits per partition for this lucene query * and returns an RDD with all hits sorted by score in descendent order. * * @param query Lucene query * @param topKByPartition topK to return per partition * @param minScore minimum score of matching documents * @return topK per partition hits RDD sorted by score in descendent order */ def searchQuery(query: StaticQueryProvider, topKByPartition: Int = defaultTopK, minScore: Double = 0): RDD[SearchRecord[S]] /** * Searches join for this input RDD elements matches against these ones * by building a lucene query string per doc * and returns matching hit as tuples. * * @param other to match with * @param queryBuilder builds the query string to match with the searched document * @param topKByPartition – topK to return by partition * @param minScore minimum score of matching documents * @tparam W Doc type to match with * @return matches doc and related hit RDD */ def searchJoin[W: ClassTag](other: RDD[W], queryBuilder: W => String, topKByPartition: Int = defaultTopK, minScore: Double = 0): RDD[(W, Option[SearchRecord[S]])] = searchJoinQuery(other, queryStringBuilder(queryBuilder, options), topKByPartition, minScore) /** * Searches join for this input RDD elements matches against these ones * by building a lucene query per doc * and returns matching hit as tuples. * * @param other to match with * @param queryBuilder builds the query string to match with the searched document * @param topKByPartition – topK to return by partition * @param minScore minimum score of matching documents * @tparam W Doc type to match with * @return matches doc and related hit RDD */ def searchJoinQuery[W: ClassTag](other: RDD[W], queryBuilder: W => Query, topKByPartition: Int = defaultTopK, minScore: Double = 0): RDD[(W, Option[SearchRecord[S]])] /** * Searches for this input RDD elements matches against these ones * by building a lucene query string per doc * and returns matching hits per doc. * * @param rdd to match with * @param queryBuilder builds the query string to match with the searched document * @param topK topK to return * @param minScore minimum score of matching documents * @tparam K key type of doc used in the top k monoid * @tparam V Doc type to match with * @return matches doc and related hits RDD */ def matches[K, V](rdd: RDD[(K, V)], queryBuilder: V => String, topK: Int = defaultTopK, minScore: Double = 0) (implicit kClassTag: ClassTag[K], vClassTag: ClassTag[V]): RDD[(K, (V, Array[SearchRecord[S]]))] = matchesQuery(rdd, queryStringBuilder(queryBuilder, options), topK, minScore) /** * Searches for this input RDD elements matches against these ones * by building a lucene query per doc * and returns matching hits per doc. * * @param rdd to match with * @param queryBuilder builds the lucene query to join with the searched document * @param topK topK to return * @param minScore minimum score of matching documents * @tparam K key type of doc used in the top k monoid * @tparam V Doc type to match with * @return matches doc and related hits RDD */ def matchesQuery[K, V](rdd: RDD[(K, V)], queryBuilder: V => Query, topK: Int = defaultTopK, minScore: Double = 0) (implicit kClassTag: ClassTag[K], vClassTag: ClassTag[V]): RDD[(K, (V, Array[SearchRecord[S]]))] private[spark] def elementClassTag: ClassTag[S] /** * Alias for searchDropDuplicates */ def distinct(numPartitions: Int): RDD[S] = { implicit val classTagS: ClassTag[S] = elementClassTag searchDropDuplicates[Long, S]() } /** * Drops duplicated records by applying lookup for matching hits of the query against this RDD. * * @param queryBuilder builds the lucene query to search for duplicate * @param minScore minimum score of matching documents */ def searchDropDuplicates[K: ClassTag, C: ClassTag]( queryBuilder: S => Query = null, createKey: S => K = (s: S) => s.hashCode.toLong.asInstanceOf[K], minScore: Double = 0, createCombiner: Seq[SearchRecord[S]] => C = (ss: Seq[SearchRecord[S]]) => ss.head.source.asInstanceOf[C], mergeValue: (C, Seq[SearchRecord[S]]) => C = (c: C, _: Seq[SearchRecord[S]]) => c, mergeCombiners: (C, C) => C = (c: C, _: C) => c, numPartitionInJoin: Int = getNumPartitions, topKToDeduplicate: Int = defaultTopK )(implicit ord: Ordering[K]): RDD[C] /** * Save the current indexed RDD onto hdfs * in order to be able to reload it later on. * * @param path Path on the spark file system (hdfs) to save on */ def save(path: String): Unit /** * @return the number of partitions of this RDD. */ def getNumPartitions: Int /** * @return current search options */ def options: SearchOptions[S] } object SearchRDD { /** * Reload an indexed RDD from spark FS. * * @param sc current spark context * @param path Path where the index were saved * @param options Search option * @tparam T Type of beans or case class this RDD is binded to * @return Restored RDD */ def load[T: ClassTag](sc: SparkContext, path: String, options: SearchOptions[T] = defaultOpts[T] ): SearchRDD[T] = new SearchRDDLucene[T](sc, new SearchIndexReloadedRDD[T](sc, path, options), options, Nil) }
phymbert/spark-search
core/src/main/scala/org/apache/spark/search/rdd/SearchRDDImpl.scala
<filename>core/src/main/scala/org/apache/spark/search/rdd/SearchRDDImpl.scala /* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import java.io.{IOException, ObjectOutputStream} import org.apache.hadoop.conf.Configuration import org.apache.hadoop.fs.{FileSystem, Path} import org.apache.lucene.search.Query import org.apache.spark._ import org.apache.spark.rdd.RDD import org.apache.spark.search._ import org.apache.spark.storage.StorageLevel import org.apache.spark.util.{BoundedPriorityQueue, Utils} import scala.collection.JavaConverters.asScalaIteratorConverter import scala.reflect.ClassTag /** * A search RDD indexes parent RDD partitions to lucene indexes. * It builds for all parent RDD partitions a one-2-one volatile Lucene index * available during the lifecycle of the spark session across executors local directories and RAM. * * @author <NAME> */ private[search] class SearchRDDLucene[S: ClassTag](sc: SparkContext, val indexerRDD: SearchRDDIndexer[S], val options: SearchOptions[S], val deps: Seq[Dependency[_]]) extends RDD[S](sc, Seq(new OneToOneDependency(indexerRDD)) ++ deps) with SearchRDD[S] { def this(rdd: RDD[S], options: SearchOptions[S]) { this(rdd.sparkContext, new SearchRDDIndexer(rdd, options), options, Seq(new OneToOneDependency(rdd))) } if (options.getIndexationOptions.isCacheSearchIndexRDD) { indexerRDD.persist(StorageLevel.DISK_ONLY) } override def count(): Long = runSearchJob[Long, Long](spr => spr.count(), _.sum) override def count(query: StaticQueryProvider): Long = runSearchJob[Long, Long](spr => spr.count(query()), _.sum) override def searchListQuery(query: StaticQueryProvider, topK: Int = defaultTopK, minScore: Double = 0 ): Array[SearchRecord[S]] = runSearchJob[Array[SearchRecord[S]], Array[SearchRecord[S]]]( spr => _partitionReaderSearchList(spr, query(), topK, minScore), reduceSearchRecordsByTopK(topK)) override def searchQuery(query: StaticQueryProvider, topKByPartition: Int = defaultTopK, minScore: Double = 0 ): RDD[SearchRecord[S]] = { val indexDirectoryByPartition = indexerRDD._indexDirectoryByPartition indexerRDD.mapPartitionsWithIndex( (index, _) => tryAndClose(reader(indexDirectoryByPartition, index)) { spr => _partitionReaderSearchList(spr, query(), topKByPartition, minScore) }.iterator ).sortBy(_.score, ascending = false) } override def matchesQuery[K, V](other: RDD[(K, V)], queryBuilder: V => Query, topK: Int = 10, minScore: Double = 0 ) (implicit kClassTag: ClassTag[K], vClassTag: ClassTag[V]): RDD[(K, (V, Array[SearchRecord[S]]))] = { val unwrapDoc = sparkContext.clean((kv: (K, V)) => queryBuilder(kv._2)) val cartesianRDD: RDD[((K, V), Option[SearchRecord[S]])] = new SearchRDDCartesian[(K, V), S]( indexerRDD, other, unwrapDoc, options.getReaderOptions, topK, minScore ) val pairedRDD: RDD[(K, (V, Option[SearchRecord[S]]))] = cartesianRDD.map { case ((k: K, v: V), Some(sr)) => (k, (v, Some(sr))) case ((k: K, v: V), None) => (k, (v, None)) } // TopK monoid implicit val ord: Ordering[(V, Option[SearchRecord[S]])] = Ordering.by(_._2.map(_.score)) val topKByKey = pairedRDD .aggregateByKey(new BoundedPriorityQueue[(V, Option[SearchRecord[S]])](topK)(ord))( seqOp = (topK, searchRecord) => topK += searchRecord, combOp = (topK1, topK2) => topK1 ++= topK2 ) val matchesByKey = topKByKey .mapValues(matches => (matches.head._1, matches.filter(_._2.isDefined).map(_._2.get).toArray.reverse)) matchesByKey } override def searchJoinQuery[W: ClassTag](other: RDD[W], queryBuilder: W => Query, topKByPartition: Int = defaultTopK, minScore: Double = 0): RDD[(W, Option[SearchRecord[S]])] = { new SearchRDDCartesian[W, S]( indexerRDD, other, queryBuilder, options.getReaderOptions, topKByPartition, minScore ) } /** * Alias for * [[org.apache.spark.search.rdd.SearchRDD#searchDropDuplicates(scala.Function1, int, double, int)}]] */ override def distinct(numPartitions: Int): RDD[S] = searchDropDuplicates[Long, S]() /** * Drops duplicated records by applying lookup for matching hits of the query against this RDD. * * @param queryBuilder builds the lucene query to search for duplicate * @param minScore minimum score of matching documents */ override def searchDropDuplicates[K: ClassTag, C: ClassTag]( queryBuilder: S => Query = null, // Default query builder createKey: S => K = (s: S) => s.hashCode.toLong.asInstanceOf[K], minScore: Double = 0, createCombiner: Seq[SearchRecord[S]] => C = (ss: Seq[SearchRecord[S]]) => ss.head.source.asInstanceOf[C], mergeValue: (C, Seq[SearchRecord[S]]) => C = (c: C, _: Seq[SearchRecord[S]]) => c, mergeCombiners: (C, C) => C = (c: C, _: C) => c, numPartitionInJoin: Int = getNumPartitions, topKToDeduplicate: Int = defaultTopK )(implicit ord: Ordering[K]): RDD[C] = { val cleanedKey = sparkContext.clean(createKey) val unwrapDoc = sparkContext.clean((ks: (K, S)) => (queryBuilder match { case null => defaultQueryBuilder[S](options)(elementClassTag) case _ => queryBuilder }) (ks._2)) val pairedRDD = map(s => (cleanedKey(s), s)).repartition(numPartitionInJoin) val cartesianRDD: RDD[((K, S), Option[SearchRecord[S]])] = new SearchRDDCartesian[(K, S), S]( indexerRDD, pairedRDD, unwrapDoc, options.getReaderOptions, topKToDeduplicate, minScore) val pairedWithSearchedRDD: RDD[(K, (S, Option[SearchRecord[S]]))] = cartesianRDD.map { case ((k: K, s: S), Some(sr)) => (k, (s, Some(sr))) case ((k: K, s: S), None) => (k, (s, None)) } val hitsByKey: RDD[(K, List[(S, Option[SearchRecord[S]])])] = pairedWithSearchedRDD.aggregateByKey(List[(S, Option[SearchRecord[S]])]())( seqOp = (matches, searchRecord) => { matches ++ List(searchRecord) }, combOp = (matches1, matches2) => { matches1 ++ matches2 } ) val keysAndDocs: RDD[(Seq[K], Seq[SearchRecord[S]])] = hitsByKey.map { case (key: K, matches: List[(S, Option[SearchRecord[S]])]) => val doc: S = matches.head._1 // assumed left join val otherMatches = matches .filter(_._2.isDefined) .map(m => (createKey(m._2.get.source), m._2.get)) // Remove self matching if exists (depending on score filter) .filter((ks) => ks._1 != key) val keys = (Seq(key) ++ otherMatches.map(_._1)).sorted (keys, Seq(new SearchRecord[S](-1, -1, 0, -1, doc)) ++ otherMatches.map(_._2)) } keysAndDocs.combineByKey(createCombiner, mergeValue, mergeCombiners) .values } override def save(pathString: String): Unit = { logInfo(s"Saving index with $getNumPartitions partitions to $pathString ...") // Be sure we are indexed count() val hadoopConf = new Configuration() val hdfs = FileSystem.get(hadoopConf) val path = new Path(pathString) if (hdfs.exists(path)) { // FIXME issue github #77 https://github.com/phymbert/spark-search/issues/77 throw new SearchException(s"HDFS path $path already exists, delete it first") } indexerRDD.save(pathString) logInfo(s"Index with $getNumPartitions partitions saved to $path") } private[spark] override def elementClassTag: ClassTag[S] = super.elementClassTag override val partitioner: Option[Partitioner] = indexerRDD.partitioner override def getPreferredLocations(split: Partition): Seq[String] = firstParent[S].asInstanceOf[SearchRDDIndexer[S]] .getPreferredLocations(split.asInstanceOf[SearchPartition[S]].searchIndexPartition) override def repartition(numPartitions: Int)(implicit ord: Ordering[S]): RDD[S] = new SearchRDDLucene[S](firstParent.firstParent.repartition(numPartitions), options) def _partitionReaderSearchList(r: SearchPartitionReader[S], query: Query, topK: Int, minScore: Double): Array[SearchRecord[S]] = r.search(query, topK, minScore).map(searchRecordJavaToProduct) protected[rdd] def reduceSearchRecordsByTopK(topK: Int): Iterator[Array[SearchRecord[S]]] => Array[SearchRecord[S]] = _.reduce(_ ++ _).sortBy(_.score)(Ordering[Double].reverse).take(topK) protected[rdd] def runSearchJob[R: ClassTag, A: ClassTag](searchByPartition: SearchPartitionReader[S] => R, reducer: Iterator[R] => A): A = runSearchJobWithContext((_searchByPartition, _) => searchByPartition(_searchByPartition), reducer) protected[rdd] def runSearchJobWithContext[R: ClassTag, A: ClassTag](searchByPartitionWithContext: (SearchPartitionReader[S], TaskContext) => R, reducer: Iterator[R] => A): A = { val indexDirectoryByPartition = indexerRDD._indexDirectoryByPartition val ret = sparkContext.runJob(indexerRDD, (context: TaskContext, it: Iterator[Array[Byte]]) => { val index = context.partitionId() // Unzip if needed ZipUtils.unzipPartition(indexDirectoryByPartition(index), it) tryAndClose(reader(indexDirectoryByPartition, index)) { r => searchByPartitionWithContext(r, context) } }) reducer(ret.toIterator) } private def reader(indexDirectoryByPartition: Map[Int, String], index: Int): SearchPartitionReader[S] = reader(index, indexDirectoryByPartition(index)) private def reader(index: Int, indexDirectory: String): SearchPartitionReader[S] = new SearchPartitionReader[S](index, indexDirectory, elementClassTag.runtimeClass.asInstanceOf[Class[S]], options.getReaderOptions) override def compute(split: Partition, context: TaskContext): Iterator[S] = { val partition = split.asInstanceOf[SearchPartition[S]] val indexedRDD = firstParent[Array[Byte]].asInstanceOf[SearchRDDIndexer[S]] // Trigger indexation if not done yet on parent rdd partition node val it: Iterator[Array[Byte]] = indexedRDD.iterator(partition.searchIndexPartition, context) val indexDirectory = partition.searchIndexPartition.indexDir // Unzip if needed ZipUtils.unzipPartition(indexDirectory, it) val spr = reader(partition.searchIndexPartition.index, partition.searchIndexPartition.indexDir) context.addTaskCompletionListener[Unit](ctx => { spr.close() }) spr.docs().asScala.map(searchRecordJavaToProduct).map(_.source) } override protected def getPartitions: Array[Partition] = { // One-2-One partition firstParent.partitions.map(p => new SearchPartition(p.index, indexerRDD)).toArray } override def clearDependencies(): Unit = { super.clearDependencies() if (options.getIndexationOptions.isCacheSearchIndexRDD) { indexerRDD.unpersist() } } } class SearchPartition[T](val idx: Int, @transient private val searchRDD: SearchRDDIndexer[T]) extends Partition { override def index: Int = idx var searchIndexPartition: SearchPartitionIndex[T] = searchRDD.partitions(idx).asInstanceOf[SearchPartitionIndex[T]] @throws(classOf[IOException]) private def writeObject(oos: ObjectOutputStream): Unit = Utils.tryOrIOException { // Update the reference to parent split at the time of task serialization searchIndexPartition = searchRDD.partitions(idx).asInstanceOf[SearchPartitionIndex[T]] oos.defaultWriteObject() } }
phymbert/spark-search
sql/src/test/scala/org/apache/spark/search/sql/LocalSparkSession.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.sql import io.netty.util.internal.logging.{InternalLoggerFactory, Slf4JLoggerFactory} import org.apache.spark.sql.SparkSession import org.scalatest.{BeforeAndAfterAll, BeforeAndAfterEach, Suite} trait LocalSparkSession extends BeforeAndAfterEach with BeforeAndAfterAll { self: Suite => @transient var _spark: SparkSession = _ override def beforeAll() { super.beforeAll() InternalLoggerFactory.setDefaultFactory(Slf4JLoggerFactory.INSTANCE) } override def afterEach() { try { resetSparkSession() } finally { super.afterEach() } } def resetSparkSession(): Unit = { LocalSparkSession.stop(_spark) _spark = null } override def beforeEach() { _spark = SparkSession.builder() .master("local[*]") .appName("Spark Search Test") .getOrCreate() } } object LocalSparkSession { def stop(sc: SparkSession) { if (sc != null) { sc.stop() } // To avoid RPC rebinding to the same port, since it doesn't unbind immediately on shutdown System.clearProperty("spark.driver.port") } /** Runs `f` by passing in `sc` and ensures that `sc` is stopped. */ def withSpark[T](sc: SparkSession)(f: SparkSession => T): T = { try { f(sc) } finally { stop(sc) } } }
phymbert/spark-search
core/src/test/scala/org/apache/spark/search/rdd/DocumentBeanUpdaterSuite.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import org.apache.spark.search.reflect.DocumentBeanUpdater import org.apache.spark.search._ import org.scalatest.funsuite.AnyFunSuite class DocumentBeanUpdaterSuite extends AnyFunSuite { test("document bean updater should support case class") { val documentBeanUpdater = new DocumentBeanUpdater[TestData.Person] val indexingDocument = new DocumentUpdater.IndexingDocument[TestData.Person](IndexationOptions.defaultOptions .asInstanceOf[IndexationOptions[TestData.Person]]) val doc = indexingDocument.doc indexingDocument.element = TestData.Person("John", "Doe", 34) documentBeanUpdater.update(indexingDocument) assertResult(3)(doc.getFields.size) assertResult("John")(doc.get("firstName")) } }
phymbert/spark-search
benchmark/src/main/scala-2.11/benchmark/LuceneRDDBenchmark.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package benchmark import org.apache.lucene.analysis.standard.StandardAnalyzer import org.apache.spark.rdd.RDD import org.zouzias.spark.lucenerdd.LuceneRDD object LuceneRDDBenchmark extends BaseBenchmark("LuceneRDD") { def main(args: Array[String]): Unit = run() override def countNameMatches(companies: RDD[Company], name: String): RDD[(Double, String)] = { import spark.implicits._ val luceneRDD = LuceneRDD(companies.toDF(), classOf[StandardAnalyzer].getName, classOf[StandardAnalyzer].getName, "classic") luceneRDD.query(s"name:${name}", Int.MaxValue) .map(r => (r.getAs[Float]("__score__").toDouble, r.getAs[String]("name"))) .sortBy(_._1, ascending = false) // Not sorted by RDD but by partition } override def joinMatch(companies: RDD[Company], secEdgarCompanies: RDD[SecEdgarCompanyInfo]): RDD[(String, Double, String)] = { import spark.implicits._ val luceneRDD = LuceneRDD(companies.toDF(), classOf[StandardAnalyzer].getName, classOf[StandardAnalyzer].getName, "classic") val prefixLinker = (company: SecEdgarCompanyInfo) => { val skipped = company.companyName.slice(0, 64).replaceAll("([+\\-=&|<>!(){}\\[\\]^\"~*?:/])", "\\\\$1") s"name:${"\"" + skipped + "\""}" } luceneRDD.link(secEdgarCompanies, prefixLinker, 1) .filter(_._2.nonEmpty) .map(t => (t._1.companyName, t._2.head.getAs[Double]("__score__"), t._2.head.getAs[String]("name"))) } }
phymbert/spark-search
examples/src/main/scala/all/examples/org/apache/spark/search/rdd/SearchRDDExamples.scala
<filename>examples/src/main/scala/all/examples/org/apache/spark/search/rdd/SearchRDDExamples.scala /* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package all.examples.org.apache.spark.search.rdd import org.apache.lucene.analysis.en.EnglishAnalyzer import org.apache.spark.search._ import org.apache.spark.sql.SparkSession import ExampleData._ import org.apache.hadoop.conf.Configuration import org.apache.hadoop.fs.{FileSystem, Path} import org.apache.spark.rdd.RDD /** * Spark Search RDD examples. */ object SearchRDDExamples { def main(args: Array[String]): Unit = { val spark = SparkSession.builder() .appName("Spark Search Examples") .getOrCreate() val sc = spark.sparkContext sc.setLogLevel("ERROR") Console.setOut(Console.err) // Amazon computers customer reviews val computersReviews: RDD[Review] = loadReviews(spark, "http://snap.stanford.edu/data/amazon/productGraph/categoryFiles/reviews_Computers.json.gz") // Search RDD API import org.apache.spark.search.rdd._ // to implicitly enhance RDD with search features // Count positive review: indexation + count matched doc val happyReview = computersReviews.count("reviewText:happy OR reviewText:best OR reviewText:good OR reviewText:\"sounds great\"~") println(s"$happyReview positive reviews :)") // Search for key words println(s"Full text search results:") computersReviews.searchList("reviewText:\"World of Warcraft\" OR reviewText:\"Civilization IV\"", topK = 10, minScore = 10) .foreach(println) // /!\ Important lucene indexation is done each time a SearchRDD is computed, // if you do multiple operations on the same parent RDD, you might have a variable in the driver: val computersReviewsSearchRDD: SearchRDD[Review] = computersReviews.searchRDD( SearchOptions.builder[Review]() // See all other options SearchRDDOptions, IndexationOptions and ReaderOptions .read((r: ReaderOptions.Builder[Review]) => r.defaultFieldName("reviewText")) .analyzer(classOf[EnglishAnalyzer]) .build()) println("All reviews speaking about hardware:") computersReviewsSearchRDD.searchList("(RAM OR memory) AND (CPU OR processor~)^4", topK = 10, minScore = 15).foreach(println) // Fuzzy matching println("Some typo in names:") computersReviews.search("(reviewerName:Mikey~0.8) OR (reviewerName:\"Patrik\"~0.4) OR (reviewerName:jonh~0.2)", topKByPartition = 10) .map(doc => s"${doc.source.reviewerName}=${doc.score}") .collect() .map(_.toLowerCase) .distinct .foreach(println) // Amazon software customer reviews println("Downloading software reviews...") val softwareReviews: RDD[Review] = loadReviews(spark, "http://snap.stanford.edu/data/amazon/productGraph/categoryFiles/reviews_Software_10.json.gz") // Match software and computer reviewers println("Joined software and computer reviews by reviewer names:") val matchesReviewers: RDD[(Review, Array[SearchRecord[Review]])] = computersReviews.matches[String, Review]( softwareReviews.filter(_.reviewerName != null).map(sr => (sr.asin, sr)), (sr: Review) => "reviewerName:\"" + sr.reviewerName.replace('"', ' ') + "\"~0.4", 10) .values matchesReviewers .filter(_._2.nonEmpty) .sortBy(_._2.length, ascending = false) .map(m => (s"Reviewer ${m._1.reviewerName} reviews computer ${m._1.asin} but also on software:", m._2.map(h => s"${h.source.reviewerName}=${h.score}=${h.source.asin}").toList)) .collect() .take(20) .foreach(println) // Drop duplicates println("Dropping duplicated reviewers:") val distinctReviewers: RDD[String] = computersReviews .filter(_.reviewerName != null) .searchDropDuplicates[Long, Review]( queryBuilder = queryStringBuilder[Review](sr => "reviewerName:\"" + sr.reviewerName.replace('"', ' ') + "\"~0.4") ).map(sr => sr.reviewerName) distinctReviewers.collect().sorted.take(20).foreach(println) // Save & restore example FileSystem.get(new Configuration).delete(new Path("/hdfs-tmp/hdfs-pathname"), true) println(s"Saving index to hdfs...") computersReviews.save("/hdfs-tmp/hdfs-pathname") println(s"Restoring from previous indexation:") val restoredSearchRDD: SearchRDD[Review] = SearchRDD.load[Review](sc, "/hdfs-tmp/hdfs-pathname") val happyReview2 = restoredSearchRDD.count("reviewText:happy OR reviewText:best OR reviewText:good") println(s"$happyReview2 positive reviews after restoration") // Restored index can be use as classical rdd val topReviewer = restoredSearchRDD.map(r => (r.reviewerID, 1)) .reduceByKey(_ + _) .sortBy(_._2, ascending = false) .take(1).head println(s"${topReviewer._1} has submitted ${topReviewer._2} reviews") val topReviewNameFiltered = restoredSearchRDD.filter(_.reviewerID.equals(topReviewer._1)) .take(1).head println(s"He is named ${topReviewNameFiltered.reviewerName}") spark.stop() } }
phymbert/spark-search
sql/src/main/scala/org/apache/spark/search/sql/expression.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.sql import org.apache.spark.sql.catalyst.InternalRow import org.apache.spark.sql.catalyst.expressions.codegen.Block._ import org.apache.spark.sql.catalyst.expressions.codegen.{CodeGenerator, CodegenContext, ExprCode, FalseLiteral, JavaCode} import org.apache.spark.sql.catalyst.expressions.{BinaryExpression, Expression, LeafExpression} import org.apache.spark.sql.types.{DataType, _} trait SearchExpression case class ScoreExpression() extends LeafExpression with SearchExpression { override def toString: String = s"SCORE" override def dataType: DataType = DoubleType override def nullable: Boolean = false override def eval(input: InternalRow): Any = { if (input != null) input.getDouble(input.numFields - 1) else Double.NaN } override protected def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = { /* if (ctx.currentVars != null && ctx.currentVars.last != null) { val oev = ctx.currentVars.head ev.isNull = oev.isNull ev.value = oev.value ev.copy(code = oev.code) } else { assert(ctx.INPUT_ROW != null, "INPUT_ROW and currentVars cannot both be null.") val javaType = JavaCode.javaType(dataType) ev.copy(code = code"$javaType ${ev.value} = ${ctx.INPUT_ROW}.getDouble(i.numFields() - 1);", isNull = FalseLiteral) }*/ ev.copy(code = code""" final ${CodeGenerator.javaType(dataType)} ${ev.value} = 10000d;""", isNull = FalseLiteral) } } /** * @author <NAME> */ case class MatchesExpression(left: Expression, right: Expression) extends BinaryExpression with SearchExpression { override def toString: String = s"$left MATCHES $right" override protected def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = { ev.copy(code = code""" final ${CodeGenerator.javaType(dataType)} ${ev.value} = true;""", isNull = FalseLiteral) } override def dataType: DataType = BooleanType }
phymbert/spark-search
core/src/main/scala/org/apache/spark/search/rdd/SearchJavaBaseRDD.scala
<filename>core/src/main/scala/org/apache/spark/search/rdd/SearchJavaBaseRDD.scala /* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import org.apache.spark.SparkContext import org.apache.spark.api.java.{JavaPairRDD, JavaRDD} import org.apache.spark.search._ import org.apache.spark.search.rdd.SearchRDDJava.{QueryBuilder, QueryStringBuilder} import scala.reflect.ClassTag /** * Java friendly version of [[SearchRDD]]. */ trait SearchRDDJavaWrapper[S] extends SearchRDDJava[S] { val searchRDD: SearchRDD[S] val classTag: ClassTag[S] override def count(): Long = searchRDD.count() override def count(query: String): Long = searchRDD.count(parseQueryString(query)) override def searchList(query: String, topK: Int): Array[SearchRecordJava[S]] = searchRDD.searchListQuery(parseQueryString(query), topK).map(searchRecordAsJava(_)) override def searchList(query: String, topK: Int, minScore: Double): Array[SearchRecordJava[S]] = searchRDD.searchListQuery(parseQueryString(query), topK, minScore).map(searchRecordAsJava) override def search(query: String, topK: Int, minScore: Double): JavaRDD[SearchRecordJava[S]] = searchRDD.search(query, topK).map(searchRecordAsJava(_)).toJavaRDD() override def matches[K, V](rdd: JavaPairRDD[K, V], queryBuilder: QueryStringBuilder[V], topK: Int, minScore: Double): JavaPairRDD[K, (V, Array[SearchRecordJava[S]])] = { implicit val kClassTag: ClassTag[K] = rdd.kClassTag implicit val vClassTag: ClassTag[V] = rdd.vClassTag new JavaPairRDD( searchRDD.matches[K, V](rdd.rdd, v => queryBuilder.build(v), topK, minScore) .mapValues(m => (m._1, m._2.map(searchRecordAsJava(_))))) } override def save(path: String): Unit = searchRDD.save(path) override def matchesQuery[K, V](rdd: JavaPairRDD[K, V], queryBuilder: QueryBuilder[V], topK: Int, minScore: Double): JavaPairRDD[K, (V, Array[SearchRecordJava[S]])] = { implicit val kClassTag: ClassTag[K] = rdd.kClassTag implicit val vClassTag: ClassTag[V] = rdd.vClassTag new JavaPairRDD( searchRDD.matchesQuery[K, V](rdd.rdd, v => queryBuilder.build(v), topK, minScore) .mapValues(m => (m._1, m._2.map(searchRecordAsJava(_))))) } override def javaRDD(): JavaRDD[S] = { implicit val classTag: ClassTag[S] = this.classTag new JavaRDD(searchAsRDD(searchRDD)) } private def searchRecordAsJava(sr: SearchRecord[S]) = new SearchRecordJava[S](sr.id, sr.partitionIndex, sr.score, sr.shardIndex, sr.source) } class SearchJavaBaseRDD[T: ClassTag](rdd: JavaRDD[T], opts: SearchOptions[T]) extends JavaRDD[T](rdd.rdd) with SearchRDDJavaWrapper[T] { override lazy val searchRDD: SearchRDD[T] = rdd.rdd.searchRDD(opts) override val classTag: ClassTag[T] = searchRDD.elementClassTag override def count(): Long = searchRDD.count() } class SearchRDDReloadedJava[T: ClassTag](val searchRdd: SearchRDD[T]) extends SearchRDDJavaWrapper[T] with Serializable { override lazy val searchRDD: SearchRDD[T] = searchRdd override val classTag: ClassTag[T] = searchRDD.elementClassTag } object SearchRDDReloadedJava { def load[T: ClassTag](sc: SparkContext, path: String, options: SearchOptions[T] ): SearchRDDJava[T] = new SearchRDDReloadedJava(SearchRDD.load(sc, path, options)) }
phymbert/spark-search
core/src/main/scala/org/apache/spark/search/rdd/ZipUtils.scala
<reponame>phymbert/spark-search /* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import java.io.{File, FileInputStream, InputStream, OutputStream} import java.nio.file.{Files, StandardCopyOption} import java.util.function.Consumer import java.util.zip.{ZipEntry, ZipInputStream, ZipOutputStream} private[rdd] object ZipUtils { class IteratorInputStream(it: Iterator[Array[Byte]]) extends InputStream { var offset: Int = 0 var buff: Array[Byte] = Array.empty override def read(): Int = { // FIXME the slowest implementation ever if (offset < buff.length) { val b = buff(offset) offset = offset + 1 b & 0xFF // Make sure value is between > 0..255 } else if (it.hasNext) { offset = 1 buff = it.next buff(0) & 0xFF } else { -1 } } } class FileInputStreamIterator(filePath: File) extends Iterator[Array[Byte]] { override def hasNext: Boolean = { if (!finished) { if (is == null) { is = new FileInputStream(filePath) } read = is.read(_next) finished = read < 0 if (finished) { is.close() } } !finished } var is: InputStream = _ var read: Int = -1 val _next: Array[Byte] = new Array[Byte](8192) var finished = false override def next(): Array[Byte] = { _next.slice(0, read) } } def unzipPartition(indexDir: String, it: Iterator[Array[Byte]]): Unit = { unzipPartition(indexDir, new IteratorInputStream(it)) } def unzipPartition(indexDir: String, is: InputStream): Unit = { val parentLocalFile = new File(indexDir) if (parentLocalFile.mkdirs()) { val zis = new ZipInputStream(is) Stream.continually(zis.getNextEntry).takeWhile(_ != null).foreach { file => Files.copy(zis, new File(parentLocalFile, file.getName).toPath, StandardCopyOption.REPLACE_EXISTING) zis.closeEntry() } zis.close() } is.close() } def zipPartition(localIndexDirPath: java.nio.file.Path, fos: OutputStream): Unit = { val zip = new ZipOutputStream(fos) val files = Files.list(localIndexDirPath) files.forEach { new Consumer[java.nio.file.Path] { override def accept(file: java.nio.file.Path): Unit = { zip.putNextEntry(new ZipEntry(file.toFile.getName)) Files.copy(file, zip) zip.closeEntry() } } } files.close() zip.close() fos.close() } }
phymbert/spark-search
core/src/main/scala/org/apache/spark/search/rdd/SearchRDDCartesian.scala
<filename>core/src/main/scala/org/apache/spark/search/rdd/SearchRDDCartesian.scala<gh_stars>10-100 /* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import java.io.{IOException, ObjectOutputStream} import org.apache.lucene.search.Query import org.apache.spark._ import org.apache.spark.rdd.RDD import org.apache.spark.search.{ReaderOptions, SearchException, SearchRecord} import org.apache.spark.util.Utils import collection.JavaConverters._ import scala.reflect.{ClassTag, classTag} /** * Result RDD of a cartesian search. * * @author <NAME> */ class SearchRDDCartesian[V: ClassTag, S: ClassTag]( @transient var searchRDDLuceneIndexer: SearchRDDIndexer[S], @transient var other: RDD[V], queryBuilder: V => Query, readerOptions: ReaderOptions[S], topK: Int = Int.MaxValue, minScore: Double = 0) extends RDD[(V, Option[SearchRecord[S]])](searchRDDLuceneIndexer.context, Nil) with Serializable { override val partitioner: Option[Partitioner] = searchRDDLuceneIndexer.partitioner override protected def getPreferredLocations(split: Partition): Seq[String] = firstParent[S].asInstanceOf[SearchRDDIndexer[S]] .getPreferredLocations(split.asInstanceOf[MatchRDDPartition].searchIndexPartition) override def compute(split: Partition, context: TaskContext): Iterator[(V, Option[SearchRecord[S]])] = { val matchPartition = split.asInstanceOf[MatchRDDPartition] // Be sure partition is indexed in our worker val it = firstParent[Array[Byte]].iterator(matchPartition.searchIndexPartition, context) // Unzip if needed ZipUtils.unzipPartition(matchPartition.searchIndexPartition.indexDir, it) val spr = reader(matchPartition.searchIndexPartition.index, matchPartition.searchIndexPartition.indexDir) context.addTaskCompletionListener[Unit](ctx => { spr.close() }) // Match other partition against our one parent[V](1).iterator(matchPartition.otherPartition, context) .flatMap(searchFor => { try { val searchResultIt = spr.iterator(queryBuilder(searchFor), topK, minScore) val result = if (searchResultIt.size() == 0) Iterator((searchFor, None)) // FIXME left join always else searchResultIt.asScala .map(searchRecordJavaToProduct) .map(s => (searchFor, Some(s))) result } catch { case e: SearchException => throw new SearchException(s"error during matching $searchFor: $e", e) } }) } override def clearDependencies(): Unit = { super.clearDependencies() searchRDDLuceneIndexer = null other = null } private val numPartitionsInOtherRdd = other.partitions.length override protected def getPartitions: Array[Partition] = { val parts = new Array[Partition](searchRDDLuceneIndexer.partitions.length * numPartitionsInOtherRdd) for (s1 <- searchRDDLuceneIndexer.partitions; s2 <- other.partitions) { val idx = s1.index * numPartitionsInOtherRdd + s2.index parts(idx) = new MatchRDDPartition(idx, searchRDDLuceneIndexer, other, s1.index, s2.index) } parts } override def getDependencies: Seq[Dependency[_]] = List( new NarrowDependency(searchRDDLuceneIndexer) { def getParents(id: Int): Seq[Int] = List(id / numPartitionsInOtherRdd) }, new NarrowDependency(other) { def getParents(id: Int): Seq[Int] = List(id % numPartitionsInOtherRdd) } ) private def reader(index: Int, indexDirectory: String): SearchPartitionReader[S] = new SearchPartitionReader[S](index, indexDirectory, classTag[S].runtimeClass.asInstanceOf[Class[S]], readerOptions) class MatchRDDPartition(val idx: Int, @transient private val searchRDDLuceneIndexer: SearchRDDIndexer[S], @transient private val other: RDD[V], val searchRDDIndex: Int, val otherIndex: Int ) extends Partition { override val index: Int = idx var searchIndexPartition: SearchPartitionIndex[S] = searchRDDLuceneIndexer.partitions(searchRDDIndex).asInstanceOf[SearchPartitionIndex[S]] var otherPartition: Partition = other.partitions(otherIndex) @throws(classOf[IOException]) private def writeObject(oos: ObjectOutputStream): Unit = Utils.tryOrIOException { // Update the reference to parent split at the time of task serialization searchIndexPartition = searchRDDLuceneIndexer.partitions(searchRDDIndex).asInstanceOf[SearchPartitionIndex[S]] otherPartition = other.partitions(otherIndex) oos.defaultWriteObject() } } }
phymbert/spark-search
core/src/main/scala/org/apache/spark/search/rdd/RDDWithSearch.scala
<reponame>phymbert/spark-search<filename>core/src/main/scala/org/apache/spark/search/rdd/RDDWithSearch.scala<gh_stars>10-100 /* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import org.apache.lucene.search.Query import org.apache.spark.rdd.RDD import org.apache.spark.search.{SearchOptions, _} import scala.reflect.ClassTag /** * Add search features to [[org.apache.spark.rdd.RDD]] * using [[org.apache.spark.search.rdd.SearchRDDLucene]]. * * @author <NAME> */ private[rdd] class RDDWithSearch[S: ClassTag](val rdd: RDD[S], val opts: SearchOptions[S] = defaultOpts[S] ) extends SearchRDD[S] { private[rdd] lazy val _searchRDD: SearchRDD[S] = searchRDD(opts) override def count(): Long = _searchRDD.count() override def count(query: StaticQueryProvider): Long = _searchRDD.count(query) override def searchListQuery(query: StaticQueryProvider, topK: Int = defaultTopK, minScore: Double = 0): Array[SearchRecord[S]] = _searchRDD.searchListQuery(query, topK, minScore) override def searchQuery(query: StaticQueryProvider, topKByPartition: Int = defaultTopK, minScore: Double = 0): RDD[SearchRecord[S]] = _searchRDD.searchQuery(query, topKByPartition, minScore) override def matchesQuery[K, V](rdd: RDD[(K, V)], queryBuilder: V => Query, topK: Int = 10, minScore: Double = 0 ) (implicit kClassTag: ClassTag[K], vClassTag: ClassTag[V]): RDD[(K, (V, Array[SearchRecord[S]]))]= _searchRDD.matchesQuery(rdd, queryBuilder, topK, minScore) override def save(path: String): Unit = _searchRDD.save(path) override def getNumPartitions: Int = _searchRDD.getNumPartitions override def options: SearchOptions[S] = _searchRDD.options /** * Builds a search rdd with that custom search options. * * @param opts Search options * @return Dependent RDD with configurable search features */ def searchRDD(opts: SearchOptions[S] = defaultOpts): SearchRDD[S] = new SearchRDDLucene[S](rdd, opts) override def searchJoinQuery[W: ClassTag](other: RDD[W], queryBuilder: W => Query, topKByPartition: Int, minScore: Double): RDD[(W, Option[SearchRecord[S]])] = _searchRDD.searchJoinQuery(other, queryBuilder, topKByPartition, minScore) override def distinct(numPartitions: Int): RDD[S] = _searchRDD.distinct(numPartitions) override def searchDropDuplicates[K: ClassTag, C: ClassTag](queryBuilder: S => Query = null, createKey: S => K = (s: S) => s.hashCode.toLong.asInstanceOf[K], minScore: Double = 0, createCombiner: Seq[SearchRecord[S]] => C = (ss: Seq[SearchRecord[S]]) => ss.head.source.asInstanceOf[C], mergeValue: (C, Seq[SearchRecord[S]]) => C = (c: C, _: Seq[SearchRecord[S]]) => c, mergeCombiners: (C, C) => C = (c: C, _: C) => c, numPartitionInJoin: Int = getNumPartitions, topKToDeduplicate: Int = defaultTopK ) (implicit ord: Ordering[K]): RDD[C] = _searchRDD.searchDropDuplicates(queryBuilder, createKey, minScore, createCombiner, mergeValue, mergeCombiners) private[spark] override def elementClassTag: ClassTag[S] = _searchRDD.elementClassTag }
phymbert/spark-search
core/src/main/scala/org/apache/spark/search/package.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark import java.util.function.{Function => JFunction} import org.apache.lucene.analysis.Analyzer import org.apache.lucene.analysis.standard.StandardAnalyzer import org.apache.lucene.queryparser.classic.QueryParser import org.apache.lucene.search.Query import org.apache.lucene.util.QueryBuilder import org.apache.spark.search.reflect.DefaultQueryBuilder import scala.collection.Iterable import scala.language.implicitConversions import scala.reflect.ClassTag /** * Spark Search brings advanced full text search features * to your Dataframe, Dataset and RDD. Powered by Apache Lucene. */ package object search { /** * Search record. */ case class SearchRecord[S](id: Long, partitionIndex: Long, score: Double, shardIndex: Long, source: S) extends Ordering[SearchRecord[S]] { override def compare(sr1: SearchRecord[S], sr2: SearchRecord[S]): Int = sr2.score.compare(sr1.score) // Reverse } /** * Default search options. */ def defaultOpts[S]: SearchOptions[S] = SearchOptions.defaultOptions.asInstanceOf[SearchOptions[S]] /** * Abstract class to ease building lucene queries using query string lucene syntax with spark search RDD. * * @param queryStringBuilder Generate lucene query string for this input element * @tparam S Type of input class */ class QueryStringBuilderWithAnalyzer[S](val queryStringBuilder: S => String, val defaultFieldName: String = ReaderOptions.DEFAULT_FIELD_NAME, override val analyzerClass: Class[_ <: Analyzer] = classOf[StandardAnalyzer]) extends CanBuildQueryWithAnalyzer[S](analyzerClass) { override def apply(s: S): Query = new QueryParser(defaultFieldName, _analyzer).parse(queryStringBuilder.apply(s)) } /** * Abstract class to ease building lucene query with spark search RDD, support serialization * and query builder creation in a distributed world. * * @param queryBuilder Generate lucene query for this input element * @tparam S Type of input class */ class QueryBuilderWithAnalyzer[S](queryBuilder: (S, QueryBuilder) => Query, override val analyzerClass: Class[_ <: Analyzer] = classOf[StandardAnalyzer] ) extends CanBuildQueryWithAnalyzer[S](analyzerClass) { @transient private lazy val _luceneQueryBuilder: QueryBuilder = new QueryBuilder(_analyzer) override def apply(s: S): Query = queryBuilder.apply(s, _luceneQueryBuilder) } /** * Abstract class to ease building lucene query with spark search RDD, support serialization * and analyzer creation in a distributed world. * * @param analyzerClass Type of the analyzer to use with the query * @tparam S Type of input class */ abstract class CanBuildQueryWithAnalyzer[S](val analyzerClass: Class[_ <: Analyzer] = classOf[StandardAnalyzer]) extends (S => Query) with Serializable { @transient lazy val _analyzer: Analyzer = analyzerClass.newInstance() } def defaultQueryBuilder[S](opts: SearchOptions[_] = defaultOpts)(implicit cls: ClassTag[S]): S => Query = new QueryBuilderWithAnalyzer[S](new DefaultQueryBuilder[S](cls.runtimeClass.asInstanceOf[Class[_ <: S]]).asInstanceOf[(S, QueryBuilder) => Query], opts.getReaderOptions.analyzer) def queryBuilder[S](builder: (S, QueryBuilder) => Query, opts: SearchOptions[_] = defaultOpts): S => Query = new QueryBuilderWithAnalyzer[S](builder, opts.getReaderOptions.analyzer) def queryStringBuilder[S](builder: S => String, opts: SearchOptions[_] = defaultOpts): S => Query = new QueryStringBuilderWithAnalyzer[S](builder, opts.getReaderOptions.getDefaultFieldName, opts.getReaderOptions.analyzer) implicit def indexOptions[S](optionsBuilderFunc: Function[IndexationOptions.Builder[S], IndexationOptions.Builder[S]]): JFunction[IndexationOptions.Builder[S], IndexationOptions.Builder[S]] = new JFunction[IndexationOptions.Builder[S], IndexationOptions.Builder[S]] { override def apply(opts: IndexationOptions.Builder[S]): IndexationOptions.Builder[S] = { optionsBuilderFunc.apply(opts) } } implicit def readerOptions[S](optionsBuilderFunc: Function[ReaderOptions.Builder[S], ReaderOptions.Builder[S]]): JFunction[ReaderOptions.Builder[S], ReaderOptions.Builder[S]] = new JFunction[ReaderOptions.Builder[S], ReaderOptions.Builder[S]] { override def apply(opts: ReaderOptions.Builder[S]): ReaderOptions.Builder[S] = { optionsBuilderFunc.apply(opts) } } private[search] def searchRecordJavaToProduct[S](sr: SearchRecordJava[S]) = { SearchRecord(sr.id, sr.partitionIndex, sr.score, sr.shardIndex, sr.source) } }
phymbert/spark-search
sql/src/test/scala/org/apache/spark/search/sql/SearchDatasetSuite.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.sql import org.apache.spark.search.sql.TestData._ import org.scalatest.flatspec.AnyFlatSpec class SearchDatasetSuite extends AnyFlatSpec with LocalSparkSession { ignore should "be searchable" in { val spark = _spark import spark.sqlContext.implicits._ val appleCompany = companies1DS(spark) .where($"name".matches("apple") && score() > 1d) assertResult(Company("Apple, Inc")) { appleCompany.collect() } } }
phymbert/spark-search
sql/src/main/scala/org/apache/spark/search/sql/SearchRule.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.sql import org.apache.spark.sql.catalyst.expressions.{And, BinaryExpression, Expression, Or} import org.apache.spark.sql.catalyst.plans.logical._ import org.apache.spark.sql.catalyst.rules.Rule /** * Spark Search SQL rule. * * Rewrite logical plan involving search expression to their equivalent in search RDD. * * @author <NAME> */ object SearchRule extends Rule[LogicalPlan] { override def apply(plan: LogicalPlan): LogicalPlan = plan transformUp { case filter@Filter(conditions, _) if hasSearchCondition(conditions) => extractSearchFilter(filter) } private def hasSearchCondition(e: Expression): Boolean = { e match { case _: MatchesExpression => true case _ => e.children.exists(hasSearchCondition) } } def extractSearchFilter(filter: Filter): LogicalPlan = { val (searchExpression, other) = searchConditions(filter.condition) val rddPlan = SearchIndexPlan(filter.child, searchExpression.get) val joinPlan = SearchJoin(filter.child, rddPlan, searchExpression.get) /*if (other.children.nonEmpty) Filter(other, joinPlan) else*/ joinPlan // FIXME lost } private def searchConditions(e: Expression): (Option[Expression], Expression) = { e match { case m: MatchesExpression => (Option(e), m) case be: BinaryExpression => val (leftSearchExpression, _) = searchConditions(be.left) val (rightSearchExpression, _) = searchConditions(be.right) val searchExpression = if (leftSearchExpression.nonEmpty && rightSearchExpression.nonEmpty) Option(be match { case _: And => And(leftSearchExpression.get, rightSearchExpression.get) case _: Or => Or(leftSearchExpression.get, rightSearchExpression.get) case _ => throw new UnsupportedOperationException }) else if (leftSearchExpression.nonEmpty) leftSearchExpression else if (rightSearchExpression.nonEmpty) rightSearchExpression else Option.empty (searchExpression, be) case _ => (Option.empty, e) } } }
phymbert/spark-search
core/src/test/scala/org/apache/spark/search/rdd/SearchRDDSuite.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import java.io.File import org.apache.commons.io.FileUtils import org.apache.lucene.analysis.en.EnglishAnalyzer import org.apache.lucene.search.BooleanClause.Occur import org.apache.lucene.util.QueryBuilder import org.apache.spark.RangePartitioner import org.apache.spark.rdd.RDD import org.apache.spark.search._ import org.apache.spark.search.rdd.TestData._ import org.scalatest.funsuite.AnyFunSuite import org.scalatest.matchers.must.Matchers.{convertToAnyMustWrapper, sorted} import scala.language.implicitConversions class SearchRDDSuite extends AnyFunSuite with LocalSparkContext { test("count all indexed documents") { assertResult(3)(sc.parallelize(persons).count) } test("count matched indexed documents") { assertResult(1)(sc.parallelize(persons) .count("firstName:bob")) } test("search list hits matching query") { assertResult(Array(new SearchRecord[Person](0, 1, 0.3150668740272522f, 0, Person("Bob", "Marley", 37))))(sc.parallelize(persons) .searchList("firstName:bob", 10)) } test("search RDD hits matching query") { assertResult(Array(new SearchRecord[Person](0, 1, 0.3150668740272522f, 0, Person("Bob", "Marley", 37))))( sc.parallelize(persons) .search("firstName:bob", 10).take(10) ) } test("search RDD hits matching fuzzy query should return topK per partition") { val personsKeyedBy = sc.parallelize(personsDuplicated) .union(sc.parallelize(personsDuplicated)) .zipWithIndex() .keyBy(_._2) val personsRDD = personsKeyedBy.partitionBy(new RangePartitioner(2, personsKeyedBy)) .map(_._2._1) personsRDD.getNumPartitions mustBe 2 val matches = personsRDD // <NAME> will have a better score than <NAME> and she should not be hit .search("firstName:agnès~ OR firstName:bob~ lastName:marley~", topKByPartition = 2) // Assert we have only one lucene index per partition matches.map(_.partitionIndex).distinct().sortBy(identity).collect() mustBe Array(0, 1) // Assert we only have bob matches.map(_.source.firstName).map(_.toLowerCase.substring(0, 3)).collect().distinct mustBe Array("bob") // Assert we have all bobs matched matches.count() mustBe 4 // Should be sorted by score score descending matches.map(_.score).collect().reverse mustBe sorted } test("search RDD query must use default field") { assertResult(Array(new SearchRecord[Person](0, 1, 0.3150668740272522f, 0, Person("Bob", "Marley", 37))))(sc.parallelize(persons).searchRDD( SearchOptions.builder[Person]() .read((r: ReaderOptions.Builder[Person]) => r.defaultFieldName("firstName")) .analyzer(classOf[EnglishAnalyzer]) .build()).search("bob", 10).take(10)) } test("Matching RDD") { val persons2 = Seq( Person("George", "Michal", 0), Person("Georgee", "Michall", 0), Person("Bobb", "Marley", 0), Person("Bob", "Marlley", 0), Person("Agnes", "Bartol", 0), Person("Agnec", "Barttol", 0)) val searchRDD = sc.parallelize(persons2).repartition(2).searchRDD() val matchingRDD = sc.parallelize(persons) val personsKeyed: RDD[(Long, Person)] = matchingRDD.zipWithIndex().map(_.swap) val matches = searchRDD.matches(personsKeyed, (p: Person) => s"firstName:${p.firstName}~0.5 AND lastName:${p.lastName}~0.5", 2).collect assertResult(3)(matches.length) assertResult(3)(matches.map(m => m._2._2.length).count(_ == 2)) } test("self matches with one query value should returns all documents with one match") { val rdd = sc.parallelize(persons) val result = rdd .matches(rdd.zipWithIndex().map(_.swap), (_: Person) => "lastName:Marley", 1) val resultFiltered = result .filter(_._2._2.count(_.source.lastName.equals("Marley")) == 1) assertResult(3)(resultFiltered.count) } test("self matches with self query value should returns all document with self match") { val rdd = sc.parallelize(persons) val searchRDD = rdd.matchesQuery(rdd.zipWithIndex().map(_.swap), queryBuilder((c: Person, lqb: QueryBuilder) => lqb.createBooleanQuery("firstName", c.firstName, Occur.MUST)), topK = 1) .map(_._2) .filter(m => m._2.count(h => h.source.firstName.equals(h.source.firstName)) == 1) .cache assertResult(3)(searchRDD.count) } test("search join should work") { val opts = SearchOptions.builder[Person]().analyzer(classOf[TestPersonAnalyzer]).build() val searchRDD = sc.parallelize(personsDuplicated).repartition(1) .searchRDD(opts) .searchJoin(sc.parallelize(personsDuplicated), (c: Person) => s"firstName:${c.firstName}", topKByPartition = 2) assertResult(12)(searchRDD.count) // FIXME add good unit test } test("search matches should work") { val opts = SearchOptions.builder[Person]().analyzer(classOf[TestPersonAnalyzer]).build() val searchRDD = sc.parallelize(personsDuplicated).repartition(1) .searchRDD(opts) .matchesQuery(sc.parallelize(persons).zipWithIndex().map(_.swap), queryBuilder((c: Person, lqb: QueryBuilder) => lqb.createBooleanQuery("firstName", c.firstName), opts), topK = 2) .filter(_._2._2.length == 2) assertResult(3)(searchRDD.count) } test("Distinct with no minimum score") { val searchRDD = sc.parallelize(personsDuplicated) .repartition(1) .searchRDD() val deduplicated = searchRDD.distinct(1).collect assertResult(1)(deduplicated.length) } test("Drop duplicates with min score") { val searchRDD = sc.parallelize(personsDuplicated).repartition(1) .searchRDD(opts = SearchOptions.builder().analyzer(classOf[TestPersonAnalyzer]).build()) val deduplicated = searchRDD.searchDropDuplicates[Long, Person](minScore = 8).collect assertResult(3)(deduplicated.length) } test("Drop duplicate with query builder and min score") { val searchRDD = sc.parallelize(personsDuplicated).repartition(1) val deduplicated = searchRDD.searchDropDuplicates[Long, Person]( queryBuilder = queryStringBuilder(p => s"firstName:${p.firstName}~0.5 AND lastName:${p.lastName}~0.5"), minScore = 1 ).collect assertResult(3)(deduplicated.length) } test("SearchRDD should be iterable as an RDD") { val searchRDD = sc.parallelize(persons).searchRDD() assertResult(Array("Agnès", "Bob", "Geoorge"))(searchRDD .filter(_.firstName.nonEmpty) .map(_.firstName) .collect().sorted) } test("Save and restore index from/to hdfs") { FileUtils.deleteDirectory(new File("target/test-save")) val searchRDD = sc.parallelize(persons).searchRDD() assertResult(3)(searchRDD.count()) searchRDD.save("target/test-save") val restoredSearchRDD = SearchRDD.load[Person](sc, "target/test-save") assertResult(3)(restoredSearchRDD.count()) assertResult(Array("Bartoll", "Marley", "Michael"))(restoredSearchRDD .map(_.lastName) .collect().sorted) } }
phymbert/spark-search
core/src/main/scala/org/apache/spark/search/rdd/SearchRDDIndexer.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import java.io._ import java.util import org.apache.commons.io.FileUtils import org.apache.hadoop.conf.Configuration import org.apache.hadoop.fs.{FileSystem, Path} import org.apache.spark._ import org.apache.spark.rdd.RDD import org.apache.spark.search._ import org.apache.spark.search.rdd.ZipUtils._ import scala.collection.JavaConverters._ import scala.reflect.ClassTag /** * RDD responsible to index element to lucene index on local disk where the partition resides, * it then sends the partition as zipped byte array. * * @author <NAME> */ private[search] class SearchRDDIndexer[S: ClassTag](sc: SparkContext, val options: SearchOptions[S], val deps: Seq[Dependency[_]]) extends RDD[Array[Byte]](sc, deps) { def this(rdd: RDD[S], options: SearchOptions[S]) { this(rdd.context, options, Seq(new OneToOneDependency(rdd))) } override def compute(split: Partition, context: TaskContext): Iterator[Array[Byte]] = { val searchRDDPartition = split.asInstanceOf[SearchPartitionIndex[S]] val elements = firstParent.iterator(searchRDDPartition.parent, context).asJava .asInstanceOf[util.Iterator[S]] searchRDDPartition.index(elements, options.getIndexationOptions) streamPartitionIndexZip(context, searchRDDPartition) } protected def streamPartitionIndexZip(context: TaskContext, searchRDDPartition: SearchPartitionIndex[S]): Iterator[Array[Byte]] = { val localIndexDirPath = new File(searchRDDPartition.indexDir) val targetPath = new File(localIndexDirPath.getParent, s"${localIndexDirPath.getName}.zip") zipPartition(localIndexDirPath.toPath, new FileOutputStream(targetPath)) new InterruptibleIterator[Array[Byte]](context, new FileInputStreamIterator(targetPath)) } override protected def getPartitions: Array[Partition] = { val parentPartitions = firstParent.partitions // One-2-One partition parentPartitions.map(p => new SearchPartitionIndex[S](p.index, rootDir, getPreferredLocation(context, p.index, parentPartitions.length, super.getPreferredLocations(p)), p)).toArray } protected val rootDir: String = s"${options.getIndexationOptions.getRootIndexDirectory}${File.separator}${sc.applicationId}-sparksearch-rdd$id" override protected[rdd] def getPreferredLocations(split: Partition): Seq[String] = split.asInstanceOf[SearchPartitionIndex[S]].preferredLocations lazy val _indexDirectoryByPartition: Map[Int, String] = partitions.map(_.asInstanceOf[SearchPartitionIndex[S]]).map(t => (t.index, t.indexDir)).toMap def save(pathString: String): Unit = { val indexDirectoryByPartition = _indexDirectoryByPartition mapPartitionsWithIndex((index, _) => { val hadoopConf = new Configuration() val hdfs = FileSystem.get(hadoopConf) val localIndexDirPath = new File(indexDirectoryByPartition(index)).toPath val targetPath = new Path(pathString, s"${localIndexDirPath.getFileName}.zip") logInfo(s"Saving partition $localIndexDirPath to $targetPath") val fos = hdfs.create(targetPath) zipPartition(localIndexDirPath, fos) logInfo(s"Partition $localIndexDirPath saved to $targetPath") Iterator() }).collect } override def unpersist(blocking: Boolean): SearchRDDIndexer.this.type = { // TODO support non blocking val indexDirectoryByPartition = _indexDirectoryByPartition sparkContext.runJob(this, (context: TaskContext, _: Iterator[Array[Byte]]) => { val indexDir = new File(indexDirectoryByPartition(context.partitionId())) FileUtils.deleteDirectory(indexDir) FileUtils.deleteQuietly(new File(indexDir.getParent, s"${indexDir.getName}.zip")) }) super.unpersist(blocking) } } private[rdd] object SearchRDDIndexer { }
phymbert/spark-search
core/src/test/scala/org/apache/spark/search/rdd/TestData.scala
<filename>core/src/test/scala/org/apache/spark/search/rdd/TestData.scala /* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import org.apache.lucene.analysis.Analyzer import org.apache.lucene.analysis.Analyzer.TokenStreamComponents import org.apache.lucene.analysis.ngram.NGramTokenizer object TestData { case class Person(firstName: String, lastName: String, age: Int) def persons = Seq( Person("Geoorge", "Michael", 53), Person("Bob", "Marley", 37), Person("Agnès", "Bartoll", -1)) def personsDuplicated = Seq( Person("George", "Michal", 0), Person("Georgee", "Michall", 0), Person("Bobb", "Marley", 0), Person("Bob", "Marlley", 0), Person("Agnes", "Bartol", 0), Person("Agnec", "Barttol", 0)) class TestPersonAnalyzer extends Analyzer { override def createComponents(fieldName: String): TokenStreamComponents = { new TokenStreamComponents(new NGramTokenizer(1, 3)) } } }
phymbert/spark-search
examples/src/main/scala/all/examples/org/apache/spark/search/rdd/ExampleData.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package all.examples.org.apache.spark.search.rdd import java.io.File import java.net.URL import org.apache.hadoop.conf.Configuration import org.apache.hadoop.fs.{FileSystem, Path} import org.apache.spark.rdd.RDD import org.apache.spark.sql.SparkSession import scala.language.postfixOps import scala.sys.process._ object ExampleData { case class Review(asin: String, helpful: Array[Long], overall: Double, reviewText: String, reviewTime: String, reviewerID: String, reviewerName: String, summary: String, unixReviewTime: Long) def loadReviews(spark: SparkSession, reviewURL: String): RDD[Review] = { import spark.implicits._ val hadoopConf = new Configuration() val hdfs = FileSystem.get(hadoopConf) val reviewsFile = File.createTempFile("reviews_", ".json.gz") reviewsFile.deleteOnExit() new URL(reviewURL) #> reviewsFile !! val dstPathName = "/hdfs-tmp/reviews.json.gz" hdfs.copyFromLocalFile(new Path(reviewsFile.getAbsolutePath), new Path(dstPathName)) hdfs.deleteOnExit(new Path(dstPathName)) spark.read.json(dstPathName).as[Review].rdd.repartition(4) } }
phymbert/spark-search
core/src/test/scala/org/apache/spark/search/rdd/ZipUtilsSuite.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import java.io.{ByteArrayOutputStream, File, FileInputStream, FileOutputStream} import org.apache.commons.io.{FileUtils, IOUtils} import org.apache.lucene.analysis.standard.StandardAnalyzer import org.apache.lucene.document.Field.Store import org.apache.lucene.document.{Document, StringField} import org.apache.lucene.index.{IndexWriter, IndexWriterConfig} import org.apache.lucene.store.MMapDirectory import org.apache.spark.search.rdd.ZipUtils._ import org.scalatest.flatspec.AnyFlatSpec class ZipUtilsSuite extends AnyFlatSpec { it should "create an iterator from a file input stream" in { val f = File.createTempFile("test", "test") f.deleteOnExit() val expected = "abcdef\n" * 100000 FileUtils.write(f, expected) val it = new FileInputStreamIterator(f) var actual = "" it.foreach { a => actual += new String(a) } assertResult(expected) { actual } } it should "create an input stream from iterator" in { val bos = new ByteArrayOutputStream() val it = Seq(Seq('a', 'b'), Seq('c', 'd', 'e'), Seq('f')) .map(s => s.map(c => c.toByte).toArray) .iterator IOUtils.copy(new IteratorInputStream(it), bos) assertResult("abcdef") { new String(bos.toByteArray) } } it should "creates valid zip and able to unzip it" in { val testDir = new File(System.getProperty("java.io.tmpdir"), "test-zipping") testDir.delete() testDir.mkdir() val f = new File(testDir, "test") f.deleteOnExit() val expected = "abcdef\n" * 100000 FileUtils.write(f, expected) val zipped = File.createTempFile("zipped", ".zip") f.deleteOnExit() zipPartition(testDir.toPath, new FileOutputStream(zipped)) f.delete() testDir.delete() unzipPartition(testDir.getAbsolutePath, new FileInputStreamIterator(zipped)) zipped.delete() val bos = new ByteArrayOutputStream() IOUtils.copy(new FileInputStream(f), bos) assertResult(expected) { new String(bos.toByteArray) } } }
phymbert/spark-search
benchmark/src/main/scala/benchmark/SparkRDDRegexBenchmark.scala
<reponame>phymbert/spark-search<gh_stars>10-100 /* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package benchmark import org.apache.spark.rdd.RDD object SparkRDDRegexBenchmark extends BaseBenchmark("Spark RDD Regex") { def main(args: Array[String]): Unit = run() override def countNameMatches(companies: RDD[Company], name: String): RDD[(Double, String)] = { val re = s".*\\Q${name.toLowerCase}\\E.*" companies .filter(_.name != null) .filter(_.name.toLowerCase.matches(re)) .map(c => (0, c.name)) } override def joinMatch(companies: RDD[Company], secEdgarCompanies: RDD[SecEdgarCompanyInfo]): RDD[(String, Double, String)] = { secEdgarCompanies .filter(_.companyName != null) .zipWithIndex() .map(_.swap) .cartesian(companies.filter(_.name != null)) .map(c => (c._1._1, (c._1._2.companyName, c._2.name))) .filter(t => t._2._2.toLowerCase.matches(s".*\\Q${t._2._1.replaceAllLiterally(" ", "\\E\\s+\\Q")}\\E.*")) .reduceByKey((c1, _) => c1) .map(t => (t._2._1, 0d, t._2._2)) } }
phymbert/spark-search
core/src/main/scala/org/apache/spark/search/rdd/SearchRDDReloaded.scala
<reponame>phymbert/spark-search<filename>core/src/main/scala/org/apache/spark/search/rdd/SearchRDDReloaded.scala /* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import org.apache.hadoop.conf.Configuration import org.apache.hadoop.fs.{FileSystem, Path, PathFilter} import org.apache.spark.search.SearchOptions import org.apache.spark.{Partition, SparkContext, TaskContext} import scala.reflect.ClassTag /** * Reloaded index from hdfs. * * @author <NAME> */ private[search] class SearchIndexReloadedRDD[S: ClassTag](sc: SparkContext, path: String, override val options: SearchOptions[S]) extends SearchRDDIndexer[S](sc, options, Nil) { override protected def getPartitions: Array[Partition] = { val hadoopConf = new Configuration() val hdfs = FileSystem.get(hadoopConf) val partitionsZipped = hdfs.listStatus(new Path(path), new PathFilter { override def accept(path: Path): Boolean = path.getName.endsWith(".zip") }).zipWithIndex partitionsZipped.map(p => new SearchIndexReloadedPartition(p._2, rootDir, p._1.getPath.toUri.toString, getPreferredLocation(context, p._2, partitionsZipped.length, Seq()))) } override def compute(split: Partition, context: TaskContext): Iterator[Array[Byte]] = { val part = split.asInstanceOf[SearchIndexReloadedPartition] val hadoopConf = new Configuration() val hdfs = FileSystem.get(hadoopConf) ZipUtils.unzipPartition(part.indexDir, hdfs.open(new Path(part.zipPath))) streamPartitionIndexZip(context, part.asInstanceOf[SearchPartitionIndex[S]]) } } class SearchIndexReloadedPartition(val idx: Int, val rootDir: String, val zipPath: String, val preferredLocations2: Array[String]) extends SearchPartitionIndex(idx, rootDir, preferredLocations2, null) { }
phymbert/spark-search
sql/src/main/scala/org/apache/spark/search/sql/SearchStrategy.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.sql import org.apache.spark.sql.catalyst.expressions.Expression import org.apache.spark.sql.catalyst.plans.logical.LogicalPlan import org.apache.spark.sql.execution.{SparkPlan, SparkStrategy} /** * Search strategy. * * @author <NAME> */ object SearchStrategy extends SparkStrategy { override def apply(plan: LogicalPlan): Seq[SparkPlan] = { plan match { case SearchJoin(left, right, searchExpression) => SearchJoinExec(planLater(left), planLater(right), searchExpression):: Nil case p: SearchIndexPlan => SearchRDDExec(planLater(p.child), p.searchExpression) :: Nil case _ => Seq.empty } } }
phymbert/spark-search
sql/src/main/scala/org/apache/spark/search/sql/physical.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.sql import org.apache.lucene.util.QueryBuilder import org.apache.spark.rdd.RDD import org.apache.spark.search.rdd.SearchRDDLucene import org.apache.spark.search.{IndexationOptions, ReaderOptions, SearchOptions, _} import org.apache.spark.sql.catalyst.InternalRow import org.apache.spark.sql.catalyst.expressions.{Attribute, AttributeReference, Expression, Literal, UnsafeRow} import org.apache.spark.sql.execution._ import org.apache.spark.sql.types.{DataTypes, StringType, StructField, StructType} import org.apache.spark.unsafe.types.UTF8String case class SearchJoinExec(left: SparkPlan, right: SparkPlan, searchExpression: Expression) extends BinaryExecNode { override protected def doExecute(): RDD[InternalRow] = { val leftRDD = left.execute() val searchRDD = right.execute().asInstanceOf[SearchRDDLucene[InternalRow]] val opts = searchRDD.options val qb = searchExpression match { // FIXME support AND / OR case MatchesExpression(left, right) => right match { case Literal(value, dataType) => dataType match { case StringType => left match { case a: AttributeReference => queryBuilder[InternalRow]((_: InternalRow, lqb: QueryBuilder) => lqb.createBooleanQuery(a.name, value.asInstanceOf[UTF8String].toString), opts) case _ => throw new UnsupportedOperationException } case _ => throw new UnsupportedOperationException } case _ => throw new UnsupportedOperationException } case _ => throw new IllegalArgumentException } searchRDD.matchesQuery(leftRDD.zipWithIndex().map(_.swap), qb, 1) .filter(_._2._2.nonEmpty) // TODO move this filter at partition level .values.map(m => toRow(m._1.asInstanceOf[UnsafeRow], m._2.head.score)) } private def toRow(doc: UnsafeRow, score: Double): InternalRow = { val row = new UnsafeRow(doc.numFields + 1) val bs = new Array[Byte](row.getSizeInBytes) row.pointTo(bs, bs.length) row.copyFrom(doc) row.setDouble(doc.numFields, score) row } override def output: Seq[Attribute] = left.output ++ Seq(scoreAttribute) } case class SearchRDDExec(child: SparkPlan, searchExpression: Expression) extends UnaryExecNode { override protected def doExecute(): RDD[InternalRow] = { val inputRDDs = child match { case wsce: WholeStageCodegenExec => wsce.child.asInstanceOf[CodegenSupport].inputRDDs() case cs: CodegenSupport => cs.inputRDDs() case _ => throw new UnsupportedOperationException("no input rdd supported") } if (inputRDDs.length != 1) { throw new UnsupportedOperationException("one input RDD expected") } val rdd = inputRDDs.head val schema = StructType(Seq(searchExpression match { // FIXME support AND / OR case MatchesExpression(left, _) => left match { case a: AttributeReference => StructField(a.name, DataTypes.StringType) case _ => throw new UnsupportedOperationException } case _ => throw new IllegalArgumentException })) val opts = SearchOptions.builder[InternalRow]() .read((readOptsBuilder: ReaderOptions.Builder[InternalRow]) => readOptsBuilder.documentConverter(new DocumentRowConverter(schema))) .index((indexOptsBuilder: IndexationOptions.Builder[InternalRow]) => indexOptsBuilder.documentUpdater(new DocumentRowUpdater(schema))) .build() new SearchRDDLucene[InternalRow](rdd, opts) } override def output: Seq[Attribute] = Seq() }
phymbert/spark-search
sql/src/main/scala/org/apache/spark/search/sql/logical.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.sql import org.apache.spark.sql.catalyst.expressions.{Attribute, AttributeReference, Expression} import org.apache.spark.sql.catalyst.plans.logical.{BinaryNode, LeafNode, LogicalPlan, Statistics} import org.apache.spark.sql.types.DoubleType trait SearchLogicalPlan case class SearchJoin(left: LogicalPlan, right: SearchIndexPlan, searchExpression: Expression) extends BinaryNode with SearchLogicalPlan { override def output: Seq[Attribute] = left.output ++ right.output } case class SearchIndexPlan(child: LogicalPlan, searchExpression: Expression) extends LeafNode with SearchLogicalPlan { override def output: Seq[Attribute] = Seq(AttributeReference(SCORE, DoubleType, nullable = false)()) override def computeStats(): Statistics = Statistics( sizeInBytes = BigInt(Long.MaxValue) // Broadcast forbidden ) }
phymbert/spark-search
core/src/main/scala/org/apache/spark/search/rdd/package.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search import org.apache.lucene.queryparser.classic.QueryParser import org.apache.lucene.search.Query import org.apache.spark.rdd.RDD import org.apache.spark.{ExecutorAllocationClient, SparkContext} import scala.language.implicitConversions import scala.reflect.ClassTag import scala.util.{Failure, Success, Try} /** * Spark Search RDD. */ package object rdd { implicit def rddWithSearch[S: ClassTag](rdd: RDD[S]): RDDWithSearch[S] = new RDDWithSearch[S](rdd) implicit def searchAsRDD[S: ClassTag](searchRDD: SearchRDD[S]): RDD[S] = searchRDD.asInstanceOf[RDD[S]] /** * Provide a static query to pass to SearchRDD serializable. * * /!\ Important as Lucene Query is not serializable. */ type StaticQueryProvider = () => Query def parseQueryString[S](queryString: String, opts: SearchOptions[_] = defaultOpts): StaticQueryProvider = // Query parser is not thread safe () => new QueryParser(opts.getReaderOptions.getDefaultFieldName, opts.getReaderOptions.analyzer.newInstance()) .parse(queryString) private[rdd] def searchRecordJavaToProduct[S](sr: SearchRecordJava[S]) = { SearchRecord(sr.id, sr.partitionIndex, sr.score, sr.shardIndex, sr.source) } private[rdd] def tryAndClose[A <: AutoCloseable, B](resource: A)(block: A => B): B = { Try(block(resource)) match { case Success(result) => resource.close() result case Failure(e) => resource.close() throw e } } private[rdd] def getPreferredLocation(sc: SparkContext, index: Int, numPartition: Int, parentPreferredLocation: Seq[String]): Array[String] = sc.schedulerBackend match { case b: ExecutorAllocationClient => // Try to balance partitions across executors val allIds = sc.getExecutorIds() val ids = allIds.grouped(numPartition).toList ids(index % ids.length).toArray case _ => parentPreferredLocation.toArray } }
phymbert/spark-search
sql/src/main/scala/org/apache/spark/search/sql/ColumnWithSearch.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.sql import org.apache.spark.sql.catalyst.expressions.Expression import org.apache.spark.sql.functions.lit import org.apache.spark.sql.{Column, SparkSession} /** * * @author <NAME> */ class ColumnWithSearch(col: Column) { @transient private final val sqlContext = SparkSession.getActiveSession.map(_.sqlContext).orNull def matches(literal: String): Column = withSearchExpr { MatchesExpression(col.expr, lit(literal).expr) } def matches(other: Column): Column = withSearchExpr { MatchesExpression(col.expr, other.expr) } /** Creates a column based on the given expression. */ private def withSearchExpr(newExpr: Expression): Column = { val searchSQLEnabled = sqlContext.experimental.extraStrategies.contains(SearchStrategy) if (!searchSQLEnabled) { sqlContext.experimental.extraStrategies = Seq(SearchStrategy) ++ sqlContext.experimental.extraStrategies sqlContext.experimental.extraOptimizations = Seq(SearchRule) ++ sqlContext.experimental.extraOptimizations } new Column(newExpr) } }
phymbert/spark-search
sql/src/main/scala/org/apache/spark/search/sql/package.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search import org.apache.lucene.search.Query import org.apache.spark.search import org.apache.spark.sql._ import org.apache.spark.sql.catalyst.expressions.{Attribute, AttributeReference} import org.apache.spark.sql.types.DoubleType import scala.language.implicitConversions import scala.reflect.ClassTag import scala.reflect.runtime.universe.TypeTag /** * Search SQL package provides search features to spark [[org.apache.spark.sql.Dataset]]. * * @author <NAME> */ package object sql { /** * Score column name. */ val SCORE: String = "__score__" private[sql] val scoreAttribute: Attribute = AttributeReference(SCORE, DoubleType, nullable = false)() /** * Score of the hit in the search request. */ def score(): Column = new Column(ScoreExpression()) /** * Default query builder. */ def defaultQueryBuilder[S: ClassTag](implicit enc: Encoder[S]): S => Query = search.defaultQueryBuilder[S]() /** * Add search feature to column. */ implicit def columnWithSearch(col: Column): ColumnWithSearch = new ColumnWithSearch(col) /** * Allow search record rdd transformation to Row. */ implicit def searchRecordEncoder[S <: Product : TypeTag](implicit enc: Encoder[S]): Encoder[SearchRecord[S]] = Encoders.product[SearchRecord[S]] }
phymbert/spark-search
benchmark/src/main/scala/benchmark/BaseBenchmark.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package benchmark import org.apache.spark.rdd.RDD import org.apache.spark.sql.{SaveMode, SparkSession} abstract class BaseBenchmark(appName: String) extends Serializable { val spark: SparkSession = SparkSession.builder().appName(appName).getOrCreate() protected def run(): Unit = { import spark.implicits._ // Convert CSV to parquet // https://www.kaggle.com/peopledatalabssf/free-7-million-company-dataset spark.read .option("header", "true") .option("inferSchema", "true") .csv("hdfs:///companies_sorted.csv") .withColumnRenamed("year founded", "yearFounded") .withColumnRenamed("size range", "sizeRange") .withColumnRenamed("linkedin url", "linkedinUrl") .withColumnRenamed("current employee estimate", "currentEmployeeEstimate") .withColumnRenamed("total employee estimate", "totalEmployeeEstimate") .withColumnRenamed("_c0", "id") .na.fill("", Seq("domain", "yearFounded", "industry", "sizeRange", "locality", "country", "linkedinUrl", "currentEmployeeEstimate", "totalEmployeeEstimate")) .write .mode(SaveMode.Ignore) .parquet("hdfs:///companies_sorted.parquet") // https://www.kaggle.com/dattapiy/sec-edgar-companies-list spark.read.option("header", "true") .option("inferSchema", "true") .csv("hdfs:///sec__edgar_company_info.csv") .withColumnRenamed("Line Number", "lineNumber") .withColumnRenamed("Company Name", "companyName") .withColumnRenamed("Company CIK Key", "companyCIKKey") .write .mode(SaveMode.Ignore) .parquet("hdfs:///sec__edgar_company_info.parquet") def loadCompanies = { spark.read .load("hdfs:///companies_sorted.parquet") .as[Company] .rdd } def loadSecEdgarCompanies = { spark.read.load("hdfs:///sec__edgar_company_info.parquet") .as[SecEdgarCompanyInfo] .rdd } // Join matches val (countJoinedMatches, joinedMatches) = spark.time({ val jm = joinMatch(loadCompanies, loadSecEdgarCompanies).cache (jm.count, jm) }) println(s"for joined ${countJoinedMatches} matches") joinedMatches.take(10).foreach(println(_)) joinedMatches.unpersist() val (countMatches, matches) = spark.time({ val m = countNameMatches(loadCompanies, "IBM").cache (m.count, m) }) println(s"for count ${countMatches} matches") matches.take(10).foreach(println(_)) matches.unpersist() spark.stop() } def countNameMatches(companies: RDD[Company], name: String): RDD[(Double, String)] def joinMatch(companies: RDD[Company], secEdgarCompany: RDD[SecEdgarCompanyInfo]): RDD[(String, Double, String)] }
phymbert/spark-search
benchmark/src/main/scala/benchmark/SearchRDDBenchmark.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package benchmark import org.apache.lucene.analysis.standard.StandardAnalyzer import org.apache.lucene.util.QueryBuilder import org.apache.spark.rdd.RDD import org.apache.spark.search.rdd._ import org.apache.spark.search.{SearchOptions, _} object SearchRDDBenchmark extends BaseBenchmark("SearchRDD") { def main(args: Array[String]): Unit = run() override def countNameMatches(companies: RDD[Company], name: String): RDD[(Double, String)] = { val opts = SearchOptions .builder[Company] .analyzer(classOf[StandardAnalyzer]).build companies.searchRDD(opts).search(s"name:${name}").map(sr => (sr.score, sr.source.name)) } override def joinMatch(companies: RDD[Company], secEdgarCompanies: RDD[SecEdgarCompanyInfo]): RDD[(String, Double, String)] = { val opts = SearchOptions .builder[Company] .analyzer(classOf[StandardAnalyzer]).build companies.searchRDD(opts) .searchJoinQuery(secEdgarCompanies, queryBuilder[SecEdgarCompanyInfo]((c: SecEdgarCompanyInfo, lqb: QueryBuilder) => lqb.createPhraseQuery("name", c.companyName.slice(0, 64)), opts), 1, 0d) .filter(_._2.isDefined) .map(m => (m._1.companyName, m._2.get.score, m._2.get.source.name)) } }
phymbert/spark-search
benchmark/src/main/scala/benchmark/ElasticsearchBenchmark.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package benchmark import org.apache.http.client.methods.HttpDelete import org.apache.http.impl.client.DefaultHttpClient import org.apache.spark.rdd.RDD import org.apache.spark.sql.functions._ import org.elasticsearch.spark._ object ElasticsearchBenchmark extends BaseBenchmark("Elasticsearch") { val esSQLSource = "org.elasticsearch.spark.sql" val esOpts = Map("es.nodes.wan.only" -> "true") def main(args: Array[String]): Unit = run() override def countNameMatches(companies: RDD[Company], name: String): RDD[(Double, String)] = { import spark.implicits._ clearES() companies.saveToEs("companies", esOpts) // FIXME maybe adjust replica/shard preferences spark.read.format(esSQLSource) .option("es.nodes.wan.only", "true") .load("companies") .filter($"name".equalTo(name)) .as[Company] .map(c => (0d, c.name)) // FIXME no score ? .rdd } override def joinMatch(companies: RDD[Company], secEdgarCompanies: RDD[SecEdgarCompanyInfo]): RDD[(String, Double, String)] = { clearES() import spark.implicits._ companies.saveToEs("companies", esOpts) secEdgarCompanies.saveToEs("sec_edgar_companies", esOpts) val companiesES = spark.read .format(esSQLSource) .option("es.nodes.wan.only", "true") .load("companies") .as[Company] val secEdgarCompaniesES = spark.read.format(esSQLSource) .option("es.nodes.wan.only", "true") .load("sec_edgar_companies") .as[SecEdgarCompanyInfo] companiesES.join(secEdgarCompaniesES, $"name".equalTo($"companyName")) //FIXME check pushdown filters .select($"companyName", lit(0d), $"name") .distinct .as[(String, Double, String)] .rdd } private def clearES() = { spark.conf.set("es.nodes", spark.conf.get("spark.es.nodes")) spark.conf.set("es.port", "80") val deleteIndices = new HttpDelete(s"http://${spark.conf.get("es.nodes")}/companies,sec_edgar_companies") println(s"Deleting indices ${deleteIndices}...") val resp = (new DefaultHttpClient).execute(deleteIndices) println(s"Indices deleted: ${resp}") } }
phymbert/spark-search
core/src/test/scala/org/apache/spark/search/rdd/SearchIndexRDDSuite.scala
/* * Copyright © 2020 Spark Search (The Spark Search Contributors) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.apache.spark.search.rdd import java.io.File import org.apache.commons.io.FileUtils import org.apache.spark.search.rdd.TestData._ import org.scalatest.flatspec.AnyFlatSpec import ZipUtils._ class SearchIndexRDDSuite extends AnyFlatSpec with LocalSparkContext { it should "creates zip index and stream to the next rdd" in { val searchIndexedRDD = sc.parallelize(persons) .searchRDD() .asInstanceOf[SearchRDDLucene[Person]] .indexerRDD searchIndexedRDD.count() val indexDirectoryByPartition = searchIndexedRDD._indexDirectoryByPartition indexDirectoryByPartition.foreach(t => { val indexDir = new File(t._2) FileUtils.deleteDirectory(indexDir) FileUtils.deleteQuietly(new File(indexDir.getParent, s"${indexDir.getName}.zip")) }) searchIndexedRDD.mapPartitionsWithIndex((t, p) => { unzipPartition(indexDirectoryByPartition(t), p) Iterator() }, preservesPartitioning = true).count() } }
leozc/lambda_java_example_pantsbuild
lambda/src/scala/com/example/slack/main/slack.scala
<reponame>leozc/lambda_java_example_pantsbuild<filename>lambda/src/scala/com/example/slack/main/slack.scala package com.example.slack.main import java.io.{InputStream, OutputStream} import java.nio.charset.StandardCharsets.UTF_8 import org.json4s._ import org.json4s.JsonDSL._ import org.json4s.jackson.JsonMethods._ case class Command( operation : String, message : String ) case class LongResponse( text : String, response_type : String ) object Slack{ def main(args: Array[String]): Unit = { println("Hello, world!") } } class Slack { implicit val formats = DefaultFormats def time(in: InputStream, out: OutputStream): Unit = { val payload = scala.io.Source.fromInputStream(in).mkString("") val cmd : Command = parse(payload).extract[Command] val lr = LongResponse(text = cmd.message, response_type = cmd.operation) val json = ("text" -> lr.text) ~ ("response_type" -> lr.response_type) out.write(compact(render(json)).getBytes(UTF_8)) } }
foursquare/util
util-core/src/main/scala/com/twitter/util/Future.scala
<gh_stars>1-10 package com.twitter.util import com.twitter.concurrent.{Offer, IVar, Tx, Scheduler} import java.util.concurrent.atomic.{ AtomicBoolean, AtomicInteger, AtomicReference, AtomicReferenceArray} import java.util.concurrent.{Future => JavaFuture, TimeUnit} import scala.annotation.tailrec import scala.collection.JavaConversions.{asScalaBuffer, seqAsJavaList} import scala.collection.mutable object Future { val DEFAULT_TIMEOUT = Duration.Top val Unit = apply(()) val Void = apply[Void](null) val Done = Unit val None: Future[Option[Nothing]] = new ConstFuture(Return(Option.empty)) val Nil: Future[Seq[Nothing]] = new ConstFuture(Return(Seq.empty)) /** * Makes a Future with a constant result. */ def const[A](result: Try[A]): Future[A] = new ConstFuture[A](result) /** * Make a Future with a constant value. E.g., Future.value(1) is a Future[Int]. */ def value[A](a: A): Future[A] = const[A](Return(a)) /** * Make a Future with an error. E.g., Future.exception(new * Exception("boo")). */ def exception[A](e: Throwable): Future[A] = const[A](Throw(e)) /** * Make a Future with an error. E.g., Future.exception(new * Exception("boo")). The exception is not wrapped in any way. */ def rawException[A](e: Throwable): Future[A] = const[A](Throw(e)) /** * A new future that can never complete. */ def never: Future[Nothing] = new NoFuture @deprecated("Prefer static Future.Void.", "5.x") def void(): Future[Void] = value[Void](null) /** * A factory function to "lift" computations into the Future monad. * It will catch nonfatal (see: [[com.twitter.util.NonFatal]]) * exceptions and wrap them in the Throw[_] type. Non-exceptional * values are wrapped in the Return[_] type. */ def apply[A](a: => A): Future[A] = const(Try(a)) def unapply[A](f: Future[A]): Option[Try[A]] = f.poll /** * Run the computation {{mkFuture}} while installing a monitor that * translates any exception thrown into an encoded one. If an * exception is thrown anywhere, the underlying computation is * interrupted with that exception. * * This function is usually called to wrap a computation that * returns a Future (f0) whose value is satisfied by the invocation * of an onSuccess/onFailure/ensure callbacks of another future * (f1). If an exception happens in the callbacks on f1, f0 is * never satisfied. In this example, `Future.monitored { f1 * onSuccess g; f0 }` will cancel f0 so that f0 never hangs. */ def monitored[A](mkFuture: => Future[A]): Future[A] = { // We define this outside the scope of the following // Promise to guarantee that it is not captured by any // closures. val promiseRef = new AtomicReference[Promise[A]] val monitor = Monitor.mk { case exc => promiseRef.getAndSet(null) match { case null => false case p => p.raise(exc) p.setException(exc) true } } val p = new Promise[A] promiseRef.set(p) monitor { val f = mkFuture p.forwardInterruptsTo(f) f respond { r => promiseRef.getAndSet(null) match { case null => () case p => p.update(r) } } } p } /** * Flattens a nested future. Same as ffa.flatten, but easier to call from Java. */ def flatten[A](ffa: Future[Future[A]]): Future[A] = ffa.flatten /** * Take a sequence of Futures, wait till they all complete * successfully. The future fails immediately if any of the joined * Futures do, mimicking the semantics of exceptions. * * @param fs a sequence of Futures * @return a Future[Unit] whose value is populated when all of the fs return. */ def join[A](fs: Seq[Future[A]]): Future[Unit] = { if (fs.isEmpty) Unit else { val count = new AtomicInteger(fs.size) val p = Promise.interrupts[Unit](fs:_*) for (f <- fs) { f onSuccess { _ => if (count.decrementAndGet() == 0) p.update(Return(())) } onFailure { cause => p.updateIfEmpty(Throw(cause)) } } p } } /* The following joins are generated with this code: scala -e ' val meths = for (end <- ''b'' to ''v''; ps = ''a'' to end) yield """/** * Join %d futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[%s](%s): Future[(%s)] = join(Seq(%s)) map { _ => (%s) }""".format( ps.size, ps map (_.toUpper) mkString ",", ps map(p => "%c: Future[%c]".format(p, p.toUpper)) mkString ",", ps map (_.toUpper) mkString ",", ps mkString ",", ps map(p => "Await.result("+p+")") mkString "," ) meths foreach println ' */ /** * Join 2 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B](a: Future[A],b: Future[B]): Future[(A,B)] = join(Seq(a,b)) map { _ => (Await.result(a),Await.result(b)) } /** * Join 3 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C](a: Future[A],b: Future[B],c: Future[C]): Future[(A,B,C)] = join(Seq(a,b,c)) map { _ => (Await.result(a),Await.result(b),Await.result(c)) } /** * Join 4 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D](a: Future[A],b: Future[B],c: Future[C],d: Future[D]): Future[(A,B,C,D)] = join(Seq(a,b,c,d)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d)) } /** * Join 5 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E]): Future[(A,B,C,D,E)] = join(Seq(a,b,c,d,e)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e)) } /** * Join 6 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F]): Future[(A,B,C,D,E,F)] = join(Seq(a,b,c,d,e,f)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f)) } /** * Join 7 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G]): Future[(A,B,C,D,E,F,G)] = join(Seq(a,b,c,d,e,f,g)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g)) } /** * Join 8 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H]): Future[(A,B,C,D,E,F,G,H)] = join(Seq(a,b,c,d,e,f,g,h)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h)) } /** * Join 9 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I]): Future[(A,B,C,D,E,F,G,H,I)] = join(Seq(a,b,c,d,e,f,g,h,i)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i)) } /** * Join 10 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J]): Future[(A,B,C,D,E,F,G,H,I,J)] = join(Seq(a,b,c,d,e,f,g,h,i,j)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j)) } /** * Join 11 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K]): Future[(A,B,C,D,E,F,G,H,I,J,K)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k)) } /** * Join 12 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L]): Future[(A,B,C,D,E,F,G,H,I,J,K,L)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l)) } /** * Join 13 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L,M](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L],m: Future[M]): Future[(A,B,C,D,E,F,G,H,I,J,K,L,M)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l,m)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l),Await.result(m)) } /** * Join 14 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L,M,N](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L],m: Future[M],n: Future[N]): Future[(A,B,C,D,E,F,G,H,I,J,K,L,M,N)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l,m,n)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l),Await.result(m),Await.result(n)) } /** * Join 15 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L,M,N,O](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L],m: Future[M],n: Future[N],o: Future[O]): Future[(A,B,C,D,E,F,G,H,I,J,K,L,M,N,O)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l,m,n,o)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l),Await.result(m),Await.result(n),Await.result(o)) } /** * Join 16 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L],m: Future[M],n: Future[N],o: Future[O],p: Future[P]): Future[(A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l),Await.result(m),Await.result(n),Await.result(o),Await.result(p)) } /** * Join 17 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L],m: Future[M],n: Future[N],o: Future[O],p: Future[P],q: Future[Q]): Future[(A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l),Await.result(m),Await.result(n),Await.result(o),Await.result(p),Await.result(q)) } /** * Join 18 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L],m: Future[M],n: Future[N],o: Future[O],p: Future[P],q: Future[Q],r: Future[R]): Future[(A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l),Await.result(m),Await.result(n),Await.result(o),Await.result(p),Await.result(q),Await.result(r)) } /** * Join 19 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L],m: Future[M],n: Future[N],o: Future[O],p: Future[P],q: Future[Q],r: Future[R],s: Future[S]): Future[(A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l),Await.result(m),Await.result(n),Await.result(o),Await.result(p),Await.result(q),Await.result(r),Await.result(s)) } /** * Join 20 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L],m: Future[M],n: Future[N],o: Future[O],p: Future[P],q: Future[Q],r: Future[R],s: Future[S],t: Future[T]): Future[(A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l),Await.result(m),Await.result(n),Await.result(o),Await.result(p),Await.result(q),Await.result(r),Await.result(s),Await.result(t)) } /** * Join 21 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L],m: Future[M],n: Future[N],o: Future[O],p: Future[P],q: Future[Q],r: Future[R],s: Future[S],t: Future[T],u: Future[U]): Future[(A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l),Await.result(m),Await.result(n),Await.result(o),Await.result(p),Await.result(q),Await.result(r),Await.result(s),Await.result(t),Await.result(u)) } /** * Join 22 futures. The returned future is complete when all * underlying futures complete. It fails immediately if any of them * do. */ def join[A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U,V](a: Future[A],b: Future[B],c: Future[C],d: Future[D],e: Future[E],f: Future[F],g: Future[G],h: Future[H],i: Future[I],j: Future[J],k: Future[K],l: Future[L],m: Future[M],n: Future[N],o: Future[O],p: Future[P],q: Future[Q],r: Future[R],s: Future[S],t: Future[T],u: Future[U],v: Future[V]): Future[(A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U,V)] = join(Seq(a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v)) map { _ => (Await.result(a),Await.result(b),Await.result(c),Await.result(d),Await.result(e),Await.result(f),Await.result(g),Await.result(h),Await.result(i),Await.result(j),Await.result(k),Await.result(l),Await.result(m),Await.result(n),Await.result(o),Await.result(p),Await.result(q),Await.result(r),Await.result(s),Await.result(t),Await.result(u),Await.result(v)) } /** * Take a sequence of Futures, wait till they all complete * successfully. The future fails immediately if any of the joined * Futures do, mimicking the semantics of exceptions. * * @param fs a java.util.List of Futures * @return a Future[Unit] whose value is populated when all of the fs return. */ def join[A](fs: java.util.List[Future[A]]): Future[Unit] = join(asScalaBuffer(fs)) /** * Collect the results from the given futures into a new future of * Seq[A]. * * @param fs a sequence of Futures * @return a Future[Seq[A]] containing the collected values from fs. */ def collect[A](fs: Seq[Future[A]]): Future[Seq[A]] = { if (fs.isEmpty) { Future(Seq[A]()) } else { val results = new AtomicReferenceArray[A](fs.size) val count = new AtomicInteger(fs.size) val p = Promise.interrupts[Seq[A]](fs:_*) for (i <- 0 until fs.size) { val f = fs(i) f onSuccess { x => results.set(i, x) if (count.decrementAndGet() == 0) { val resultsArray = new mutable.ArrayBuffer[A](fs.size) for (j <- 0 until fs.size) resultsArray += results.get(j) p.setValue(resultsArray) } } onFailure { cause => p.updateIfEmpty(Throw(cause)) } } p } } /** * Collect the results from the given futures into a new future of * Seq[A]. * * @param fs a java.util.List of Futures * @return a Future[java.util.List[A]] containing the collected values from fs. */ def collect[A](fs: java.util.List[Future[A]]): Future[java.util.List[A]] = collect(asScalaBuffer(fs)) map(seqAsJavaList(_)) /** * "Select" off the first future to be satisfied. Return this as a * result, with the remainder of the Futures as a sequence. * * @param fs a scala.collection.Seq */ def select[A](fs: Seq[Future[A]]): Future[(Try[A], Seq[Future[A]])] = { if (fs.isEmpty) { Future.exception(new IllegalArgumentException("empty future list!")) } else { val p = Promise.interrupts[(Try[A], Seq[Future[A]])](fs:_*) @tailrec def stripe(heads: Seq[Future[A]], elem: Future[A], tail: Seq[Future[A]]) { elem respond { res => if (!p.isDefined) { p.updateIfEmpty(Return((res, heads ++ tail))) } } if (!tail.isEmpty) stripe(heads ++ Seq(elem), tail.head, tail.tail) } stripe(Seq(), fs.head, fs.tail) p } } /** * "Select" off the first future to be satisfied. Return this as a * result, with the remainder of the Futures as a sequence. * * @param fs a java.util.List * @return a Future[Tuple2[Try[A], java.util.List[Future[A]]]] representing the first future * to be satisfied and the rest of the futures. */ def select[A](fs: java.util.List[Future[A]]): Future[(Try[A], java.util.List[Future[A]])] = { select(asScalaBuffer(fs)) map { case (first, rest) => (first, seqAsJavaList(rest)) } } /** * Repeat a computation that returns a Future some number of times, after each * computation completes. */ def times[A](n: Int)(f: => Future[A]): Future[Unit] = { val count = new AtomicInteger(0) whileDo(count.getAndIncrement() < n)(f) } /** * Repeat a computation that returns a Future while some predicate obtains, * after each computation completes. */ def whileDo[A](p: => Boolean)(f: => Future[A]): Future[Unit] = { def loop(): Future[Unit] = { if (p) f flatMap { _ => loop() } else Future.Unit } loop() } def parallel[A](n: Int)(f: => Future[A]): Seq[Future[A]] = { (0 until n) map { i => f } } } class FutureCancelledException extends Exception("The future was cancelled with Future.cancel") /** * An alternative interface for handling Future Events. This * interface is designed to be friendly to Java users since it does * not require creating many small Function objects. */ trait FutureEventListener[T] { /** * Invoked if the computation completes successfully */ def onSuccess(value: T): Unit /** * Invoked if the computation completes unsuccessfully */ def onFailure(cause: Throwable): Unit } /** * An alternative interface for performing Future transformations; * that is, converting a Future[A] to a Future[B]. This interface is * designed to be friendly to Java users since it does not require * creating many small Function objects. It is used in conjunction * with `transformedBy`. * * You must override one of `{map, flatMap}`. If you override both * `map` and `flatMap`, `flatMap` takes precedence. If you fail to * override one of `{map, flatMap}`, an `AbstractMethodError` will be * thrown at Runtime. * * '''Note:''' an exception e thrown in any of * map/flatMap/handle/rescue will make the result of transformedBy be * equivalent to Future.exception(e). */ abstract class FutureTransformer[-A, +B] { /** * Invoked if the computation completes successfully. Returns the * new transformed value in a Future. */ def flatMap(value: A): Future[B] = Future.value(map(value)) /** * Invoked if the computation completes successfully. Returns the * new transformed value. * * ''Note'': this method will throw an `AbstractMethodError` if it is not overridden. */ def map(value: A): B = throw new AbstractMethodError /** * Invoked if the computation completes unsuccessfully. Returns the * new Future value. */ def rescue(throwable: Throwable): Future[B] = Future.value(handle(throwable)) /** * Invoked if the computation fails. Returns the new transformed * value. */ def handle(throwable: Throwable): B = throw throwable } /** * A computation evaluated asynchronously. This implementation of * Future does not assume any concrete implementation; in particular, * it does not couple the user to a specific executor or event loop. * * Note that this class extends Try[_] indicating that the results of * the computation may succeed or fail. * * Futures are also [[com.twitter.util.Cancellable]], but with * special semantics: the cancellation signal is only guaranteed to * be delivered when the promise has not yet completed. */ abstract class Future[+A] extends Awaitable[A] { import Future.DEFAULT_TIMEOUT /** * When the computation completes, invoke the given callback * function. Respond() yields a Try (either a Return or a Throw). * This method is most useful for very generic code (like * libraries). Otherwise, it is a best practice to use one of the * alternatives (onSuccess(), onFailure(), etc.). Note that almost * all methods on Future[_] are written in terms of respond(), so * this is the essential template method for use in concrete * subclasses. * * @return a chained Future[A] */ def respond(k: Try[A] => Unit): Future[A] /** * Invoked regardless of whether the computation completed successfully or unsuccessfully. * Implemented in terms of `respond` so that subclasses control evaluation order. Returns a * chained Future. */ def ensure(f: => Unit): Future[A] = respond { _ => f } /** * Block indefinitely, wait for the result of the Future to be available. */ @deprecated("Use Await.result", "6.2.x") def apply(): A = apply(DEFAULT_TIMEOUT) /** * Block, but only as long as the given Timeout. */ @deprecated("Use Await.result", "6.2.x") def apply(timeout: Duration): A = get(timeout)() /** * Alias for apply(). */ @deprecated("Use Await.result", "6.2.x") def get() = apply() @deprecated("Use Await.result", "6.2.x") def isReturn = get(DEFAULT_TIMEOUT) isReturn @deprecated("Use Await.result", "6.2.x") def isThrow = get(DEFAULT_TIMEOUT) isThrow /** * Is the result of the Future available yet? */ def isDefined: Boolean = poll.isDefined /** * Demands that the result of the future be available within * `timeout`. The result is a Return[_] or Throw[_] depending upon * whether the computation finished in time. */ @deprecated("Use Await.result", "6.2.x") final def get(timeout: Duration): Try[A] = { Await.ready(this, timeout) try Return(Await.result(this, Duration.Zero)) catch { // For legacy reasons, we catch even // fatal exceptions. case e => Throw(e) } } /** * Polls for an available result. If the Future has been * satisfied, returns Some(result), otherwise None. */ def poll: Option[Try[A]] /** * Raise the given throwable as an interrupt. Interrupts are * one-shot and latest-interrupt wins. That is, the last interrupt * to have been raised is delivered exactly once to the Promise * responsible for making progress on the future (multiple such * promises may be involed in `flatMap` chains). * * Raising an interrupt does not alter the externally observable * state of the Future. They are used to signal to the ''producer'' * of the future's value that the result is no longer desired (for * whatever reason given in the passed Throwable). */ def raise(interrupt: Throwable) @deprecated("Provided for API compatibility; use raise() instead.", "6.0.0") def cancel() { raise(new FutureCancelledException) } /** * Same as the other within, but with an implicit timer. Sometimes this is more convenient. */ def within(timeout: Duration)(implicit timer: Timer): Future[A] = within(timer, timeout) /** * Returns a new Future that will error if this Future does not return in time. * * @param timeout indicates how long you are willing to wait for the result to be available. */ def within(timer: Timer, timeout: Duration): Future[A] = { if (timeout == Duration.Top) return this val p = Promise.interrupts[A](this) val task = timer.schedule(timeout.fromNow) { p.updateIfEmpty(Throw(new TimeoutException(timeout.toString))) } respond { r => task.cancel() p.updateIfEmpty(r) } p } def transform[B](f: Try[A] => Future[B]): Future[B] /** * If this, the original future, succeeds, run f on the result. * * The returned result is a Future that is satisfied when the original future * and the callback, f, are done. * If the original future fails, this one will also fail, without executing f. * * @see map() */ def flatMap[B](f: A => Future[B]): Future[B] = transform({ case Return(v) => f(v) case Throw(t) => Future.rawException(t) }) def rescue[B >: A]( rescueException: PartialFunction[Throwable, Future[B]] ): Future[B] = transform({ case Throw(t) if rescueException.isDefinedAt(t) => rescueException(t) case _ => this }) /** * Invoke the callback only if the Future returns successfully. Useful for Scala `for` * comprehensions. Use `onSuccess` instead of this method for more readable code. */ def foreach(k: A => Unit) = onSuccess(k) /** * If this, the original future, succeeds, run f on the result. * * The returned result is a Future that is satisfied when the original future * and the callback, f, are done. * If the original future fails, this one will also fail, without executing f. * * @see flatMap() */ def map[B](f: A => B): Future[B] = flatMap { a => Future { f(a) } } def filter(p: A => Boolean): Future[A] = transform { x: Try[A] => Future.const(x.filter(p)) } def withFilter(p: A => Boolean): Future[A] = filter(p) /** * Invoke the function on the result, if the computation was * successful. Returns a chained Future as in `respond`. * * @return chained Future */ def onSuccess(f: A => Unit): Future[A] = respond({ case Return(value) => f(value) case _ => }) /** * Invoke the function on the error, if the computation was * unsuccessful. Returns a chained Future as in `respond`. * * @return chained Future */ def onFailure(rescueException: Throwable => Unit): Future[A] = respond({ case Throw(throwable) => rescueException(throwable) case _ => }) /** * Register a FutureEventListener to be invoked when the * computation completes. This method is typically used by Java * programs because it avoids the use of small Function objects. * * Compare this method to `transformedBy`. The difference is that * `addEventListener` is used to perform a simple action when a * computation completes, such as recording data in a log-file. It * analogous to a `void` method in Java: it has side-effects and no * return value. `transformedBy`, on the other hand, is used to * transform values from one type to another, or to chain a series * of asynchronous calls and return the result. It is analogous to * methods in Java that have a return-type. Note that * `transformedBy` and `addEventListener` are not mutually * exclusive and may be profitably combined. */ def addEventListener(listener: FutureEventListener[_ >: A]) = respond({ case Throw(cause) => listener.onFailure(cause) case Return(value) => listener.onSuccess(value) }) /** * Transform the Future[A] into a Future[B] using the * FutureTransformer. The FutureTransformer handles both success * (Return) and failure (Throw) values by implementing map/flatMap * and handle/rescue. This method is typically used by Java * programs because it avoids the use of small Function objects. * * Compare this method to `addEventListener`. The difference is * that `addEventListener` is used to perform a simple action when * a computation completes, such as recording data in a log-file. * It analogous to a `void` method in Java: it has side-effects and * no return value. `transformedBy`, on the other hand, is used to * transform values from one type to another, or to chain a series * of asynchronous calls and return the result. It is analogous to * methods in Java that have a return-type. Note that * `transformedBy` and `addEventListener` are not mutually * exclusive and may be profitably combined. * * ''Note'': The FutureTransformer must implement either `flatMap` * or `map` and may optionally implement `handle`. Failing to * implement a method will result in a run-time (AbstractMethod) * error. */ def transformedBy[B](transformer: FutureTransformer[A, B]): Future[B] = transform { case Return(v) => transformer.flatMap(v) case Throw(t) => transformer.rescue(t) } def handle[B >: A](rescueException: PartialFunction[Throwable, B]): Future[B] = rescue { case e: Throwable if rescueException.isDefinedAt(e) => Future(rescueException(e)) case e: Throwable => this } /** * Choose the first Future to succeed. * * @param other another Future * @return a new Future whose result is that of the first of this and other to return */ def select[U >: A](other: Future[U]): Future[U] = { val p = Promise.interrupts[U](other, this) other respond { p.updateIfEmpty(_) } this respond { p.updateIfEmpty(_) } p } /** * A synonym for select(): Choose the first Future to succeed. */ def or[U >: A](other: Future[U]): Future[U] = select(other) /** * Combines two Futures into one Future of the Tuple of the two results. */ def join[B](other: Future[B]): Future[(A, B)] = { val p = Promise.interrupts[(A, B)](this, other) this.respond { case Throw(t) => p() = Throw(t) case Return(a) => other respond { case Throw(t) => p() = Throw(t) case Return(b) => p() = Return((a, b)) } } p } /** * Convert this Future[A] to a Future[Unit] by discarding the result. */ def unit: Future[Unit] = map(_ => ()) /** * Convert this Future[A] to a Future[Void] by discarding the result. */ def voided: Future[Void] = map(_ => null.asInstanceOf[Void]) @deprecated("'void' is a reserved word in javac.", "5.x") def void: Future[Void] = voided /** * Send updates from this Future to the other. * ``other'' must not yet be satisfied. */ def proxyTo[B >: A](other: Promise[B]) { respond { other() = _ } } /** * An offer for this future. The offer is activated when the future * is satisfied. */ def toOffer: Offer[Try[A]] = new Offer[Try[A]] { def prepare() = transform { res: Try[A] => val tx = new Tx[Try[A]] { def ack() = Future.value(Tx.Commit(res)) def nack() {} } Future.value(tx) } } /** * Convert a Twitter Future to a Java native Future. This should * match the semantics of a Java Future as closely as possible to * avoid issues with the way another API might use them. See: * * http://download.oracle.com/javase/6/docs/api/java/util/concurrent/Future.html#cancel(boolean) */ def toJavaFuture: JavaFuture[_ <: A] = { val f = this new JavaFuture[A] { val wasCancelled = new AtomicBoolean(false) override def cancel(mayInterruptIfRunning: Boolean): Boolean = { if (wasCancelled.compareAndSet(false, true)) f.raise(new java.util.concurrent.CancellationException) true } override def isCancelled: Boolean = wasCancelled.get override def isDone: Boolean = isCancelled || f.isDefined override def get(): A = { if (isCancelled) throw new java.util.concurrent.CancellationException() Await.result(f) } override def get(time: Long, timeUnit: TimeUnit): A = { if (isCancelled) throw new java.util.concurrent.CancellationException() Await.result(f, Duration.fromTimeUnit(time, timeUnit)) } } } /** * Converts a Future[Future[B]] into a Future[B] */ def flatten[B](implicit ev: A <:< Future[B]): Future[B] = flatMap[B] { x => x } /** * Returns a Future[Boolean] indicating whether two Futures are equivalent. Note that * Future.exception(e).willEqual(Future.exception(e)) == Future.value(true). */ def willEqual[B](that: Future[B]) = { val areEqual = new Promise[Boolean] this respond { thisResult => that respond { thatResult => areEqual.setValue(thisResult == thatResult) } } areEqual } } /** * A Future that is already completed. These are cheap in * construction compared to Promises. */ class ConstFuture[A](result: Try[A]) extends Future[A] { def respond(k: Try[A] => Unit): Future[A] = { val saved = Local.save() Scheduler.submit(new Runnable { def run() { val current = Local.save() Local.restore(saved) try Monitor { k(result) } finally Local.restore(current) } }) this } def raise(interrupt: Throwable) {} def transform[B](f: Try[A] => Future[B]): Future[B] = { val p = new Promise[B] respond({ r => val result = try f(r) catch { case NonFatal(e) => Future.exception(e) } p.become(result) }) p } def poll: Option[Try[A]] = Some(result) // Awaitable @throws(classOf[TimeoutException]) @throws(classOf[InterruptedException]) def ready(timeout: Duration)(implicit permit: Awaitable.CanAwait): this.type = this @throws(classOf[Exception]) def result(timeout: Duration)(implicit permit: Awaitable.CanAwait): A = result() } /** * A future with no future (never completes). */ class NoFuture extends Future[Nothing] { def respond(k: Try[Nothing] => Unit): Future[Nothing] = this def transform[B](f: Try[Nothing] => Future[B]): Future[B] = this def raise(interrupt: Throwable) {} // Awaitable @throws(classOf[TimeoutException]) @throws(classOf[InterruptedException]) def ready(timeout: Duration)(implicit permit: Awaitable.CanAwait): this.type = { Thread.sleep(timeout.inMilliseconds) throw new TimeoutException(timeout.toString) } @throws(classOf[Exception]) def result(timeout: Duration)(implicit permit: Awaitable.CanAwait): Nothing = { Thread.sleep(timeout.inMilliseconds) throw new TimeoutException(timeout.toString) } def poll: Option[Try[Nothing]] = None } class FutureTask[A](fn: => A) extends Promise[A] with Runnable { def run() { update(Try(fn)) } } object FutureTask { def apply[A](fn: => A) = new FutureTask[A](fn) } private[util] object FutureBenchmark { /** * Admittedly, this is not very good microbenchmarking technique. */ import com.twitter.conversions.storage._ private[this] val NumIters = 100.million private[this] def bench[A](numIters: Long)(f: => A): Long = { val begin = System.currentTimeMillis() (0L until numIters) foreach { _ => f } System.currentTimeMillis() - begin } private[this] def run[A](name: String)(work: => A) { printf("Warming up %s.. ", name) val warmupTime = bench(NumIters)(work) printf("%d ms\n", warmupTime) printf("Running .. ") val runTime = bench(NumIters)(work) printf( "%d ms, %d %s/sec\n", runTime, 1000 * NumIters / runTime, name) } def main(args: Array[String]) { run("respond") { val promise = new Promise[Unit] promise respond { res => () } promise() = Return(()) } run("flatMaps") { val promise = new Promise[Unit] promise flatMap { _ => Future.value(()) } promise() = Return(()) } } } private[util] object Leaky { def main(args: Array[String]) { def loop(i: Int): Future[Int] = Future.value(i) flatMap { count => if (count % 1000000 == 0) { System.gc() println("iter %d %dMB".format( count, Runtime.getRuntime().totalMemory()>>20)) } loop(count + 1) } loop(1) } }
foursquare/util
util-app/src/test/scala/com/twitter/app/FlagTest.scala
package com.twitter.app import org.scalatest.FunSuite import org.scalatest.junit.JUnitRunner import org.junit.runner.RunWith object MyGlobalFlag extends GlobalFlag("a test flag", "a global test flag") @RunWith(classOf[JUnitRunner]) class FlagTest extends FunSuite { test("Flaggable: parse booleans") { assert(Flaggable.ofBoolean.parse("true")) assert(!Flaggable.ofBoolean.parse("false")) intercept[Throwable] { Flaggable.ofBoolean.parse("") } intercept[Throwable] { Flaggable.ofBoolean.parse("gibberish") } } test("Flaggable: parse strings") { assert(Flaggable.ofString.parse("blah") === "blah") } test("Flaggable: parse/show inet addresses") { val local8080 = Flaggable.ofInetSocketAddress.parse(":8080") assert(local8080.getAddress.isAnyLocalAddress) assert(local8080.getPort === 8080) val ip8080 = Flaggable.ofInetSocketAddress.parse("172.16.31.10:8080") assert(ip8080.getHostName === "172.16.31.10") assert(ip8080.getPort === 8080) assert(Flaggable.ofInetSocketAddress.show(local8080) === ":8080") assert(Flaggable.ofInetSocketAddress.show(ip8080) ==="172.16.31.10:8080") } test("Flaggable: parse seqs") { assert(Flaggable.ofSeq[Int].parse("1,2,3,4") === Seq(1,2,3,4)) } test("Flaggable: parse tuples") { assert(Flaggable.ofTuple[Int, String].parse("1,hello") === (1, "hello")) intercept[IllegalArgumentException] { Flaggable.ofTuple[Int, String].parse("1") } } class Ctx { val flag = new Flags("test") val fooFlag = flag("foo", 123, "The foo value") val barFlag = flag("bar", "okay", "The bar value") } test("Flag: defaults") { val ctx = new Ctx import ctx._ assert(fooFlag() === 123) assert(barFlag() === "okay") } test("Flag: add and parse flags") { val ctx = new Ctx import ctx._ assert(flag.parse(Array("-foo", "973", "-bar", "hello there")).isEmpty) assert(fooFlag() === 973) assert(barFlag() === "hello there") } class Bctx extends Ctx { val yesFlag = flag("yes", false, "Just say yes.") } test("Boolean: default") { val ctx = new Bctx import ctx._ assert(!yesFlag()) } test("Boolean: -yes") { val ctx = new Bctx import ctx._ assert(flag.parse(Array("-yes")).isEmpty) assert(yesFlag()) } test("Boolean: -yes=true") { val ctx = new Bctx import ctx._ assert(flag.parse(Array("-yes=true")).isEmpty) assert(yesFlag()) } test("Boolean: -yes=false") { val ctx = new Bctx import ctx._ assert(flag.parse(Array("-yes=false")).isEmpty) assert(!yesFlag()) } test("Flag: handle remainders (sequential)") { val ctx = new Ctx import ctx._ assert(flag.parse(Array("-foo", "333", "arg0", "arg1")) === Seq("arg0", "arg1")) } test("Flag: handle remainders (interpspersed)") { val ctx = new Ctx import ctx._ assert(flag.parse(Array("arg0", "-foo", "333", "arg1")) === Seq("arg0", "arg1")) } test("Flag: give nice parse errors") { val ctx = new Ctx import ctx._ val thr = intercept[Exception] { flag.parse(Array("-foo", "blah")) } } test("Flag: handle -help") { val ctx = new Ctx import ctx._ intercept[FlagUsageError] { flag.parse(Array("-help")) } } test("Flag: mandatory flag without argument") { val ctx = new Ctx import ctx._ val thr = intercept[FlagParseException] { flag.parse(Array("-foo")) } } test("Flag: undefined") { val ctx = new Ctx import ctx._ val thr = intercept[FlagParseException] { flag.parse(Array("-undefined")) } assert(flag.parse(Array("-undefined"), true) === Seq("-undefined")) } test("GlobalFlag") { assert(MyGlobalFlag() === "a test flag") val flag = new Flags("my", includeGlobal=true) flag.parse(Array("-com.twitter.app.MyGlobalFlag", "okay")) assert(MyGlobalFlag() === "okay") System.setProperty("com.twitter.app.MyGlobalFlag", "not okay") assert(MyGlobalFlag() === "okay") MyGlobalFlag.reset() assert(MyGlobalFlag() === "not okay") } }
foursquare/util
util-core/src/main/scala/com/twitter/concurrent/Spool.scala
<gh_stars>1-10 package com.twitter.concurrent import scala.collection.mutable.ArrayBuffer import com.twitter.util.{Future, Promise, Return, Await, Duration} /** * A spool is an asynchronous stream. It more or less * mimics the scala {{Stream}} collection, but with cons * cells that have deferred tails. * * Construction is done with Spool.cons, Spool.empty. Convenience * syntax like that of [[scala.Stream]] is provided. In order to use * these operators for deconstruction, they must be imported * explicitly ({{import Spool.{*::, **::}}}) * * {{{ * def fill(rest: Promise[Spool[Int]]) { * asyncProcess foreach { result => * if (result.last) { * rest() = Return(result **:: Spool.empty) * } else { * val next = new Promise[Spool[Int]] * rest() = Return(result *:: next) * fill(next) * } * } * } * val rest = new Promise[Spool[Int]] * fill(rest) * firstElem *:: rest * }}} */ sealed trait Spool[+A] { import Spool.{cons, empty} def isEmpty: Boolean /** * The first element of the spool. Invalid for empty spools. */ def head: A /** * The (deferred) tail of the spool. Invalid for empty spools. */ def tail: Future[Spool[A]] /** * Apply {{f}} for each item in the spool, until the end. {{f}} is * applied as the items become available. */ def foreach[B](f: A => B) = foreachElem { _ foreach(f) } /** * A version of {{foreach}} that wraps each element in an * {{Option}}, terminating the stream (EOF or failure) with * {{None}}. */ def foreachElem[B](f: Option[A] => B) { if (!isEmpty) { f(Some(head)) tail onSuccess { s => s.foreachElem(f) } onFailure { _ => f(None) } } else { f(None) } } /** * The standard Scala collect, in order to implement map & filter. * * It may seem unnatural to return a Future[…] here, but we cannot * know whether the first element exists until we have applied its * filter. */ def collect[B](f: PartialFunction[A, B]): Future[Spool[B]] def map[B](f: A => B): Spool[B] = { val s = collect { case x => f(x) } Await.result(s, Duration.Zero) } def filter(f: A => Boolean): Future[Spool[A]] = collect { case x if f(x) => x } /** * Concatenates two spools. */ def ++[B >: A](that: Spool[B]): Spool[B] = if (isEmpty) that else cons(head: B, tail map { _ ++ that }) /** * Concatenates two spools. */ def ++[B >: A](that: Future[Spool[B]]): Future[Spool[B]] = if (isEmpty) that else Future.value(cons(head: B, tail flatMap { _ ++ that })) /** * Applies a function that generates a spool to each element in this spool, * flattening the result into a single spool. */ def flatMap[B](f: A => Future[Spool[B]]): Future[Spool[B]] = if (isEmpty) Future.value(empty[B]) else f(head) flatMap { _ ++ (tail flatMap { _ flatMap f }) } /** * Fully buffer the spool to a {{Seq}}. The returned future is * satisfied when the entire result is ready. */ def toSeq: Future[Seq[A]] = { val p = new Promise[Seq[A]] val as = new ArrayBuffer[A] foreachElem { case Some(a) => as += a case None => p() = Return(as) } p } } object Spool { case class Cons[A](value: A, next: Future[Spool[A]]) extends Spool[A] { def isEmpty = false def head = value def tail = next def collect[B](f: PartialFunction[A, B]) = { val next_ = next flatMap { _.collect(f) } if (f.isDefinedAt(head)) Future.value(Cons(f(head), next_)) else next_ } override def toString = "Cons(%s, %c)".format(head, if (tail.isDefined) '*' else '?') } object Empty extends Spool[Nothing] { def isEmpty = true def head = throw new NoSuchElementException("spool is empty") def tail = Future.exception(new NoSuchElementException("spool is empty")) def collect[B](f: PartialFunction[Nothing, B]) = Future.value(this) override def toString = "Empty" } /** * Cons a value & (possibly deferred) tail to a new {{Spool}}. */ def cons[A](value: A, next: Future[Spool[A]]): Spool[A] = Cons(value, next) def cons[A](value: A, nextSpool: Spool[A]): Spool[A] = Cons(value, Future.value(nextSpool)) /** * The empty spool. */ def empty[A]: Spool[A] = Empty /** * Syntax support. We retain different constructors for future * resolving vs. not. * * *:: constructs and deconstructs deferred tails * **:: constructs and deconstructs eager tails */ class Syntax[A](tail: => Future[Spool[A]]) { def *::(head: A) = cons(head, tail) } implicit def syntax[A](s: Future[Spool[A]]) = new Syntax(s) object *:: { def unapply[A](s: Spool[A]): Option[(A, Future[Spool[A]])] = { if (s.isEmpty) None else Some((s.head, s.tail)) } } class Syntax1[A](tail: => Spool[A]) { def **::(head: A) = cons(head, tail) } implicit def syntax1[A](s: Spool[A]) = new Syntax1(s) object **:: { def unapply[A](s: Spool[A]): Option[(A, Spool[A])] = { if (s.isEmpty) None else Some((s.head, Await.result(s.tail))) } } }
foursquare/util
util-core/src/main/scala/com/twitter/util/Timer.scala
<filename>util-core/src/main/scala/com/twitter/util/Timer.scala package com.twitter.util import com.twitter.concurrent.NamedPoolThreadFactory import com.twitter.concurrent.Serialized import com.twitter.conversions.time._ import java.util.concurrent.atomic.AtomicBoolean import java.util.concurrent.{CancellationException, ExecutorService, RejectedExecutionHandler, ScheduledThreadPoolExecutor, ThreadFactory, TimeUnit} import scala.collection.mutable.ArrayBuffer trait TimerTask { def cancel() } trait Timer { def schedule(when: Time)(f: => Unit): TimerTask def schedule(when: Time, period: Duration)(f: => Unit): TimerTask def schedule(period: Duration)(f: => Unit): TimerTask = { schedule(period.fromNow, period)(f) } /** * Performs an operation after the specified delay. Interrupting the Future * will cancel the scheduled timer task, if not too late. */ def doLater[A](delay: Duration)(f: => A): Future[A] = { doAt(Time.now + delay)(f) } /** * Performs an operation at the specified time. Interrupting the Future * will cancel the scheduled timer task, if not too late. */ def doAt[A](time: Time)(f: => A): Future[A] = { val pending = new AtomicBoolean(true) val p = new Promise[A] val task = schedule(time) { if (pending.compareAndSet(true, false)) p.update(Try(f)) } p.setInterruptHandler { case cause => if (pending.compareAndSet(true, false)) { task.cancel() val exc = new CancellationException exc.initCause(cause) p.setException(exc) } } p } def stop() } /** * A NullTimer is not a timer at all: it invokes all tasks immediately * and inline. */ class NullTimer extends Timer { def schedule(when: Time)(f: => Unit): TimerTask = { f NullTimerTask } def schedule(when: Time, period: Duration)(f: => Unit): TimerTask = { f NullTimerTask } def stop() {} } object NullTimerTask extends TimerTask { def cancel() {} } class ThreadStoppingTimer(underlying: Timer, executor: ExecutorService) extends Timer { def schedule(when: Time)(f: => Unit): TimerTask = underlying.schedule(when)(f) def schedule(when: Time, period: Duration)(f: => Unit): TimerTask = underlying.schedule(when, period)(f) def stop() { executor.submit(new Runnable { def run() = underlying.stop() }) } } trait ReferenceCountedTimer extends Timer { def acquire() } class ReferenceCountingTimer(factory: () => Timer) extends ReferenceCountedTimer { private[this] var refcount = 0 private[this] var underlying: Timer = null def acquire() = synchronized { refcount += 1 if (refcount == 1) { require(underlying == null) underlying = factory() } } def stop() = synchronized { refcount -= 1 if (refcount == 0) { underlying.stop() underlying = null } } // Just dispatch to the underlying timer. It's the responsibility of // the API consumer to not call into the timer once it has been // stopped. def schedule(when: Time)(f: => Unit) = underlying.schedule(when)(f) def schedule(when: Time, period: Duration)(f: => Unit) = underlying.schedule(when, period)(f) } class JavaTimer(isDaemon: Boolean) extends Timer { def this() = this(false) private[this] val underlying = new java.util.Timer(isDaemon) def schedule(when: Time)(f: => Unit) = { val task = toJavaTimerTask(f) underlying.schedule(task, when.toDate) toTimerTask(task) } def schedule(when: Time, period: Duration)(f: => Unit) = { val task = toJavaTimerTask(f) underlying.schedule(task, when.toDate, period.inMillis) toTimerTask(task) } def stop() = underlying.cancel() /** * log any Throwables caught by the internal TimerTask. * * By default we log to System.err but users may subclass and log elsewhere. * * This method MUST NOT throw or else your Timer will die. */ def logError(t: Throwable) { System.err.println("WARNING: JavaTimer caught exception running task: %s".format(t)) t.printStackTrace(System.err) } private[this] def toJavaTimerTask(f: => Unit) = new java.util.TimerTask { def run { try { f } catch { case NonFatal(t) => logError(t) case fatal: Throwable => logError(fatal) throw fatal } } } private[this] def toTimerTask(task: java.util.TimerTask) = new TimerTask { def cancel() { task.cancel() } } } class ScheduledThreadPoolTimer( poolSize: Int, threadFactory: ThreadFactory, rejectedExecutionHandler: Option[RejectedExecutionHandler]) extends Timer { def this(poolSize: Int, threadFactory: ThreadFactory) = this(poolSize, threadFactory, None) def this(poolSize: Int, threadFactory: ThreadFactory, handler: RejectedExecutionHandler) = this(poolSize, threadFactory, Some(handler)) /** Construct a ScheduledThreadPoolTimer with a NamedPoolThreadFactory. */ def this(poolSize: Int = 2, name: String = "timer", makeDaemons: Boolean = false) = this(poolSize, new NamedPoolThreadFactory(name, makeDaemons), None) private[this] val underlying = rejectedExecutionHandler match { case None => new ScheduledThreadPoolExecutor(poolSize, threadFactory) case Some(h: RejectedExecutionHandler) => new ScheduledThreadPoolExecutor(poolSize, threadFactory, h) } def schedule(when: Time)(f: => Unit): TimerTask = { val runnable = new Runnable { def run = f } val javaFuture = underlying.schedule(runnable, when.sinceNow.inMillis, TimeUnit.MILLISECONDS) new TimerTask { def cancel() { javaFuture.cancel(true) underlying.remove(runnable) } } } def schedule(when: Time, period: Duration)(f: => Unit): TimerTask = schedule(when.sinceNow, period)(f) def schedule(wait: Duration, period: Duration)(f: => Unit): TimerTask = { val runnable = new Runnable { def run = f } val javaFuture = underlying.scheduleAtFixedRate(runnable, wait.inMillis, period.inMillis, TimeUnit.MILLISECONDS) new TimerTask { def cancel() { javaFuture.cancel(true) underlying.remove(runnable) } } } def stop() = underlying.shutdown() } // Exceedingly useful for writing well-behaved tests. class MockTimer extends Timer { // These are weird semantics admittedly, but there may // be a bunch of tests that rely on them already. case class Task(var when: Time, runner: () => Unit) extends TimerTask { var isCancelled = false def cancel() { isCancelled = true; nCancelled += 1; when = Time.now; tick() } } var isStopped = false var tasks = ArrayBuffer[Task]() var nCancelled = 0 def tick() { if (isStopped) throw new IllegalStateException("timer is stopped already") val now = Time.now val (toRun, toQueue) = tasks.partition { task => task.when <= now } tasks = toQueue toRun filter { !_.isCancelled } foreach { _.runner() } } def schedule(when: Time)(f: => Unit): TimerTask = { val task = Task(when, () => f) tasks += task task } def schedule(when: Time, period: Duration)(f: => Unit): TimerTask = throw new Exception("periodic scheduling not supported") def stop() { isStopped = true } }
foursquare/util
util-core/src/main/scala/com/twitter/util/Closable.scala
package com.twitter.util /** * Closable is a mixin trait to describe a closable ``resource``. */ trait Closable { self => /** * Close the resource. The returned Future is completed when * the resource has been fully relinquished. */ final def close(): Future[Unit] = close(Time.Bottom) /** * Close the resource with the given deadline. This deadline is advisory, * giving the callee some leeway, for example to drain clients or finish * up other tasks. */ def close(deadline: Time): Future[Unit] } object Closable { /** * Concurrent composition: creates a new closable which, when * closed, closes all of the underlying resources simultaneously. */ def all(closables: Closable*): Closable = new Closable { def close(deadline: Time) = Future.join(closables map(_.close(deadline))) } /** * Sequential composition: create a new Closable which, when * closed, closes all of the underlying ones in sequence: that is, * resource ''n+1'' is not closed until resource ''n'' is. */ def sequence(closables: Closable*): Closable = new Closable { private final def closeSeq(deadline: Time, closables: Seq[Closable]): Future[Unit] = closables match { case Seq() => Future.Done case Seq(hd, tl@_*) => hd.close(deadline) flatMap { _ => closeSeq(deadline, tl) } } def close(deadline: Time) = closeSeq(deadline, closables) } }
foursquare/util
util-jvm/src/main/scala/com/twitter/jvm/ContentionSnapshot.scala
package com.twitter.jvm import java.lang.management.{ManagementFactory, ThreadInfo} import java.lang.Thread.State._ import scala.collection.mutable /** * A thread contention summary. This provides a brief overview of threads * that are blocked or otherwise waiting. * * While this could be an object, we use instantiation as a signal of intent * and enable contention monitoring. */ class ContentionSnapshot { ManagementFactory.getThreadMXBean.setThreadContentionMonitoringEnabled(true) case class Snapshot( blockedThreads: Seq[String], lockOwners: Seq[String]) private[this] object Blocked { def unapply(t: ThreadInfo): Option[ThreadInfo] = { t.getThreadState match { case BLOCKED | WAITING | TIMED_WAITING => Some(t) case _ => None } } } def snap(): Snapshot = { val bean = ManagementFactory.getThreadMXBean val lockOwners = mutable.Set[Long]() val blocked = bean.getThreadInfo(bean.getAllThreadIds, true, true) .filter(_ != null) .collect { case Blocked(info) => info } val ownerIds = blocked map(_.getLockOwnerId) filter(_ != -1) val owners = if (ownerIds.size == 0) Seq[String]() else bean.getThreadInfo(ownerIds.toArray, true, true).map(_.toString).toSeq Snapshot( blockedThreads = blocked.map(_.toString).toSeq, lockOwners = owners) } }
foursquare/util
util-core/src/test/scala/com/twitter/concurrent/SpoolSpec.scala
<gh_stars>1-10 package com.twitter.concurrent import scala.collection.mutable.ArrayBuffer import org.specs.SpecificationWithJUnit import com.twitter.util.{Await, Future, Promise, Return, Throw} import Spool.{*::, **::} class SpoolSpec extends SpecificationWithJUnit { "Empty Spool" should { val s = Spool.empty[Int] "iterate over all elements" in { val xs = new ArrayBuffer[Int] s foreach { xs += _ } xs.size must be_==(0) } "map" in { (s map { _ * 2 } ) must be_==(Spool.empty[Int]) } "deconstruct" in { s must beLike { case x **:: rest => false case _ => true } } "append via ++" in { (s ++ Spool.empty[Int]) must be_==(Spool.empty[Int]) (Spool.empty[Int] ++ s) must be_==(Spool.empty[Int]) val s2 = s ++ (3 **:: 4 **:: Spool.empty[Int]) Await.result(s2.toSeq) must be_==(Seq(3, 4)) } "append via ++ with Future rhs" in { Await.result(s ++ Future(Spool.empty[Int])) must be_==(Spool.empty[Int]) Await.result(Spool.empty[Int] ++ Future(s)) must be_==(Spool.empty[Int]) val s2 = s ++ Future(3 **:: 4 **:: Spool.empty[Int]) Await.result(s2 flatMap (_.toSeq)) must be_==(Seq(3, 4)) } "flatMap" in { val f = (x: Int) => Future(x.toString **:: (x * 2).toString **:: Spool.empty) Await.result(s flatMap f) must be_==(Spool.empty[Int]) } } "Simple resolved Spool" should { val s = 1 **:: 2 **:: Spool.empty "iterate over all elements" in { val xs = new ArrayBuffer[Int] s foreach { xs += _ } xs.toSeq must be_==(Seq(1,2)) } "buffer to a sequence" in { Await.result(s.toSeq) must be_==(Seq(1, 2)) } "map" in { Await.result(s map { _ * 2 } toSeq) must be_==(Seq(2, 4)) } "deconstruct" in { s must beLike { case x **:: rest => x must be_==(1) rest must beLike { case y **:: rest if y == 2 && rest.isEmpty => true } } } "append via ++" in { Await.result((s ++ Spool.empty[Int]).toSeq) must be_==(Seq(1, 2)) Await.result((Spool.empty[Int] ++ s).toSeq) must be_==(Seq(1, 2)) val s2 = s ++ (3 **:: 4 **:: Spool.empty) Await.result(s2.toSeq) must be_==(Seq(1, 2, 3, 4)) } "append via ++ with Future rhs" in { Await.result(s ++ Future(Spool.empty[Int]) flatMap (_.toSeq)) must be_==(Seq(1, 2)) Await.result(Spool.empty[Int] ++ Future(s) flatMap (_.toSeq)) must be_==(Seq(1, 2)) val s2 = s ++ Future(3 **:: 4 **:: Spool.empty) Await.result(s2 flatMap (_.toSeq)) must be_==(Seq(1, 2, 3, 4)) } "flatMap" in { val f = (x: Int) => Future(x.toString **:: (x * 2).toString **:: Spool.empty) val s2 = s flatMap f Await.result(s2 flatMap (_.toSeq)) must be_==(Seq("1", "2", "2", "4")) } } "Simple resolved spool with error" should { val p = new Promise[Spool[Int]](Throw(new Exception("sad panda"))) val s = 1 **:: 2 *:: p "EOF iteration on error" in { val xs = new ArrayBuffer[Option[Int]] s foreachElem { xs += _ } xs.toSeq must be_==(Seq(Some(1), Some(2), None)) } } "Simple delayed Spool" should { val p = new Promise[Spool[Int]] val p1 = new Promise[Spool[Int]] val p2 = new Promise[Spool[Int]] val s = 1 *:: p "iterate as results become available" in { val xs = new ArrayBuffer[Int] s foreach { xs += _ } xs.toSeq must be_==(Seq(1)) p() = Return(2 *:: p1) xs.toSeq must be_==(Seq(1, 2)) p1() = Return(Spool.empty) xs.toSeq must be_==(Seq(1, 2)) } "EOF iteration on failure" in { val xs = new ArrayBuffer[Option[Int]] s foreachElem { xs += _ } xs.toSeq must be_==(Seq(Some(1))) p() = Throw(new Exception("sad panda")) xs.toSeq must be_==(Seq(Some(1), None)) } "return a buffered seq when complete" in { val f = s.toSeq f.isDefined must beFalse p() = Return(2 *:: p1) f.isDefined must beFalse p1() = Return(Spool.empty) f.isDefined must beTrue Await.result(f) must be_==(Seq(1,2)) } "deconstruct" in { s must beLike { case fst *:: rest if fst == 1 && !rest.isDefined => true } } "collect" in { val f = s collect { case x if x % 2 == 0 => x * 2 } f.isDefined must beFalse // 1 != 2 mod 0 p() = Return(2 *:: p1) f.isDefined must beTrue val s1 = Await.result(f) s1 must beLike { case x *:: rest if x == 4 && !rest.isDefined => true } p1() = Return(3 *:: p2) s1 must beLike { case x *:: rest if x == 4 && !rest.isDefined => true } p2() = Return(4 **:: Spool.empty) val s1s = s1.toSeq s1s.isDefined must beTrue Await.result(s1s) must be_==(Seq(4, 8)) } } }
foursquare/util
util-core/src/test/scala/com/twitter/concurrent/AsyncSemaphoreSpec.scala
package com.twitter.concurrent import org.specs.SpecificationWithJUnit import org.specs.mock.Mockito import java.util.concurrent.{ConcurrentLinkedQueue, RejectedExecutionException} import com.twitter.util.Await class AsyncSemaphoreSpec extends SpecificationWithJUnit with Mockito { "AsyncSemaphore" should { var count = 0 val s = new AsyncSemaphore(2) val permits = new ConcurrentLinkedQueue[Permit] def acquire() { s.acquire() onSuccess { permit => count += 1 permits add permit } } "execute immediately while permits are available" in { acquire() count must be_==(1) acquire() count must be_==(2) acquire() count must be_==(2) } "execute deferred computations when permits are released" in { acquire() acquire() acquire() acquire() count must be_==(2) permits.poll().release() count must be_==(3) permits.poll().release() count must be_==(4) permits.poll().release() count must be_==(4) } "bound the number of waiters" in { val s2 = new AsyncSemaphore(2, 3) def acquire2() = { s2.acquire() onSuccess { permit => count += 1 permits add permit } } // The first two acquires obtain a permit. acquire2() acquire2() count must be_==(2) // The next three acquires wait. acquire2() acquire2() acquire2() count must be_==(2) s2.numWaiters mustEqual(3) // The next acquire should be rejected. val futurePermit = acquire2() s2.numWaiters mustEqual(3) Await.result(futurePermit) must throwA[RejectedExecutionException] // Waiting tasks should still execute once permits are available. permits.poll().release() permits.poll().release() permits.poll().release() count must be_==(5) } } }
foursquare/util
util-logging/src/main/scala/com/twitter/logging/App.scala
<reponame>foursquare/util package com.twitter.logging import com.twitter.app.{App, Flaggable} import com.twitter.util.{Throw, Return} object Logging { implicit object LevelFlaggable extends Flaggable[Level] { def parse(s: String) = if (Logger.levelNames contains s) Logger.levelNames(s) else throw new Exception("Invalid log level: "+s) } } /** * A [[com.twitter.app.App]] mixin to use for logging. Defines flags * to configure the (default) logger setup. */ trait Logging { self: App => import Logging._ lazy val log = Logger(name) def defaultLogLevel: Level = Level.INFO private val levelFlag = flag("log.level", defaultLogLevel, "Log level") private val outputFlag = flag("log.output", "/dev/stderr", "Output file") premain { val factory = LoggerFactory( node = "", level = Some(levelFlag()), handlers = FileHandler(outputFlag()) :: Nil ) Logger.configure(factory :: Nil) } }
foursquare/util
util-zk-common/src/test/scala/com/twitter/zk/CommonConnectorSpec.scala
<filename>util-zk-common/src/test/scala/com/twitter/zk/CommonConnectorSpec.scala package com.twitter.zk import com.twitter.common.net.InetSocketAddressHelper import com.twitter.common.zookeeper.ZooKeeperClient import com.twitter.conversions.common.quantity._ import com.twitter.conversions.common.zookeeper._ import com.twitter.conversions.time._ import com.twitter.util.{Await, FuturePool} import org.specs.SpecificationWithJUnit import java.net.InetSocketAddress import scala.collection.JavaConverters._ class CommonConnectorSpec extends SpecificationWithJUnit { val timeout = 2.seconds val addresses = new InetSocketAddress("localhost", 2181) :: Nil "CommonConnector" should { "initialize" in { "with addresses" in { implicit val pool = FuturePool.immediatePool CommonConnector(addresses, timeout) must notBeNull } "with a ZooKeeperClient instance" in { implicit val pool = FuturePool.immediatePool val zookeeper = new ZooKeeperClient(timeout.toIntAmount, addresses.asJava) val connector = CommonConnector(zookeeper, timeout) connector.underlying mustBe zookeeper } } } // A simple live test Option { System.getProperty("com.twitter.zk.TEST_CONNECT") } foreach { connectString => val address = InetSocketAddressHelper.parse(connectString) "A live server @ %s".format(connectString) should { val commonClient: ZooKeeperClient = new ZooKeeperClient(timeout.toIntAmount, address) val zkClient = commonClient.toZkClient(timeout)(FuturePool.immediatePool) doAfter { Await.ready(zkClient.release()) } "have 'zookeeper' in '/'" in { Await.result(zkClient("/").getChildren(), timeout).children map { _.name } mustContain("zookeeper") } } } }
foursquare/util
util-core/src/main/scala/com/twitter/util/Awaitable.scala
<reponame>foursquare/util package com.twitter.util import java.util.concurrent.atomic.AtomicBoolean /** * Wait for the result of some action. Awaitable is not used * directly, but through the `Await` object. */ trait Awaitable[+T] { import Awaitable._ /** * Support for `Await.ready`. The use of the implicit permit is an * access control mechanism: only `Await.ready` may call this * method. */ @throws(classOf[TimeoutException]) @throws(classOf[InterruptedException]) def ready(timeout: Duration)(implicit permit: CanAwait): this.type /** * Support for `Await.result`. The use of the implicit permit is an * access control mechanism: only `Await.result` may call this * method. */ @throws(classOf[Exception]) def result(timeout: Duration)(implicit permit: CanAwait): T } object Awaitable { sealed trait CanAwait } /** * Await the result of some action. * * @define ready * * Returns the object when the action has completed. * * @define result * * Returns the result of the action when it has completed. */ object Await { import Awaitable._ private object AwaitPermit extends CanAwait /** $ready */ @throws(classOf[TimeoutException]) @throws(classOf[InterruptedException]) def ready[T <: Awaitable[_]](awaitable: T): T = ready(awaitable, Duration.Top) /** $ready */ @throws(classOf[TimeoutException]) @throws(classOf[InterruptedException]) def ready[T <: Awaitable[_]](awaitable: T, timeout: Duration): T = awaitable.ready(timeout)(AwaitPermit) /** $result */ @throws(classOf[Exception]) def result[T](awaitable: Awaitable[T]): T = result(awaitable, Duration.Top) /** $result */ @throws(classOf[Exception]) def result[T](awaitable: Awaitable[T], timeout: Duration): T = awaitable.result(timeout)(AwaitPermit) } /** * A mixin to make an [[com.twitter.util.Awaitable]] out * of a [[com.twitter.util.Closable]]. * * Use `closeAwaitably` in the definition of `close`: * * {{{ * class MyClosable extends Closable with CloseAwaitably { * def close(deadline: Time) = closeAwaitably { * // close the resource * } * } * }}} */ trait CloseAwaitably extends Awaitable[Unit] { private[this] val onClose = new Promise[Unit] private[this] val closed = new AtomicBoolean(false) /** * closeAwaitably is intended to be used as a wrapper for * `close`. The underlying `f` will be called at most once. */ protected def closeAwaitably(f: => Future[Unit]): Future[Unit] = { if (closed.compareAndSet(false, true)) onClose.become(f) onClose } def ready(timeout: Duration)(implicit permit: Awaitable.CanAwait): this.type = { Await.ready(onClose, timeout) this } def result(timeout: Duration)(implicit permit: Awaitable.CanAwait): Unit = Await.result(onClose, timeout) }
foursquare/util
project/Build.scala
<filename>project/Build.scala import sbt._ import Keys._ object Util extends Build { val zkVersion = "3.3.4" val sharedSettings = Seq( version := "6.3.0", organization := "com.twitter", crossScalaVersions := Seq("2.9.2", "2.10.0"), // Workaround for a scaladoc bug which causes it to choke on // empty classpaths. unmanagedClasspath in Compile += Attributed.blank(new java.io.File("doesnotexist")), libraryDependencies ++= Seq( "junit" % "junit" % "4.8.1" % "test" withSources(), "org.scalatest" %% "scalatest" %"1.9.1" % "test", "org.scala-tools.testing" %% "specs" % "1.6.9" % "test" withSources() cross CrossVersion.binaryMapped { case "2.9.2" => "2.9.1" case "2.10.0" => "2.10" case x => x }, "org.mockito" % "mockito-all" % "1.8.5" % "test" withSources() ), resolvers += "twitter repo" at "http://maven.twttr.com", ivyXML := <dependencies> <exclude org="com.sun.jmx" module="jmxri" /> <exclude org="com.sun.jdmk" module="jmxtools" /> <exclude org="javax.jms" module="jms" /> </dependencies>, scalacOptions ++= Seq("-encoding", "utf8"), scalacOptions += "-deprecation", javacOptions ++= Seq("-source", "1.6", "-target", "1.6"), javacOptions in doc := Seq("-source", "1.6"), // This is bad news for things like com.twitter.util.Time parallelExecution in Test := false, // Sonatype publishing publishArtifact in Test := false, pomIncludeRepository := { _ => false }, publishMavenStyle := true, pomExtra := ( <url>https://github.com/twitter/util</url> <licenses> <license> <name>Apache License, Version 2.0</name> <url>http://www.apache.org/licenses/LICENSE-2.0</url> </license> </licenses> <scm> <url><EMAIL>:twitter/util.git</url> <connection>scm:git:<EMAIL>:twitter/util.git</connection> </scm> <developers> <developer> <id>twitter</id> <name>Twitter Inc.</name> <url>https://www.twitter.com/</url> </developer> </developers>), publishTo <<= version { (v: String) => val nexus = "https://oss.sonatype.org/" if (v.trim.endsWith("SNAPSHOT")) Some("snapshots" at nexus + "content/repositories/snapshots") else Some("releases" at nexus + "service/local/staging/deploy/maven2") } ) val jmockSettings = Seq( libraryDependencies ++= Seq( "org.jmock" % "jmock" % "2.4.0" % "test", "cglib" % "cglib" % "2.1_3" % "test", "asm" % "asm" % "1.5.3" % "test", "org.objenesis" % "objenesis" % "1.1" % "test", "org.hamcrest" % "hamcrest-all" % "1.1" % "test" ) ) lazy val util = Project( id = "util", base = file("."), settings = Project.defaultSettings ++ sharedSettings ) aggregate( utilCore, utilEval, utilCodec, utilCollection, utilReflect, utilLogging, utilThrift, utilHashing, utilJvm, utilZk, utilZkCommon, utilClassPreloader, utilBenchmark, utilApp ) lazy val utilCore = Project( id = "util-core", base = file("util-core"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-core", libraryDependencies ++= Seq( "com.twitter.common" % "objectsize" % "0.0.7" % "test" ), testOptions in Test <<= scalaVersion map { // There seems to be an issue with mockito spies, // specs1, and scala 2.10 case "2.10" | "2.10.0" => Seq(Tests.Filter(s => !s.endsWith("MonitorSpec"))) case _ => Seq() } ) lazy val utilEval = Project( id = "util-eval", base = file("util-eval"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-eval", libraryDependencies <+= scalaVersion { "org.scala-lang" % "scala-compiler" % _ % "compile" } ).dependsOn(utilCore) lazy val utilCodec = Project( id = "util-codec", base = file("util-codec"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-codec", libraryDependencies ++= Seq( "commons-codec" % "commons-codec" % "1.5" ) ).dependsOn(utilCore) lazy val utilCollection = Project( id = "util-collection", base = file("util-collection"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-collection", libraryDependencies ++= Seq( "com.google.guava" % "guava" % "11.0.2", "commons-collections" % "commons-collections" % "3.2.1" ) ).dependsOn(utilCore) lazy val utilReflect = Project( id = "util-reflect", base = file("util-reflect"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-reflect", libraryDependencies ++= Seq( "asm" % "asm" % "3.3.1", "asm" % "asm-util" % "3.3.1", "asm" % "asm-commons" % "3.3.1", "cglib" % "cglib" % "2.2" ) ).dependsOn(utilCore) lazy val utilLogging = Project( id = "util-logging", base = file("util-logging"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-logging", libraryDependencies ++= Seq( "org.scala-tools.testing" % "specs" % "1.6.9" % "provided" cross CrossVersion.binaryMapped { case "2.9.2" => "2.9.1" case "2.10.0" => "2.10" case x => x } ) ).dependsOn(utilCore, utilApp) lazy val utilThrift = Project( id = "util-thrift", base = file("util-thrift"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-thrift", libraryDependencies ++= Seq( "thrift" % "libthrift" % "0.5.0", "org.slf4j" % "slf4j-nop" % "1.5.8" % "provided", "org.codehaus.jackson" % "jackson-core-asl" % "1.8.1", "org.codehaus.jackson" % "jackson-mapper-asl" % "1.8.1" ) ).dependsOn(utilCodec) lazy val utilHashing = Project( id = "util-hashing", base = file("util-hashing"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-hashing", libraryDependencies ++= Seq( "commons-codec" % "commons-codec" % "1.5" % "test" ) ).dependsOn(utilCore) lazy val utilJvm = Project( id = "util-jvm", base = file("util-jvm"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-jvm" ).dependsOn(utilCore, utilLogging % "test") lazy val utilZk = Project( id = "util-zk", base = file("util-zk"), settings = Project.defaultSettings ++ sharedSettings ++ jmockSettings ).settings( name := "util-zk", ivyXML := <dependencies> <exclude org="com.sun.jmx" module="jmxri" /> <exclude org="com.sun.jdmk" module="jmxtools" /> <exclude org="javax.jms" module="jms" /> <override org="junit" rev="4.8.1"/> </dependencies>, libraryDependencies ++= Seq( "org.apache.zookeeper" % "zookeeper" % zkVersion ) ).dependsOn(utilCore, utilCollection, utilLogging) lazy val utilZkCommon = Project( id = "util-zk-common", base = file("util-zk-common"), settings = Project.defaultSettings ++ sharedSettings ++ jmockSettings ).settings( name := "util-zk-common", libraryDependencies ++= Seq( "com.twitter.common.zookeeper" % "client" % "0.0.21", "com.twitter.common.zookeeper" % "group" % "0.0.14", "com.twitter.common.zookeeper" % "server-set" % "1.0.3", "org.apache.zookeeper" % "zookeeper" % zkVersion ) ).dependsOn(utilCore, utilLogging, utilZk, // These are dependended on to provide transitive dependencies // that would otherwise cause incompatibilities. See above comment. utilEval, utilCollection, utilHashing ) lazy val utilClassPreloader = Project( id = "util-class-preloader", base = file("util-class-preloader"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-class-preloader" ).dependsOn(utilCore) lazy val utilBenchmark = Project( id = "util-benchmark", base = file("util-benchmark"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-benchmark", libraryDependencies ++= Seq( "com.google.caliper" % "caliper" % "0.5-rc1" ) ).dependsOn(utilCore, utilJvm) lazy val utilApp = Project( id = "util-app", base = file("util-app"), settings = Project.defaultSettings ++ sharedSettings ).settings( name := "util-app" ).dependsOn(utilCore) }
foursquare/util
util-jvm/src/main/scala/com/twitter/jvm/Estimator.scala
<reponame>foursquare/util package com.twitter.jvm import scala.util.Random /** * An estimator for values of type T. */ trait Estimator[T] { /** A scalar measurement `m` was taken */ def measure(m: T) /** Estimate the current value */ def estimate: T } /** * A simple Kalman filter to estimate a scalar value. */ class Kalman(N: Int) { private[this] val mbuf = new Array[Double](N) private[this] val ebuf = new Array[Double](N) private[this] var est: Double = _ private[this] var n = 0L private[this] var weight: Double = 0.9 /** * Update the filter with measurement `m` * and measurement error `e`. */ def measure(m: Double, e: Double) { val i = (n%N).toInt mbuf(i) = m ebuf(i) = e if (n == 0) est = m est += weight*(m-est) val mv = mvar val ev = evar if (mv + ev == 0) weight = 1D else weight = mv / (mv+ev) n += 1 } private[this] def mvar = variance( if (n < N) mbuf take n.toInt else mbuf ) private[this] def evar = variance( if (n < N) ebuf take n.toInt else ebuf ) def estimate = est private[this] def variance(samples: Array[Double]): Double = { if (samples.size == 1) return 0D val sum = samples.sum val mean = sum / samples.size val diff = (samples map { x => (x-mean)*(x-mean) }).sum diff/(samples.size-1) } override def toString = "Kalman<estimate=%f, weight=%f, mvar=%f, evar=%f>".format(estimate, weight, mvar, evar) } /** * A Kalman filter in which measurement errors are normally * distributed over the given range (as a fraction of the measured * value). */ class KalmanGaussianError(N: Int, range: Double) extends Kalman(N) with Estimator[Double] { require(range >= 0D && range <1D) private[this] val rng = new Random def measure(m: Double) { measure(m, rng.nextGaussian()*range*m) } } /** * An estimator for weighted windows of means. */ class WindowedMeans(N: Int, windows: Seq[(Int, Int)]) extends Estimator[Double] { require(windows forall { case (_, i) => i <= N }) private[this] val normalized = { val sum = (windows map { case (w, _) => w }).sum windows map { case (w, i) => (w.toDouble / sum, i) } } private[this] val buf = new Array[Double](N) private[this] var n = 0L private[this] def mean(from: Long, count: Int): Double = { require(count <= N && count > 0) val i = { val x = ((from-count)%N).toInt if (x < 0) (x + N) else x } val j = (from%N).toInt val sum = if (i == j) buf.sum else if (i < j) buf.slice(i, j).sum else buf.slice(i, N).sum + buf.slice(0, j).sum sum/count } def measure(m: Double) { if (n == 0) java.util.Arrays.fill(buf, m) else buf((n%N).toInt) = m n += 1 } def estimate = { require(n > 0) val weightedMeans = normalized map { case (w, i) => w*mean(n, i) } weightedMeans.sum } } /** * Unix-like load average, an exponentially weighted moving average, * smoothed to the given interval (counted in number of * measurements). See: * http://web.mit.edu/saltzer/www/publications/instrumentation.html */ class LoadAverage(interval: Double) extends Estimator[Double] { private[this] val a = math.exp(-1D/interval) private[this] var load = Double.NaN private[this] var first = true def measure(m: Double) { load = if (load.isNaN) m else load*a + m*(1-a) } def estimate = load } /** * Take a GC log produced by: * * {{{ * $ jstat -gc $PID 250 ... * }}} * * And report on GC prediction accuracy. Time is * indexed by the jstat output, and the columns are, * in order: current time, next gc, estimated next GC. */ object EstimatorTest extends App { import com.twitter.conversions.storage._ val estimator = args match { case Array("kalman", n, error) => new KalmanGaussianError(n.toInt, error.toDouble) case Array("windowed", n, windows) => new WindowedMeans(n.toInt, windows.split(",") map { w => w.split(":") match { case Array(w, i) => (w.toInt, i.toInt) case _ => throw new IllegalArgumentException("bad weight, count pair "+w) } } ) case Array("load", interval) => new LoadAverage(interval.toDouble) case _ => throw new IllegalArgumentException("bad args ") } val lines = scala.io.Source.stdin.getLines().drop(1) val states = lines.toArray map(_.split(" ") filter(_ != "") map(_.toDouble)) collect { case Array(s0c, s1c, s0u, s1u, ec, eu, oc, ou, pc, pu, ygc, ygct, fgc, fgct, gct) => PoolState(ygc.toLong, ec.toLong.bytes, eu.toLong.bytes) } var elapsed = 1 for (List(begin, end) <- states.toList.sliding(2)) { val allocated = (end - begin).used estimator.measure(allocated.inBytes) val r = end.capacity - end.used val i = (r.inBytes/estimator.estimate.toLong) + elapsed val j = states.indexWhere(_.numCollections > end.numCollections) if (j > 0) println("%d %d %d".format(elapsed, j, i)) elapsed += 1 } } /* The following script is useful for plotting results from EstimatorTest: % scala ... com.twitter.jvm.EstimatorTest [ARGS] > /tmp/out #!/usr/bin/env gnuplot set terminal png size 800,600 set title "GC predictor" set macros set grid set timestamp "Generated on %Y-%m-%d by `whoami`" font "Helvetica-Oblique, 8pt" set noclip set xrange [0:1100] set key reverse left Left # box set ylabel "Time of GC" textcolor lt 3 set yrange [0:1100] set mytics 5 set y2tics set xlabel "Time" textcolor lt 4 set mxtics 5 set boxwidth 0.5 set style fill transparent pattern 4 bo plot "< awk '{print $1 \" \" $2}' /tmp/out" title "Actual", \ "< awk '{print $1 \" \" $3}' /tmp/out" title "Predicted", \ "< awk '{print $1 \" \" $1}' /tmp/out" title "time" with lines */
foursquare/util
util-logging/src/main/scala/com/twitter/logging/SyslogHandler.scala
/* * Copyright 2010 Twitter, Inc. * * Licensed under the Apache License, Version 2.0 (the "License"); you may * not use this file except in compliance with the License. You may obtain * a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package com.twitter.logging import com.twitter.concurrent.NamedPoolThreadFactory import com.twitter.conversions.string._ import java.net.{DatagramPacket, DatagramSocket, InetAddress, InetSocketAddress, SocketAddress} import java.text.SimpleDateFormat import java.util.concurrent.Executors import java.util.{logging => javalog} object SyslogHandler { val DEFAULT_PORT = 514 val PRIORITY_USER = 8 val PRIORITY_DAEMON = 24 val PRIORITY_LOCAL0 = 128 val PRIORITY_LOCAL1 = 136 val PRIORITY_LOCAL2 = 144 val PRIORITY_LOCAL3 = 152 val PRIORITY_LOCAL4 = 160 val PRIORITY_LOCAL5 = 168 val PRIORITY_LOCAL6 = 176 val PRIORITY_LOCAL7 = 184 private val SEVERITY_EMERGENCY = 0 private val SEVERITY_ALERT = 1 private val SEVERITY_CRITICAL = 2 private val SEVERITY_ERROR = 3 private val SEVERITY_WARNING = 4 private val SEVERITY_NOTICE = 5 private val SEVERITY_INFO = 6 private val SEVERITY_DEBUG = 7 /** * Convert the java/scala log level into its closest syslog-ng severity. */ private[logging] def severityForLogLevel(level: Int): Int = { if (level >= Level.FATAL.value) { SEVERITY_ALERT } else if (level >= Level.CRITICAL.value) { SEVERITY_CRITICAL } else if (level >= Level.ERROR.value) { SEVERITY_ERROR } else if (level >= Level.WARNING.value) { SEVERITY_WARNING } else if (level >= Level.INFO.value) { SEVERITY_INFO } else { SEVERITY_DEBUG } } val ISO_DATE_FORMAT = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss") val OLD_SYSLOG_DATE_FORMAT = new SimpleDateFormat("MMM dd HH:mm:ss") /** * Generates a HandlerFactory that returns a SyslogHandler. * * @param server * Syslog server hostname. * * @param port * Syslog server port. */ def apply( server: String = "localhost", port: Int = SyslogHandler.DEFAULT_PORT, formatter: Formatter = new Formatter(), level: Option[Level] = None ) = () => new SyslogHandler(server, port, formatter, level) } class SyslogHandler( val server: String, val port: Int, formatter: Formatter, level: Option[Level]) extends Handler(formatter, level) { private val socket = new DatagramSocket private[logging] val dest = new InetSocketAddress(server, port) def flush() = { } def close() = { } def publish(record: javalog.LogRecord) = { val data = formatter.format(record).getBytes val packet = new DatagramPacket(data, data.length, dest) SyslogFuture { try { socket.send(packet) } catch { case e => System.err.println(Formatter.formatStackTrace(e, 30).mkString("\n")) } } } } /** * @param hostname * Hostname to prepend to log lines. * * @param serverName * Optional server name to insert before log entries. * * @param useIsoDateFormat * Use new standard ISO-format timestamps instead of old BSD-format? * * @param priority * Priority level in syslog numbers. * * @param timezone * Should dates in log messages be reported in a different time zone rather than local time? * If set, the time zone name must be one known by the java `TimeZone` class. * * @param truncateAt * Truncate log messages after N characters. 0 = don't truncate (the default). * * @param truncateStackTracesAt * Truncate stack traces in exception logging (line count). */ class SyslogFormatter( val hostname: String = InetAddress.getLocalHost().getHostName(), val serverName: Option[String] = None, val useIsoDateFormat: Boolean = true, val priority: Int = SyslogHandler.PRIORITY_USER, timezone: Option[String] = None, truncateAt: Int = 0, truncateStackTracesAt: Int = Formatter.DefaultStackTraceSizeLimit) extends Formatter( timezone, truncateAt, truncateStackTracesAt, useFullPackageNames = false, prefix = "") { override def dateFormat = if (useIsoDateFormat) { SyslogHandler.ISO_DATE_FORMAT } else { SyslogHandler.OLD_SYSLOG_DATE_FORMAT } override def lineTerminator = "" override def formatPrefix(level: javalog.Level, date: String, name: String): String = { val syslogLevel = level match { case x: Level => SyslogHandler.severityForLogLevel(x.value) case x: javalog.Level => SyslogHandler.severityForLogLevel(x.intValue) } serverName match { case None => "<%d>%s %s %s: ".format(priority | syslogLevel, date, hostname, name) case Some(serverName) => "<%d>%s %s [%s] %s: ".format(priority | syslogLevel, date, hostname, serverName, name) } } } object SyslogFuture { private val executor = Executors.newSingleThreadExecutor( new NamedPoolThreadFactory("TWITTER-UTIL-SYSLOG", true/*daemon*/)) private val noop = new Runnable { def run() {} } def apply(action: => Unit) = executor.submit(new Runnable { def run() { action } }) def sync() { val f = executor.submit(noop) f.get() } }
foursquare/util
util-core/src/test/scala/com/twitter/concurrent/BrokerSpec.scala
package com.twitter.concurrent import org.specs.SpecificationWithJUnit import org.specs.mock.Mockito import org.mockito.{Matchers, ArgumentCaptor} import com.twitter.util.{Await, Return} import com.twitter.common.objectsize.ObjectSizeCalculator class BrokerSpec extends SpecificationWithJUnit with Mockito { "Broker" should { "send data (send, recv)" in { val br = new Broker[Int] val sendF = br.send(123).sync() sendF.isDefined must beFalse val recvF = br.recv.sync() recvF.isDefined must beTrue Await.result(recvF) must be_==(123) sendF.isDefined must beTrue } "send data (recv, send)" in { val br = new Broker[Int] val recvF = br.recv.sync() recvF.isDefined must beFalse val sendF = br.send(123).sync() sendF.isDefined must beTrue recvF.isDefined must beTrue Await.result(recvF) must be_==(123) } "queue receivers (recv, recv, send, send)" in { val br = new Broker[Int] val r0, r1 = br.recv.sync() r0.isDefined must beFalse r1.isDefined must beFalse val s = br.send(123) s.sync().poll must beSome(Return(())) r0.poll must beSome(Return(123)) r1.isDefined must beFalse s.sync().poll must beSome(Return(())) r1.poll must beSome(Return(123)) s.sync().isDefined must beFalse } "queue senders (send, send, recv, recv)" in { val br = new Broker[Int] val s0, s1 = br.send(123).sync() s0.isDefined must beFalse s1.isDefined must beFalse val r = br.recv r.sync().poll must beSome(Return(123)) s0.poll must beSome(Return(())) s1.isDefined must beFalse r.sync().poll must beSome(Return(123)) s1.poll must beSome(Return(())) r.sync().isDefined must beFalse } "interrupts" in { "removes queued receiver" in { val br = new Broker[Int] val recvF = br.recv.sync() recvF.raise(new Exception) br.send(123).sync().poll must beNone recvF.poll must beNone } "removes queued sender" in { val br = new Broker[Int] val sendF = br.send(123).sync() sendF.raise(new Exception) br.recv.sync().poll must beNone sendF.poll must beNone } "doesn't result in space leaks" in { val br = new Broker[Int] Offer.select(Offer.const(1), br.recv).poll must beSome(Return(1)) val initial = ObjectSizeCalculator.getObjectSize(br) for (_ <- 0 until 1000) { Offer.select(Offer.const(1), br.recv).poll must beSome(Return(1)) ObjectSizeCalculator.getObjectSize(br) must be_==(initial) } } "works with orElse" in { val b0, b1 = new Broker[Int] val o = b0.recv orElse b1.recv val f = o.sync() f.isDefined must beFalse val sendf0 = b0.send(12).sync() sendf0.isDefined must beFalse val sendf1 = b1.send(32).sync() sendf1.isDefined must beTrue f.poll must beSome(Return(32)) o.sync().poll must beSome(Return(12)) sendf0.poll must beSome(Return(())) } } "integrate" in { val br = new Broker[Int] val offer = Offer.choose(br.recv, Offer.const(999)) offer.sync().poll must beSome(Return(999)) val item = br.recv.sync() item.isDefined must beFalse br.send(123).sync().poll must beSome(Return(())) item.poll must beSome(Return(123)) } } }
scarraway/scala-item-microservice
src/main/scala/AkkaHttpMicroservice.scala
import akka.actor.ActorSystem import akka.http.scaladsl.server.{Directives, PathMatchers, Route, RouteResult} import akka.event.{Logging, LoggingAdapter} import akka.http.scaladsl.Http import akka.stream.{ActorMaterializer, Materializer} import akka.http.scaladsl.marshallers.sprayjson.SprayJsonSupport import akka.http.scaladsl.model.StatusCodes import ch.megard.akka.http.cors.scaladsl.settings.CorsSettings import ch.megard.akka.http.cors.scaladsl.CorsDirectives.cors import com.mongodb.client.model.{Filters, UpdateOptions} import com.typesafe.config.Config import com.typesafe.config.ConfigFactory import org.mongodb.scala.{Completed, MongoClient, MongoCollection, Observer} import org.mongodb.scala.bson.codecs.Macros._ import org.mongodb.scala.bson.codecs.DEFAULT_CODEC_REGISTRY import org.bson.codecs.configuration.CodecRegistries.{fromProviders, fromRegistries} import org.bson.types.ObjectId import org.mongodb.scala.model.Updates import scala.concurrent.{ExecutionContext, ExecutionContextExecutor, Future} import spray.json.DefaultJsonProtocol import scala.util.{Failure, Success, Try} object Item { def apply(name: String): Item = Item(Some(new ObjectId().toString), name) } final case class Item(_id: Option[String] = None, name: String) final case class ItemNameUpdate(name: String) trait JsonSupport extends SprayJsonSupport with DefaultJsonProtocol { implicit val itemFormat = jsonFormat(Item.apply, "_id", "name") implicit val itemNameUpdateFormat = jsonFormat1(ItemNameUpdate) } trait Service extends Directives with JsonSupport { implicit val system: ActorSystem implicit def executor: ExecutionContextExecutor implicit val materializer: Materializer // Use a Connection String val mongoClient: MongoClient = MongoClient()//"mongodb://mongo" def config: Config val logger: LoggingAdapter val settings = CorsSettings.defaultSettings.copy( ) val routes = cors(settings) { logRequestResult("akka-http-microservice") { path("items") { get { completeWith(instanceOf[Seq[Item]]) { c => getItemsCollection.find().collect().head().map(a=>a) onComplete { case Success(result) => { c(result) }// use result for something case Failure(t) => t.printStackTrace() } } } ~ post { decodeRequest { entity(as[List[Item]]){ itemList => completeWith(instanceOf[String]) { c => getItemsCollection.insertMany(addMissingIdsToItemList(itemList)).toFuture().map(a=>a) onComplete { case Success(result) => { c(result.toString()) }// use result for something case Failure(t) => t.printStackTrace() } } } } } } ~ pathPrefix("items" / PathMatchers.RemainingPath) { id => val possibleItem = getItem(id.toString()) onSuccess(possibleItem) { case Some(item) => specificItemRoutes(item) case None => complete(StatusCodes.NotFound) } } } } private def specificItemRoutes(item: Item): Route = { get { complete(item) } ~ delete { val possibleDeletion:Future[Item] = deleteItem(item._id.get) onComplete(possibleDeletion){ deletionResult => complete(deletionResult) } } ~ put { decodeRequest { entity(as[ItemNameUpdate]) { itemNameUpdate => val possibleUpdate:Future[Item] = updateItem(item._id.get, itemNameUpdate.name) onComplete(possibleUpdate){ updateResult => complete(updateResult) } } } } } private def addMissingIdsToItemList(itemList: List[Item]) = { itemList.map(i => if (i._id == None) Item.apply(i.name) else i) } private def getItemsCollection:MongoCollection[Item] = { val codecRegistry = fromRegistries(fromProviders(classOf[Item]), DEFAULT_CODEC_REGISTRY ) mongoClient.getDatabase("item").withCodecRegistry(codecRegistry).getCollection("items") } private def getItem(id: String):Future[Option[Item]] = { getItemsCollection.find(Filters.eq("_id", id)) .toFuture() .recoverWith { case e: Throwable => { Future.failed(e) } } .map(_.headOption) } private def deleteItem(id: String):Future[Item] = { getItemsCollection.findOneAndDelete(Filters.eq("_id", id)) .toFuture() .recoverWith { case e: Throwable => { Future.failed(e) } } } private def updateItem(id: String, newName: String):Future[Item] = { getItemsCollection.findOneAndUpdate(Filters.eq("_id", id), Updates.set("name", newName)) .toFuture() .recoverWith { case e: Throwable => { Future.failed(e) } } } } object AkkaHttpMicroservice extends App with Service { override implicit val system = ActorSystem() override implicit val executor = system.dispatcher override implicit val materializer = ActorMaterializer() override val config = ConfigFactory.load() override val logger = Logging(system, getClass) Http().bindAndHandle(routes, config.getString("http.interface"), config.getInt("http.port")) }
scarraway/scala-item-microservice
src/test/scala/ServiceSpec.scala
import akka.event.NoLogging import akka.http.scaladsl.model.ContentTypes._ import akka.http.scaladsl.model.StatusCodes._ import akka.http.scaladsl.testkit.ScalatestRouteTest import org.scalatest._ class ServiceSpec extends FlatSpec with Matchers with ScalatestRouteTest with Service { override def testConfigSource = "akka.loglevel = WARNING" override def config = testConfig override val logger = NoLogging val baseRoute = "/items" val testIdentifier = "testId" val testItemRoute = baseRoute + "/" + testIdentifier val invalidItemRoute = baseRoute + "/" + "invalid" val testItem = new Item(Some(testIdentifier), "TestItem") val testItemUpdate = new ItemNameUpdate("Test Item Updated") override def afterAll(): Unit ={ Delete(testItemRoute) ~> routes ~> check { println("Double checked deletion of " + testIdentifier) } } //Must be first since these tests are stateful "The service" should "allow inserting new Items via POST" in { Post(baseRoute, List(testItem)) ~> routes ~> check { status shouldBe OK } } "The service" should "return all items for GET requests to the root path" in { Get(baseRoute) ~> routes ~> check { status shouldBe OK contentType shouldBe `application/json` responseAs[Seq[Item]] shouldBe a [Seq[_]] responseAs[Seq[Item]] contains testItem } } "The service" should "return 200 for GET test item" in { Get(testItemRoute) ~> routes ~> check { status shouldBe OK } } "The service" should "return 200 for updating(PUT) test item" in { Put(testItemRoute, testItemUpdate) ~> routes ~> check { status shouldBe OK } } "The item" should "be updated" in { Get(testItemRoute) ~> routes ~> check { status shouldBe OK responseAs[Item] should have ( 'name (testItemUpdate.name) ) } } "The service" should "return 200 for DELETE test item" in { Delete(testItemRoute) ~> routes ~> check { status shouldBe OK } } "The service" should "return 404 for GET nonsense id" in { Get(invalidItemRoute) ~> routes ~> check { status shouldBe NotFound } } "The service" should "return 404 for PUT nonsense id" in { Put(invalidItemRoute) ~> routes ~> check { status shouldBe NotFound } } "The service" should "return 404 for DELETE nonsense id" in { Delete(invalidItemRoute) ~> routes ~> check { status shouldBe NotFound } } }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/Commands.scala
<gh_stars>1-10 package com.wix.pay.creditguard.model object Commands { /** The doDeal command is used to process transactions in the CG Gateway. */ val doDeal = "doDeal" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/TransactionCodes.scala
package com.wix.pay.creditguard.model object TransactionCodes { /** Swiping of magnetic card. */ val regular = "Regular" /** Self service. */ val selfService = "SelfService" /** Fuel self service. */ val fuelSelfService = "FuelSelfService" val contactless = "Contactless" val contactlessSelfService = "ContactlessSelfService" val contactlessSelfServiceInGasStation = "ContactlessSelfServiceInGasStation" /** Transaction through Internet/phone with card number. */ val phone = "Phone" /** Card holder is present, however card is not swiped. */ val signature = "Signature" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/Currencies.scala
<filename>libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/Currencies.scala package com.wix.pay.creditguard.model object Currencies { /** New Israeli Shekel USD linked. */ val ilsByUsd = "IlsByUsd" /** New Israeli Shekel index linked. */ val ilsbyIndex = "IlsbyIndex" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/Conversions.scala
package com.wix.pay.creditguard.model import java.math.{BigDecimal => JBigDecimal} object Conversions { def toCreditguardAmount(amount: Double): Int = { JBigDecimal.valueOf(amount).movePointRight(2).intValueExact() } def toCreditguardYearMonth(year: Int, month: Int): String = { f"$month%02d${year % 100}%02d" } }
wix/libpay-creditguard
libpay-creditguard/src/main/scala/com/wix/pay/creditguard/CreditguardMerchantParser.scala
<filename>libpay-creditguard/src/main/scala/com/wix/pay/creditguard/CreditguardMerchantParser.scala package com.wix.pay.creditguard trait CreditguardMerchantParser { def parse(merchantKey: String): CreditguardMerchant def stringify(merchant: CreditguardMerchant): String }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/RequestTransactionTypes.scala
<reponame>wix/libpay-creditguard package com.wix.pay.creditguard.model object RequestTransactionTypes { /** Card holder is charged. */ val debit = "Debit" /** Card holder is credited. */ val credit = "Credit" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/Versions.scala
package com.wix.pay.creditguard.model object Versions { val standard = "1001" }
wix/libpay-creditguard
libpay-creditguard/src/test/scala/com/wix/pay/creditguard/JsonCreditguardAuthorizationParserTest.scala
package com.wix.pay.creditguard import com.wix.pay.creditguard.CreditguardMatchers._ import org.specs2.mutable.SpecWithJUnit import org.specs2.specification.Scope class JsonCreditguardAuthorizationParserTest extends SpecWithJUnit { trait Ctx extends Scope { val authorizationParser: CreditguardAuthorizationParser = new JsonCreditguardAuthorizationParser } "stringify and then parse" should { "yield an authorization similar to the original one" in new Ctx { val someAuthorization = CreditguardAuthorization( authNumber = "some authorization number", currency = "some currency", tranId = "some transaction ID", cardId = "some card ID", cardExpiration = "some card expiration", user = "some user" ) val authorizationKey = authorizationParser.stringify(someAuthorization) authorizationParser.parse(authorizationKey) must beAuthorization( authNumber = ===(someAuthorization.authNumber), currency = ===(someAuthorization.currency), tranId = ===(someAuthorization.tranId), cardId = ===(someAuthorization.cardId), cardExpiration = ===(someAuthorization.cardExpiration), user = ===(someAuthorization.user) ) } } }
wix/libpay-creditguard
libpay-creditguard/src/test/scala/com/wix/pay/creditguard/JsonCreditguardMerchantParserTest.scala
<reponame>wix/libpay-creditguard package com.wix.pay.creditguard import com.wix.pay.creditguard.CreditguardMatchers._ import org.specs2.mutable.SpecWithJUnit import org.specs2.specification.Scope class JsonCreditguardMerchantParserTest extends SpecWithJUnit { trait Ctx extends Scope { val merchantParser: CreditguardMerchantParser = new JsonCreditguardMerchantParser } "stringify and then parse" should { "yield a merchant similar to the original one" in new Ctx { val someMerchant = CreditguardMerchant( user = "some user", password = "<PASSWORD>", terminalNumber = "some terminal number", supplierNumber = "some supplier number", idPrefix = "some ID prefix" ) val merchantKey = merchantParser.stringify(someMerchant) merchantParser.parse(merchantKey) must beMerchant( user = ===(someMerchant.user), password = === (<PASSWORD>), terminalNumber = ===(someMerchant.terminalNumber), supplierNumber = ===(someMerchant.supplierNumber), idPrefix = ===(someMerchant.idPrefix) ) } } }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/Cvvs.scala
package com.wix.pay.creditguard.model object Cvvs { /** Merchant chooses not to pass CVV. */ val notProvided = "0" /** CVV is not readable. */ val unreadable = "2" /** card does not have CVV. */ val doesNotExist = "9" }
wix/libpay-creditguard
libpay-creditguard/src/main/scala/com/wix/pay/creditguard/CreditguardMerchant.scala
package com.wix.pay.creditguard /** * @param idPrefix Prefix for ''com.wix.pay.creditguard.model.DoDealRequest#user'' field */ case class CreditguardMerchant(user: String, password: String, terminalNumber: String, supplierNumber: String, idPrefix: String)
wix/libpay-creditguard
libpay-creditguard/src/main/scala/com/wix/pay/creditguard/CreditguardAuthorizationParser.scala
package com.wix.pay.creditguard trait CreditguardAuthorizationParser { def parse(authorizationKey: String): CreditguardAuthorization def stringify(authorization: CreditguardAuthorization): String }
wix/libpay-creditguard
libpay-creditguard/src/main/scala/com/wix/pay/creditguard/CreditguardAuthorization.scala
package com.wix.pay.creditguard case class CreditguardAuthorization(authNumber: String, currency: String, tranId: String, cardId: String, cardExpiration: String, user: String)
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/AuthSources.scala
<filename>libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/AuthSources.scala package com.wix.pay.creditguard.model object AuthSources { val shva = "Shva" val creditCompany = "Credit Company" val voiceMail = "VoiceMail" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/Validations.scala
package com.wix.pay.creditguard.model object Validations { /** * J1: Verifies card locally. * If the card is ok and the total amount of the deal is under the ceiling, a debit is made without communication to Shva. * If it's above the ceiling, an error occurs. */ val noComm = "NoComm" /** * J2: A local check on the CG Gateway for the validity of the credit card number and if it exist in the blocked cards list. * No actual debit occurs. */ val normal = "Normal" /** * J3: Same as J2 (Normal). * It also returns ceiling limit in the total field. for Israeli cards Only * A positive response results in actual settlement. */ val creditLimit = "CreditLimit" /** * J4: Verifies card locally or in credit company; depends on ceiling ZFL terminal parameters * A positive response results in actual settlement. */ val autoComm = "AutoComm" /** * J5: Verifies card by credit company regardless of the ceiling ZFL terminal parameters. * No settlement is performed; the amount of verify without settlement is held in card holder's obligor. * (This is used for authorization purposes only.) */ val verify = "Verify" /** * J6: Verifies card by credit company regardless of the ceiling ZFL terminal parameters; settlement is performed. */ val dealer = "Dealer" /** * J9: Performs a J4 transaction. Yet the transaction will not be deposited. * The method of depositing the transactions can be configured per merchant or by releasing the transaction with AutoCommRelease command. */ val autoCommHold = "AutoCommHold" /** * J102: A local check on the CG Gateway for the validity of the credit card number for tokenization purposes. * Perform an actual J2 request and return cardId when terminal is configured to do so. */ val token = "Token" /** * J109: Used for releasing a transaction (previously performed by using J9). * Realeasing a transaction can be done by using the original card number, the cardId (when supported on the terminal) * or track2 when the original transaction was performed with track2. */ val autoCommRelease = "AutoCommRelease" /** * J201: Used for retrieving card number of an existing card id that was generated for the merchant. * This option is configuration dependent. */ val cardNo = "cardNo" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/Languages.scala
<gh_stars>1-10 package com.wix.pay.creditguard.model object Languages { val hebrew = "Heb" val english = "Eng" }
wix/libpay-creditguard
libpay-creditguard/src/main/scala/com/wix/pay/creditguard/JsonCreditguardMerchantParser.scala
package com.wix.pay.creditguard import org.json4s.DefaultFormats import org.json4s.native.Serialization class JsonCreditguardMerchantParser() extends CreditguardMerchantParser { private implicit val formats = DefaultFormats override def parse(merchantKey: String): CreditguardMerchant = { Serialization.read[CreditguardMerchant](merchantKey) } override def stringify(merchant: CreditguardMerchant): String = { Serialization.write(merchant) } }
wix/libpay-creditguard
libpay-creditguard/src/test/scala/com/wix/pay/creditguard/CreditguardMatchers.scala
package com.wix.pay.creditguard import org.specs2.matcher.{AlwaysMatcher, Matcher, Matchers} trait CreditguardMatchers extends Matchers { def authorizationParser: CreditguardAuthorizationParser def beMerchant(user: Matcher[String] = AlwaysMatcher(), password: Matcher[String] = AlwaysMatcher(), terminalNumber: Matcher[String] = AlwaysMatcher(), supplierNumber: Matcher[String] = AlwaysMatcher(), idPrefix: Matcher[String] = AlwaysMatcher()): Matcher[CreditguardMerchant] = { user ^^ { (_: CreditguardMerchant).user aka "user" } and password ^^ { (_: CreditguardMerchant).password aka "password" } and terminalNumber ^^ { (_: CreditguardMerchant).terminalNumber aka "terminal number" } and supplierNumber ^^ { (_: CreditguardMerchant).supplierNumber aka "supplier number" } and idPrefix ^^ { (_: CreditguardMerchant).idPrefix aka "ID prefix" } } def beAuthorization(authNumber: Matcher[String] = AlwaysMatcher(), currency: Matcher[String] = AlwaysMatcher(), tranId: Matcher[String] = AlwaysMatcher(), cardId: Matcher[String] = AlwaysMatcher(), cardExpiration: Matcher[String] = AlwaysMatcher(), user: Matcher[String] = AlwaysMatcher()): Matcher[CreditguardAuthorization] = { authNumber ^^ { (_: CreditguardAuthorization).authNumber aka "authorization number" } and currency ^^ { (_: CreditguardAuthorization).currency aka "currency" } and tranId ^^ { (_: CreditguardAuthorization).tranId aka "transaction ID" } and cardId ^^ { (_: CreditguardAuthorization).cardId aka "card ID" } and cardExpiration ^^ { (_: CreditguardAuthorization).cardExpiration aka "card expiration" } and user ^^ { (_: CreditguardAuthorization).user aka "user" } } def beAuthorizationKey(authorization: Matcher[CreditguardAuthorization]): Matcher[String] = { authorization ^^ { authorizationParser.parse(_: String) aka "parsed authorization"} } } object CreditguardMatchers extends CreditguardMatchers { override val authorizationParser = new JsonCreditguardAuthorizationParser() }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/Fields.scala
package com.wix.pay.creditguard.model object Fields { val user = "user" val password = "password" val int_in = "int_in" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/CommReasons.scala
package com.wix.pay.creditguard.model object CommReasons { val random = "1" val creditLimit = "2" val confidentialNumer = "3" // sic. val serviceCode = "4" val verifyOnly = "5" val ambiguousBlocked = "6" val zfl = "7" val initiative = "8" val charging = "9" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/ResponseTransactionTypes.scala
<reponame>wix/libpay-creditguard package com.wix.pay.creditguard.model object ResponseTransactionTypes { /** Card holder is charged. */ val blocked = "Blocked" /** Card holder is charged. */ val regularDebit = "RegularDebit" /** Card holder is charged. */ val authDebit = "AuthDebit" /** Card holder is charged. */ val forcedDebit = "ForcedDebit" /** Card holder is credited. */ val regularCredit = "RegularCredit" /** Card holder is credited. */ val refund = "Refund" /** Card holder is credited. */ val authCredit = "AuthCredit" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/CardBrands.scala
package com.wix.pay.creditguard.model object CardBrands { val privateLabel = "0" val mastercard = "1" val visa = "2" val maestro = "3" val isracard = "5" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/CardAcquirers.scala
package com.wix.pay.creditguard.model object CardAcquirers { val isracard = "Isracard" val visa = "Visa" val diners = "Diners" val amex = "Amex" val jcb = "Jcb" val alphacard = "Alphacard" }
wix/libpay-creditguard
libpay-creditguard-common/src/main/scala/com/wix/pay/creditguard/model/Statuses.scala
<gh_stars>1-10 package com.wix.pay.creditguard.model object Statuses { val absent = "Absent" val valid = "Valid" val invalid = "Invalid" val notValidated = "NotValidated" }
wix/libpay-creditguard
libpay-creditguard/src/test/scala/com/wix/pay/creditguard/it/CreditguardGatewayIT.scala
<filename>libpay-creditguard/src/test/scala/com/wix/pay/creditguard/it/CreditguardGatewayIT.scala package com.wix.pay.creditguard.it import com.google.api.client.http.javanet.NetHttpTransport import com.wix.pay.creditcard.{CreditCard, CreditCardOptionalFields, YearMonth} import com.wix.pay.creditguard.CreditguardMatchers._ import com.wix.pay.creditguard.testkit.CreditguardDriver import com.wix.pay.creditguard.{CreditguardAuthorization, CreditguardMerchant, _} import com.wix.pay.model.{CurrencyAmount, Deal, Payment} import com.wix.pay.{PaymentErrorException, PaymentGateway, PaymentRejectedException} import org.specs2.mutable.SpecWithJUnit import org.specs2.specification.Scope class CreditguardGatewayIT extends SpecWithJUnit { val creditguardPort = 10010 val requestFactory = new NetHttpTransport().createRequestFactory() val driver = new CreditguardDriver(port = creditguardPort) step { driver.start() } sequential trait Ctx extends Scope { val merchantParser = new JsonCreditguardMerchantParser() val authorizationParser = new JsonCreditguardAuthorizationParser() val someMerchant = CreditguardMerchant( user = "some user", password = "<PASSWORD>", terminalNumber = "some terminal number", supplierNumber = "some supplier number", idPrefix = "some ID prefix" ) val merchantKey = merchantParser.stringify(someMerchant) val somePayment = Payment(currencyAmount = CurrencyAmount("ILS", 33.3)) val someCreditCard = CreditCard( number = "4012888818888", expiration = YearMonth(2020, 12), additionalFields = Some(CreditCardOptionalFields.withFields( csc = Some("123"), holderId = Some("some holder ID")) ) ) val someDeal = Deal(id = "some deal ID") val someAuthorization = CreditguardAuthorization( authNumber = "someAuthorizationNumber", currency = "someCurrency", tranId = "someTransactionId", cardId = "someCardId", cardExpiration = "someCardExpiration", user = "someUser" ) val authorizationKey = authorizationParser.stringify(someAuthorization) val someCaptureAmount = 11.1 val creditguard: PaymentGateway = new CreditguardGateway( requestFactory = requestFactory, endpointUrl = s"http://localhost:$creditguardPort/", merchantParser = merchantParser, authorizationParser = authorizationParser ) driver.reset() } "sale request via CreditGuard gateway" should { "gracefully fail on invalid merchant" in new Ctx { val someErrorMessage = "some error message" driver.aSaleFor( user = someMerchant.user, password = <PASSWORD>, terminalNumber = someMerchant.terminalNumber, supplierNumber = someMerchant.supplierNumber, idPrefix = someMerchant.idPrefix, orderId = Some(someDeal.id), card = someCreditCard, currencyAmount = somePayment.currencyAmount ) failsOnInvalidMerchant(someErrorMessage) creditguard.sale( merchantKey = merchantKey, creditCard = someCreditCard, payment = somePayment, deal = Some(someDeal) ) must beAFailedTry.like { case e: PaymentErrorException => e.message must contain(someErrorMessage) } } "gracefully fail on rejected card" in new Ctx { val someErrorMessage = "some error message" driver.aSaleFor( user = someMerchant.user, password = <PASSWORD>, terminalNumber = someMerchant.terminalNumber, supplierNumber = someMerchant.supplierNumber, idPrefix = someMerchant.idPrefix, orderId = Some(someDeal.id), card = someCreditCard, currencyAmount = somePayment.currencyAmount ) getsRejected(someErrorMessage) creditguard.sale( merchantKey = merchantKey, creditCard = someCreditCard, payment = somePayment, deal = Some(someDeal) ) must beAFailedTry.like { case e: PaymentRejectedException => e.message must contain(someErrorMessage) } } "successfully yield a transaction ID on valid request" in new Ctx { driver.aSaleFor( user = someMerchant.user, password = <PASSWORD>, terminalNumber = someMerchant.terminalNumber, supplierNumber = someMerchant.supplierNumber, idPrefix = someMerchant.idPrefix, orderId = Some(someDeal.id), card = someCreditCard, somePayment.currencyAmount ) returns( transactionId = someAuthorization.tranId ) creditguard.sale( merchantKey = merchantKey, creditCard = someCreditCard, payment = somePayment, deal = Some(someDeal) ) must beASuccessfulTry( check = ===(someAuthorization.tranId) ) } } "authorize request via CreditGuard gateway" should { "gracefully fail on invalid merchant" in new Ctx { val someErrorMessage = "some error message" driver.anAuthorizeFor( user = someMerchant.user, password = <PASSWORD>, terminalNumber = someMerchant.terminalNumber, supplierNumber = someMerchant.supplierNumber, idPrefix = someMerchant.idPrefix, orderId = Some(someDeal.id), card = someCreditCard, currencyAmount = somePayment.currencyAmount ) failsOnInvalidMerchant(someErrorMessage) creditguard.authorize( merchantKey = merchantKey, creditCard = someCreditCard, payment = somePayment, deal = Some(someDeal) ) must beAFailedTry.like { case e: PaymentErrorException => e.message must contain(someErrorMessage) } } "gracefully fail on rejected card" in new Ctx { val someErrorMessage = "some error message" driver.anAuthorizeFor( user = someMerchant.user, password = <PASSWORD>, terminalNumber = someMerchant.terminalNumber, supplierNumber = someMerchant.supplierNumber, idPrefix = someMerchant.idPrefix, orderId = Some(someDeal.id), card = someCreditCard, somePayment.currencyAmount ) getsRejected(someErrorMessage) creditguard.authorize( merchantKey = merchantKey, creditCard = someCreditCard, payment = somePayment, deal = Some(someDeal) ) must beAFailedTry.like { case e: PaymentRejectedException => e.message must contain(someErrorMessage) } } "successfully yield an authorization key on valid request" in new Ctx { driver.anAuthorizeFor( user = someMerchant.user, password = <PASSWORD>, terminalNumber = someMerchant.terminalNumber, supplierNumber = someMerchant.supplierNumber, idPrefix = someMerchant.idPrefix, orderId = Some(someDeal.id), card = someCreditCard, currencyAmount = somePayment.currencyAmount ) returns( authNumber = someAuthorization.authNumber, cardId = someAuthorization.cardId, cardExpiration = someAuthorization.cardExpiration, currency = someAuthorization.currency, transactionId = someAuthorization.tranId ) creditguard.authorize( merchantKey = merchantKey, creditCard = someCreditCard, payment = somePayment, deal = Some(someDeal) ) must beASuccessfulTry( check = beAuthorizationKey( authorization = beAuthorization( authNumber = ===(someAuthorization.authNumber), currency = ===(someAuthorization.currency), tranId = ===(someAuthorization.tranId), cardId = ===(someAuthorization.cardId), cardExpiration = ===(someAuthorization.cardExpiration) ) ) ) } } "capture request via CreditGuard gateway" should { "gracefully fail on invalid merchant" in new Ctx { val someErrorMessage = "some error message" driver.aCaptureFor( user = someMerchant.user, password = <PASSWORD>, terminalNumber = someMerchant.terminalNumber, supplierNumber = someMerchant.supplierNumber, authNumber = someAuthorization.authNumber, currency = someAuthorization.currency, amount = someCaptureAmount, cardId = someAuthorization.cardId, cardExpiration = someAuthorization.cardExpiration, userField = someAuthorization.user ) failsOnInvalidMerchant(someErrorMessage) creditguard.capture( merchantKey = merchantKey, authorizationKey = authorizationKey, amount = someCaptureAmount ) must beAFailedTry.like { case e: PaymentErrorException => e.message must contain(someErrorMessage) } } "successfully yield a transaction ID on valid request" in new Ctx { driver.aCaptureFor( user = someMerchant.user, password = <PASSWORD>, terminalNumber = someMerchant.terminalNumber, supplierNumber = someMerchant.supplierNumber, authNumber = someAuthorization.authNumber, currency = someAuthorization.currency, amount = someCaptureAmount, cardId = someAuthorization.cardId, cardExpiration = someAuthorization.cardExpiration, userField = someAuthorization.user ) returns( transactionId = someAuthorization.tranId ) creditguard.capture( merchantKey = merchantKey, authorizationKey = authorizationKey, amount = someCaptureAmount ) must beASuccessfulTry( check = ===(someAuthorization.tranId) ) } } step { driver.stop() } }
wix/libpay-creditguard
libpay-creditguard/src/main/scala/com/wix/pay/creditguard/CreditguardGateway.scala
<reponame>wix/libpay-creditguard<filename>libpay-creditguard/src/main/scala/com/wix/pay/creditguard/CreditguardGateway.scala package com.wix.pay.creditguard import com.google.api.client.http._ import com.wix.pay.creditcard.CreditCard import com.wix.pay.creditguard.model.Fields import com.wix.pay.model.{Customer, Deal, Payment} import com.wix.pay.shva.model.{IsShvaRejectedStatusCode, StatusCodes} import com.wix.pay.{PaymentErrorException, PaymentException, PaymentGateway, PaymentRejectedException} import scala.collection.JavaConversions._ import scala.concurrent.duration.Duration import scala.util.{Failure, Success, Try} object Endpoints { val test = "https://cguat2.creditguard.co.il/xpo/Relay" val caspitProduction = "https://payment.caspitgroup.co.il:8443/xpo/Relay" } class CreditguardGateway(requestFactory: HttpRequestFactory, connectTimeout: Option[Duration] = None, readTimeout: Option[Duration] = None, numberOfRetries: Int = 0, endpointUrl: String = Endpoints.caspitProduction, merchantParser: CreditguardMerchantParser = new JsonCreditguardMerchantParser, authorizationParser: CreditguardAuthorizationParser = new JsonCreditguardAuthorizationParser) extends PaymentGateway { override def authorize(merchantKey: String, creditCard: CreditCard, payment: Payment, customer: Option[Customer], deal: Option[Deal]): Try[String] = { Try { require(payment.installments == 1, "Installments are not implemented yet") val merchant = merchantParser.parse(merchantKey) val request = CreditguardHelper.createAuthorizeRequest( terminalNumber = merchant.terminalNumber, supplierNumber = merchant.supplierNumber, idPrefix = merchant.idPrefix, orderId = deal.map { _.id }, card = creditCard, currencyAmount = payment.currencyAmount ) val requestXml = RequestParser.stringify(request) val responseXml = doRequest( user = merchant.user, password = <PASSWORD>, requestXml = requestXml ) val response = ResponseParser.parse(responseXml) verifyShvaStatusCode( statusCode = response.response.doDeal.status, errorMessage = response.response.doDeal.statusText ) val authorization = CreditguardAuthorization( authNumber = response.response.doDeal.authNumber, currency = response.response.doDeal.currency, tranId = response.response.tranId, cardId = response.response.doDeal.cardId, cardExpiration = response.response.doDeal.cardExpiration, user = response.response.doDeal.user ) authorizationParser.stringify(authorization) } match { case Success(authorizationKey) => Success(authorizationKey) case Failure(e: PaymentException) => Failure(e) case Failure(e) => Failure(new PaymentErrorException(e.getMessage, e)) } } override def capture(merchantKey: String, authorizationKey: String, amount: Double): Try[String] = { Try { val merchant = merchantParser.parse(merchantKey) val authorization = authorizationParser.parse(authorizationKey) val request = CreditguardHelper.createCaptureRequest( terminalNumber = merchant.terminalNumber, supplierNumber = merchant.supplierNumber, cardId = authorization.cardId, cardExpiration = authorization.cardExpiration, authNumber = authorization.authNumber, currency = authorization.currency, amount = amount, user = authorization.user ) val requestXml = RequestParser.stringify(request) val responseXml = doRequest( user = merchant.user, password = <PASSWORD>, requestXml = requestXml ) val response = ResponseParser.parse(responseXml) verifyShvaStatusCode( statusCode = response.response.doDeal.status, errorMessage = response.response.doDeal.statusText ) response.response.tranId } match { case Success(transactionId) => Success(transactionId) case Failure(e: PaymentException) => Failure(e) case Failure(e) => Failure(new PaymentErrorException(e.getMessage, e)) } } override def sale(merchantKey: String, creditCard: CreditCard, payment: Payment, customer: Option[Customer], deal: Option[Deal]): Try[String] = { Try { require(payment.installments == 1, "Installments are not implemented yet") val merchant = merchantParser.parse(merchantKey) val request = CreditguardHelper.createSaleRequest( terminalNumber = merchant.terminalNumber, supplierNumber = merchant.supplierNumber, idPrefix = merchant.idPrefix, orderId = deal.map { _.id }, card = creditCard, currencyAmount = payment.currencyAmount ) val requestXml = RequestParser.stringify(request) val responseXml = doRequest( user = merchant.user, password = <PASSWORD>, requestXml = requestXml ) val response = ResponseParser.parse(responseXml) verifyShvaStatusCode( statusCode = response.response.doDeal.status, errorMessage = response.response.doDeal.statusText ) response.response.tranId } match { case Success(transactionId) => Success(transactionId) case Failure(e: PaymentException) => Failure(e) case Failure(e) => Failure(new PaymentErrorException(e.getMessage, e)) } } override def voidAuthorization(merchantKey: String, authorizationKey: String): Try[String] = { Try { // val merchant = merchantParser.parse(merchantKey) val authorization = authorizationParser.parse(authorizationKey) // CreditGuard doesn't support voiding an authorization. Authorizations should be automatically voided after a while. authorization.tranId } } private def doRequest(user: String, password: String, requestXml: String): String = { val params = Map( Fields.user -> user, Fields.password -> password, Fields.int_in -> requestXml ) val httpRequest = requestFactory.buildPostRequest( new GenericUrl(endpointUrl), new UrlEncodedContent(mapAsJavaMap(params)) ) connectTimeout foreach (duration => httpRequest.setConnectTimeout(duration.toMillis.toInt)) readTimeout foreach (duration => httpRequest.setReadTimeout(duration.toMillis.toInt)) httpRequest.setNumberOfRetries(numberOfRetries) val httpResponse = httpRequest.execute() try { httpResponse.parseAsString() } finally { httpResponse.ignore() } } private def verifyShvaStatusCode(statusCode: String, errorMessage: String): Unit = { statusCode match { case StatusCodes.success => // Operation successful. case IsShvaRejectedStatusCode(rejectedStatusCode) => throw PaymentRejectedException(s"$errorMessage (code = $rejectedStatusCode)") case _ => throw PaymentErrorException(s"$errorMessage (code = $statusCode)") } } }