folly/io/test/CompressionTest.cpp

   1 /*
   2  * Copyright 2017 Facebook, Inc.
   3  *
   4  * Licensed under the Apache License, Version 2.0 (the "License");
   5  * you may not use this file except in compliance with the License.
   6  * You may obtain a copy of the License at
   7  *
   8  *   http://www.apache.org/licenses/LICENSE-2.0
   9  *
  10  * Unless required by applicable law or agreed to in writing, software
  11  * distributed under the License is distributed on an "AS IS" BASIS,
  12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13  * See the License for the specific language governing permissions and
  14  * limitations under the License.
  15  */
  16
  17 #include <folly/io/Compression.h>
  18
  19 #include <random>
  20 #include <set>
  21 #include <thread>
  22 #include <unordered_map>
  23
  24 #include <boost/noncopyable.hpp>
  25 #include <glog/logging.h>
  26
  27 #include <folly/Benchmark.h>
  28 #include <folly/Hash.h>
  29 #include <folly/Random.h>
  30 #include <folly/Varint.h>
  31 #include <folly/io/IOBufQueue.h>
  32 #include <folly/portability/GTest.h>
  33
  34 namespace folly { namespace io { namespace test {
  35
  36 class DataHolder : private boost::noncopyable {
  37  public:
  38   uint64_t hash(size_t size) const;
  39   ByteRange data(size_t size) const;
  40
  41  protected:
  42   explicit DataHolder(size_t sizeLog2);
  43   const size_t size_;
  44   std::unique_ptr<uint8_t[]> data_;
  45   mutable std::unordered_map<uint64_t, uint64_t> hashCache_;
  46 };
  47
  48 DataHolder::DataHolder(size_t sizeLog2)
  49   : size_(size_t(1) << sizeLog2),
  50     data_(new uint8_t[size_]) {
  51 }
  52
  53 uint64_t DataHolder::hash(size_t size) const {
  54   CHECK_LE(size, size_);
  55   auto p = hashCache_.find(size);
  56   if (p != hashCache_.end()) {
  57     return p->second;
  58   }
  59
  60   uint64_t h = folly::hash::fnv64_buf(data_.get(), size);
  61   hashCache_[size] = h;
  62   return h;
  63 }
  64
  65 ByteRange DataHolder::data(size_t size) const {
  66   CHECK_LE(size, size_);
  67   return ByteRange(data_.get(), size);
  68 }
  69
  70 uint64_t hashIOBuf(const IOBuf* buf) {
  71   uint64_t h = folly::hash::FNV_64_HASH_START;
  72   for (auto& range : *buf) {
  73     h = folly::hash::fnv64_buf(range.data(), range.size(), h);
  74   }
  75   return h;
  76 }
  77
  78 class RandomDataHolder : public DataHolder {
  79  public:
  80   explicit RandomDataHolder(size_t sizeLog2);
  81 };
  82
  83 RandomDataHolder::RandomDataHolder(size_t sizeLog2)
  84   : DataHolder(sizeLog2) {
  85   constexpr size_t numThreadsLog2 = 3;
  86   constexpr size_t numThreads = size_t(1) << numThreadsLog2;
  87
  88   uint32_t seed = randomNumberSeed();
  89
  90   std::vector<std::thread> threads;
  91   threads.reserve(numThreads);
  92   for (size_t t = 0; t < numThreads; ++t) {
  93     threads.emplace_back(
  94         [this, seed, t, numThreadsLog2, sizeLog2] () {
  95           std::mt19937 rng(seed + t);
  96           size_t countLog2 = sizeLog2 - numThreadsLog2;
  97           size_t start = size_t(t) << countLog2;
  98           for (size_t i = 0; i < countLog2; ++i) {
  99             this->data_[start + i] = rng();
 100           }
 101         });
 102   }
 103
 104   for (auto& t : threads) {
 105     t.join();
 106   }
 107 }
 108
 109 class ConstantDataHolder : public DataHolder {
 110  public:
 111   explicit ConstantDataHolder(size_t sizeLog2);
 112 };
 113
 114 ConstantDataHolder::ConstantDataHolder(size_t sizeLog2)
 115   : DataHolder(sizeLog2) {
 116   memset(data_.get(), 'a', size_);
 117 }
 118
 119 constexpr size_t dataSizeLog2 = 27;  // 128MiB
 120 RandomDataHolder randomDataHolder(dataSizeLog2);
 121 ConstantDataHolder constantDataHolder(dataSizeLog2);
 122
 123 TEST(CompressionTestNeedsUncompressedLength, Simple) {
 124   EXPECT_FALSE(getCodec(CodecType::NO_COMPRESSION)->needsUncompressedLength());
 125   EXPECT_TRUE(getCodec(CodecType::LZ4)->needsUncompressedLength());
 126   EXPECT_FALSE(getCodec(CodecType::SNAPPY)->needsUncompressedLength());
 127   EXPECT_FALSE(getCodec(CodecType::ZLIB)->needsUncompressedLength());
 128   EXPECT_FALSE(getCodec(CodecType::LZ4_VARINT_SIZE)->needsUncompressedLength());
 129   EXPECT_TRUE(getCodec(CodecType::LZMA2)->needsUncompressedLength());
 130   EXPECT_FALSE(getCodec(CodecType::LZMA2_VARINT_SIZE)
 131     ->needsUncompressedLength());
 132   EXPECT_FALSE(getCodec(CodecType::ZSTD)->needsUncompressedLength());
 133   EXPECT_FALSE(getCodec(CodecType::GZIP)->needsUncompressedLength());
 134 }
 135
 136 class CompressionTest
 137     : public testing::TestWithParam<std::tr1::tuple<int, int, CodecType>> {
 138  protected:
 139   void SetUp() override {
 140     auto tup = GetParam();
 141     uncompressedLength_ = uint64_t(1) << std::tr1::get<0>(tup);
 142     chunks_ = std::tr1::get<1>(tup);
 143     codec_ = getCodec(std::tr1::get<2>(tup));
 144   }
 145
 146   void runSimpleTest(const DataHolder& dh);
 147
 148  private:
 149   std::unique_ptr<IOBuf> split(std::unique_ptr<IOBuf> data) const;
 150
 151   uint64_t uncompressedLength_;
 152   size_t chunks_;
 153   std::unique_ptr<Codec> codec_;
 154 };
 155
 156 void CompressionTest::runSimpleTest(const DataHolder& dh) {
 157   const auto original = split(IOBuf::wrapBuffer(dh.data(uncompressedLength_)));
 158   const auto compressed = split(codec_->compress(original.get()));
 159   if (!codec_->needsUncompressedLength()) {
 160     auto uncompressed = codec_->uncompress(compressed.get());
 161     EXPECT_EQ(uncompressedLength_, uncompressed->computeChainDataLength());
 162     EXPECT_EQ(dh.hash(uncompressedLength_), hashIOBuf(uncompressed.get()));
 163   }
 164   {
 165     auto uncompressed = codec_->uncompress(compressed.get(),
 166                                            uncompressedLength_);
 167     EXPECT_EQ(uncompressedLength_, uncompressed->computeChainDataLength());
 168     EXPECT_EQ(dh.hash(uncompressedLength_), hashIOBuf(uncompressed.get()));
 169   }
 170 }
 171
 172 // Uniformly split data into (potentially empty) chunks.
 173 std::unique_ptr<IOBuf> CompressionTest::split(
 174     std::unique_ptr<IOBuf> data) const {
 175   if (data->isChained()) {
 176     data->coalesce();
 177   }
 178
 179   const size_t size = data->computeChainDataLength();
 180
 181   std::multiset<size_t> splits;
 182   for (size_t i = 1; i < chunks_; ++i) {
 183     splits.insert(Random::rand64(size));
 184   }
 185
 186   folly::IOBufQueue result;
 187
 188   size_t offset = 0;
 189   for (size_t split : splits) {
 190     result.append(IOBuf::copyBuffer(data->data() + offset, split - offset));
 191     offset = split;
 192   }
 193   result.append(IOBuf::copyBuffer(data->data() + offset, size - offset));
 194
 195   return result.move();
 196 }
 197
 198 TEST_P(CompressionTest, RandomData) {
 199   runSimpleTest(randomDataHolder);
 200 }
 201
 202 TEST_P(CompressionTest, ConstantData) {
 203   runSimpleTest(constantDataHolder);
 204 }
 205
 206 INSTANTIATE_TEST_CASE_P(
 207     CompressionTest,
 208     CompressionTest,
 209     testing::Combine(
 210         testing::Values(0, 1, 12, 22, 25, 27),
 211         testing::Values(1, 2, 3, 8, 65),
 212         testing::Values(
 213             CodecType::NO_COMPRESSION,
 214             CodecType::LZ4,
 215             CodecType::SNAPPY,
 216             CodecType::ZLIB,
 217             CodecType::LZ4_VARINT_SIZE,
 218             CodecType::LZMA2,
 219             CodecType::LZMA2_VARINT_SIZE,
 220             CodecType::ZSTD,
 221             CodecType::GZIP)));
 222
 223 class CompressionVarintTest
 224     : public testing::TestWithParam<std::tr1::tuple<int, CodecType>> {
 225  protected:
 226   void SetUp() override {
 227     auto tup = GetParam();
 228     uncompressedLength_ = uint64_t(1) << std::tr1::get<0>(tup);
 229     codec_ = getCodec(std::tr1::get<1>(tup));
 230   }
 231
 232   void runSimpleTest(const DataHolder& dh);
 233
 234   uint64_t uncompressedLength_;
 235   std::unique_ptr<Codec> codec_;
 236 };
 237
 238 inline uint64_t oneBasedMsbPos(uint64_t number) {
 239   uint64_t pos = 0;
 240   for (; number > 0; ++pos, number >>= 1) {
 241   }
 242   return pos;
 243 }
 244
 245 void CompressionVarintTest::runSimpleTest(const DataHolder& dh) {
 246   auto original = IOBuf::wrapBuffer(dh.data(uncompressedLength_));
 247   auto compressed = codec_->compress(original.get());
 248   auto breakPoint =
 249       1UL +
 250       Random::rand64(
 251           std::max(uint64_t(9), oneBasedMsbPos(uncompressedLength_)) / 9UL);
 252   auto tinyBuf = IOBuf::copyBuffer(compressed->data(),
 253                                    std::min(compressed->length(), breakPoint));
 254   compressed->trimStart(breakPoint);
 255   tinyBuf->prependChain(std::move(compressed));
 256   compressed = std::move(tinyBuf);
 257
 258   auto uncompressed = codec_->uncompress(compressed.get());
 259
 260   EXPECT_EQ(uncompressedLength_, uncompressed->computeChainDataLength());
 261   EXPECT_EQ(dh.hash(uncompressedLength_), hashIOBuf(uncompressed.get()));
 262 }
 263
 264 TEST_P(CompressionVarintTest, RandomData) {
 265   runSimpleTest(randomDataHolder);
 266 }
 267
 268 TEST_P(CompressionVarintTest, ConstantData) {
 269   runSimpleTest(constantDataHolder);
 270 }
 271
 272 INSTANTIATE_TEST_CASE_P(
 273     CompressionVarintTest,
 274     CompressionVarintTest,
 275     testing::Combine(
 276         testing::Values(0, 1, 12, 22, 25, 27),
 277         testing::Values(
 278             CodecType::LZ4_VARINT_SIZE,
 279             CodecType::LZMA2_VARINT_SIZE)));
 280
 281 class CompressionCorruptionTest : public testing::TestWithParam<CodecType> {
 282  protected:
 283   void SetUp() override { codec_ = getCodec(GetParam()); }
 284
 285   void runSimpleTest(const DataHolder& dh);
 286
 287   std::unique_ptr<Codec> codec_;
 288 };
 289
 290 void CompressionCorruptionTest::runSimpleTest(const DataHolder& dh) {
 291   constexpr uint64_t uncompressedLength = 42;
 292   auto original = IOBuf::wrapBuffer(dh.data(uncompressedLength));
 293   auto compressed = codec_->compress(original.get());
 294
 295   if (!codec_->needsUncompressedLength()) {
 296     auto uncompressed = codec_->uncompress(compressed.get());
 297     EXPECT_EQ(uncompressedLength, uncompressed->computeChainDataLength());
 298     EXPECT_EQ(dh.hash(uncompressedLength), hashIOBuf(uncompressed.get()));
 299   }
 300   {
 301     auto uncompressed = codec_->uncompress(compressed.get(),
 302                                            uncompressedLength);
 303     EXPECT_EQ(uncompressedLength, uncompressed->computeChainDataLength());
 304     EXPECT_EQ(dh.hash(uncompressedLength), hashIOBuf(uncompressed.get()));
 305   }
 306
 307   EXPECT_THROW(codec_->uncompress(compressed.get(), uncompressedLength + 1),
 308                std::runtime_error);
 309
 310   // Corrupt the first character
 311   ++(compressed->writableData()[0]);
 312
 313   if (!codec_->needsUncompressedLength()) {
 314     EXPECT_THROW(codec_->uncompress(compressed.get()),
 315                  std::runtime_error);
 316   }
 317
 318   EXPECT_THROW(codec_->uncompress(compressed.get(), uncompressedLength),
 319                std::runtime_error);
 320 }
 321
 322 TEST_P(CompressionCorruptionTest, RandomData) {
 323   runSimpleTest(randomDataHolder);
 324 }
 325
 326 TEST_P(CompressionCorruptionTest, ConstantData) {
 327   runSimpleTest(constantDataHolder);
 328 }
 329
 330 INSTANTIATE_TEST_CASE_P(
 331     CompressionCorruptionTest,
 332     CompressionCorruptionTest,
 333     testing::Values(
 334         // NO_COMPRESSION can't detect corruption
 335         // LZ4 can't detect corruption reliably (sigh)
 336         CodecType::SNAPPY,
 337         CodecType::ZLIB));
 338
 339 }}}  // namespaces
 340
 341 int main(int argc, char *argv[]) {
 342   testing::InitGoogleTest(&argc, argv);
 343   gflags::ParseCommandLineFlags(&argc, &argv, true);
 344
 345   auto ret = RUN_ALL_TESTS();
 346   if (!ret) {
 347     folly::runBenchmarksOnFlag();
 348   }
 349   return ret;
 350 }