snappy 0.3.0-java → 0.5.0-java

data/vendor/snappy/snappy.h

@@ -39,7 +39,9 @@
 #ifndef THIRD_PARTY_SNAPPY_SNAPPY_H__
 #define THIRD_PARTY_SNAPPY_SNAPPY_H__
 
-#include <cstddef>
+#include <stddef.h>
+#include <stdint.h>
+
 #include <string>
 
 #include "snappy-stubs-public.h"
@@ -48,13 +50,38 @@ namespace snappy {
   class Source;
   class Sink;
 
+  struct CompressionOptions {
+    // Compression level.
+    // Level 1 is the fastest
+    // Level 2 is a little slower but provides better compression. Level 2 is
+    // **EXPERIMENTAL** for the time being. It might happen that we decide to
+    // fall back to level 1 in the future.
+    // Levels 3+ are currently not supported. We plan to support levels up to
+    // 9 in the future.
+    // If you played with other compression algorithms, level 1 is equivalent to
+    // fast mode (level 1) of LZ4, level 2 is equivalent to LZ4's level 2 mode
+    // and compresses somewhere around zstd:-3 and zstd:-2 but generally with
+    // faster decompression speeds than snappy:1 and zstd:-3.
+    int level = DefaultCompressionLevel();
+
+    constexpr CompressionOptions() = default;
+    constexpr CompressionOptions(int compression_level)
+        : level(compression_level) {}
+    static constexpr int MinCompressionLevel() { return 1; }
+    static constexpr int MaxCompressionLevel() { return 2; }
+    static constexpr int DefaultCompressionLevel() { return 1; }
+  };
+
   // ------------------------------------------------------------------------
   // Generic compression/decompression routines.
   // ------------------------------------------------------------------------
 
-  // Compress the bytes read from "*source" and append to "*sink". Return the
+  // Compress the bytes read from "*reader" and append to "*writer". Return the
   // number of bytes written.
-  size_t Compress(Source* source, Sink* sink);
+  // First version is to preserve ABI.
+  size_t Compress(Source* reader, Sink* writer);
+  size_t Compress(Source* reader, Sink* writer,
+                  CompressionOptions options);
 
   // Find the uncompressed length of the given stream, as given by the header.
   // Note that the true length could deviate from this; the stream could e.g.
@@ -63,20 +90,34 @@ namespace snappy {
   // Also note that this leaves "*source" in a state that is unsuitable for
   // further operations, such as RawUncompress(). You will need to rewind
   // or recreate the source yourself before attempting any further calls.
-  bool GetUncompressedLength(Source* source, uint32* result);
+  bool GetUncompressedLength(Source* source, uint32_t* result);
 
   // ------------------------------------------------------------------------
   // Higher-level string based routines (should be sufficient for most users)
   // ------------------------------------------------------------------------
 
-  // Sets "*compressed" to the compressed version of "input[0,input_length-1]".
+  // Sets "*compressed" to the compressed version of "input[0..input_length-1]".
   // Original contents of *compressed are lost.
   //
   // REQUIRES: "input[]" is not an alias of "*compressed".
+  // First version is to preserve ABI.
   size_t Compress(const char* input, size_t input_length,
                   std::string* compressed);
+  size_t Compress(const char* input, size_t input_length,
+                  std::string* compressed, CompressionOptions options);
+
+  // Same as `Compress` above but taking an `iovec` array as input. Note that
+  // this function preprocesses the inputs to compute the sum of
+  // `iov[0..iov_cnt-1].iov_len` before reading. To avoid this, use
+  // `RawCompressFromIOVec` below.
+  // First version is to preserve ABI.
+  size_t CompressFromIOVec(const struct iovec* iov, size_t iov_cnt,
+                           std::string* compressed);
+  size_t CompressFromIOVec(const struct iovec* iov, size_t iov_cnt,
+                           std::string* compressed,
+                           CompressionOptions options);
 
-  // Decompresses "compressed[0,compressed_length-1]" to "*uncompressed".
+  // Decompresses "compressed[0..compressed_length-1]" to "*uncompressed".
   // Original contents of "*uncompressed" are lost.
   //
   // REQUIRES: "compressed[]" is not an alias of "*uncompressed".
@@ -117,10 +158,19 @@ namespace snappy {
   //    RawCompress(input, input_length, output, &output_length);
   //    ... Process(output, output_length) ...
   //    delete [] output;
-  void RawCompress(const char* input,
-                   size_t input_length,
-                   char* compressed,
+  void RawCompress(const char* input, size_t input_length, char* compressed,
                    size_t* compressed_length);
+  void RawCompress(const char* input, size_t input_length, char* compressed,
+                   size_t* compressed_length, CompressionOptions options);
+
+  // Same as `RawCompress` above but taking an `iovec` array as input. Note that
+  // `uncompressed_length` is the total number of bytes to be read from the
+  // elements of `iov` (_not_ the number of elements in `iov`).
+  void RawCompressFromIOVec(const struct iovec* iov, size_t uncompressed_length,
+                            char* compressed, size_t* compressed_length);
+  void RawCompressFromIOVec(const struct iovec* iov, size_t uncompressed_length,
+                            char* compressed, size_t* compressed_length,
+                            CompressionOptions options);
 
   // Given data in "compressed[0..compressed_length-1]" generated by
   // calling the Snappy::Compress routine, this routine
@@ -200,7 +250,7 @@ namespace snappy {
   static constexpr int kMinHashTableBits = 8;
   static constexpr size_t kMinHashTableSize = 1 << kMinHashTableBits;
 
-  static constexpr int kMaxHashTableBits = 14;
+  static constexpr int kMaxHashTableBits = 15;
   static constexpr size_t kMaxHashTableSize = 1 << kMaxHashTableBits;
 }  // end namespace snappy
 

data/vendor/snappy/snappy_benchmark.cc

@@ -0,0 +1,398 @@
+// Copyright 2020 Google Inc. All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+#include <cstddef>
+#include <cstdint>
+#include <string>
+#include <vector>
+
+#include "benchmark/benchmark.h"
+#include "snappy-internal.h"
+#include "snappy-sinksource.h"
+#include "snappy-test.h"
+#include "snappy.h"
+#include "snappy_test_data.h"
+
+namespace snappy {
+
+namespace {
+
+void FilesAndLevels(benchmark::internal::Benchmark* benchmark) {
+  for (int i = 0; i < ARRAYSIZE(kTestDataFiles); ++i) {
+    for (int level = snappy::CompressionOptions::MinCompressionLevel();
+         level <= snappy::CompressionOptions::MaxCompressionLevel(); ++level) {
+      benchmark->ArgPair(i, level);
+    }
+  }
+}
+
+void BM_UFlat(benchmark::State& state) {
+  // Pick file to process based on state.range(0).
+  int file_index = state.range(0);
+
+  CHECK_GE(file_index, 0);
+  CHECK_LT(file_index, ARRAYSIZE(kTestDataFiles));
+  std::string contents =
+      ReadTestDataFile(kTestDataFiles[file_index].filename,
+                       kTestDataFiles[file_index].size_limit);
+
+  std::string zcontents;
+  snappy::Compress(
+      contents.data(), contents.size(), &zcontents,
+      snappy::CompressionOptions{/*level=*/static_cast<int>(state.range(1))});
+  char* dst = new char[contents.size()];
+
+  for (auto s : state) {
+    CHECK(snappy::RawUncompress(zcontents.data(), zcontents.size(), dst));
+    benchmark::DoNotOptimize(dst);
+  }
+  state.SetBytesProcessed(static_cast<int64_t>(state.iterations()) *
+                          static_cast<int64_t>(contents.size()));
+  state.SetLabel(kTestDataFiles[file_index].label);
+
+  delete[] dst;
+}
+BENCHMARK(BM_UFlat)->Apply(FilesAndLevels);
+
+struct SourceFiles {
+  SourceFiles() {
+    for (int i = 0; i < kFiles; i++) {
+      std::string contents = ReadTestDataFile(kTestDataFiles[i].filename,
+                                              kTestDataFiles[i].size_limit);
+      max_size = std::max(max_size, contents.size());
+      sizes[i] = contents.size();
+      snappy::Compress(contents.data(), contents.size(), &zcontents[i]);
+    }
+  }
+  static constexpr int kFiles = ARRAYSIZE(kTestDataFiles);
+  std::string zcontents[kFiles];
+  size_t sizes[kFiles];
+  size_t max_size = 0;
+};
+
+void BM_UFlatMedley(benchmark::State& state) {
+  static const SourceFiles* const source = new SourceFiles();
+
+  std::vector<char> dst(source->max_size);
+
+  for (auto s : state) {
+    for (int i = 0; i < SourceFiles::kFiles; i++) {
+      CHECK(snappy::RawUncompress(source->zcontents[i].data(),
+                                  source->zcontents[i].size(), dst.data()));
+      benchmark::DoNotOptimize(dst);
+    }
+  }
+
+  int64_t source_sizes = 0;
+  for (int i = 0; i < SourceFiles::kFiles; i++) {
+    source_sizes += static_cast<int64_t>(source->sizes[i]);
+  }
+  state.SetBytesProcessed(static_cast<int64_t>(state.iterations()) *
+                          source_sizes);
+}
+BENCHMARK(BM_UFlatMedley);
+
+void BM_UValidate(benchmark::State& state) {
+  // Pick file to process based on state.range(0).
+  int file_index = state.range(0);
+
+  CHECK_GE(file_index, 0);
+  CHECK_LT(file_index, ARRAYSIZE(kTestDataFiles));
+  std::string contents =
+      ReadTestDataFile(kTestDataFiles[file_index].filename,
+                       kTestDataFiles[file_index].size_limit);
+
+  std::string zcontents;
+  snappy::Compress(
+      contents.data(), contents.size(), &zcontents,
+      snappy::CompressionOptions{/*level=*/static_cast<int>(state.range(1))});
+
+  for (auto s : state) {
+    CHECK(snappy::IsValidCompressedBuffer(zcontents.data(), zcontents.size()));
+  }
+  state.SetBytesProcessed(static_cast<int64_t>(state.iterations()) *
+                          static_cast<int64_t>(contents.size()));
+  state.SetLabel(kTestDataFiles[file_index].label);
+}
+BENCHMARK(BM_UValidate)->Apply(FilesAndLevels);
+
+void BM_UValidateMedley(benchmark::State& state) {
+  static const SourceFiles* const source = new SourceFiles();
+
+  for (auto s : state) {
+    for (int i = 0; i < SourceFiles::kFiles; i++) {
+      CHECK(snappy::IsValidCompressedBuffer(source->zcontents[i].data(),
+                                            source->zcontents[i].size()));
+    }
+  }
+
+  int64_t source_sizes = 0;
+  for (int i = 0; i < SourceFiles::kFiles; i++) {
+    source_sizes += static_cast<int64_t>(source->sizes[i]);
+  }
+  state.SetBytesProcessed(static_cast<int64_t>(state.iterations()) *
+                          source_sizes);
+}
+BENCHMARK(BM_UValidateMedley);
+
+void BM_UIOVecSource(benchmark::State& state) {
+  // Pick file to process based on state.range(0).
+  int file_index = state.range(0);
+  int level = state.range(1);
+
+  CHECK_GE(file_index, 0);
+  CHECK_LT(file_index, ARRAYSIZE(kTestDataFiles));
+  std::string contents =
+      ReadTestDataFile(kTestDataFiles[file_index].filename,
+                       kTestDataFiles[file_index].size_limit);
+
+  // Create `iovec`s of the `contents`.
+  const int kNumEntries = 10;
+  struct iovec iov[kNumEntries];
+  size_t used_so_far = 0;
+  for (int i = 0; i < kNumEntries; ++i) {
+    iov[i].iov_base = const_cast<char*>(contents.data()) + used_so_far;
+    if (used_so_far == contents.size()) {
+      iov[i].iov_len = 0;
+      continue;
+    }
+    if (i == kNumEntries - 1) {
+      iov[i].iov_len = contents.size() - used_so_far;
+    } else {
+      iov[i].iov_len = contents.size() / kNumEntries;
+    }
+    used_so_far += iov[i].iov_len;
+  }
+
+  char* dst = new char[snappy::MaxCompressedLength(contents.size())];
+  size_t zsize = 0;
+  for (auto s : state) {
+    snappy::RawCompressFromIOVec(iov, contents.size(), dst, &zsize,
+                                 snappy::CompressionOptions{/*level=*/level});
+    benchmark::DoNotOptimize(iov);
+  }
+  state.SetBytesProcessed(static_cast<int64_t>(state.iterations()) *
+                          static_cast<int64_t>(contents.size()));
+  const double compression_ratio =
+      static_cast<double>(zsize) / std::max<size_t>(1, contents.size());
+  state.SetLabel(StrFormat("%s (%.2f %%)", kTestDataFiles[file_index].label,
+                           100.0 * compression_ratio));
+  VLOG(0) << StrFormat("compression for %s: %d -> %d bytes",
+                       kTestDataFiles[file_index].label, contents.size(),
+                       zsize);
+
+  delete[] dst;
+}
+BENCHMARK(BM_UIOVecSource)->Apply(FilesAndLevels);
+
+void BM_UIOVecSink(benchmark::State& state) {
+  // Pick file to process based on state.range(0).
+  int file_index = state.range(0);
+
+  CHECK_GE(file_index, 0);
+  CHECK_LT(file_index, ARRAYSIZE(kTestDataFiles));
+  std::string contents =
+      ReadTestDataFile(kTestDataFiles[file_index].filename,
+                       kTestDataFiles[file_index].size_limit);
+
+  std::string zcontents;
+  snappy::Compress(contents.data(), contents.size(), &zcontents);
+
+  // Uncompress into an iovec containing ten entries.
+  const int kNumEntries = 10;
+  struct iovec iov[kNumEntries];
+  char* dst = new char[contents.size()];
+  size_t used_so_far = 0;
+  for (int i = 0; i < kNumEntries; ++i) {
+    iov[i].iov_base = dst + used_so_far;
+    if (used_so_far == contents.size()) {
+      iov[i].iov_len = 0;
+      continue;
+    }
+
+    if (i == kNumEntries - 1) {
+      iov[i].iov_len = contents.size() - used_so_far;
+    } else {
+      iov[i].iov_len = contents.size() / kNumEntries;
+    }
+    used_so_far += iov[i].iov_len;
+  }
+
+  for (auto s : state) {
+    CHECK(snappy::RawUncompressToIOVec(zcontents.data(), zcontents.size(), iov,
+                                       kNumEntries));
+    benchmark::DoNotOptimize(iov);
+  }
+  state.SetBytesProcessed(static_cast<int64_t>(state.iterations()) *
+                          static_cast<int64_t>(contents.size()));
+  state.SetLabel(kTestDataFiles[file_index].label);
+
+  delete[] dst;
+}
+BENCHMARK(BM_UIOVecSink)->DenseRange(0, 4);
+
+void BM_UFlatSink(benchmark::State& state) {
+  // Pick file to process based on state.range(0).
+  int file_index = state.range(0);
+
+  CHECK_GE(file_index, 0);
+  CHECK_LT(file_index, ARRAYSIZE(kTestDataFiles));
+  std::string contents =
+      ReadTestDataFile(kTestDataFiles[file_index].filename,
+                       kTestDataFiles[file_index].size_limit);
+
+  std::string zcontents;
+  snappy::Compress(
+      contents.data(), contents.size(), &zcontents,
+      snappy::CompressionOptions{/*level=*/static_cast<int>(state.range(1))});
+  char* dst = new char[contents.size()];
+
+  for (auto s : state) {
+    snappy::ByteArraySource source(zcontents.data(), zcontents.size());
+    snappy::UncheckedByteArraySink sink(dst);
+    CHECK(snappy::Uncompress(&source, &sink));
+    benchmark::DoNotOptimize(sink);
+  }
+  state.SetBytesProcessed(static_cast<int64_t>(state.iterations()) *
+                          static_cast<int64_t>(contents.size()));
+  state.SetLabel(kTestDataFiles[file_index].label);
+
+  std::string s(dst, contents.size());
+  CHECK_EQ(contents, s);
+
+  delete[] dst;
+}
+
+BENCHMARK(BM_UFlatSink)->Apply(FilesAndLevels);
+
+void BM_ZFlat(benchmark::State& state) {
+  // Pick file to process based on state.range(0).
+  int file_index = state.range(0);
+  int level = state.range(1);
+
+  CHECK_GE(file_index, 0);
+  CHECK_LT(file_index, ARRAYSIZE(kTestDataFiles));
+  std::string contents =
+      ReadTestDataFile(kTestDataFiles[file_index].filename,
+                       kTestDataFiles[file_index].size_limit);
+  char* dst = new char[snappy::MaxCompressedLength(contents.size())];
+
+  size_t zsize = 0;
+  for (auto s : state) {
+    snappy::RawCompress(contents.data(), contents.size(), dst, &zsize,
+                        snappy::CompressionOptions{/*level=*/level});
+    benchmark::DoNotOptimize(dst);
+  }
+  state.SetBytesProcessed(static_cast<int64_t>(state.iterations()) *
+                          static_cast<int64_t>(contents.size()));
+  const double compression_ratio =
+      static_cast<double>(zsize) / std::max<size_t>(1, contents.size());
+  state.SetLabel(StrFormat("%s (%.2f %%)", kTestDataFiles[file_index].label,
+                           100.0 * compression_ratio));
+  VLOG(0) << StrFormat("compression for %s: %d -> %d bytes",
+                       kTestDataFiles[file_index].label, contents.size(),
+                       zsize);
+  delete[] dst;
+}
+
+BENCHMARK(BM_ZFlat)->Apply(FilesAndLevels);
+
+void BM_ZFlatAll(benchmark::State& state) {
+  const int num_files = ARRAYSIZE(kTestDataFiles);
+  int level = state.range(0);
+
+  std::vector<std::string> contents(num_files);
+  std::vector<char*> dst(num_files);
+
+  int64_t total_contents_size = 0;
+  for (int i = 0; i < num_files; ++i) {
+    contents[i] = ReadTestDataFile(kTestDataFiles[i].filename,
+                                   kTestDataFiles[i].size_limit);
+    dst[i] = new char[snappy::MaxCompressedLength(contents[i].size())];
+    total_contents_size += contents[i].size();
+  }
+
+  size_t zsize = 0;
+  for (auto s : state) {
+    for (int i = 0; i < num_files; ++i) {
+      snappy::RawCompress(contents[i].data(), contents[i].size(), dst[i],
+                          &zsize, snappy::CompressionOptions{/*level=*/level});
+      benchmark::DoNotOptimize(dst);
+    }
+  }
+
+  state.SetBytesProcessed(static_cast<int64_t>(state.iterations()) *
+                          total_contents_size);
+
+  for (char* dst_item : dst) {
+    delete[] dst_item;
+  }
+  state.SetLabel(StrFormat("%d kTestDataFiles", num_files));
+}
+BENCHMARK(BM_ZFlatAll)->DenseRange(1, 2);
+
+void BM_ZFlatIncreasingTableSize(benchmark::State& state) {
+  CHECK_GT(ARRAYSIZE(kTestDataFiles), 0);
+  int level = state.range(0);
+  const std::string base_content = ReadTestDataFile(
+      kTestDataFiles[0].filename, kTestDataFiles[0].size_limit);
+
+  std::vector<std::string> contents;
+  std::vector<char*> dst;
+  int64_t total_contents_size = 0;
+  for (int table_bits = kMinHashTableBits; table_bits <= kMaxHashTableBits;
+       ++table_bits) {
+    std::string content = base_content;
+    content.resize(1 << table_bits);
+    dst.push_back(new char[snappy::MaxCompressedLength(content.size())]);
+    total_contents_size += content.size();
+    contents.push_back(std::move(content));
+  }
+
+  size_t zsize = 0;
+  for (auto s : state) {
+    for (size_t i = 0; i < contents.size(); ++i) {
+      snappy::RawCompress(contents[i].data(), contents[i].size(), dst[i],
+                          &zsize, snappy::CompressionOptions{/*level=*/level});
+      benchmark::DoNotOptimize(dst);
+    }
+  }
+
+  state.SetBytesProcessed(static_cast<int64_t>(state.iterations()) *
+                          total_contents_size);
+
+  for (char* dst_item : dst) {
+    delete[] dst_item;
+  }
+  state.SetLabel(StrFormat("%d tables", contents.size()));
+}
+BENCHMARK(BM_ZFlatIncreasingTableSize)->DenseRange(1, 2);
+
+}  // namespace
+
+}  // namespace snappy

data/vendor/snappy/snappy_compress_fuzzer.cc

@@ -28,9 +28,10 @@
 //
 // libFuzzer harness for fuzzing snappy compression code.
 
+#include <stddef.h>
+#include <stdint.h>
+
 #include <cassert>
-#include <cstddef>
-#include <cstdint>
 #include <string>
 
 #include "snappy.h"
@@ -38,22 +39,26 @@
 // Entry point for LibFuzzer.
 extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
   std::string input(reinterpret_cast<const char*>(data), size);
+  for (int level = snappy::CompressionOptions::MinCompressionLevel();
+       level <= snappy::CompressionOptions::MaxCompressionLevel(); ++level) {
+    std::string compressed;
+    size_t compressed_size =
+        snappy::Compress(input.data(), input.size(), &compressed,
+                         snappy::CompressionOptions{/*level=*/level});
 
-  std::string compressed;
-  size_t compressed_size =
-      snappy::Compress(input.data(), input.size(), &compressed);
-
-  (void)compressed_size;  // Variable only used in debug builds.
-  assert(compressed_size == compressed.size());
-  assert(compressed.size() <= snappy::MaxCompressedLength(input.size()));
-  assert(snappy::IsValidCompressedBuffer(compressed.data(), compressed.size()));
+    (void)compressed_size;  // Variable only used in debug builds.
+    assert(compressed_size == compressed.size());
+    assert(compressed.size() <= snappy::MaxCompressedLength(input.size()));
+    assert(
+        snappy::IsValidCompressedBuffer(compressed.data(), compressed.size()));
 
-  std::string uncompressed_after_compress;
-  bool uncompress_succeeded = snappy::Uncompress(
-      compressed.data(), compressed.size(), &uncompressed_after_compress);
+    std::string uncompressed_after_compress;
+    bool uncompress_succeeded = snappy::Uncompress(
+        compressed.data(), compressed.size(), &uncompressed_after_compress);
 
-  (void)uncompress_succeeded;  // Variable only used in debug builds.
-  assert(uncompress_succeeded);
-  assert(input == uncompressed_after_compress);
+    (void)uncompress_succeeded;  // Variable only used in debug builds.
+    assert(uncompress_succeeded);
+    assert(input == uncompressed_after_compress);
+  }
   return 0;
 }

data/vendor/snappy/snappy_test_data.cc

@@ -0,0 +1,57 @@
+// Copyright 2020 Google Inc. All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Support code for reading test data.
+
+#include "snappy_test_data.h"
+
+#include <cstddef>
+#include <cstdlib>
+#include <string>
+
+#include "snappy-test.h"
+
+namespace snappy {
+
+std::string ReadTestDataFile(const char* base, size_t size_limit) {
+  std::string srcdir;
+  const char* srcdir_env = std::getenv("srcdir");  // This is set by Automake.
+  if (srcdir_env) {
+    srcdir = std::string(srcdir_env) + "/";
+  }
+
+  std::string contents;
+  CHECK_OK(file::GetContents(srcdir + "testdata/" + base, &contents,
+                             file::Defaults()));
+  if (size_limit > 0) {
+    contents = contents.substr(0, size_limit);
+  }
+  return contents;
+}
+
+}  // namespace snappy