aboutsummaryrefslogtreecommitdiff
path: root/src/zenutil/chunkedfile.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'src/zenutil/chunkedfile.cpp')
-rw-r--r--src/zenutil/chunkedfile.cpp525
1 files changed, 525 insertions, 0 deletions
diff --git a/src/zenutil/chunkedfile.cpp b/src/zenutil/chunkedfile.cpp
new file mode 100644
index 000000000..a2c041ffd
--- /dev/null
+++ b/src/zenutil/chunkedfile.cpp
@@ -0,0 +1,525 @@
+// Copyright Epic Games, Inc. All Rights Reserved.
+
+#include <zenutil/chunkedfile.h>
+
+#include <zencore/basicfile.h>
+#include <zencore/trace.h>
+
+#include "chunking.h"
+
+ZEN_THIRD_PARTY_INCLUDES_START
+#include <tsl/robin_map.h>
+#include <gsl/gsl-lite.hpp>
+ZEN_THIRD_PARTY_INCLUDES_END
+
+namespace zen {
+
+namespace {
+ struct ChunkedHeader
+ {
+ static constexpr uint32_t ExpectedMagic = 0x646b6863; // chkd
+ static constexpr uint32_t CurrentVersion = 1;
+
+ uint32_t Magic = ExpectedMagic;
+ uint32_t Version = CurrentVersion;
+ uint32_t ChunkSequenceLength;
+ uint32_t ChunkHashCount;
+ uint64_t ChunkSequenceOffset;
+ uint64_t ChunkHashesOffset;
+ uint64_t RawSize = 0;
+ IoHash RawHash;
+ };
+} // namespace
+
+IoBuffer
+SerializeChunkedInfo(const ChunkedInfo& Info)
+{
+ ZEN_TRACE_CPU("SerializeChunkedInfo");
+ size_t HeaderSize = RoundUp(sizeof(ChunkedHeader), 16) + RoundUp(sizeof(uint32_t) * Info.ChunkSequence.size(), 16) +
+ RoundUp(sizeof(IoHash) * Info.ChunkHashes.size(), 16);
+ IoBuffer HeaderData(HeaderSize);
+
+ ChunkedHeader Header;
+ Header.ChunkSequenceLength = gsl::narrow<uint32_t>(Info.ChunkSequence.size());
+ Header.ChunkHashCount = gsl::narrow<uint32_t>(Info.ChunkHashes.size());
+ Header.ChunkSequenceOffset = RoundUp(sizeof(ChunkedHeader), 16);
+ Header.ChunkHashesOffset = RoundUp(Header.ChunkSequenceOffset + sizeof(uint32_t) * Header.ChunkSequenceLength, 16);
+ Header.RawSize = Info.RawSize;
+ Header.RawHash = Info.RawHash;
+
+ MutableMemoryView WriteView = HeaderData.GetMutableView();
+ {
+ MutableMemoryView HeaderWriteView = WriteView.Left(sizeof(Header));
+ HeaderWriteView.CopyFrom(MemoryView(&Header, sizeof(Header)));
+ }
+ {
+ MutableMemoryView ChunkSequenceWriteView = WriteView.Mid(Header.ChunkSequenceOffset, sizeof(uint32_t) * Header.ChunkSequenceLength);
+ ChunkSequenceWriteView.CopyFrom(MemoryView(Info.ChunkSequence.data(), ChunkSequenceWriteView.GetSize()));
+ }
+ {
+ MutableMemoryView ChunksWriteView = WriteView.Mid(Header.ChunkHashesOffset, sizeof(IoHash) * Header.ChunkHashCount);
+ ChunksWriteView.CopyFrom(MemoryView(Info.ChunkHashes.data(), ChunksWriteView.GetSize()));
+ }
+
+ return HeaderData;
+}
+
+ChunkedInfo
+DeserializeChunkedInfo(IoBuffer& Buffer)
+{
+ ZEN_TRACE_CPU("DeserializeChunkedInfo");
+ MemoryView View = Buffer.GetView();
+ ChunkedHeader Header;
+ {
+ MutableMemoryView HeaderWriteView(&Header, sizeof(Header));
+ HeaderWriteView.CopyFrom(View.Left(sizeof(Header)));
+ }
+ if (Header.Magic != ChunkedHeader::ExpectedMagic)
+ {
+ return {};
+ }
+ if (Header.Version != ChunkedHeader::CurrentVersion)
+ {
+ return {};
+ }
+ ChunkedInfo Info;
+ Info.RawSize = Header.RawSize;
+ Info.RawHash = Header.RawHash;
+ Info.ChunkSequence.resize(Header.ChunkSequenceLength);
+ Info.ChunkHashes.resize(Header.ChunkHashCount);
+ {
+ MutableMemoryView ChunkSequenceWriteView(Info.ChunkSequence.data(), sizeof(uint32_t) * Header.ChunkSequenceLength);
+ ChunkSequenceWriteView.CopyFrom(View.Mid(Header.ChunkSequenceOffset, ChunkSequenceWriteView.GetSize()));
+ }
+ {
+ MutableMemoryView ChunksWriteView(Info.ChunkHashes.data(), sizeof(IoHash) * Header.ChunkHashCount);
+ ChunksWriteView.CopyFrom(View.Mid(Header.ChunkHashesOffset, ChunksWriteView.GetSize()));
+ }
+
+ return Info;
+}
+
+void
+Reconstruct(const ChunkedInfo& Info, const std::filesystem::path& TargetPath, std::function<IoBuffer(const IoHash& ChunkHash)> GetChunk)
+{
+ ZEN_TRACE_CPU("Reconstruct");
+ BasicFile Reconstructed;
+ Reconstructed.Open(TargetPath, BasicFile::Mode::kTruncate);
+ BasicFileWriter ReconstructedWriter(Reconstructed, 64 * 1024);
+ uint64_t Offset = 0;
+ for (uint32_t SequenceIndex : Info.ChunkSequence)
+ {
+ IoBuffer Chunk = GetChunk(Info.ChunkHashes[SequenceIndex]);
+ ReconstructedWriter.Write(Chunk.GetData(), Chunk.GetSize(), Offset);
+ Offset += Chunk.GetSize();
+ }
+}
+
+ChunkedInfoWithSource
+ChunkData(BasicFile& RawData,
+ uint64_t Offset,
+ uint64_t Size,
+ ChunkedParams Params,
+ std::atomic<uint64_t>* BytesProcessed,
+ std::atomic<bool>* AbortFlag)
+{
+ ZEN_TRACE_CPU("ChunkData");
+
+ ChunkedInfoWithSource Result;
+ tsl::robin_map<IoHash, uint32_t, IoHash::Hasher> FoundChunks;
+
+ ZenChunkHelper Chunker;
+ Chunker.SetUseThreshold(Params.UseThreshold);
+ Chunker.SetChunkSize(Params.MinSize, Params.MaxSize, Params.AvgSize);
+ size_t End = Offset + Size;
+ const size_t ScanBufferSize = Max(1u * 1024 * 1024, Params.MaxSize);
+ BasicFileBuffer RawBuffer(RawData, ScanBufferSize);
+ MemoryView SliceView = RawBuffer.MakeView(Min(End - Offset, ScanBufferSize), Offset);
+ ZEN_ASSERT(!SliceView.IsEmpty());
+ size_t SliceSize = SliceView.GetSize();
+ IoHashStream RawHashStream;
+ while (Offset < End)
+ {
+ if (AbortFlag != nullptr && AbortFlag->load())
+ {
+ return {};
+ }
+ size_t ScanLength = Chunker.ScanChunk(SliceView.GetData(), SliceSize);
+ if (ScanLength == ZenChunkHelper::kNoBoundaryFound)
+ {
+ if (Offset + SliceSize == End)
+ {
+ ScanLength = SliceSize;
+ }
+ else
+ {
+ SliceView = RawBuffer.MakeView(Min(End - Offset, ScanBufferSize), Offset);
+ SliceSize = SliceView.GetSize();
+ Chunker.Reset();
+ continue;
+ }
+ }
+ uint32_t ChunkLength = gsl::narrow<uint32_t>(ScanLength); // +HashedLength);
+ MemoryView ChunkView = SliceView.Left(ScanLength);
+ RawHashStream.Append(ChunkView);
+ IoHash ChunkHash = IoHash::HashBuffer(ChunkView);
+ SliceView.RightChopInline(ScanLength);
+ if (auto It = FoundChunks.find(ChunkHash); It != FoundChunks.end())
+ {
+ Result.Info.ChunkSequence.push_back(It->second);
+ }
+ else
+ {
+ uint32_t ChunkIndex = gsl::narrow<uint32_t>(Result.Info.ChunkHashes.size());
+ FoundChunks.insert_or_assign(ChunkHash, ChunkIndex);
+ Result.Info.ChunkHashes.push_back(ChunkHash);
+ Result.ChunkSources.push_back(ChunkSource{.Offset = Offset, .Size = ChunkLength});
+ Result.Info.ChunkSequence.push_back(ChunkIndex);
+ }
+
+ SliceSize = SliceView.GetSize();
+ Offset += ChunkLength;
+ if (BytesProcessed != nullptr)
+ {
+ BytesProcessed->fetch_add(ChunkLength);
+ }
+ }
+ Result.Info.RawSize = Size;
+ Result.Info.RawHash = RawHashStream.GetHash();
+ return Result;
+}
+
+} // namespace zen
+
+#if ZEN_WITH_TESTS
+# include <zencore/filesystem.h>
+# include <zencore/fmtutils.h>
+# include <zencore/iohash.h>
+# include <zencore/logging.h>
+# include <zencore/scopeguard.h>
+# include <zencore/timer.h>
+# include <zencore/testing.h>
+# include <zencore/testutils.h>
+# include <zencore/workthreadpool.h>
+
+# include "chunking.h"
+
+ZEN_THIRD_PARTY_INCLUDES_START
+# include <tsl/robin_map.h>
+# include <tsl/robin_set.h>
+ZEN_THIRD_PARTY_INCLUDES_END
+
+namespace zen {
+# if 0
+TEST_CASE("chunkedfile.findparams")
+{
+# if 1
+ DirectoryContent SourceContent1;
+ GetDirectoryContent("E:\\Temp\\ChunkingTestData\\31379208", DirectoryContentFlags::IncludeFiles, SourceContent1);
+ const std::vector<std::filesystem::path>& SourceFiles1 = SourceContent1.Files;
+ DirectoryContent SourceContent2;
+ GetDirectoryContent("E:\\Temp\\ChunkingTestData\\31379208_2", DirectoryContentFlags::IncludeFiles, SourceContent2);
+ const std::vector<std::filesystem::path>& SourceFiles2 = SourceContent2.Files;
+# else
+ std::filesystem::path SourcePath1 =
+ "E:\\Temp\\ChunkingTestData\\31375996\\ShaderArchive-FortniteGame_Chunk10-PCD3D_SM6-PCD3D_SM6.ushaderbytecode";
+ std::filesystem::path SourcePath2 =
+ "E:\\Temp\\ChunkingTestData\\31379208\\ShaderArchive-FortniteGame_Chunk10-PCD3D_SM6-PCD3D_SM6.ushaderbytecode";
+ const std::vector<std::filesystem::path>& SourceFiles1 = {SourcePath1};
+ const std::vector<std::filesystem::path>& SourceFiles2 = {SourcePath2};
+# endif
+ ChunkedParams Params[] = {ChunkedParams{.UseThreshold = false, .MinSize = 17280, .MaxSize = 139264, .AvgSize = 36340},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15456, .MaxSize = 122880, .AvgSize = 35598},
+ ChunkedParams{.UseThreshold = false, .MinSize = 16848, .MaxSize = 135168, .AvgSize = 39030},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14256, .MaxSize = 114688, .AvgSize = 36222},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15744, .MaxSize = 126976, .AvgSize = 36600},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15264, .MaxSize = 122880, .AvgSize = 35442},
+ ChunkedParams{.UseThreshold = false, .MinSize = 16464, .MaxSize = 131072, .AvgSize = 37950},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15408, .MaxSize = 122880, .AvgSize = 38914},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15408, .MaxSize = 122880, .AvgSize = 35556},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15360, .MaxSize = 122880, .AvgSize = 35520},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15312, .MaxSize = 122880, .AvgSize = 35478},
+ ChunkedParams{.UseThreshold = false, .MinSize = 16896, .MaxSize = 135168, .AvgSize = 39072},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15360, .MaxSize = 122880, .AvgSize = 38880},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15840, .MaxSize = 126976, .AvgSize = 36678},
+ ChunkedParams{.UseThreshold = false, .MinSize = 16800, .MaxSize = 135168, .AvgSize = 38994},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15888, .MaxSize = 126976, .AvgSize = 36714},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15792, .MaxSize = 126976, .AvgSize = 36636},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14880, .MaxSize = 118784, .AvgSize = 37609},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15936, .MaxSize = 126976, .AvgSize = 36756},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15456, .MaxSize = 122880, .AvgSize = 38955},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15984, .MaxSize = 126976, .AvgSize = 36792},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14400, .MaxSize = 114688, .AvgSize = 36338},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14832, .MaxSize = 118784, .AvgSize = 37568},
+ ChunkedParams{.UseThreshold = false, .MinSize = 16944, .MaxSize = 135168, .AvgSize = 39108},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14352, .MaxSize = 114688, .AvgSize = 36297},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14208, .MaxSize = 114688, .AvgSize = 36188},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14448, .MaxSize = 114688, .AvgSize = 36372},
+ ChunkedParams{.UseThreshold = false, .MinSize = 13296, .MaxSize = 106496, .AvgSize = 36592},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15264, .MaxSize = 122880, .AvgSize = 38805},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14304, .MaxSize = 114688, .AvgSize = 36263},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14784, .MaxSize = 118784, .AvgSize = 37534},
+ ChunkedParams{.UseThreshold = false, .MinSize = 15312, .MaxSize = 122880, .AvgSize = 38839},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14256, .MaxSize = 114688, .AvgSize = 39360},
+ ChunkedParams{.UseThreshold = false, .MinSize = 13776, .MaxSize = 110592, .AvgSize = 37976},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14736, .MaxSize = 118784, .AvgSize = 37493},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14928, .MaxSize = 118784, .AvgSize = 37643},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14448, .MaxSize = 114688, .AvgSize = 39504},
+ ChunkedParams{.UseThreshold = false, .MinSize = 13392, .MaxSize = 106496, .AvgSize = 36664},
+ ChunkedParams{.UseThreshold = false, .MinSize = 13872, .MaxSize = 110592, .AvgSize = 38048},
+ ChunkedParams{.UseThreshold = false, .MinSize = 14352, .MaxSize = 114688, .AvgSize = 39432},
+ ChunkedParams{.UseThreshold = false, .MinSize = 13200, .MaxSize = 106496, .AvgSize = 36520},
+ ChunkedParams{.UseThreshold = false, .MinSize = 17328, .MaxSize = 139264, .AvgSize = 36378},
+ ChunkedParams{.UseThreshold = false, .MinSize = 17376, .MaxSize = 139264, .AvgSize = 36421},
+ ChunkedParams{.UseThreshold = false, .MinSize = 17424, .MaxSize = 139264, .AvgSize = 36459},
+ ChunkedParams{.UseThreshold = false, .MinSize = 17472, .MaxSize = 139264, .AvgSize = 36502},
+ ChunkedParams{.UseThreshold = false, .MinSize = 17520, .MaxSize = 139264, .AvgSize = 36540},
+ ChunkedParams{.UseThreshold = false, .MinSize = 17808, .MaxSize = 143360, .AvgSize = 37423},
+ ChunkedParams{.UseThreshold = false, .MinSize = 17856, .MaxSize = 143360, .AvgSize = 37466},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18000, .MaxSize = 143360, .AvgSize = 25834},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18000, .MaxSize = 143360, .AvgSize = 21917},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18000, .MaxSize = 143360, .AvgSize = 29751},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18000, .MaxSize = 143360, .AvgSize = 33668},
+ ChunkedParams{.UseThreshold = false, .MinSize = 17952, .MaxSize = 143360, .AvgSize = 37547},
+ ChunkedParams{.UseThreshold = false, .MinSize = 17904, .MaxSize = 143360, .AvgSize = 37504},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18336, .MaxSize = 147456, .AvgSize = 22371},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18000, .MaxSize = 143360, .AvgSize = 37585},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18336, .MaxSize = 147456, .AvgSize = 26406},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18384, .MaxSize = 147456, .AvgSize = 26450},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18528, .MaxSize = 147456, .AvgSize = 30615},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18336, .MaxSize = 147456, .AvgSize = 30441},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18384, .MaxSize = 147456, .AvgSize = 22417},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18528, .MaxSize = 147456, .AvgSize = 22557},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18432, .MaxSize = 147456, .AvgSize = 30528},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18816, .MaxSize = 151552, .AvgSize = 27112},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18528, .MaxSize = 147456, .AvgSize = 34644},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18336, .MaxSize = 147456, .AvgSize = 34476},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18816, .MaxSize = 151552, .AvgSize = 35408},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18432, .MaxSize = 147456, .AvgSize = 38592},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18384, .MaxSize = 147456, .AvgSize = 30483},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18528, .MaxSize = 147456, .AvgSize = 26586},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18432, .MaxSize = 147456, .AvgSize = 26496},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18864, .MaxSize = 151552, .AvgSize = 31302},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18384, .MaxSize = 147456, .AvgSize = 34516},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18816, .MaxSize = 151552, .AvgSize = 22964},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18864, .MaxSize = 151552, .AvgSize = 35448},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18480, .MaxSize = 147456, .AvgSize = 38630},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18864, .MaxSize = 151552, .AvgSize = 23010},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18816, .MaxSize = 151552, .AvgSize = 31260},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18480, .MaxSize = 147456, .AvgSize = 34600},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18864, .MaxSize = 151552, .AvgSize = 27156},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18480, .MaxSize = 147456, .AvgSize = 30570},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18384, .MaxSize = 147456, .AvgSize = 38549},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18480, .MaxSize = 147456, .AvgSize = 22510},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18528, .MaxSize = 147456, .AvgSize = 38673},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18432, .MaxSize = 147456, .AvgSize = 34560},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18432, .MaxSize = 147456, .AvgSize = 22464},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18480, .MaxSize = 147456, .AvgSize = 26540},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18336, .MaxSize = 147456, .AvgSize = 38511},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18912, .MaxSize = 151552, .AvgSize = 23057},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18912, .MaxSize = 151552, .AvgSize = 27202},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18912, .MaxSize = 151552, .AvgSize = 31347},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18912, .MaxSize = 151552, .AvgSize = 35492},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18960, .MaxSize = 151552, .AvgSize = 31389},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18960, .MaxSize = 151552, .AvgSize = 27246},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18960, .MaxSize = 151552, .AvgSize = 23103},
+ ChunkedParams{.UseThreshold = false, .MinSize = 18960, .MaxSize = 151552, .AvgSize = 35532},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19008, .MaxSize = 151552, .AvgSize = 23150},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19008, .MaxSize = 151552, .AvgSize = 27292},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19008, .MaxSize = 151552, .AvgSize = 31434},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19008, .MaxSize = 151552, .AvgSize = 35576},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19056, .MaxSize = 151552, .AvgSize = 27336},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19056, .MaxSize = 151552, .AvgSize = 23196},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19056, .MaxSize = 151552, .AvgSize = 31476},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19056, .MaxSize = 151552, .AvgSize = 35616},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19344, .MaxSize = 155648, .AvgSize = 27862},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19344, .MaxSize = 155648, .AvgSize = 32121},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19344, .MaxSize = 155648, .AvgSize = 23603},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19344, .MaxSize = 155648, .AvgSize = 36380},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19392, .MaxSize = 155648, .AvgSize = 27908},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19392, .MaxSize = 155648, .AvgSize = 23650},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19392, .MaxSize = 155648, .AvgSize = 32166},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19392, .MaxSize = 155648, .AvgSize = 36424},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19440, .MaxSize = 155648, .AvgSize = 23696},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19488, .MaxSize = 155648, .AvgSize = 32253},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19440, .MaxSize = 155648, .AvgSize = 32208},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19488, .MaxSize = 155648, .AvgSize = 23743},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19536, .MaxSize = 155648, .AvgSize = 36548},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19536, .MaxSize = 155648, .AvgSize = 28042},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19536, .MaxSize = 155648, .AvgSize = 23789},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19536, .MaxSize = 155648, .AvgSize = 32295},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19488, .MaxSize = 155648, .AvgSize = 36508},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19440, .MaxSize = 155648, .AvgSize = 27952},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19488, .MaxSize = 155648, .AvgSize = 27998},
+ ChunkedParams{.UseThreshold = false, .MinSize = 19440, .MaxSize = 155648, .AvgSize = 36464}};
+
+ static const size_t ParamsCount = sizeof(Params) / sizeof(ChunkedParams);
+ std::vector<ChunkedInfoWithSource> Infos1(SourceFiles1.size());
+ std::vector<ChunkedInfoWithSource> Infos2(SourceFiles2.size());
+
+ WorkerThreadPool WorkerPool(32);
+
+ for (size_t I = 0; I < ParamsCount; I++)
+ {
+ for (int UseThreshold = 0; UseThreshold < 2; UseThreshold++)
+ {
+ Latch WorkLatch(1);
+ ChunkedParams Param = Params[I];
+ Param.UseThreshold = UseThreshold == 1;
+ Stopwatch Timer;
+ for (size_t F = 0; F < SourceFiles1.size(); F++)
+ {
+ WorkLatch.AddCount(1);
+ WorkerPool.ScheduleWork([&WorkLatch, F, Param, &SourceFiles1, &Infos1]() {
+ auto _ = MakeGuard([&WorkLatch]() { WorkLatch.CountDown(); });
+ BasicFile SourceData1;
+ SourceData1.Open(SourceFiles1[F], BasicFile::Mode::kRead);
+ Infos1[F] = ChunkData(SourceData1, 0, SourceData1.FileSize(), Param);
+ });
+ }
+ for (size_t F = 0; F < SourceFiles2.size(); F++)
+ {
+ WorkLatch.AddCount(1);
+ WorkerPool.ScheduleWork([&WorkLatch, F, Param, &SourceFiles2, &Infos2]() {
+ auto _ = MakeGuard([&WorkLatch]() { WorkLatch.CountDown(); });
+ BasicFile SourceData2;
+ SourceData2.Open(SourceFiles2[F], BasicFile::Mode::kRead);
+ Infos2[F] = ChunkData(SourceData2, 0, SourceData2.FileSize(), Param);
+ });
+ }
+ WorkLatch.CountDown();
+ WorkLatch.Wait();
+ uint64_t ChunkTimeMS = Timer.GetElapsedTimeMs();
+
+ uint64_t Raw1Size = 0;
+ tsl::robin_set<IoHash> Chunks1;
+ size_t ChunkedSize1 = 0;
+ for (size_t F = 0; F < SourceFiles1.size(); F++)
+ {
+ const ChunkedInfoWithSource& Info = Infos1[F];
+ Raw1Size += Info.Info.RawSize;
+ for (uint32_t Chunk1Index = 0; Chunk1Index < Info.Info.ChunkHashes.size(); ++Chunk1Index)
+ {
+ const IoHash ChunkHash = Info.Info.ChunkHashes[Chunk1Index];
+ if (Chunks1.insert(ChunkHash).second)
+ {
+ ChunkedSize1 += Info.ChunkSources[Chunk1Index].Size;
+ }
+ }
+ }
+
+ uint64_t Raw2Size = 0;
+ tsl::robin_set<IoHash> Chunks2;
+ size_t ChunkedSize2 = 0;
+ size_t DiffSize = 0;
+ for (size_t F = 0; F < SourceFiles2.size(); F++)
+ {
+ const ChunkedInfoWithSource& Info = Infos2[F];
+ Raw2Size += Info.Info.RawSize;
+ for (uint32_t Chunk2Index = 0; Chunk2Index < Info.Info.ChunkHashes.size(); ++Chunk2Index)
+ {
+ const IoHash ChunkHash = Info.Info.ChunkHashes[Chunk2Index];
+ if (Chunks2.insert(ChunkHash).second)
+ {
+ ChunkedSize2 += Info.ChunkSources[Chunk2Index].Size;
+ if (!Chunks1.contains(ChunkHash))
+ {
+ DiffSize += Info.ChunkSources[Chunk2Index].Size;
+ }
+ }
+ }
+ }
+
+ ZEN_INFO(
+ "Diff = {}, Chunks1 = {}, Chunks2 = {}, .UseThreshold = {}, .MinSize = {}, .MaxSize = {}, .AvgSize = {}, RawSize(1) = {}, "
+ "RawSize(2) = {}, "
+ "Saved(1) = {}, Saved(2) = {} in {}",
+ NiceBytes(DiffSize),
+ Chunks1.size(),
+ Chunks2.size(),
+ Param.UseThreshold,
+ Param.MinSize,
+ Param.MaxSize,
+ Param.AvgSize,
+ NiceBytes(Raw1Size),
+ NiceBytes(Raw2Size),
+ NiceBytes(Raw1Size - ChunkedSize1),
+ NiceBytes(Raw2Size - ChunkedSize2),
+ NiceTimeSpanMs(ChunkTimeMS));
+ }
+ }
+
+# if 0
+ for (int64_t MinSizeBase = (12u * 1024u); MinSizeBase <= (32u * 1024u); MinSizeBase += 512)
+ {
+ for (int64_t Wiggle = -132; Wiggle < 126; Wiggle += 2)
+ {
+ // size_t MinSize = 7 * 1024 - 61; // (size_t)(MinSizeBase + Wiggle);
+ // size_t MaxSize = 16 * (7 * 1024); // 8 * 7 * 1024;// MinSizeBase * 6;
+ // size_t AvgSize = MaxSize / 2; // 4 * 7 * 1024;// MinSizeBase * 3;
+ size_t MinSize = (size_t)(MinSizeBase + Wiggle);
+ //for (size_t MaxSize = (MinSize * 4) - 768; MaxSize < (MinSize * 5) + 768; MaxSize += 64)
+ size_t MaxSize = 8u * MinSizeBase;
+ {
+ for (size_t AvgSize = (MaxSize - MinSize) / 32 + MinSize; AvgSize < (MaxSize - MinSize) / 4 + MinSize; AvgSize += (MaxSize - MinSize) / 32)
+// size_t AvgSize = (MaxSize - MinSize) / 4 + MinSize;
+ {
+ WorkLatch.AddCount(1);
+ WorkerPool.ScheduleWork([&WorkLatch, MinSize, MaxSize, AvgSize, SourcePath1, SourcePath2]()
+ {
+ auto _ = MakeGuard([&WorkLatch]() { WorkLatch.CountDown(); });
+ ChunkedParams Params{ .UseThreshold = true, .MinSize = MinSize, .MaxSize = MaxSize, .AvgSize = AvgSize };
+ BasicFile SourceData1;
+ SourceData1.Open(SourcePath1, BasicFile::Mode::kRead);
+ BasicFile SourceData2;
+ SourceData2.Open(SourcePath2, BasicFile::Mode::kRead);
+ ChunkedInfoWithSource Info1 = ChunkData(SourceData1, Params);
+ ChunkedInfoWithSource Info2 = ChunkData(SourceData2, Params);
+
+ tsl::robin_set<IoHash> Chunks1;
+ Chunks1.reserve(Info1.Info.ChunkHashes.size());
+ Chunks1.insert(Info1.Info.ChunkHashes.begin(), Info1.Info.ChunkHashes.end());
+ size_t ChunkedSize1 = 0;
+ for (uint32_t Chunk1Index = 0; Chunk1Index < Info1.Info.ChunkHashes.size(); ++Chunk1Index)
+ {
+ ChunkedSize1 += Info1.ChunkSources[Chunk1Index].Size;
+ }
+ size_t DiffSavedSize = 0;
+ size_t ChunkedSize2 = 0;
+ for (uint32_t Chunk2Index = 0; Chunk2Index < Info2.Info.ChunkHashes.size(); ++Chunk2Index)
+ {
+ ChunkedSize2 += Info2.ChunkSources[Chunk2Index].Size;
+ if (Chunks1.find(Info2.Info.ChunkHashes[Chunk2Index]) == Chunks1.end())
+ {
+ DiffSavedSize += Info2.ChunkSources[Chunk2Index].Size;
+ }
+ }
+ ZEN_INFO("Diff {}, Chunks1: {}, Chunks2: {}, Min: {}, Max: {}, Avg: {}, Saved(1) {}, Saved(2) {}",
+ NiceBytes(DiffSavedSize),
+ Info1.Info.ChunkHashes.size(),
+ Info2.Info.ChunkHashes.size(),
+ MinSize,
+ MaxSize,
+ AvgSize,
+ NiceBytes(Info1.Info.RawSize - ChunkedSize1),
+ NiceBytes(Info2.Info.RawSize - ChunkedSize2));
+ });
+ }
+ }
+ }
+ }
+# endif // 0
+
+ // WorkLatch.CountDown();
+ // WorkLatch.Wait();
+}
+# endif // 0
+
+void
+chunkedfile_forcelink()
+{
+}
+
+} // namespace zen
+
+#endif