diff options
Diffstat (limited to 'src/zenutil/chunkingcontroller.cpp')
| -rw-r--r-- | src/zenutil/chunkingcontroller.cpp | 359 |
1 files changed, 359 insertions, 0 deletions
diff --git a/src/zenutil/chunkingcontroller.cpp b/src/zenutil/chunkingcontroller.cpp new file mode 100644 index 000000000..6fb4182c0 --- /dev/null +++ b/src/zenutil/chunkingcontroller.cpp @@ -0,0 +1,359 @@ +// Copyright Epic Games, Inc. All Rights Reserved. + +#include <zenutil/chunkingcontroller.h> + +#include <zencore/basicfile.h> +#include <zencore/compactbinarybuilder.h> +#include <zencore/filesystem.h> +#include <zencore/trace.h> + +ZEN_THIRD_PARTY_INCLUDES_START +#include <tsl/robin_map.h> +ZEN_THIRD_PARTY_INCLUDES_END + +namespace zen { +using namespace std::literals; + +namespace { + std::vector<std::string> ReadStringArray(CbArrayView StringArray) + { + std::vector<std::string> Result; + Result.reserve(StringArray.Num()); + for (CbFieldView FieldView : StringArray) + { + Result.emplace_back(FieldView.AsString()); + } + return Result; + } + + ChunkedParams ReadChunkParams(CbObjectView Params) + { + bool UseThreshold = Params["UseThreshold"sv].AsBool(true); + size_t MinSize = Params["MinSize"sv].AsUInt64(DefaultChunkedParams.MinSize); + size_t MaxSize = Params["MaxSize"sv].AsUInt64(DefaultChunkedParams.MaxSize); + size_t AvgSize = Params["AvgSize"sv].AsUInt64(DefaultChunkedParams.AvgSize); + + return ChunkedParams{.UseThreshold = UseThreshold, .MinSize = MinSize, .MaxSize = MaxSize, .AvgSize = AvgSize}; + } + + void WriteChunkParams(CbObjectWriter& Writer, const ChunkedParams& Params) + { + Writer.BeginObject("ChunkingParams"sv); + { + Writer.AddBool("UseThreshold"sv, Params.UseThreshold); + + Writer.AddInteger("MinSize"sv, (uint64_t)Params.MinSize); + Writer.AddInteger("MaxSize"sv, (uint64_t)Params.MaxSize); + Writer.AddInteger("AvgSize"sv, (uint64_t)Params.AvgSize); + } + Writer.EndObject(); // ChunkingParams + } + + bool IsElfFile(BasicFile& Buffer) + { + if (Buffer.FileSize() > 4) + { + uint32_t ElfCheck = 0; + Buffer.Read(&ElfCheck, 4, 0); + if (ElfCheck == 0x464c457f) + { + return true; + } + } + return false; + } + + bool IsMachOFile(BasicFile& Buffer) + { + if (Buffer.FileSize() > 4) + { + uint32_t MachOCheck = 0; + Buffer.Read(&MachOCheck, 4, 0); + if ((MachOCheck == 0xfeedface) || (MachOCheck == 0xcefaedfe)) + { + return true; + } + } + return false; + } +} // namespace + +class BasicChunkingController : public ChunkingController +{ +public: + BasicChunkingController(const BasicChunkingControllerSettings& Settings) : m_Settings(Settings) {} + + BasicChunkingController(CbObjectView Parameters) : m_Settings(ReadSettings(Parameters)) {} + + virtual bool ProcessFile(const std::filesystem::path& InputPath, + uint64_t RawSize, + ChunkedInfoWithSource& OutChunked, + std::atomic<uint64_t>& BytesProcessed, + std::atomic<bool>& AbortFlag) const override + { + ZEN_TRACE_CPU("BasicChunkingController::ProcessFile"); + const bool ExcludeFromChunking = + std::find(m_Settings.ExcludeExtensions.begin(), m_Settings.ExcludeExtensions.end(), InputPath.extension()) != + m_Settings.ExcludeExtensions.end(); + + if (ExcludeFromChunking || (RawSize < m_Settings.ChunkFileSizeLimit)) + { + return false; + } + + BasicFile Buffer(InputPath, BasicFile::Mode::kRead); + if (m_Settings.ExcludeElfFiles && IsElfFile(Buffer)) + { + return false; + } + if (m_Settings.ExcludeMachOFiles && IsMachOFile(Buffer)) + { + return false; + } + + OutChunked = ChunkData(Buffer, 0, RawSize, m_Settings.ChunkingParams, &BytesProcessed, &AbortFlag); + return true; + } + + virtual std::string_view GetName() const override { return Name; } + + virtual CbObject GetParameters() const override + { + CbObjectWriter Writer; + Writer.BeginArray("ChunkExcludeExtensions"sv); + { + for (const std::string& Extension : m_Settings.ExcludeExtensions) + { + Writer.AddString(Extension); + } + } + Writer.EndArray(); // ChunkExcludeExtensions + + Writer.AddBool("ExcludeElfFiles"sv, m_Settings.ExcludeElfFiles); + Writer.AddBool("ExcludeMachOFiles"sv, m_Settings.ExcludeMachOFiles); + Writer.AddInteger("ChunkFileSizeLimit"sv, m_Settings.ChunkFileSizeLimit); + + WriteChunkParams(Writer, m_Settings.ChunkingParams); + + return Writer.Save(); + } + static constexpr std::string_view Name = "BasicChunkingController"sv; + +private: + static BasicChunkingControllerSettings ReadSettings(CbObjectView Parameters) + { + return BasicChunkingControllerSettings{ + .ExcludeExtensions = ReadStringArray(Parameters["ChunkExcludeExtensions"sv].AsArrayView()), + .ExcludeElfFiles = Parameters["ExcludeElfFiles"sv].AsBool(DefaultChunkingExcludeElfFiles), + .ExcludeMachOFiles = Parameters["ExcludeMachOFiles"sv].AsBool(DefaultChunkingExcludeMachOFiles), + .ChunkFileSizeLimit = Parameters["ChunkFileSizeLimit"sv].AsUInt64(DefaultChunkingFileSizeLimit), + .ChunkingParams = ReadChunkParams(Parameters["ChunkingParams"sv].AsObjectView())}; + } + + const BasicChunkingControllerSettings m_Settings; +}; + +class ChunkingControllerWithFixedChunking : public ChunkingController +{ +public: + ChunkingControllerWithFixedChunking(const ChunkingControllerWithFixedChunkingSettings& Settings) : m_Settings(Settings) {} + + ChunkingControllerWithFixedChunking(CbObjectView Parameters) : m_Settings(ReadSettings(Parameters)) {} + + virtual bool ProcessFile(const std::filesystem::path& InputPath, + uint64_t RawSize, + ChunkedInfoWithSource& OutChunked, + std::atomic<uint64_t>& BytesProcessed, + std::atomic<bool>& AbortFlag) const override + { + ZEN_TRACE_CPU("ChunkingControllerWithFixedChunking::ProcessFile"); + const bool ExcludeFromChunking = + std::find(m_Settings.ExcludeExtensions.begin(), m_Settings.ExcludeExtensions.end(), InputPath.extension()) != + m_Settings.ExcludeExtensions.end(); + + if (ExcludeFromChunking || (RawSize < m_Settings.ChunkFileSizeLimit)) + { + return false; + } + + const bool FixedChunkingExtension = + std::find(m_Settings.FixedChunkingExtensions.begin(), m_Settings.FixedChunkingExtensions.end(), InputPath.extension()) != + m_Settings.FixedChunkingExtensions.end(); + + if (FixedChunkingExtension) + { + if (RawSize < m_Settings.MinSizeForFixedChunking) + { + return false; + } + ZEN_TRACE_CPU("FixedChunking"); + IoHashStream FullHasher; + BasicFile Source(InputPath, BasicFile::Mode::kRead); + uint64_t Offset = 0; + tsl::robin_map<IoHash, uint32_t, IoHash::Hasher> ChunkHashToChunkIndex; + const uint64_t ExpectedChunkCount = 1 + (RawSize / m_Settings.FixedChunkingChunkSize); + ChunkHashToChunkIndex.reserve(ExpectedChunkCount); + OutChunked.Info.ChunkHashes.reserve(ExpectedChunkCount); + OutChunked.Info.ChunkSequence.reserve(ExpectedChunkCount); + OutChunked.ChunkSources.reserve(ExpectedChunkCount); + + static const uint64_t BufferingSize = 256u * 1024u; + + IoHashStream ChunkHasher; + + while (Offset < RawSize) + { + if (AbortFlag) + { + return false; + } + + ChunkHasher.Reset(); + + uint64_t ChunkSize = std::min<uint64_t>(RawSize - Offset, m_Settings.FixedChunkingChunkSize); + if (ChunkSize >= (BufferingSize + BufferingSize / 2)) + { + ScanFile(Source.Handle(), + Offset, + ChunkSize, + BufferingSize, + [&FullHasher, &ChunkHasher, &BytesProcessed](const void* Data, size_t Size) { + FullHasher.Append(Data, Size); + ChunkHasher.Append(Data, Size); + BytesProcessed.fetch_add(Size); + }); + } + else + { + IoBuffer ChunkData = Source.ReadRange(Offset, ChunkSize); + FullHasher.Append(ChunkData); + ChunkHasher.Append(ChunkData); + BytesProcessed.fetch_add(ChunkSize); + } + + const IoHash ChunkHash = ChunkHasher.GetHash(); + if (auto It = ChunkHashToChunkIndex.find(ChunkHash); It != ChunkHashToChunkIndex.end()) + { + OutChunked.Info.ChunkSequence.push_back(It->second); + } + else + { + uint32_t ChunkIndex = gsl::narrow<uint32_t>(OutChunked.Info.ChunkHashes.size()); + OutChunked.Info.ChunkHashes.push_back(ChunkHash); + OutChunked.Info.ChunkSequence.push_back(ChunkIndex); + OutChunked.ChunkSources.push_back({.Offset = Offset, .Size = gsl::narrow<uint32_t>(ChunkSize)}); + } + Offset += ChunkSize; + } + OutChunked.Info.RawSize = RawSize; + OutChunked.Info.RawHash = FullHasher.GetHash(); + return true; + } + else + { + BasicFile Buffer(InputPath, BasicFile::Mode::kRead); + if (m_Settings.ExcludeElfFiles && IsElfFile(Buffer)) + { + return false; + } + if (m_Settings.ExcludeMachOFiles && IsMachOFile(Buffer)) + { + return false; + } + + OutChunked = ChunkData(Buffer, 0, RawSize, m_Settings.ChunkingParams, &BytesProcessed, &AbortFlag); + return true; + } + } + + virtual std::string_view GetName() const override { return Name; } + + virtual CbObject GetParameters() const override + { + CbObjectWriter Writer; + Writer.BeginArray("FixedChunkingExtensions"); + { + for (const std::string& Extension : m_Settings.FixedChunkingExtensions) + { + Writer.AddString(Extension); + } + } + Writer.EndArray(); // ChunkExcludeExtensions + + Writer.BeginArray("ChunkExcludeExtensions"sv); + { + for (const std::string& Extension : m_Settings.ExcludeExtensions) + { + Writer.AddString(Extension); + } + } + Writer.EndArray(); // ChunkExcludeExtensions + + Writer.AddBool("ExcludeElfFiles"sv, m_Settings.ExcludeElfFiles); + Writer.AddBool("ExcludeMachOFiles"sv, m_Settings.ExcludeMachOFiles); + + Writer.AddInteger("ChunkFileSizeLimit"sv, m_Settings.ChunkFileSizeLimit); + + WriteChunkParams(Writer, m_Settings.ChunkingParams); + + Writer.AddInteger("FixedChunkingChunkSize"sv, m_Settings.FixedChunkingChunkSize); + Writer.AddInteger("MinSizeForFixedChunking"sv, m_Settings.MinSizeForFixedChunking); + return Writer.Save(); + } + + static constexpr std::string_view Name = "ChunkingControllerWithFixedChunking"sv; + +private: + static ChunkingControllerWithFixedChunkingSettings ReadSettings(CbObjectView Parameters) + { + return ChunkingControllerWithFixedChunkingSettings{ + .FixedChunkingExtensions = ReadStringArray(Parameters["FixedChunkingExtensions"sv].AsArrayView()), + .ExcludeExtensions = ReadStringArray(Parameters["ChunkExcludeExtensions"sv].AsArrayView()), + .ExcludeElfFiles = Parameters["ExcludeElfFiles"sv].AsBool(DefaultChunkingExcludeElfFiles), + .ExcludeMachOFiles = Parameters["ExcludeMachOFiles"sv].AsBool(DefaultChunkingExcludeMachOFiles), + .ChunkFileSizeLimit = Parameters["ChunkFileSizeLimit"sv].AsUInt64(DefaultChunkingFileSizeLimit), + .ChunkingParams = ReadChunkParams(Parameters["ChunkingParams"sv].AsObjectView()), + .FixedChunkingChunkSize = Parameters["FixedChunkingChunkSize"sv].AsUInt64(DefaultFixedChunkingChunkSize), + .MinSizeForFixedChunking = Parameters["MinSizeForFixedChunking"sv].AsUInt64(DefaultFixedChunkingChunkSize)}; + } + + const ChunkingControllerWithFixedChunkingSettings m_Settings; +}; + +std::unique_ptr<ChunkingController> +CreateBasicChunkingController(const BasicChunkingControllerSettings& Settings) +{ + return std::make_unique<BasicChunkingController>(Settings); +} +std::unique_ptr<ChunkingController> +CreateBasicChunkingController(CbObjectView Parameters) +{ + return std::make_unique<BasicChunkingController>(Parameters); +} + +std::unique_ptr<ChunkingController> +CreateChunkingControllerWithFixedChunking(const ChunkingControllerWithFixedChunkingSettings& Setting) +{ + return std::make_unique<ChunkingControllerWithFixedChunking>(Setting); +} +std::unique_ptr<ChunkingController> +CreateChunkingControllerWithFixedChunking(CbObjectView Parameters) +{ + return std::make_unique<ChunkingControllerWithFixedChunking>(Parameters); +} + +std::unique_ptr<ChunkingController> +CreateChunkingController(std::string_view Name, CbObjectView Parameters) +{ + if (Name == BasicChunkingController::Name) + { + return CreateBasicChunkingController(Parameters); + } + else if (Name == ChunkingControllerWithFixedChunking::Name) + { + return CreateChunkingControllerWithFixedChunking(Parameters); + } + return {}; +} + +} // namespace zen |