aboutsummaryrefslogtreecommitdiff
path: root/src/zencore/compress.cpp
diff options
context:
space:
mode:
authorDan Engelbrecht <[email protected]>2025-02-25 15:48:43 +0100
committerGitHub Enterprise <[email protected]>2025-02-25 15:48:43 +0100
commit5bc5b0dd59c0f02afe553e5074dfe57951b19044 (patch)
tree625d46a9ef656cd6dd5f2879182f686b0299f44b /src/zencore/compress.cpp
parent5.5.18 (diff)
downloadzen-5bc5b0dd59c0f02afe553e5074dfe57951b19044.tar.xz
zen-5bc5b0dd59c0f02afe553e5074dfe57951b19044.zip
improvements and infrastructure for upcoming builds api command line (#284)
* add modification tick to filesystem traversal * add ShowDetails option to ProgressBar * log callstack if we terminate process * handle chunking if MaxSize > 1MB * BasicFile write helpers and WriteToTempFile simplifications * bugfix for CompositeBuffer::IterateRange when using DecompressToComposite for actually comrpessed data revert of earlier optimization * faster compress/decompress for large disk-based files * enable progress feedback in IoHash::HashBuffer * add payload validation in HttpClient::Get * fix range requests (range is including end byte) * remove BuildPartId for blob/block related operations in builds api
Diffstat (limited to 'src/zencore/compress.cpp')
-rw-r--r--src/zencore/compress.cpp214
1 files changed, 161 insertions, 53 deletions
diff --git a/src/zencore/compress.cpp b/src/zencore/compress.cpp
index 29c1d9256..0e2ce2b54 100644
--- a/src/zencore/compress.cpp
+++ b/src/zencore/compress.cpp
@@ -2,6 +2,7 @@
#include <zencore/compress.h>
+#include <zencore/basicfile.h>
#include <zencore/blake3.h>
#include <zencore/compositebuffer.h>
#include <zencore/crc32.h>
@@ -314,37 +315,77 @@ BlockEncoder::Compress(const CompositeBuffer& RawData, const uint64_t BlockSize)
CompressedBlockSizes.reserve(BlockCount);
uint64_t CompressedSize = 0;
{
- UniqueBuffer RawBlockCopy;
MutableMemoryView CompressedBlocksView = CompressedData.GetMutableView() + sizeof(BufferHeader) + MetaSize;
- CompositeBuffer::Iterator It = RawData.GetIterator(0);
-
- for (uint64_t RawOffset = 0; RawOffset < RawSize;)
+ IoBufferFileReference FileRef = {nullptr, 0, 0};
+ if ((RawData.GetSegments().size() == 1) && RawData.GetSegments()[0].AsIoBuffer().GetFileReference(FileRef))
{
- const uint64_t RawBlockSize = zen::Min(RawSize - RawOffset, BlockSize);
- const MemoryView RawBlock = RawData.ViewOrCopyRange(It, RawBlockSize, RawBlockCopy);
- RawHash.Append(RawBlock);
-
- MutableMemoryView CompressedBlock = CompressedBlocksView;
- if (!CompressBlock(CompressedBlock, RawBlock))
+ ZEN_ASSERT(FileRef.FileHandle != nullptr);
+ UniqueBuffer RawBlockCopy = UniqueBuffer::Alloc(BlockSize);
+ BasicFile Source;
+ Source.Attach(FileRef.FileHandle);
+ for (uint64_t RawOffset = 0; RawOffset < RawSize;)
{
- return CompositeBuffer();
- }
+ const uint64_t RawBlockSize = zen::Min(RawSize - RawOffset, BlockSize);
+ Source.Read(RawBlockCopy.GetData(), RawBlockSize, FileRef.FileChunkOffset + RawOffset);
+ const MemoryView RawBlock = RawBlockCopy.GetView().Left(RawBlockSize);
+ RawHash.Append(RawBlock);
+ MutableMemoryView CompressedBlock = CompressedBlocksView;
+ if (!CompressBlock(CompressedBlock, RawBlock))
+ {
+ Source.Detach();
+ return CompositeBuffer();
+ }
- uint64_t CompressedBlockSize = CompressedBlock.GetSize();
- if (RawBlockSize <= CompressedBlockSize)
- {
- CompressedBlockSize = RawBlockSize;
- CompressedBlocksView = CompressedBlocksView.CopyFrom(RawBlock);
+ uint64_t CompressedBlockSize = CompressedBlock.GetSize();
+ if (RawBlockSize <= CompressedBlockSize)
+ {
+ CompressedBlockSize = RawBlockSize;
+ CompressedBlocksView = CompressedBlocksView.CopyFrom(RawBlock);
+ }
+ else
+ {
+ CompressedBlocksView += CompressedBlockSize;
+ }
+
+ CompressedBlockSizes.push_back(static_cast<uint32_t>(CompressedBlockSize));
+ CompressedSize += CompressedBlockSize;
+ RawOffset += RawBlockSize;
}
- else
+ Source.Detach();
+ }
+ else
+ {
+ UniqueBuffer RawBlockCopy;
+ CompositeBuffer::Iterator It = RawData.GetIterator(0);
+
+ for (uint64_t RawOffset = 0; RawOffset < RawSize;)
{
- CompressedBlocksView += CompressedBlockSize;
- }
+ const uint64_t RawBlockSize = zen::Min(RawSize - RawOffset, BlockSize);
+ const MemoryView RawBlock = RawData.ViewOrCopyRange(It, RawBlockSize, RawBlockCopy);
+ RawHash.Append(RawBlock);
- CompressedBlockSizes.push_back(static_cast<uint32_t>(CompressedBlockSize));
- CompressedSize += CompressedBlockSize;
- RawOffset += RawBlockSize;
+ MutableMemoryView CompressedBlock = CompressedBlocksView;
+ if (!CompressBlock(CompressedBlock, RawBlock))
+ {
+ return CompositeBuffer();
+ }
+
+ uint64_t CompressedBlockSize = CompressedBlock.GetSize();
+ if (RawBlockSize <= CompressedBlockSize)
+ {
+ CompressedBlockSize = RawBlockSize;
+ CompressedBlocksView = CompressedBlocksView.CopyFrom(RawBlock);
+ }
+ else
+ {
+ CompressedBlocksView += CompressedBlockSize;
+ }
+
+ CompressedBlockSizes.push_back(static_cast<uint32_t>(CompressedBlockSize));
+ CompressedSize += CompressedBlockSize;
+ RawOffset += RawBlockSize;
+ }
}
}
@@ -560,51 +601,118 @@ BlockDecoder::TryDecompressTo(const BufferHeader& Header,
CompressedOffset += CompressedBlockSize;
}
- for (size_t BlockIndex = FirstBlockIndex; BlockIndex <= LastBlockIndex; BlockIndex++)
+ IoBufferFileReference FileRef = {nullptr, 0, 0};
+ if ((CompressedData.GetSegments().size() == 1) && CompressedData.GetSegments()[0].AsIoBuffer().GetFileReference(FileRef))
{
- const uint64_t UncompressedBlockSize = BlockIndex == Header.BlockCount - 1 ? LastBlockSize : BlockSize;
- const uint32_t CompressedBlockSize = ByteSwap(CompressedBlockSizes[BlockIndex]);
- const bool IsCompressed = CompressedBlockSize < UncompressedBlockSize;
+ ZEN_ASSERT(FileRef.FileHandle != nullptr);
+ BasicFile Source;
+ Source.Attach(FileRef.FileHandle);
- const uint64_t BytesToUncompress = OffsetInFirstBlock > 0 ? zen::Min(RawView.GetSize(), UncompressedBlockSize - OffsetInFirstBlock)
- : zen::Min(RemainingRawSize, BlockSize);
+ for (size_t BlockIndex = FirstBlockIndex; BlockIndex <= LastBlockIndex; BlockIndex++)
+ {
+ const uint64_t UncompressedBlockSize = BlockIndex == Header.BlockCount - 1 ? LastBlockSize : BlockSize;
+ const uint32_t CompressedBlockSize = ByteSwap(CompressedBlockSizes[BlockIndex]);
+ const bool IsCompressed = CompressedBlockSize < UncompressedBlockSize;
- MemoryView CompressedBlock = CompressedData.ViewOrCopyRange(CompressedOffset, CompressedBlockSize, CompressedBlockCopy);
+ const uint64_t BytesToUncompress = OffsetInFirstBlock > 0
+ ? zen::Min(RawView.GetSize(), UncompressedBlockSize - OffsetInFirstBlock)
+ : zen::Min(RemainingRawSize, BlockSize);
- if (IsCompressed)
- {
- MutableMemoryView UncompressedBlock = RawView.Left(BytesToUncompress);
+ if (CompressedBlockCopy.GetSize() < CompressedBlockSize)
+ {
+ CompressedBlockCopy = UniqueBuffer::Alloc(CompressedBlockSize);
+ }
+ Source.Read(CompressedBlockCopy.GetData(), CompressedBlockSize, FileRef.FileChunkOffset + CompressedOffset);
- const bool IsAligned = BytesToUncompress == UncompressedBlockSize;
- if (!IsAligned)
+ MemoryView CompressedBlock = CompressedBlockCopy.GetView().Left(CompressedBlockSize);
+
+ if (IsCompressed)
{
- // Decompress to a temporary buffer when the first or the last block reads are not aligned with the block boundaries.
- if (UncompressedBlockCopy.IsNull())
+ MutableMemoryView UncompressedBlock = RawView.Left(BytesToUncompress);
+
+ const bool IsAligned = BytesToUncompress == UncompressedBlockSize;
+ if (!IsAligned)
{
- UncompressedBlockCopy = UniqueBuffer::Alloc(BlockSize);
+ // Decompress to a temporary buffer when the first or the last block reads are not aligned with the block boundaries.
+ if (UncompressedBlockCopy.IsNull())
+ {
+ UncompressedBlockCopy = UniqueBuffer::Alloc(BlockSize);
+ }
+ UncompressedBlock = UncompressedBlockCopy.GetMutableView().Mid(0, UncompressedBlockSize);
}
- UncompressedBlock = UncompressedBlockCopy.GetMutableView().Mid(0, UncompressedBlockSize);
- }
- if (!DecompressBlock(UncompressedBlock, CompressedBlock))
- {
- return false;
- }
+ if (!DecompressBlock(UncompressedBlock, CompressedBlock))
+ {
+ Source.Detach();
+ return false;
+ }
- if (!IsAligned)
+ if (!IsAligned)
+ {
+ RawView.CopyFrom(UncompressedBlock.Mid(OffsetInFirstBlock, BytesToUncompress));
+ }
+ }
+ else
{
- RawView.CopyFrom(UncompressedBlock.Mid(OffsetInFirstBlock, BytesToUncompress));
+ RawView.CopyFrom(CompressedBlock.Mid(OffsetInFirstBlock, BytesToUncompress));
}
+
+ OffsetInFirstBlock = 0;
+ RemainingRawSize -= BytesToUncompress;
+ CompressedOffset += CompressedBlockSize;
+ RawView += BytesToUncompress;
}
- else
+ Source.Detach();
+ }
+ else
+ {
+ for (size_t BlockIndex = FirstBlockIndex; BlockIndex <= LastBlockIndex; BlockIndex++)
{
- RawView.CopyFrom(CompressedBlock.Mid(OffsetInFirstBlock, BytesToUncompress));
- }
+ const uint64_t UncompressedBlockSize = BlockIndex == Header.BlockCount - 1 ? LastBlockSize : BlockSize;
+ const uint32_t CompressedBlockSize = ByteSwap(CompressedBlockSizes[BlockIndex]);
+ const bool IsCompressed = CompressedBlockSize < UncompressedBlockSize;
- OffsetInFirstBlock = 0;
- RemainingRawSize -= BytesToUncompress;
- CompressedOffset += CompressedBlockSize;
- RawView += BytesToUncompress;
+ const uint64_t BytesToUncompress = OffsetInFirstBlock > 0
+ ? zen::Min(RawView.GetSize(), UncompressedBlockSize - OffsetInFirstBlock)
+ : zen::Min(RemainingRawSize, BlockSize);
+
+ MemoryView CompressedBlock = CompressedData.ViewOrCopyRange(CompressedOffset, CompressedBlockSize, CompressedBlockCopy);
+
+ if (IsCompressed)
+ {
+ MutableMemoryView UncompressedBlock = RawView.Left(BytesToUncompress);
+
+ const bool IsAligned = BytesToUncompress == UncompressedBlockSize;
+ if (!IsAligned)
+ {
+ // Decompress to a temporary buffer when the first or the last block reads are not aligned with the block boundaries.
+ if (UncompressedBlockCopy.IsNull())
+ {
+ UncompressedBlockCopy = UniqueBuffer::Alloc(BlockSize);
+ }
+ UncompressedBlock = UncompressedBlockCopy.GetMutableView().Mid(0, UncompressedBlockSize);
+ }
+
+ if (!DecompressBlock(UncompressedBlock, CompressedBlock))
+ {
+ return false;
+ }
+
+ if (!IsAligned)
+ {
+ RawView.CopyFrom(UncompressedBlock.Mid(OffsetInFirstBlock, BytesToUncompress));
+ }
+ }
+ else
+ {
+ RawView.CopyFrom(CompressedBlock.Mid(OffsetInFirstBlock, BytesToUncompress));
+ }
+
+ OffsetInFirstBlock = 0;
+ RemainingRawSize -= BytesToUncompress;
+ CompressedOffset += CompressedBlockSize;
+ RawView += BytesToUncompress;
+ }
}
return RemainingRawSize == 0;