Compress large messages with zstd

This commit is contained in:
Yousef Khadadeh 2025-07-28 17:07:39 +03:00
parent eb9db31d31
commit 1300be0227
10 changed files with 81 additions and 60 deletions

View file

@ -2,6 +2,7 @@
#include <cereal/archives/portable_binary.hpp>
#include <cereal/types/string.hpp>
#include "NetLibrary.hpp"
#include <zstd.h>
ByteMemBuf::ByteMemBuf(char* p, size_t l) {
setg(p, p, p + l);
@ -15,16 +16,30 @@ ByteMemStream::ByteMemStream(char* p, size_t l):
rdbuf(&buffer);
}
std::vector<std::shared_ptr<std::stringstream>> fragment_message(std::string_view v, uint64_t stride) {
if(v.length() <= stride)
std::vector<std::shared_ptr<std::stringstream>> fragment_message(std::string_view uncompressedView, uint64_t stride) {
if(uncompressedView.length() <= stride)
return {};
std::vector<char> compressedData(ZSTD_compressBound(uncompressedView.size()));
size_t trueCompressedSize = ZSTD_compress(compressedData.data(), compressedData.size(), uncompressedView.data(), uncompressedView.size(), ZSTD_CLEVEL_DEFAULT);
std::string_view v(compressedData.data(), trueCompressedSize);
uint64_t nextByte = 0;
std::vector<std::shared_ptr<std::stringstream>> toRet;
toRet.emplace_back(std::make_shared<std::stringstream>());
size_t messFirstStride = std::min(stride, v.size());
{
cereal::PortableBinaryOutputArchive m(*toRet.back());
m((MessageCommandType)0, (uint64_t)v.length(), cereal::binary_data(v.data() + nextByte, stride));
m((MessageCommandType)0, (uint64_t)v.length(), cereal::binary_data(v.data() + nextByte, messFirstStride));
}
if(messFirstStride == v.size())
return toRet;
nextByte += stride;
for(;;) {
if((v.length() - nextByte) <= stride) {
@ -44,3 +59,39 @@ std::vector<std::shared_ptr<std::stringstream>> fragment_message(std::string_vie
}
return toRet;
}
void uncompress_and_run_func(PartialFragmentMessage& spfm, std::function<void(cereal::PortableBinaryInputArchive& completeArchive)> runAfterDecoded) {
std::vector<char> uncompressedData(ZSTD_getFrameContentSize(spfm.partialFragmentMessage.data(), spfm.partialFragmentMessage.size()));
size_t trueUncompressedSize = ZSTD_decompress(uncompressedData.data(), uncompressedData.size(), spfm.partialFragmentMessage.data(), spfm.partialFragmentMessage.size());
ByteMemStream completeStrm(uncompressedData.data(), trueUncompressedSize);
cereal::PortableBinaryInputArchive completeArchive(completeStrm);
runAfterDecoded(completeArchive);
spfm.partialFragmentMessage.clear();
spfm.partialFragmentMessageLoc = 0;
}
void decode_fragmented_message(cereal::PortableBinaryInputArchive& inArchive, PartialFragmentMessage& spfm, size_t fragmentMessageStride, std::function<void(cereal::PortableBinaryInputArchive& completeArchive)> runAfterDecoded) {
if(spfm.partialFragmentMessage.size() == 0) {
spfm.partialFragmentMessageLoc = 0;
uint64_t messageSize;
inArchive(messageSize);
spfm.partialFragmentMessage.resize(messageSize);
size_t messStride = std::min(fragmentMessageStride, spfm.partialFragmentMessage.size());
inArchive(cereal::binary_data(spfm.partialFragmentMessage.data() + spfm.partialFragmentMessageLoc, messStride));
spfm.partialFragmentMessageLoc += messStride;
if(messStride == spfm.partialFragmentMessage.size())
uncompress_and_run_func(spfm, runAfterDecoded);
}
else if(spfm.partialFragmentMessage.size() - spfm.partialFragmentMessageLoc <= fragmentMessageStride) {
inArchive(cereal::binary_data(spfm.partialFragmentMessage.data() + spfm.partialFragmentMessageLoc, spfm.partialFragmentMessage.size() - spfm.partialFragmentMessageLoc));
uncompress_and_run_func(spfm, runAfterDecoded);
}
else {
inArchive(cereal::binary_data(spfm.partialFragmentMessage.data() + spfm.partialFragmentMessageLoc, fragmentMessageStride));
spfm.partialFragmentMessageLoc += fragmentMessageStride;
}
}