Method EstimateIdsLength

catboost/private/libs/quantized_pool/serialization.cpp:775–804

}

size_t NCB::EstimateIdsLength(const TStringBuf path) {
    const auto file = TBlob::FromFile(TString(path));
    const TConstArrayRef<ui8> blob(file.AsUnsignedCharPtr(), file.Size());

    TPoolMetainfo poolMetainfo;
    auto parseMetainfo = [&] (TConstArrayRef<ui8> bytes) {
        const auto poolMetainfoParsed = poolMetainfo.ParseFromArray(bytes.data(), bytes.size());
        CB_ENSURE(poolMetainfoParsed);
    };
    TVector<TVector<NCB::TQuantizedPool::TChunkDescription>> stringColumnChunks;
    bool isFakeColumn = false;
    auto parseColumn = [&] (ui32 columnIndex) -> bool {
        isFakeColumn = NCB::NQuantizationSchemaDetail::IsFakeIndex(columnIndex, poolMetainfo);
        return true;
    };
    size_t estimatedIdsLength = 0;
    auto parseChunk = [&] (TConstArrayRef<ui8> bytes, ui32 docOffset, ui32 docsInChunkCount) {
        const auto* const chunk = flatbuffers::GetRoot<NCB::NIdl::TQuantizedFeatureChunk>(bytes.data());
        if (isFakeColumn && docOffset == 0) {
            estimatedIdsLength += 1 + chunk->Quants()->size() / (docsInChunkCount + 1);
        }
    };
    ParseQuantizedPool(
        parseMetainfo,
        /*parseSchema*/ Nothing(),
        parseColumn,
        parseChunk,
        blob);
    return estimatedIdsLength;
}

void NCB::EstimateGroupSize(
    const TStringBuf path,