Function checkDataPart

src/Storages/MergeTree/checkDataPart.cpp:217–467

IMergeTreeDataPart::Checksums checkDataPart(
    MergeTreeData::DataPartPtr data_part,
    const DiskPtr & disk,
    const String & full_relative_path,
    const NamesAndTypesList & columns_list,
    const MergeTreeDataPartType & part_type,
    const NameSet & files_without_checksums,
    bool require_checksums,
    std::function<bool()> is_cancelled)
{
    /** Responsibility:
      * - read list of columns from columns.txt;
      * - read checksums if exist;
      * - validate list of columns and checksums
      */

    CurrentMetrics::Increment metric_increment{CurrentMetrics::ReplicatedChecks};

    String path = full_relative_path;
    if (!path.empty() && path.back() != '/')
        path += "/";

    NamesAndTypesList columns_txt;

    {
        auto buf = disk->readFile(fs::path(path) / "columns.txt");
        columns_txt.readText(*buf);
        assertEOF(*buf);
    }

    if (columns_txt != columns_list)
        throw Exception("Columns doesn't match in part " + path
            + ". Expected: " + columns_list.toString()
            + ". Found: " + columns_txt.toString(), ErrorCodes::CORRUPTED_DATA);

    /// Real checksums based on contents of data. Must correspond to checksums.txt. If not - it means the data is broken.
    IMergeTreeDataPart::Checksums checksums_data;

    /// This function calculates checksum for both compressed and decompressed contents of compressed file.
    auto checksum_compressed_file = [](const DiskPtr & disk_, const String & file_path)
    {
        auto file_buf = disk_->readFile(file_path);
        HashingReadBuffer compressed_hashing_buf(*file_buf);
        CompressedReadBuffer uncompressing_buf(compressed_hashing_buf);
        HashingReadBuffer uncompressed_hashing_buf(uncompressing_buf);

        uncompressed_hashing_buf.ignoreAll();
        return IMergeTreeDataPart::Checksums::Checksum
        {
            compressed_hashing_buf.count(), compressed_hashing_buf.getHash(),
            uncompressed_hashing_buf.count(), uncompressed_hashing_buf.getHash()
        };
    };

    /// This function calculates only checksum of file content (compressed or uncompressed).
    /// It also calculates checksum of projections.
    auto checksum_file = [&](const String & file_path, const String & file_name)
    {