Method getSampling

src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp:219–493

}

MergeTreeDataSelectSamplingData MergeTreeDataSelectExecutor::getSampling(
    const SelectQueryInfo & select_query_info,
    NamesAndTypesList available_real_columns,
    const RangesInDataParts & parts,
    KeyCondition & key_condition,
    const MergeTreeData & data,
    const StorageMetadataPtr & metadata_snapshot,
    ContextPtr context,
    LoggerPtr log)
{
    const Settings & settings = context->getSettingsRef();
    /// Sampling.
    MergeTreeDataSelectSamplingData sampling;

    RelativeSize relative_sample_size = 0;
    RelativeSize relative_sample_offset = 0;

    std::optional<ASTSampleRatio::Rational> sample_size_ratio;
    std::optional<ASTSampleRatio::Rational> sample_offset_ratio;

    if (select_query_info.table_expression_modifiers)
    {
        const auto & table_expression_modifiers = *select_query_info.table_expression_modifiers;
        sample_size_ratio = table_expression_modifiers.getSampleSizeRatio();
        sample_offset_ratio = table_expression_modifiers.getSampleOffsetRatio();
    }
    else
    {
        auto & select = select_query_info.query->as<ASTSelectQuery &>();

        auto select_sample_size = select.sampleSize();
        auto select_sample_offset = select.sampleOffset();

        if (select_sample_size)
            sample_size_ratio = select_sample_size->as<ASTSampleRatio &>().ratio;

        if (select_sample_offset)
            sample_offset_ratio = select_sample_offset->as<ASTSampleRatio &>().ratio;
    }

    if (sample_size_ratio)
    {
        relative_sample_size.assign(sample_size_ratio->numerator, sample_size_ratio->denominator);

        if (relative_sample_size < 0)
            throw Exception(ErrorCodes::ARGUMENT_OUT_OF_BOUND, "Negative sample size");

        relative_sample_offset = 0;
        if (sample_offset_ratio)
            relative_sample_offset.assign(sample_offset_ratio->numerator, sample_offset_ratio->denominator);

        if (relative_sample_offset < 0)
            throw Exception(ErrorCodes::ARGUMENT_OUT_OF_BOUND, "Negative sample offset");

        /// Convert absolute value of the sampling (in form `SAMPLE 1000000` - how many rows to
        /// read) into the relative `SAMPLE 0.1` (how much data to read).
        size_t approx_total_rows = 0;
        if (relative_sample_size > 1 || relative_sample_offset > 1)