Method work

src/Processors/Transforms/LazyFinalKeyAnalysisTransform.cpp:155–222

}

void LazyFinalKeyAnalysisTransform::work()
{
    auto reading = buildReadingStep(
        metadata_snapshot, mutations_snapshot, storage_snapshot,
        data_settings, data, max_block_numbers_to_read, ranges, query_context);

    /// Count total marks before index analysis.
    size_t total_marks = 0;
    for (const auto & part : reading->getParts())
        total_marks += part.getMarksCount();

    /// Build and apply the IN-set filter for index analysis.
    {
        const auto & primary_key = metadata_snapshot->getPrimaryKey();

        ActionsDAG filter_dag = primary_key.expression->getActionsDAG().clone();
        filter_dag.getOutputs() = filter_dag.findInOutputs(primary_key.column_names);

        auto set_type = std::make_shared<DataTypeSet>();
        ColumnConst::Ptr set_column = ColumnConst::create(ColumnSet::create(1, future_set), 0);

        const auto * key_node = filter_dag.getOutputs().at(0);
        if (filter_dag.getOutputs().size() > 1)
        {
            auto function_tuple = FunctionFactory::instance().get("tuple", query_context);
            key_node = &filter_dag.addFunction(function_tuple, filter_dag.getOutputs(), {});
        }
        const auto * set_node = &filter_dag.addColumn(std::move(set_column), std::move(set_type), "");
        auto function_in = FunctionFactory::instance().get("in", query_context);
        const auto * in_func = &filter_dag.addFunction(function_in, {key_node, set_node}, {});
        filter_dag.getOutputs().push_back(in_func);

        reading->addFilter(std::move(filter_dag), in_func->result_name);
        reading->SourceStepWithFilterBase::applyFilters();
    }

    /// Count marks after index analysis.
    auto analysis_result = reading->selectRangesToRead();
    size_t selected_marks = 0;
    if (analysis_result)
        for (const auto & part : analysis_result->parts_with_ranges)
            selected_marks += part.getMarksCount();

    /// Check if enough marks were filtered by the IN-set.
    float filtered_ratio = total_marks > 0 ? 1.0f - static_cast<float>(selected_marks) / static_cast<float>(total_marks) : 0.0f;

    if (min_filtered_ratio > 0 && filtered_ratio < min_filtered_ratio)
    {
        LOG_TRACE(log,
            "Lazy FINAL disabled: filtered ratio {:.2f} is below threshold {:.2f} "
            "(total_marks={}, selected_marks={}, set_rows={})",
            filtered_ratio, min_filtered_ratio,
            total_marks, selected_marks, future_set->get()->getTotalRowCount());
        should_signal = false;
    }
    else
    {
        LOG_TRACE(log,