Method joinBlock

src/Interpreters/DirectJoin.cpp:142–210

}

JoinResultPtr DirectKeyValueJoin::joinBlock(Block block)
{
    const String & key_name = table_join->getOnlyClause().key_names_left[0];
    const ColumnWithTypeAndName & key_col = block.getByName(key_name);
    if (!key_col.column)
        return {};

    Block original_right_block = originalRightBlock(right_sample_block, *table_join);
    Block right_block_to_use = !right_sample_block_with_storage_column_names.empty() ? right_sample_block_with_storage_column_names : original_right_block;
    const Names & attribute_names = right_block_to_use.getNames();

    bool is_all_join = table_join->strictness() == JoinStrictness::All;
    bool is_semi_join = table_join->strictness() == JoinStrictness::Semi;
    bool is_anti_join = table_join->strictness() == JoinStrictness::Anti;

    NullMap null_map;
    IColumn::Offsets offsets;
    Chunk joined_chunk = storage->getByKeys({key_col}, attribute_names, null_map, offsets);

    /// Expected right block may differ from structure in storage, because of `join_use_nulls` or we just select not all joined attributes
    Block sample_storage_block = storage->getSampleBlock(attribute_names);
    MutableColumns result_columns = convertBlockStructure(sample_storage_block, right_block_to_use, joined_chunk.mutateColumns(), null_map);

    if (!offsets.empty())
    {
        /// For ALL semantics with offsets, replicate left rows
        if (is_all_join)
        {
            MutableColumns replicated_columns = block.mutateColumns();
            for (auto && col : replicated_columns)
                col = IColumn::mutate(col->replicate(offsets));
            block.setColumns(std::move(replicated_columns));
        }
        else
        {
            /// For ANY/ANTI/SEMI semantics right columns are not replicated
            /// We need to 'unreplicate' them by keeping only the first match
            selectFirstMatchForEachKey(offsets, result_columns, null_map);
        }
    }

    for (size_t i = 0; i < result_columns.size(); ++i)
    {
        ColumnWithTypeAndName col = right_sample_block.getByPosition(i);
        col.column = std::move(result_columns[i]);
        block.insert(std::move(col));
    }

    if (is_anti_join)
    {
        /// invert null_map
        for (auto & val : null_map)
            val = !val;
    }

    size_t non_null_rows = countBytesInFilter(null_map);
    bool has_null_matches = non_null_rows < null_map.size();
    bool need_filtering = (isInner(table_join->kind()) || (isLeft(table_join->kind()) && (is_semi_join || is_anti_join)));