Function extractValuesString

src/DataTypes/DataTypeMapHelpers.cpp:278–333

Specialized value extractor for ColumnString, with optional Nullable support. Two-pass approach: first pass computes offsets, total chars size, and fills null map; second pass re-iterates matched_positions to copy string data.

/// Two-pass approach: first pass computes offsets, total chars size, and fills null map;
/// second pass re-iterates matched_positions to copy string data.
void extractValuesString(
    const ColumnString & values_column,
    ColumnString & result,
    const PaddedPODArray<size_t> & matched_positions,
    const NullMap * src_null_map = nullptr,
    NullMap * dst_null_map = nullptr)
{
    const auto & src_chars = values_column.getChars();
    const auto & src_offsets = values_column.getOffsets();
    auto & dst_chars = result.getChars();
    auto & dst_offsets = result.getOffsets();

    size_t old_offsets_size = dst_offsets.size();
    size_t num_rows = matched_positions.size();

    dst_offsets.resize(old_offsets_size + num_rows);
    if (dst_null_map)
        dst_null_map->resize(dst_null_map->size() + num_rows);

    /// First pass: compute result offsets, total chars size, and fill null map.
    /// total_chars_size must start from the existing chars size because
    /// ColumnString offsets are absolute positions into the chars array.
    size_t old_chars_size = dst_chars.size();
    size_t total_chars_size = old_chars_size;
    for (size_t i = 0; i < num_rows; ++i)
    {
        size_t pos = matched_positions[i];
        if (pos != KEY_NOT_FOUND)
        {
            total_chars_size += src_offsets[pos] - src_offsets[ssize_t(pos) - 1];
            if (dst_null_map)
                (*dst_null_map)[old_offsets_size + i] = (*src_null_map)[pos];
        }
        else
        {
            if (dst_null_map)
                (*dst_null_map)[old_offsets_size + i] = static_cast<UInt8>(1);
        }
        dst_offsets[old_offsets_size + i] = total_chars_size;
    }

    /// Second pass: resize chars once and copy string data.
    dst_chars.resize(total_chars_size);
    size_t current_offset = old_chars_size;
    for (size_t i = 0; i < num_rows; ++i)
    {
        size_t pos = matched_positions[i];
        if (pos != KEY_NOT_FOUND)
        {
            size_t src_offset = src_offsets[ssize_t(pos) - 1];
            size_t src_size = src_offsets[pos] - src_offset;
            memcpy(&dst_chars[current_offset], &src_chars[src_offset], src_size);
            current_offset += src_size;
        }
    }
}

/// Specialized value extractor for ColumnFixedString, with optional Nullable support.