Method match

src/Dictionaries/RegExpTreeDictionary.cpp:685–869

};

UnorderedMapWithMemoryTracking<String, ColumnPtr> RegExpTreeDictionary::match(
    const ColumnString::Chars & keys_data,
    const ColumnString::Offsets & keys_offsets,
    const UnorderedMapWithMemoryTracking<String, const DictionaryAttribute &> & attributes,
    DefaultMapOrFilter default_or_filter,
    std::optional<size_t> collect_values_limit) const
{
    bool is_short_circuit = std::holds_alternative<RefFilter>(default_or_filter);
    chassert(is_short_circuit || std::holds_alternative<RefDefaultMap>(default_or_filter));

#if USE_VECTORSCAN
    hs_scratch_t * scratch = nullptr;
    if (use_vectorscan)
    {
        hs_error_t err = hs_clone_scratch(origin_scratch.get(), &scratch);

        if (err != HS_SUCCESS)
        {
            throw Exception(ErrorCodes::CANNOT_ALLOCATE_MEMORY, "Could not clone scratch space for hyperscan");
        }
    }

    MultiRegexps::ScratchPtr smart_scratch(scratch);
#endif

    UnorderedMapWithMemoryTracking<String, MutableColumnPtr> columns;

    size_t input_rows_count = keys_offsets.size();

    /// initialize columns
    for (const auto & [name_, attr] : attributes)
    {
        auto col_ptr = (collect_values_limit ? std::make_shared<DataTypeArray>(attr.type) : attr.type)->createColumn();
        col_ptr->reserve(input_rows_count);
        columns[name_] = std::move(col_ptr);
    }

    std::optional<RefDefaultMap> default_map;
    std::optional<RefFilter> default_mask;
    if (is_short_circuit)
    {
        default_mask = std::get<RefFilter>(default_or_filter).get();
        default_mask.value().get().resize(keys_offsets.size());
    }
    else
    {
        default_map = std::get<RefDefaultMap>(default_or_filter).get();
    }

    UInt64 curr_offset = 0;
    for (size_t key_idx = 0; key_idx < input_rows_count; ++key_idx)
    {
        auto next_offset = keys_offsets[key_idx];
        UInt64 length = next_offset - curr_offset;

        const char * begin = reinterpret_cast<const char *>(keys_data.data()) + curr_offset;

        MatchContext match_result(regexp_ids, topology_order, begin, length, regex_nodes);