#pragma once #include #include #include namespace DB { namespace ErrorCodes { extern const int NUMBER_OF_ARGUMENTS_DOESNT_MATCH; } template struct MultiSearchImpl { using ResultType = UInt8; /// Variable for understanding, if we used offsets for the output, most /// likely to determine whether the function returns ColumnVector of ColumnArray. static constexpr bool is_column_array = false; static constexpr auto name = Name::name; static auto getReturnType() { return std::make_shared>(); } static void vectorConstant( const ColumnString::Chars & haystack_data, const ColumnString::Offsets & haystack_offsets, const Array & needles_arr, PaddedPODArray & res, [[maybe_unused]] PaddedPODArray & offsets, bool /*allow_hyperscan*/, size_t /*max_hyperscan_regexp_length*/, size_t /*max_hyperscan_regexp_total_length*/) { // For performance of Volnitsky search, it is crucial to save only one byte for pattern number. if (needles_arr.size() > std::numeric_limits::max()) throw Exception(ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, "Number of arguments for function {} doesn't match: passed {}, should be at most {}", name, std::to_string(needles_arr.size()), std::to_string(std::numeric_limits::max())); std::vector needles; needles.reserve(needles_arr.size()); for (const auto & needle : needles_arr) needles.emplace_back(needle.get()); auto searcher = Impl::createMultiSearcherInBigHaystack(needles); const size_t haystack_string_size = haystack_offsets.size(); res.resize(haystack_string_size); size_t iteration = 0; while (searcher.hasMoreToSearch()) { size_t prev_offset = 0; for (size_t j = 0; j < haystack_string_size; ++j) { const auto * haystack = &haystack_data[prev_offset]; const auto * haystack_end = haystack + haystack_offsets[j] - prev_offset - 1; if (iteration == 0 || !res[j]) res[j] = searcher.searchOne(haystack, haystack_end); prev_offset = haystack_offsets[j]; } ++iteration; } if (iteration == 0) std::fill(res.begin(), res.end(), 0); } }; }