/root/doris/be/src/vec/functions/match.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | // Licensed to the Apache Software Foundation (ASF) under one |
2 | | // or more contributor license agreements. See the NOTICE file |
3 | | // distributed with this work for additional information |
4 | | // regarding copyright ownership. The ASF licenses this file |
5 | | // to you under the Apache License, Version 2.0 (the |
6 | | // "License"); you may not use this file except in compliance |
7 | | // with the License. You may obtain a copy of the License at |
8 | | // |
9 | | // http://www.apache.org/licenses/LICENSE-2.0 |
10 | | // |
11 | | // Unless required by applicable law or agreed to in writing, |
12 | | // software distributed under the License is distributed on an |
13 | | // "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY |
14 | | // KIND, either express or implied. See the License for the |
15 | | // specific language governing permissions and limitations |
16 | | // under the License. |
17 | | |
18 | | #include "vec/functions/match.h" |
19 | | |
20 | | #include <hs/hs.h> |
21 | | |
22 | | #include "runtime/query_context.h" |
23 | | #include "runtime/runtime_state.h" |
24 | | #include "util/debug_points.h" |
25 | | |
26 | | namespace doris::vectorized { |
27 | | |
28 | | Status FunctionMatchBase::execute_impl(FunctionContext* context, Block& block, |
29 | | const ColumnNumbers& arguments, size_t result, |
30 | 0 | size_t input_rows_count) { |
31 | 0 | ColumnPtr& column_ptr = block.get_by_position(arguments[1]).column; |
32 | 0 | DataTypePtr& type_ptr = block.get_by_position(arguments[1]).type; |
33 | 0 | auto match_query_str = type_ptr->to_string(*column_ptr, 0); |
34 | 0 | std::string column_name = block.get_by_position(arguments[0]).name; |
35 | 0 | auto match_pred_column_name = |
36 | 0 | BeConsts::BLOCK_TEMP_COLUMN_PREFIX + column_name + "_match_" + match_query_str; |
37 | 0 | if (!block.has(match_pred_column_name)) { |
38 | 0 | VLOG_DEBUG << "begin to execute match directly, column_name=" << column_name |
39 | 0 | << ", match_query_str=" << match_query_str; |
40 | 0 | InvertedIndexCtx* inverted_index_ctx = reinterpret_cast<InvertedIndexCtx*>( |
41 | 0 | context->get_function_state(FunctionContext::THREAD_LOCAL)); |
42 | 0 | if (inverted_index_ctx == nullptr) { |
43 | 0 | inverted_index_ctx = reinterpret_cast<InvertedIndexCtx*>( |
44 | 0 | context->get_function_state(FunctionContext::FRAGMENT_LOCAL)); |
45 | 0 | } |
46 | |
|
47 | 0 | const ColumnPtr source_col = |
48 | 0 | block.get_by_position(arguments[0]).column->convert_to_full_column_if_const(); |
49 | 0 | const auto* values = check_and_get_column<ColumnString>(source_col.get()); |
50 | 0 | const ColumnArray* array_col = nullptr; |
51 | 0 | if (source_col->is_column_array()) { |
52 | 0 | if (source_col->is_nullable()) { |
53 | 0 | auto* nullable = check_and_get_column<ColumnNullable>(source_col.get()); |
54 | 0 | array_col = check_and_get_column<ColumnArray>(*nullable->get_nested_column_ptr()); |
55 | 0 | } else { |
56 | 0 | array_col = check_and_get_column<ColumnArray>(source_col.get()); |
57 | 0 | } |
58 | 0 | if (array_col && !array_col->get_data().is_column_string()) { |
59 | 0 | return Status::NotSupported( |
60 | 0 | fmt::format("unsupported nested array of type {} for function {}", |
61 | 0 | is_column_nullable(array_col->get_data()) |
62 | 0 | ? array_col->get_data().get_name() |
63 | 0 | : array_col->get_data().get_family_name(), |
64 | 0 | get_name())); |
65 | 0 | } |
66 | | |
67 | 0 | if (is_column_nullable(array_col->get_data())) { |
68 | 0 | const auto& array_nested_null_column = |
69 | 0 | reinterpret_cast<const ColumnNullable&>(array_col->get_data()); |
70 | 0 | values = check_and_get_column<ColumnString>( |
71 | 0 | *(array_nested_null_column.get_nested_column_ptr())); |
72 | 0 | } else { |
73 | | // array column element is always set Nullable for now. |
74 | 0 | values = check_and_get_column<ColumnString>(*(array_col->get_data_ptr())); |
75 | 0 | } |
76 | 0 | } else if (auto* nullable = check_and_get_column<ColumnNullable>(source_col.get())) { |
77 | | // match null |
78 | 0 | if (type_ptr->is_nullable()) { |
79 | 0 | if (column_ptr->only_null()) { |
80 | 0 | block.get_by_position(result).column = nullable->get_null_map_column_ptr(); |
81 | 0 | return Status::OK(); |
82 | 0 | } |
83 | 0 | } else { |
84 | 0 | values = check_and_get_column<ColumnString>(*nullable->get_nested_column_ptr()); |
85 | 0 | } |
86 | 0 | } |
87 | | |
88 | 0 | if (!values) { |
89 | 0 | LOG(WARNING) << "Illegal column " << source_col->get_name(); |
90 | 0 | return Status::InternalError("Not supported input column types"); |
91 | 0 | } |
92 | | // result column |
93 | 0 | auto res = ColumnUInt8::create(); |
94 | 0 | ColumnUInt8::Container& vec_res = res->get_data(); |
95 | | // set default value to 0, and match functions only need to set 1/true |
96 | 0 | vec_res.resize_fill(input_rows_count); |
97 | 0 | RETURN_IF_ERROR(execute_match( |
98 | 0 | context, column_name, match_query_str, input_rows_count, values, inverted_index_ctx, |
99 | 0 | (array_col ? &(array_col->get_offsets()) : nullptr), vec_res)); |
100 | 0 | block.replace_by_position(result, std::move(res)); |
101 | 0 | } else { |
102 | 0 | auto match_pred_column = |
103 | 0 | block.get_by_name(match_pred_column_name).column->convert_to_full_column_if_const(); |
104 | 0 | block.replace_by_position(result, std::move(match_pred_column)); |
105 | 0 | } |
106 | | |
107 | 0 | return Status::OK(); |
108 | 0 | } |
109 | | |
110 | 0 | inline doris::segment_v2::InvertedIndexQueryType FunctionMatchBase::get_query_type_from_fn_name() { |
111 | 0 | std::string fn_name = get_name(); |
112 | 0 | if (fn_name == MATCH_ANY_FUNCTION) { |
113 | 0 | return doris::segment_v2::InvertedIndexQueryType::MATCH_ANY_QUERY; |
114 | 0 | } else if (fn_name == MATCH_ALL_FUNCTION) { |
115 | 0 | return doris::segment_v2::InvertedIndexQueryType::MATCH_ALL_QUERY; |
116 | 0 | } else if (fn_name == MATCH_PHRASE_FUNCTION) { |
117 | 0 | return doris::segment_v2::InvertedIndexQueryType::MATCH_PHRASE_QUERY; |
118 | 0 | } else if (fn_name == MATCH_PHRASE_PREFIX_FUNCTION) { |
119 | 0 | return doris::segment_v2::InvertedIndexQueryType::MATCH_PHRASE_PREFIX_QUERY; |
120 | 0 | } else if (fn_name == MATCH_PHRASE_REGEXP_FUNCTION) { |
121 | 0 | return doris::segment_v2::InvertedIndexQueryType::MATCH_REGEXP_QUERY; |
122 | 0 | } |
123 | 0 | return doris::segment_v2::InvertedIndexQueryType::UNKNOWN_QUERY; |
124 | 0 | } |
125 | | |
126 | | void FunctionMatchBase::analyse_query_str_token(std::vector<std::string>* query_tokens, |
127 | | InvertedIndexCtx* inverted_index_ctx, |
128 | | const std::string& match_query_str, |
129 | 0 | const std::string& column_name) { |
130 | 0 | VLOG_DEBUG << "begin to run " << get_name() << ", parser_type: " |
131 | 0 | << inverted_index_parser_type_to_string(inverted_index_ctx->parser_type); |
132 | 0 | if (inverted_index_ctx->parser_type == InvertedIndexParserType::PARSER_NONE) { |
133 | 0 | query_tokens->emplace_back(match_query_str); |
134 | 0 | return; |
135 | 0 | } |
136 | 0 | auto reader = doris::segment_v2::InvertedIndexReader::create_reader(inverted_index_ctx, |
137 | 0 | match_query_str); |
138 | 0 | doris::segment_v2::InvertedIndexReader::get_analyse_result( |
139 | 0 | *query_tokens, reader.get(), inverted_index_ctx->analyzer, column_name, |
140 | 0 | get_query_type_from_fn_name()); |
141 | 0 | } |
142 | | |
143 | | inline std::vector<std::string> FunctionMatchBase::analyse_data_token( |
144 | | const std::string& column_name, InvertedIndexCtx* inverted_index_ctx, |
145 | | const ColumnString* string_col, int32_t current_block_row_idx, |
146 | 0 | const ColumnArray::Offsets64* array_offsets, int32_t& current_src_array_offset) { |
147 | 0 | std::vector<std::string> data_tokens; |
148 | 0 | auto query_type = get_query_type_from_fn_name(); |
149 | 0 | if (array_offsets) { |
150 | 0 | for (auto next_src_array_offset = (*array_offsets)[current_block_row_idx]; |
151 | 0 | current_src_array_offset < next_src_array_offset; ++current_src_array_offset) { |
152 | 0 | const auto& str_ref = string_col->get_data_at(current_src_array_offset); |
153 | 0 | if (inverted_index_ctx->parser_type == InvertedIndexParserType::PARSER_NONE) { |
154 | 0 | data_tokens.emplace_back(str_ref.to_string()); |
155 | 0 | continue; |
156 | 0 | } |
157 | 0 | auto reader = doris::segment_v2::InvertedIndexReader::create_reader( |
158 | 0 | inverted_index_ctx, str_ref.to_string()); |
159 | |
|
160 | 0 | std::vector<std::string> element_tokens; |
161 | |
|
162 | 0 | doris::segment_v2::InvertedIndexReader::get_analyse_result( |
163 | 0 | element_tokens, reader.get(), inverted_index_ctx->analyzer, column_name, |
164 | 0 | query_type, false); |
165 | 0 | data_tokens.insert(data_tokens.end(), element_tokens.begin(), element_tokens.end()); |
166 | 0 | } |
167 | 0 | } else { |
168 | 0 | const auto& str_ref = string_col->get_data_at(current_block_row_idx); |
169 | 0 | if (inverted_index_ctx->parser_type == InvertedIndexParserType::PARSER_NONE) { |
170 | 0 | data_tokens.emplace_back(str_ref.to_string()); |
171 | 0 | } else { |
172 | 0 | auto reader = doris::segment_v2::InvertedIndexReader::create_reader( |
173 | 0 | inverted_index_ctx, str_ref.to_string()); |
174 | 0 | doris::segment_v2::InvertedIndexReader::get_analyse_result( |
175 | 0 | data_tokens, reader.get(), inverted_index_ctx->analyzer, column_name, |
176 | 0 | query_type, false); |
177 | 0 | } |
178 | 0 | } |
179 | 0 | return data_tokens; |
180 | 0 | } |
181 | | |
182 | 0 | Status FunctionMatchBase::check(FunctionContext* context, const std::string& function_name) const { |
183 | 0 | if (!context->state()->query_options().enable_match_without_inverted_index) { |
184 | 0 | return Status::Error<ErrorCode::INVERTED_INDEX_NOT_SUPPORTED>( |
185 | 0 | "{} not support execute_match", function_name); |
186 | 0 | } |
187 | | |
188 | 0 | DBUG_EXECUTE_IF("match.invert_index_not_support_execute_match", { |
189 | 0 | return Status::Error<ErrorCode::INVERTED_INDEX_NOT_SUPPORTED>( |
190 | 0 | "{} not support execute_match", function_name); |
191 | 0 | }); |
192 | |
|
193 | 0 | return Status::OK(); |
194 | 0 | } |
195 | | |
196 | | Status FunctionMatchAny::execute_match(FunctionContext* context, const std::string& column_name, |
197 | | const std::string& match_query_str, size_t input_rows_count, |
198 | | const ColumnString* string_col, |
199 | | InvertedIndexCtx* inverted_index_ctx, |
200 | | const ColumnArray::Offsets64* array_offsets, |
201 | 0 | ColumnUInt8::Container& result) { |
202 | 0 | RETURN_IF_ERROR(check(context, name)); |
203 | | |
204 | 0 | std::vector<std::string> query_tokens; |
205 | 0 | analyse_query_str_token(&query_tokens, inverted_index_ctx, match_query_str, column_name); |
206 | 0 | if (query_tokens.empty()) { |
207 | 0 | LOG(WARNING) << fmt::format( |
208 | 0 | "token parser result is empty for query, " |
209 | 0 | "please check your query: '{}' and index parser: '{}'", |
210 | 0 | match_query_str, |
211 | 0 | inverted_index_parser_type_to_string(inverted_index_ctx->parser_type)); |
212 | 0 | return Status::OK(); |
213 | 0 | } |
214 | | |
215 | 0 | auto current_src_array_offset = 0; |
216 | 0 | for (int i = 0; i < input_rows_count; i++) { |
217 | 0 | std::vector<std::string> data_tokens = |
218 | 0 | analyse_data_token(column_name, inverted_index_ctx, string_col, i, array_offsets, |
219 | 0 | current_src_array_offset); |
220 | | |
221 | | // TODO: more efficient impl |
222 | 0 | for (auto& token : query_tokens) { |
223 | 0 | auto it = std::find(data_tokens.begin(), data_tokens.end(), token); |
224 | 0 | if (it != data_tokens.end()) { |
225 | 0 | result[i] = true; |
226 | 0 | break; |
227 | 0 | } |
228 | 0 | } |
229 | 0 | } |
230 | |
|
231 | 0 | return Status::OK(); |
232 | 0 | } |
233 | | |
234 | | Status FunctionMatchAll::execute_match(FunctionContext* context, const std::string& column_name, |
235 | | const std::string& match_query_str, size_t input_rows_count, |
236 | | const ColumnString* string_col, |
237 | | InvertedIndexCtx* inverted_index_ctx, |
238 | | const ColumnArray::Offsets64* array_offsets, |
239 | 0 | ColumnUInt8::Container& result) { |
240 | 0 | RETURN_IF_ERROR(check(context, name)); |
241 | | |
242 | 0 | std::vector<std::string> query_tokens; |
243 | 0 | analyse_query_str_token(&query_tokens, inverted_index_ctx, match_query_str, column_name); |
244 | 0 | if (query_tokens.empty()) { |
245 | 0 | LOG(WARNING) << fmt::format( |
246 | 0 | "token parser result is empty for query, " |
247 | 0 | "please check your query: '{}' and index parser: '{}'", |
248 | 0 | match_query_str, |
249 | 0 | inverted_index_parser_type_to_string(inverted_index_ctx->parser_type)); |
250 | 0 | return Status::OK(); |
251 | 0 | } |
252 | | |
253 | 0 | auto current_src_array_offset = 0; |
254 | 0 | for (int i = 0; i < input_rows_count; i++) { |
255 | 0 | std::vector<std::string> data_tokens = |
256 | 0 | analyse_data_token(column_name, inverted_index_ctx, string_col, i, array_offsets, |
257 | 0 | current_src_array_offset); |
258 | | |
259 | | // TODO: more efficient impl |
260 | 0 | auto find_count = 0; |
261 | 0 | for (auto& token : query_tokens) { |
262 | 0 | auto it = std::find(data_tokens.begin(), data_tokens.end(), token); |
263 | 0 | if (it != data_tokens.end()) { |
264 | 0 | ++find_count; |
265 | 0 | } else { |
266 | 0 | break; |
267 | 0 | } |
268 | 0 | } |
269 | |
|
270 | 0 | if (find_count == query_tokens.size()) { |
271 | 0 | result[i] = true; |
272 | 0 | } |
273 | 0 | } |
274 | |
|
275 | 0 | return Status::OK(); |
276 | 0 | } |
277 | | |
278 | | Status FunctionMatchPhrase::execute_match(FunctionContext* context, const std::string& column_name, |
279 | | const std::string& match_query_str, |
280 | | size_t input_rows_count, const ColumnString* string_col, |
281 | | InvertedIndexCtx* inverted_index_ctx, |
282 | | const ColumnArray::Offsets64* array_offsets, |
283 | 0 | ColumnUInt8::Container& result) { |
284 | 0 | RETURN_IF_ERROR(check(context, name)); |
285 | | |
286 | 0 | std::vector<std::string> query_tokens; |
287 | 0 | analyse_query_str_token(&query_tokens, inverted_index_ctx, match_query_str, column_name); |
288 | 0 | if (query_tokens.empty()) { |
289 | 0 | VLOG_DEBUG << fmt::format( |
290 | 0 | "token parser result is empty for query, " |
291 | 0 | "please check your query: '{}' and index parser: '{}'", |
292 | 0 | match_query_str, |
293 | 0 | inverted_index_parser_type_to_string(inverted_index_ctx->parser_type)); |
294 | 0 | return Status::OK(); |
295 | 0 | } |
296 | | |
297 | 0 | auto current_src_array_offset = 0; |
298 | 0 | for (int i = 0; i < input_rows_count; i++) { |
299 | 0 | std::vector<std::string> data_tokens = |
300 | 0 | analyse_data_token(column_name, inverted_index_ctx, string_col, i, array_offsets, |
301 | 0 | current_src_array_offset); |
302 | | |
303 | | // TODO: more efficient impl |
304 | 0 | bool matched = false; |
305 | 0 | auto data_it = data_tokens.begin(); |
306 | 0 | while (data_it != data_tokens.end()) { |
307 | | // find position of first token |
308 | 0 | data_it = std::find(data_it, data_tokens.end(), query_tokens[0]); |
309 | 0 | if (data_it != data_tokens.end()) { |
310 | 0 | matched = true; |
311 | 0 | auto data_it_next = ++data_it; |
312 | 0 | auto query_it = query_tokens.begin() + 1; |
313 | | // compare query_tokens after the first to data_tokens one by one |
314 | 0 | while (query_it != query_tokens.end()) { |
315 | 0 | if (data_it_next == data_tokens.end() || *data_it_next != *query_it) { |
316 | 0 | matched = false; |
317 | 0 | break; |
318 | 0 | } |
319 | 0 | query_it++; |
320 | 0 | data_it_next++; |
321 | 0 | } |
322 | |
|
323 | 0 | if (matched) { |
324 | 0 | break; |
325 | 0 | } |
326 | 0 | } |
327 | 0 | } |
328 | | |
329 | | // check matched |
330 | 0 | if (matched) { |
331 | 0 | result[i] = true; |
332 | 0 | } |
333 | 0 | } |
334 | |
|
335 | 0 | return Status::OK(); |
336 | 0 | } |
337 | | |
338 | | Status FunctionMatchPhrasePrefix::execute_match( |
339 | | FunctionContext* context, const std::string& column_name, |
340 | | const std::string& match_query_str, size_t input_rows_count, const ColumnString* string_col, |
341 | | InvertedIndexCtx* inverted_index_ctx, const ColumnArray::Offsets64* array_offsets, |
342 | 0 | ColumnUInt8::Container& result) { |
343 | 0 | RETURN_IF_ERROR(check(context, name)); |
344 | | |
345 | 0 | std::vector<std::string> query_tokens; |
346 | 0 | analyse_query_str_token(&query_tokens, inverted_index_ctx, match_query_str, column_name); |
347 | 0 | if (query_tokens.empty()) { |
348 | 0 | VLOG_DEBUG << fmt::format( |
349 | 0 | "token parser result is empty for query, " |
350 | 0 | "please check your query: '{}' and index parser: '{}'", |
351 | 0 | match_query_str, |
352 | 0 | inverted_index_parser_type_to_string(inverted_index_ctx->parser_type)); |
353 | 0 | return Status::OK(); |
354 | 0 | } |
355 | | |
356 | 0 | int32_t current_src_array_offset = 0; |
357 | 0 | for (size_t i = 0; i < input_rows_count; i++) { |
358 | 0 | auto data_tokens = analyse_data_token(column_name, inverted_index_ctx, string_col, i, |
359 | 0 | array_offsets, current_src_array_offset); |
360 | |
|
361 | 0 | int32_t dis_count = data_tokens.size() - query_tokens.size(); |
362 | 0 | if (dis_count < 0) { |
363 | 0 | continue; |
364 | 0 | } |
365 | | |
366 | 0 | for (size_t j = 0; j < dis_count + 1; j++) { |
367 | 0 | if (data_tokens[j] == query_tokens[0] || query_tokens.size() == 1) { |
368 | 0 | bool match = true; |
369 | 0 | for (size_t k = 0; k < query_tokens.size(); k++) { |
370 | 0 | const std::string& data_token = data_tokens[j + k]; |
371 | 0 | const std::string& query_token = query_tokens[k]; |
372 | 0 | if (k == query_tokens.size() - 1) { |
373 | 0 | if (data_token.compare(0, query_token.size(), query_token) != 0) { |
374 | 0 | match = false; |
375 | 0 | break; |
376 | 0 | } |
377 | 0 | } else { |
378 | 0 | if (data_token != query_token) { |
379 | 0 | match = false; |
380 | 0 | break; |
381 | 0 | } |
382 | 0 | } |
383 | 0 | } |
384 | 0 | if (match) { |
385 | 0 | result[i] = true; |
386 | 0 | break; |
387 | 0 | } |
388 | 0 | } |
389 | 0 | } |
390 | 0 | } |
391 | |
|
392 | 0 | return Status::OK(); |
393 | 0 | } |
394 | | |
395 | | Status FunctionMatchRegexp::execute_match(FunctionContext* context, const std::string& column_name, |
396 | | const std::string& match_query_str, |
397 | | size_t input_rows_count, const ColumnString* string_col, |
398 | | InvertedIndexCtx* inverted_index_ctx, |
399 | | const ColumnArray::Offsets64* array_offsets, |
400 | 0 | ColumnUInt8::Container& result) { |
401 | 0 | RETURN_IF_ERROR(check(context, name)); |
402 | | |
403 | 0 | VLOG_DEBUG << "begin to run FunctionMatchRegexp::execute_match, parser_type: " |
404 | 0 | << inverted_index_parser_type_to_string(inverted_index_ctx->parser_type); |
405 | |
|
406 | 0 | if (match_query_str.empty()) { |
407 | 0 | VLOG_DEBUG << fmt::format( |
408 | 0 | "token parser result is empty for query, " |
409 | 0 | "please check your query: '{}' and index parser: '{}'", |
410 | 0 | match_query_str, |
411 | 0 | inverted_index_parser_type_to_string(inverted_index_ctx->parser_type)); |
412 | 0 | return Status::OK(); |
413 | 0 | } |
414 | | |
415 | 0 | const std::string& pattern = match_query_str; |
416 | |
|
417 | 0 | hs_database_t* database = nullptr; |
418 | 0 | hs_compile_error_t* compile_err = nullptr; |
419 | 0 | hs_scratch_t* scratch = nullptr; |
420 | |
|
421 | 0 | if (hs_compile(pattern.data(), HS_FLAG_DOTALL | HS_FLAG_ALLOWEMPTY | HS_FLAG_UTF8, |
422 | 0 | HS_MODE_BLOCK, nullptr, &database, &compile_err) != HS_SUCCESS) { |
423 | 0 | LOG(ERROR) << "hyperscan compilation failed: " << compile_err->message; |
424 | 0 | hs_free_compile_error(compile_err); |
425 | 0 | return Status::Error<ErrorCode::INVERTED_INDEX_INVALID_PARAMETERS>( |
426 | 0 | std::string("hyperscan compilation failed:") + compile_err->message); |
427 | 0 | } |
428 | | |
429 | 0 | if (hs_alloc_scratch(database, &scratch) != HS_SUCCESS) { |
430 | 0 | LOG(ERROR) << "hyperscan could not allocate scratch space."; |
431 | 0 | hs_free_database(database); |
432 | 0 | return Status::Error<ErrorCode::INVERTED_INDEX_INVALID_PARAMETERS>( |
433 | 0 | "hyperscan could not allocate scratch space."); |
434 | 0 | } |
435 | | |
436 | 0 | auto on_match = [](unsigned int id, unsigned long long from, unsigned long long to, |
437 | 0 | unsigned int flags, void* context) -> int { |
438 | 0 | *((bool*)context) = true; |
439 | 0 | return 0; |
440 | 0 | }; |
441 | |
|
442 | 0 | try { |
443 | 0 | auto current_src_array_offset = 0; |
444 | 0 | for (int i = 0; i < input_rows_count; i++) { |
445 | 0 | std::vector<std::string> data_tokens = |
446 | 0 | analyse_data_token(column_name, inverted_index_ctx, string_col, i, |
447 | 0 | array_offsets, current_src_array_offset); |
448 | |
|
449 | 0 | for (auto& input : data_tokens) { |
450 | 0 | bool is_match = false; |
451 | 0 | if (hs_scan(database, input.data(), input.size(), 0, scratch, on_match, |
452 | 0 | (void*)&is_match) != HS_SUCCESS) { |
453 | 0 | LOG(ERROR) << "hyperscan match failed: " << input; |
454 | 0 | break; |
455 | 0 | } |
456 | | |
457 | 0 | if (is_match) { |
458 | 0 | result[i] = true; |
459 | 0 | break; |
460 | 0 | } |
461 | 0 | } |
462 | 0 | } |
463 | 0 | } |
464 | 0 | _CLFINALLY({ |
465 | 0 | hs_free_scratch(scratch); |
466 | 0 | hs_free_database(database); |
467 | 0 | }) |
468 | |
|
469 | 0 | return Status::OK(); |
470 | 0 | } |
471 | | |
472 | 1 | void register_function_match(SimpleFunctionFactory& factory) { |
473 | 1 | factory.register_function<FunctionMatchAny>(); |
474 | 1 | factory.register_function<FunctionMatchAll>(); |
475 | 1 | factory.register_function<FunctionMatchPhrase>(); |
476 | 1 | factory.register_function<FunctionMatchPhrasePrefix>(); |
477 | 1 | factory.register_function<FunctionMatchRegexp>(); |
478 | 1 | } |
479 | | |
480 | | } // namespace doris::vectorized |