Coverage Report

Created: 2025-10-16 19:56

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/root/doris/be/src/vec/functions/like.cpp
Line
Count
Source
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 "like.h"
19
20
#include <fmt/format.h>
21
#include <hs/hs_compile.h>
22
#include <re2/stringpiece.h>
23
24
#include <cstddef>
25
#include <ostream>
26
#include <utility>
27
#include <vector>
28
29
#include "common/logging.h"
30
#include "vec/columns/column.h"
31
#include "vec/columns/column_const.h"
32
#include "vec/columns/column_vector.h"
33
#include "vec/common/string_ref.h"
34
#include "vec/core/block.h"
35
#include "vec/core/column_with_type_and_name.h"
36
#include "vec/functions/simple_function_factory.h"
37
38
namespace doris::vectorized {
39
#include "common/compile_check_begin.h"
40
// A regex to match any regex pattern is equivalent to a substring search.
41
static const RE2 SUBSTRING_RE(R"((?:\.\*)*([^\.\^\{\[\(\|\)\]\}\+\*\?\$\\]*)(?:\.\*)*)");
42
43
// A regex to match any regex pattern which is equivalent to matching a constant string
44
// at the end of the string values.
45
static const RE2 ENDS_WITH_RE(R"((?:\.\*)*([^\.\^\{\[\(\|\)\]\}\+\*\?\$\\]*)\$)");
46
47
// A regex to match any regex pattern which is equivalent to matching a constant string
48
// at the end of the string values.
49
static const RE2 STARTS_WITH_RE(R"(\^([^\.\^\{\[\(\|\)\]\}\+\*\?\$\\]*)(?:\.\*)*)");
50
51
// A regex to match any regex pattern which is equivalent to a constant string match.
52
static const RE2 EQUALS_RE(R"(\^([^\.\^\{\[\(\|\)\]\}\+\*\?\$\\]*)\$)");
53
// A regex to match .*
54
static const RE2 ALLPASS_RE(R"((\.\*)+)");
55
56
// Like patterns
57
static const re2::RE2 LIKE_SUBSTRING_RE(R"((?:%+)(((\\_)|([^%_\\]))+)(?:%+))");
58
static const re2::RE2 LIKE_ENDS_WITH_RE("(?:%+)(((\\\\_)|([^%_]))+)");
59
static const re2::RE2 LIKE_STARTS_WITH_RE(R"((((\\%)|(\\_)|([^%_\\]))+)(?:%+))");
60
static const re2::RE2 LIKE_EQUALS_RE("(((\\\\_)|([^%_]))+)");
61
static const re2::RE2 LIKE_ALLPASS_RE("%+");
62
63
struct VectorAllpassSearchState : public VectorPatternSearchState {
64
307
    VectorAllpassSearchState() : VectorPatternSearchState(FunctionLikeBase::vector_allpass_fn) {}
65
66
307
    ~VectorAllpassSearchState() override = default;
67
68
312
    void like_pattern_match(const std::string& pattern_str) override {
69
312
        if (!pattern_str.empty() && RE2::FullMatch(pattern_str, LIKE_ALLPASS_RE)) {
70
26
            _search_strings->insert_default();
71
286
        } else {
72
286
            _pattern_matched = false;
73
286
        }
74
312
    }
75
76
0
    void regexp_pattern_match(const std::string& pattern_str) override {
77
0
        if (RE2::FullMatch(pattern_str, ALLPASS_RE)) {
78
0
            _search_strings->insert_default();
79
0
        } else {
80
0
            _pattern_matched = false;
81
0
        }
82
0
    }
83
};
84
85
struct VectorEqualSearchState : public VectorPatternSearchState {
86
307
    VectorEqualSearchState() : VectorPatternSearchState(FunctionLikeBase::vector_equals_fn) {}
87
88
307
    ~VectorEqualSearchState() override = default;
89
90
315
    void like_pattern_match(const std::string& pattern_str) override {
91
315
        _search_string.clear();
92
315
        if (pattern_str.empty() || RE2::FullMatch(pattern_str, LIKE_EQUALS_RE, &_search_string)) {
93
86
            FunctionLike::remove_escape_character(&_search_string);
94
86
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
95
229
        } else {
96
229
            _pattern_matched = false;
97
229
        }
98
315
    }
99
100
0
    void regexp_pattern_match(const std::string& pattern_str) override {
101
0
        _search_string.clear();
102
0
        if (RE2::FullMatch(pattern_str, EQUALS_RE, &_search_string)) {
103
0
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
104
0
        } else {
105
0
            _pattern_matched = false;
106
0
        }
107
0
    }
108
};
109
110
struct VectorSubStringSearchState : public VectorPatternSearchState {
111
    VectorSubStringSearchState()
112
307
            : VectorPatternSearchState(FunctionLikeBase::vector_substring_fn) {}
113
114
307
    ~VectorSubStringSearchState() override = default;
115
116
321
    void like_pattern_match(const std::string& pattern_str) override {
117
321
        _search_string.clear();
118
321
        if (RE2::FullMatch(pattern_str, LIKE_SUBSTRING_RE, &_search_string)) {
119
41
            FunctionLike::remove_escape_character(&_search_string);
120
41
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
121
280
        } else {
122
280
            _pattern_matched = false;
123
280
        }
124
321
    }
125
126
0
    void regexp_pattern_match(const std::string& pattern_str) override {
127
0
        _search_string.clear();
128
0
        if (RE2::FullMatch(pattern_str, SUBSTRING_RE, &_search_string)) {
129
0
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
130
0
        } else {
131
0
            _pattern_matched = false;
132
0
        }
133
0
    }
134
};
135
136
struct VectorStartsWithSearchState : public VectorPatternSearchState {
137
    VectorStartsWithSearchState()
138
307
            : VectorPatternSearchState(FunctionLikeBase::vector_starts_with_fn) {}
139
140
307
    ~VectorStartsWithSearchState() override = default;
141
142
313
    void like_pattern_match(const std::string& pattern_str) override {
143
313
        _search_string.clear();
144
313
        if (RE2::FullMatch(pattern_str, LIKE_STARTS_WITH_RE, &_search_string)) {
145
37
            FunctionLike::remove_escape_character(&_search_string);
146
37
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
147
276
        } else {
148
276
            _pattern_matched = false;
149
276
        }
150
313
    }
151
152
0
    void regexp_pattern_match(const std::string& pattern_str) override {
153
0
        _search_string.clear();
154
0
        if (RE2::FullMatch(pattern_str, STARTS_WITH_RE, &_search_string)) {
155
0
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
156
0
        } else {
157
0
            _pattern_matched = false;
158
0
        }
159
0
    }
160
};
161
162
struct VectorEndsWithSearchState : public VectorPatternSearchState {
163
307
    VectorEndsWithSearchState() : VectorPatternSearchState(FunctionLikeBase::vector_ends_with_fn) {}
164
165
307
    ~VectorEndsWithSearchState() override = default;
166
167
312
    void like_pattern_match(const std::string& pattern_str) override {
168
312
        _search_string.clear();
169
312
        if (RE2::FullMatch(pattern_str, LIKE_ENDS_WITH_RE, &_search_string)) {
170
36
            FunctionLike::remove_escape_character(&_search_string);
171
36
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
172
276
        } else {
173
276
            _pattern_matched = false;
174
276
        }
175
312
    }
176
177
0
    void regexp_pattern_match(const std::string& pattern_str) override {
178
0
        _search_string.clear();
179
0
        if (RE2::FullMatch(pattern_str, ENDS_WITH_RE, &_search_string)) {
180
0
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
181
0
        } else {
182
0
            _pattern_matched = false;
183
0
        }
184
0
    }
185
};
186
187
0
Status LikeSearchState::clone(LikeSearchState& cloned) {
188
0
    cloned.set_search_string(search_string);
189
190
0
    std::string re_pattern;
191
0
    FunctionLike::convert_like_pattern(this, pattern_str, &re_pattern);
192
0
    if (hs_database) { // use hyperscan
193
0
        hs_database_t* database = nullptr;
194
0
        hs_scratch_t* scratch = nullptr;
195
0
        RETURN_IF_ERROR(FunctionLike::hs_prepare(nullptr, re_pattern.c_str(), &database, &scratch));
196
197
0
        cloned.hs_database.reset(database);
198
0
        cloned.hs_scratch.reset(scratch);
199
0
    } else { // fallback to re2
200
0
        cloned.hs_database.reset();
201
0
        cloned.hs_scratch.reset();
202
203
0
        RE2::Options opts;
204
0
        opts.set_never_nl(false);
205
0
        opts.set_dot_nl(true);
206
0
        cloned.regex = std::make_unique<RE2>(re_pattern, opts);
207
0
        if (!cloned.regex->ok()) {
208
0
            return Status::InternalError("Invalid regex expression: {}", re_pattern);
209
0
        }
210
0
    }
211
212
0
    return Status::OK();
213
0
}
214
215
Status FunctionLikeBase::constant_allpass_fn(const LikeSearchState* state, const ColumnString& vals,
216
                                             const StringRef& pattern,
217
14
                                             ColumnUInt8::Container& result) {
218
14
    memset(result.data(), 1, vals.size());
219
14
    return Status::OK();
220
14
}
221
222
Status FunctionLikeBase::constant_allpass_fn_scalar(const LikeSearchState* state,
223
                                                    const StringRef& val, const StringRef& pattern,
224
0
                                                    unsigned char* result) {
225
0
    *result = 1;
226
0
    return Status::OK();
227
0
}
228
229
Status FunctionLikeBase::vector_allpass_fn(const ColumnString& vals,
230
                                           const ColumnString& search_strings,
231
21
                                           ColumnUInt8::Container& result) {
232
21
    DCHECK(vals.size() == search_strings.size());
233
21
    DCHECK(vals.size() == result.size());
234
21
    memset(result.data(), 1, vals.size());
235
21
    return Status::OK();
236
21
}
237
238
Status FunctionLikeBase::constant_starts_with_fn(const LikeSearchState* state,
239
                                                 const ColumnString& val, const StringRef& pattern,
240
22
                                                 ColumnUInt8::Container& result) {
241
22
    auto sz = val.size();
242
44
    for (size_t i = 0; i < sz; i++) {
243
22
        const auto& str_ref = val.get_data_at(i);
244
22
        result[i] = (str_ref.size >= state->search_string_sv.size) &&
245
22
                    str_ref.start_with(state->search_string_sv);
246
22
    }
247
22
    return Status::OK();
248
22
}
249
250
Status FunctionLikeBase::constant_starts_with_fn_scalar(const LikeSearchState* state,
251
                                                        const StringRef& val,
252
                                                        const StringRef& pattern,
253
0
                                                        unsigned char* result) {
254
0
    *result = (val.size >= state->search_string_sv.size) &&
255
0
              (state->search_string_sv == val.substring(0, state->search_string_sv.size));
256
0
    return Status::OK();
257
0
}
258
259
Status FunctionLikeBase::vector_starts_with_fn(const ColumnString& vals,
260
                                               const ColumnString& search_strings,
261
31
                                               ColumnUInt8::Container& result) {
262
31
    DCHECK(vals.size() == search_strings.size());
263
31
    DCHECK(vals.size() == result.size());
264
31
    auto sz = vals.size();
265
67
    for (size_t i = 0; i < sz; ++i) {
266
36
        const auto& str_sv = vals.get_data_at(i);
267
36
        const auto& search_string_sv = search_strings.get_data_at(i);
268
36
        result[i] = (str_sv.size >= search_string_sv.size) && str_sv.start_with(search_string_sv);
269
36
    }
270
31
    return Status::OK();
271
31
}
272
273
Status FunctionLikeBase::constant_ends_with_fn(const LikeSearchState* state,
274
                                               const ColumnString& val, const StringRef& pattern,
275
22
                                               ColumnUInt8::Container& result) {
276
22
    auto sz = val.size();
277
44
    for (size_t i = 0; i < sz; i++) {
278
22
        const auto& str_ref = val.get_data_at(i);
279
22
        result[i] = (str_ref.size >= state->search_string_sv.size) &&
280
22
                    str_ref.end_with(state->search_string_sv);
281
22
    }
282
22
    return Status::OK();
283
22
}
284
285
Status FunctionLikeBase::constant_ends_with_fn_scalar(const LikeSearchState* state,
286
                                                      const StringRef& val,
287
                                                      const StringRef& pattern,
288
0
                                                      unsigned char* result) {
289
0
    *result = (val.size >= state->search_string_sv.size) &&
290
0
              (state->search_string_sv == val.substring(val.size - state->search_string_sv.size,
291
0
                                                        state->search_string_sv.size));
292
0
    return Status::OK();
293
0
}
294
295
Status FunctionLikeBase::vector_ends_with_fn(const ColumnString& vals,
296
                                             const ColumnString& search_strings,
297
31
                                             ColumnUInt8::Container& result) {
298
31
    DCHECK(vals.size() == search_strings.size());
299
31
    DCHECK(vals.size() == result.size());
300
31
    auto sz = vals.size();
301
67
    for (size_t i = 0; i < sz; ++i) {
302
36
        const auto& str_sv = vals.get_data_at(i);
303
36
        const auto& search_string_sv = search_strings.get_data_at(i);
304
36
        result[i] = (str_sv.size >= search_string_sv.size) && str_sv.end_with(search_string_sv);
305
36
    }
306
31
    return Status::OK();
307
31
}
308
309
Status FunctionLikeBase::constant_equals_fn(const LikeSearchState* state, const ColumnString& val,
310
                                            const StringRef& pattern,
311
47
                                            ColumnUInt8::Container& result) {
312
47
    auto sz = val.size();
313
94
    for (size_t i = 0; i < sz; i++) {
314
47
        result[i] = (val.get_data_at(i) == state->search_string_sv);
315
47
    }
316
47
    return Status::OK();
317
47
}
318
319
Status FunctionLikeBase::constant_equals_fn_scalar(const LikeSearchState* state,
320
                                                   const StringRef& val, const StringRef& pattern,
321
0
                                                   unsigned char* result) {
322
0
    *result = (val == state->search_string_sv);
323
0
    return Status::OK();
324
0
}
325
326
Status FunctionLikeBase::vector_equals_fn(const ColumnString& vals,
327
                                          const ColumnString& search_strings,
328
78
                                          ColumnUInt8::Container& result) {
329
78
    DCHECK(vals.size() == search_strings.size());
330
78
    DCHECK(vals.size() == result.size());
331
78
    auto sz = vals.size();
332
156
    for (size_t i = 0; i < sz; ++i) {
333
78
        const auto& str_sv = vals.get_data_at(i);
334
78
        const auto& search_string_sv = search_strings.get_data_at(i);
335
78
        result[i] = str_sv == search_string_sv;
336
78
    }
337
78
    return Status::OK();
338
78
}
339
340
Status FunctionLikeBase::constant_substring_fn(const LikeSearchState* state,
341
                                               const ColumnString& val, const StringRef& pattern,
342
0
                                               ColumnUInt8::Container& result) {
343
0
    auto sz = val.size();
344
0
    for (size_t i = 0; i < sz; i++) {
345
0
        if (state->search_string_sv.size == 0) {
346
0
            result[i] = true;
347
0
            continue;
348
0
        }
349
0
        result[i] = state->substring_pattern.search(val.get_data_at(i)) != -1;
350
0
    }
351
0
    return Status::OK();
352
0
}
353
354
Status FunctionLikeBase::constant_substring_fn_scalar(const LikeSearchState* state,
355
                                                      const StringRef& val,
356
                                                      const StringRef& pattern,
357
0
                                                      unsigned char* result) {
358
0
    if (state->search_string_sv.size == 0) {
359
0
        *result = true;
360
0
        return Status::OK();
361
0
    }
362
0
    *result = state->substring_pattern.search(val) != -1;
363
0
    return Status::OK();
364
0
}
365
366
Status FunctionLikeBase::vector_substring_fn(const ColumnString& vals,
367
                                             const ColumnString& search_strings,
368
27
                                             ColumnUInt8::Container& result) {
369
27
    DCHECK(vals.size() == search_strings.size());
370
27
    DCHECK(vals.size() == result.size());
371
27
    auto sz = vals.size();
372
54
    for (size_t i = 0; i < sz; ++i) {
373
27
        const auto& str_sv = vals.get_data_at(i);
374
27
        const auto& search_string_sv = search_strings.get_data_at(i);
375
27
        if (search_string_sv.size == 0) {
376
0
            result[i] = true;
377
0
            continue;
378
0
        }
379
27
        doris::StringSearch substring_search(&search_string_sv);
380
27
        result[i] = substring_search.search(str_sv) != -1;
381
27
    }
382
27
    return Status::OK();
383
27
}
384
385
Status FunctionLikeBase::constant_regex_fn_scalar(const LikeSearchState* state,
386
                                                  const StringRef& val, const StringRef& pattern,
387
53
                                                  unsigned char* result) {
388
53
    if (state->hs_database) { // use hyperscan
389
53
        auto ret = hs_scan(state->hs_database.get(), val.data, (int)val.size, 0,
390
53
                           state->hs_scratch.get(),
391
53
                           doris::vectorized::LikeSearchState::hs_match_handler, (void*)result);
392
53
        if (ret != HS_SUCCESS && ret != HS_SCAN_TERMINATED) {
393
0
            return Status::RuntimeError(fmt::format("hyperscan error: {}", ret));
394
0
        }
395
53
    } else { // fallback to re2
396
0
        *result = RE2::PartialMatch(re2::StringPiece(val.data, val.size), *state->regex);
397
0
    }
398
399
53
    return Status::OK();
400
53
}
401
402
Status FunctionLikeBase::regexp_fn_scalar(const LikeSearchState* state, const StringRef& val,
403
209
                                          const StringRef& pattern, unsigned char* result) {
404
209
    RE2::Options opts;
405
209
    opts.set_never_nl(false);
406
209
    opts.set_dot_nl(true);
407
209
    re2::RE2 re(re2::StringPiece(pattern.data, pattern.size), opts);
408
209
    if (re.ok()) {
409
209
        *result = RE2::PartialMatch(re2::StringPiece(val.data, val.size), re);
410
209
    } else {
411
0
        return Status::RuntimeError("Invalid pattern: {}", pattern.debug_string());
412
0
    }
413
414
209
    return Status::OK();
415
209
}
416
417
Status FunctionLikeBase::constant_regex_fn(const LikeSearchState* state, const ColumnString& val,
418
                                           const StringRef& pattern,
419
75
                                           ColumnUInt8::Container& result) {
420
75
    auto sz = val.size();
421
75
    if (state->hs_database) { // use hyperscan
422
150
        for (size_t i = 0; i < sz; i++) {
423
75
            const auto& str_ref = val.get_data_at(i);
424
75
            auto ret = hs_scan(state->hs_database.get(), str_ref.data, (int)str_ref.size, 0,
425
75
                               state->hs_scratch.get(),
426
75
                               doris::vectorized::LikeSearchState::hs_match_handler,
427
75
                               (void*)(result.data() + i));
428
75
            if (ret != HS_SUCCESS && ret != HS_SCAN_TERMINATED) {
429
0
                return Status::RuntimeError(fmt::format("hyperscan error: {}", ret));
430
0
            }
431
75
        }
432
75
    } else { // fallback to re2
433
0
        for (size_t i = 0; i < sz; i++) {
434
0
            const auto& str_ref = val.get_data_at(i);
435
0
            *(result.data() + i) =
436
0
                    RE2::PartialMatch(re2::StringPiece(str_ref.data, str_ref.size), *state->regex);
437
0
        }
438
0
    }
439
440
75
    return Status::OK();
441
75
}
442
443
Status FunctionLikeBase::regexp_fn(const LikeSearchState* state, const ColumnString& val,
444
0
                                   const StringRef& pattern, ColumnUInt8::Container& result) {
445
0
    std::string re_pattern(pattern.data, pattern.size);
446
447
0
    hs_database_t* database = nullptr;
448
0
    hs_scratch_t* scratch = nullptr;
449
0
    if (hs_prepare(nullptr, re_pattern.c_str(), &database, &scratch).ok()) { // use hyperscan
450
0
        auto sz = val.size();
451
0
        for (size_t i = 0; i < sz; i++) {
452
0
            const auto& str_ref = val.get_data_at(i);
453
0
            auto ret = hs_scan(database, str_ref.data, (int)str_ref.size, 0, scratch,
454
0
                               doris::vectorized::LikeSearchState::hs_match_handler,
455
0
                               (void*)(result.data() + i));
456
0
            if (ret != HS_SUCCESS && ret != HS_SCAN_TERMINATED) {
457
0
                return Status::RuntimeError(fmt::format("hyperscan error: {}", ret));
458
0
            }
459
0
        }
460
461
0
        hs_free_scratch(scratch);
462
0
        hs_free_database(database);
463
0
    } else { // fallback to re2
464
0
        RE2::Options opts;
465
0
        opts.set_never_nl(false);
466
0
        opts.set_dot_nl(true);
467
0
        re2::RE2 re(re_pattern, opts);
468
0
        if (re.ok()) {
469
0
            auto sz = val.size();
470
0
            for (size_t i = 0; i < sz; i++) {
471
0
                const auto& str_ref = val.get_data_at(i);
472
0
                *(result.data() + i) =
473
0
                        RE2::PartialMatch(re2::StringPiece(str_ref.data, str_ref.size), re);
474
0
            }
475
0
        } else {
476
0
            return Status::RuntimeError("Invalid pattern: {}", pattern.debug_string());
477
0
        }
478
0
    }
479
480
0
    return Status::OK();
481
0
}
482
483
// hyperscan compile expression to database and allocate scratch space
484
Status FunctionLikeBase::hs_prepare(FunctionContext* context, const char* expression,
485
132
                                    hs_database_t** database, hs_scratch_t** scratch) {
486
132
    hs_compile_error_t* compile_err;
487
132
    auto res = hs_compile(expression, HS_FLAG_DOTALL | HS_FLAG_ALLOWEMPTY | HS_FLAG_UTF8,
488
132
                          HS_MODE_BLOCK, nullptr, database, &compile_err);
489
490
132
    if (res != HS_SUCCESS) {
491
0
        *database = nullptr;
492
0
        std::string error_message = compile_err->message;
493
0
        hs_free_compile_error(compile_err);
494
        // Do not call FunctionContext::set_error here, since we do not want to cancel the query here.
495
0
        return Status::RuntimeError<false>("hs_compile regex pattern error:" + error_message);
496
0
    }
497
132
    hs_free_compile_error(compile_err);
498
499
132
    if (hs_alloc_scratch(*database, scratch) != HS_SUCCESS) {
500
0
        hs_free_database(*database);
501
0
        *database = nullptr;
502
0
        *scratch = nullptr;
503
        // Do not call FunctionContext::set_error here, since we do not want to cancel the query here.
504
0
        return Status::RuntimeError<false>("hs_alloc_scratch allocate scratch space error");
505
0
    }
506
507
132
    return Status::OK();
508
132
}
509
510
Status FunctionLikeBase::execute_impl(FunctionContext* context, Block& block,
511
                                      const ColumnNumbers& arguments, uint32_t result,
512
553
                                      size_t input_rows_count) const {
513
553
    const auto values_col =
514
553
            block.get_by_position(arguments[0]).column->convert_to_full_column_if_const();
515
553
    const auto* values = check_and_get_column<ColumnString>(values_col.get());
516
517
553
    if (!values) {
518
0
        return Status::InternalError("Not supported input arguments types");
519
0
    }
520
    // result column
521
553
    auto res = ColumnUInt8::create();
522
553
    ColumnUInt8::Container& vec_res = res->get_data();
523
    // set default value to 0, and match functions only need to set 1/true
524
553
    vec_res.resize_fill(input_rows_count);
525
553
    auto* state = reinterpret_cast<LikeState*>(
526
553
            context->get_function_state(FunctionContext::THREAD_LOCAL));
527
    // for constant_substring_fn, use long run length search for performance
528
553
    if (constant_substring_fn ==
529
553
        *(state->function
530
553
                  .target<doris::Status (*)(const LikeSearchState* state, const ColumnString&,
531
553
                                            const StringRef&, ColumnUInt8::Container&)>())) {
532
66
        RETURN_IF_ERROR(execute_substring(values->get_chars(), values->get_offsets(), vec_res,
533
66
                                          &state->search_state));
534
487
    } else {
535
487
        const auto pattern_col = block.get_by_position(arguments[1]).column;
536
487
        if (const auto* str_patterns = check_and_get_column<ColumnString>(pattern_col.get())) {
537
307
            RETURN_IF_ERROR(
538
307
                    vector_non_const(*values, *str_patterns, vec_res, state, input_rows_count));
539
307
        } else if (const auto* const_patterns =
540
180
                           check_and_get_column<ColumnConst>(pattern_col.get())) {
541
180
            const auto& pattern_val = const_patterns->get_data_at(0);
542
180
            RETURN_IF_ERROR(vector_const(*values, &pattern_val, vec_res, state->function,
543
180
                                         &state->search_state));
544
180
        } else {
545
0
            return Status::InternalError("Not supported input arguments types");
546
0
        }
547
487
    }
548
553
    block.replace_by_position(result, std::move(res));
549
553
    return Status::OK();
550
553
}
551
552
Status FunctionLikeBase::execute_substring(const ColumnString::Chars& values,
553
                                           const ColumnString::Offsets& value_offsets,
554
                                           ColumnUInt8::Container& result,
555
66
                                           LikeSearchState* search_state) const {
556
    // treat continuous multi string data as a long string data
557
66
    const UInt8* begin = values.data();
558
66
    const UInt8* end = begin + values.size();
559
66
    const UInt8* pos = begin;
560
561
    /// Current index in the array of strings.
562
66
    size_t i = 0;
563
66
    size_t needle_size = search_state->substring_pattern.get_pattern_length();
564
565
    /// We will search for the next occurrence in all strings at once.
566
108
    while (pos < end) {
567
        // search return matched substring start offset
568
64
        pos = (UInt8*)search_state->substring_pattern.search((char*)pos, end - pos);
569
64
        if (pos >= end) {
570
22
            break;
571
22
        }
572
573
        /// Determine which index it refers to.
574
        /// begin + value_offsets[i] is the start offset of string at i+1
575
42
        while (i < value_offsets.size() && begin + value_offsets[i] < pos) {
576
0
            ++i;
577
0
        }
578
579
        /// We check that the entry does not pass through the boundaries of strings.
580
42
        if (pos + needle_size <= begin + value_offsets[i]) {
581
42
            result[i] = 1;
582
42
        }
583
584
        // move to next string offset
585
42
        pos = begin + value_offsets[i];
586
42
        ++i;
587
42
    }
588
589
66
    return Status::OK();
590
66
}
591
592
Status FunctionLikeBase::vector_const(const ColumnString& values, const StringRef* pattern_val,
593
                                      ColumnUInt8::Container& result, const LikeFn& function,
594
180
                                      LikeSearchState* search_state) const {
595
180
    RETURN_IF_ERROR((function)(search_state, values,
596
180
                               *reinterpret_cast<const StringRef*>(pattern_val), result));
597
180
    return Status::OK();
598
180
}
599
600
template <bool LIKE_PATTERN>
601
307
VPatternSearchStateSPtr FunctionLikeBase::pattern_type_recognition(const ColumnString& patterns) {
602
307
    VPatternSearchStateSPtr allpass_state = std::make_shared<VectorAllpassSearchState>();
603
307
    VPatternSearchStateSPtr equal_state = std::make_shared<VectorEqualSearchState>();
604
307
    VPatternSearchStateSPtr substring_state = std::make_shared<VectorSubStringSearchState>();
605
307
    VPatternSearchStateSPtr starts_with_state = std::make_shared<VectorStartsWithSearchState>();
606
307
    VPatternSearchStateSPtr ends_with_state = std::make_shared<VectorEndsWithSearchState>();
607
307
    size_t size = patterns.size();
608
609
652
    for (size_t i = 0; i < size; ++i) {
610
357
        if (!allpass_state->_pattern_matched && !equal_state->_pattern_matched &&
611
357
            !substring_state->_pattern_matched && !starts_with_state->_pattern_matched &&
612
357
            !ends_with_state->_pattern_matched) {
613
12
            return nullptr;
614
12
        }
615
345
        std::string pattern_str = patterns.get_data_at(i).to_string();
616
345
        if (allpass_state->_pattern_matched) {
617
312
            if constexpr (LIKE_PATTERN) {
618
312
                allpass_state->like_pattern_match(pattern_str);
619
312
            } else {
620
0
                allpass_state->regexp_pattern_match(pattern_str);
621
0
            }
622
312
        }
623
345
        if (equal_state->_pattern_matched) {
624
315
            if constexpr (LIKE_PATTERN) {
625
315
                equal_state->like_pattern_match(pattern_str);
626
315
            } else {
627
0
                equal_state->regexp_pattern_match(pattern_str);
628
0
            }
629
315
        }
630
345
        if (substring_state->_pattern_matched) {
631
321
            if constexpr (LIKE_PATTERN) {
632
321
                substring_state->like_pattern_match(pattern_str);
633
321
            } else {
634
0
                substring_state->regexp_pattern_match(pattern_str);
635
0
            }
636
321
        }
637
345
        if (starts_with_state->_pattern_matched) {
638
313
            if constexpr (LIKE_PATTERN) {
639
313
                starts_with_state->like_pattern_match(pattern_str);
640
313
            } else {
641
0
                starts_with_state->regexp_pattern_match(pattern_str);
642
0
            }
643
313
        }
644
345
        if (ends_with_state->_pattern_matched) {
645
312
            if constexpr (LIKE_PATTERN) {
646
312
                ends_with_state->like_pattern_match(pattern_str);
647
312
            } else {
648
0
                ends_with_state->regexp_pattern_match(pattern_str);
649
0
            }
650
312
        }
651
345
    }
652
653
295
    if (allpass_state->_pattern_matched) {
654
21
        return allpass_state;
655
274
    } else if (equal_state->_pattern_matched) {
656
78
        return equal_state;
657
196
    } else if (substring_state->_pattern_matched) {
658
27
        return substring_state;
659
169
    } else if (starts_with_state->_pattern_matched) {
660
31
        return starts_with_state;
661
138
    } else if (ends_with_state->_pattern_matched) {
662
31
        return ends_with_state;
663
107
    } else {
664
107
        return nullptr;
665
107
    }
666
295
}
_ZN5doris10vectorized16FunctionLikeBase24pattern_type_recognitionILb1EEESt10shared_ptrINS0_24VectorPatternSearchStateEERKNS0_9ColumnStrIjEE
Line
Count
Source
601
307
VPatternSearchStateSPtr FunctionLikeBase::pattern_type_recognition(const ColumnString& patterns) {
602
307
    VPatternSearchStateSPtr allpass_state = std::make_shared<VectorAllpassSearchState>();
603
307
    VPatternSearchStateSPtr equal_state = std::make_shared<VectorEqualSearchState>();
604
307
    VPatternSearchStateSPtr substring_state = std::make_shared<VectorSubStringSearchState>();
605
307
    VPatternSearchStateSPtr starts_with_state = std::make_shared<VectorStartsWithSearchState>();
606
307
    VPatternSearchStateSPtr ends_with_state = std::make_shared<VectorEndsWithSearchState>();
607
307
    size_t size = patterns.size();
608
609
652
    for (size_t i = 0; i < size; ++i) {
610
357
        if (!allpass_state->_pattern_matched && !equal_state->_pattern_matched &&
611
357
            !substring_state->_pattern_matched && !starts_with_state->_pattern_matched &&
612
357
            !ends_with_state->_pattern_matched) {
613
12
            return nullptr;
614
12
        }
615
345
        std::string pattern_str = patterns.get_data_at(i).to_string();
616
345
        if (allpass_state->_pattern_matched) {
617
312
            if constexpr (LIKE_PATTERN) {
618
312
                allpass_state->like_pattern_match(pattern_str);
619
            } else {
620
                allpass_state->regexp_pattern_match(pattern_str);
621
            }
622
312
        }
623
345
        if (equal_state->_pattern_matched) {
624
315
            if constexpr (LIKE_PATTERN) {
625
315
                equal_state->like_pattern_match(pattern_str);
626
            } else {
627
                equal_state->regexp_pattern_match(pattern_str);
628
            }
629
315
        }
630
345
        if (substring_state->_pattern_matched) {
631
321
            if constexpr (LIKE_PATTERN) {
632
321
                substring_state->like_pattern_match(pattern_str);
633
            } else {
634
                substring_state->regexp_pattern_match(pattern_str);
635
            }
636
321
        }
637
345
        if (starts_with_state->_pattern_matched) {
638
313
            if constexpr (LIKE_PATTERN) {
639
313
                starts_with_state->like_pattern_match(pattern_str);
640
            } else {
641
                starts_with_state->regexp_pattern_match(pattern_str);
642
            }
643
313
        }
644
345
        if (ends_with_state->_pattern_matched) {
645
312
            if constexpr (LIKE_PATTERN) {
646
312
                ends_with_state->like_pattern_match(pattern_str);
647
            } else {
648
                ends_with_state->regexp_pattern_match(pattern_str);
649
            }
650
312
        }
651
345
    }
652
653
295
    if (allpass_state->_pattern_matched) {
654
21
        return allpass_state;
655
274
    } else if (equal_state->_pattern_matched) {
656
78
        return equal_state;
657
196
    } else if (substring_state->_pattern_matched) {
658
27
        return substring_state;
659
169
    } else if (starts_with_state->_pattern_matched) {
660
31
        return starts_with_state;
661
138
    } else if (ends_with_state->_pattern_matched) {
662
31
        return ends_with_state;
663
107
    } else {
664
107
        return nullptr;
665
107
    }
666
295
}
Unexecuted instantiation: _ZN5doris10vectorized16FunctionLikeBase24pattern_type_recognitionILb0EEESt10shared_ptrINS0_24VectorPatternSearchStateEERKNS0_9ColumnStrIjEE
667
668
Status FunctionLikeBase::vector_non_const(const ColumnString& values, const ColumnString& patterns,
669
                                          ColumnUInt8::Container& result, LikeState* state,
670
307
                                          size_t input_rows_count) const {
671
307
    ColumnString::MutablePtr replaced_patterns;
672
307
    VPatternSearchStateSPtr vector_search_state;
673
307
    if (state->is_like_pattern) {
674
307
        if (state->has_custom_escape) {
675
0
            replaced_patterns = ColumnString::create();
676
0
            for (int i = 0; i < input_rows_count; ++i) {
677
0
                std::string val =
678
0
                        replace_pattern_by_escape(patterns.get_data_at(i), state->escape_char);
679
0
                replaced_patterns->insert_data(val.c_str(), val.size());
680
0
            }
681
0
            vector_search_state = pattern_type_recognition<true>(*replaced_patterns);
682
307
        } else {
683
307
            vector_search_state = pattern_type_recognition<true>(patterns);
684
307
        }
685
307
    } else {
686
0
        vector_search_state = pattern_type_recognition<false>(patterns);
687
0
    }
688
689
307
    const ColumnString& real_pattern = state->has_custom_escape ? *replaced_patterns : patterns;
690
691
307
    if (vector_search_state == nullptr) {
692
        // pattern type recognition failed, use default case
693
381
        for (int i = 0; i < input_rows_count; ++i) {
694
262
            const auto pattern_val = real_pattern.get_data_at(i);
695
262
            const auto value_val = values.get_data_at(i);
696
262
            RETURN_IF_ERROR((state->scalar_function)(&state->search_state, value_val, pattern_val,
697
262
                                                     &result[i]));
698
262
        }
699
119
        return Status::OK();
700
119
    }
701
188
    const auto* search_strings =
702
188
            static_cast<const ColumnString*>(vector_search_state->_search_strings.get());
703
188
    return (vector_search_state->_vector_function)(values, *search_strings, result);
704
307
}
705
706
Status FunctionLike::like_fn(const LikeSearchState* state, const ColumnString& val,
707
0
                             const StringRef& pattern, ColumnUInt8::Container& result) {
708
0
    std::string re_pattern;
709
0
    convert_like_pattern(state, std::string(pattern.data, pattern.size), &re_pattern);
710
0
    return regexp_fn(state, val, {re_pattern.c_str(), re_pattern.size()}, result);
711
0
}
712
713
Status FunctionLike::like_fn_scalar(const LikeSearchState* state, const StringRef& val,
714
209
                                    const StringRef& pattern, unsigned char* result) {
715
209
    std::string re_pattern;
716
209
    convert_like_pattern(state, std::string(pattern.data, pattern.size), &re_pattern);
717
718
209
    return regexp_fn_scalar(state, StringRef(val.data, val.size),
719
209
                            {re_pattern.c_str(), re_pattern.size()}, result);
720
209
}
721
722
void FunctionLike::convert_like_pattern(const LikeSearchState* state, const std::string& pattern,
723
319
                                        std::string* re_pattern) {
724
319
    re_pattern->clear();
725
726
319
    if (pattern.empty()) {
727
14
        re_pattern->append("^$");
728
14
        return;
729
14
    }
730
731
    // add ^ to pattern head to match line head
732
305
    if (!pattern.empty() && pattern[0] != '%') {
733
229
        re_pattern->append("^");
734
229
    }
735
736
    // expect % and _, all chars should keep it literal mean.
737
1.96k
    for (size_t i = 0; i < pattern.size(); i++) {
738
1.66k
        char c = pattern[i];
739
1.66k
        if (c == '\\' && i + 1 < pattern.size()) {
740
71
            char next_c = pattern[i + 1];
741
71
            if (next_c == '%' || next_c == '_') {
742
                // convert "\%" and "\_" to literal "%" and "_"
743
35
                re_pattern->append(1, next_c);
744
35
                i++;
745
35
                continue;
746
36
            } else if (next_c == '\\') {
747
                // keep valid escape "\\"
748
18
                re_pattern->append("\\\\");
749
18
                i++;
750
18
                continue;
751
18
            }
752
71
        }
753
754
1.61k
        if (c == '%') {
755
283
            re_pattern->append(".*");
756
1.32k
        } else if (c == '_') {
757
256
            re_pattern->append(".");
758
1.07k
        } else {
759
            // special for hyperscan: [, ], (, ), {, }, -, *, +, \, |, /, :, ^, ., $, ?
760
1.07k
            if (c == '[' || c == ']' || c == '(' || c == ')' || c == '{' || c == '}' || c == '-' ||
761
1.07k
                c == '*' || c == '+' || c == '\\' || c == '|' || c == '/' || c == ':' || c == '^' ||
762
1.07k
                c == '.' || c == '$' || c == '?') {
763
37
                re_pattern->append(1, '\\');
764
37
            }
765
1.07k
            re_pattern->append(1, c);
766
1.07k
        }
767
1.61k
    }
768
769
    // add $ to pattern tail to match line tail
770
305
    if (!pattern.empty() && re_pattern->back() != '*') {
771
195
        re_pattern->append("$");
772
195
    }
773
305
}
774
775
415
void FunctionLike::remove_escape_character(std::string* search_string) {
776
415
    std::string tmp_search_string;
777
415
    tmp_search_string.swap(*search_string);
778
415
    int64_t len = tmp_search_string.length();
779
    // sometime 'like' may allowed converted to 'equals/start_with/end_with/sub_with'
780
    // so we need to remove escape from pattern to construct search string and use to do 'equals/start_with/end_with/sub_with'
781
1.87k
    for (int i = 0; i < len;) {
782
1.46k
        if (tmp_search_string[i] == '\\' && i + 1 < len &&
783
1.46k
            (tmp_search_string[i + 1] == '%' || tmp_search_string[i + 1] == '_' ||
784
44
             tmp_search_string[i + 1] == '\\')) {
785
36
            search_string->append(1, tmp_search_string[i + 1]);
786
36
            i += 2;
787
1.42k
        } else {
788
1.42k
            search_string->append(1, tmp_search_string[i]);
789
1.42k
            i++;
790
1.42k
        }
791
1.46k
    }
792
415
}
793
794
0
bool re2_full_match(const std::string& str, const RE2& re, std::vector<std::string>& results) {
795
0
    if (!re.ok()) {
796
0
        return false;
797
0
    }
798
799
0
    std::vector<RE2::Arg> arguments;
800
0
    std::vector<RE2::Arg*> arguments_ptrs;
801
0
    std::size_t args_count = re.NumberOfCapturingGroups();
802
0
    arguments.resize(args_count);
803
0
    arguments_ptrs.resize(args_count);
804
0
    results.resize(args_count);
805
0
    for (std::size_t i = 0; i < args_count; ++i) {
806
0
        arguments[i] = &results[i];
807
0
        arguments_ptrs[i] = &arguments[i];
808
0
    }
809
810
0
    return RE2::FullMatchN(str, re, arguments_ptrs.data(), (int)args_count);
811
0
}
812
813
0
void verbose_log_match(const std::string& str, const std::string& pattern_name, const RE2& re) {
814
0
    std::vector<std::string> results;
815
0
    VLOG_DEBUG << "arg str: " << str << ", size: " << str.size() << ", pattern " << pattern_name
816
0
               << ": " << re.pattern() << ", size: " << re.pattern().size();
817
0
    if (re2_full_match(str, re, results)) {
818
0
        for (int i = 0; i < results.size(); ++i) {
819
0
            VLOG_DEBUG << "match " << i << ": " << results[i] << ", size: " << results[i].size();
820
0
        }
821
0
    } else {
822
0
        VLOG_DEBUG << "no match";
823
0
    }
824
0
}
825
826
Status FunctionLike::construct_like_const_state(FunctionContext* context, const StringRef& pattern,
827
                                                std::shared_ptr<LikeState>& state,
828
347
                                                bool try_hyperscan) {
829
347
    std::string pattern_str;
830
347
    if (state->has_custom_escape) {
831
1
        pattern_str = replace_pattern_by_escape(pattern, state->escape_char);
832
346
    } else {
833
346
        pattern_str = pattern.to_string();
834
346
    }
835
347
    state->search_state.pattern_str = pattern_str;
836
347
    std::string search_string;
837
838
347
    if (!pattern_str.empty() && RE2::FullMatch(pattern_str, LIKE_ALLPASS_RE)) {
839
22
        state->search_state.set_search_string("");
840
22
        state->function = constant_allpass_fn;
841
22
        state->scalar_function = constant_allpass_fn_scalar;
842
325
    } else if (pattern_str.empty() || RE2::FullMatch(pattern_str, LIKE_EQUALS_RE, &search_string)) {
843
95
        if (VLOG_DEBUG_IS_ON) {
844
0
            verbose_log_match(pattern_str, "LIKE_EQUALS_RE", LIKE_EQUALS_RE);
845
0
            VLOG_DEBUG << "search_string : " << search_string << ", size: " << search_string.size();
846
0
        }
847
95
        remove_escape_character(&search_string);
848
95
        if (VLOG_DEBUG_IS_ON) {
849
0
            VLOG_DEBUG << "search_string escape removed: " << search_string
850
0
                       << ", size: " << search_string.size();
851
0
        }
852
95
        state->search_state.set_search_string(search_string);
853
95
        state->function = constant_equals_fn;
854
95
        state->scalar_function = constant_equals_fn_scalar;
855
230
    } else if (RE2::FullMatch(pattern_str, LIKE_STARTS_WITH_RE, &search_string)) {
856
32
        if (VLOG_DEBUG_IS_ON) {
857
0
            verbose_log_match(pattern_str, "LIKE_STARTS_WITH_RE", LIKE_STARTS_WITH_RE);
858
0
            VLOG_DEBUG << "search_string : " << search_string << ", size: " << search_string.size();
859
0
        }
860
32
        remove_escape_character(&search_string);
861
32
        if (VLOG_DEBUG_IS_ON) {
862
0
            VLOG_DEBUG << "search_string escape removed: " << search_string
863
0
                       << ", size: " << search_string.size();
864
0
        }
865
32
        state->search_state.set_search_string(search_string);
866
32
        state->function = constant_starts_with_fn;
867
32
        state->scalar_function = constant_starts_with_fn_scalar;
868
198
    } else if (RE2::FullMatch(pattern_str, LIKE_ENDS_WITH_RE, &search_string)) {
869
32
        if (VLOG_DEBUG_IS_ON) {
870
0
            verbose_log_match(pattern_str, "LIKE_ENDS_WITH_RE", LIKE_ENDS_WITH_RE);
871
0
            VLOG_DEBUG << "search_string : " << search_string << ", size: " << search_string.size();
872
0
        }
873
32
        remove_escape_character(&search_string);
874
32
        if (VLOG_DEBUG_IS_ON) {
875
0
            VLOG_DEBUG << "search_string escape removed: " << search_string
876
0
                       << ", size: " << search_string.size();
877
0
        }
878
32
        state->search_state.set_search_string(search_string);
879
32
        state->function = constant_ends_with_fn;
880
32
        state->scalar_function = constant_ends_with_fn_scalar;
881
166
    } else if (RE2::FullMatch(pattern_str, LIKE_SUBSTRING_RE, &search_string)) {
882
56
        if (VLOG_DEBUG_IS_ON) {
883
0
            verbose_log_match(pattern_str, "LIKE_SUBSTRING_RE", LIKE_SUBSTRING_RE);
884
0
            VLOG_DEBUG << "search_string : " << search_string << ", size: " << search_string.size();
885
0
        }
886
56
        remove_escape_character(&search_string);
887
56
        if (VLOG_DEBUG_IS_ON) {
888
0
            VLOG_DEBUG << "search_string escape removed: " << search_string
889
0
                       << ", size: " << search_string.size();
890
0
        }
891
56
        state->search_state.set_search_string(search_string);
892
56
        state->function = constant_substring_fn;
893
56
        state->scalar_function = constant_substring_fn_scalar;
894
110
    } else {
895
110
        std::string re_pattern;
896
110
        convert_like_pattern(&state->search_state, pattern_str, &re_pattern);
897
110
        if (VLOG_DEBUG_IS_ON) {
898
0
            VLOG_DEBUG << "hyperscan, pattern str: " << pattern_str
899
0
                       << ", size: " << pattern_str.size() << ", re pattern: " << re_pattern
900
0
                       << ", size: " << re_pattern.size();
901
0
        }
902
903
110
        hs_database_t* database = nullptr;
904
110
        hs_scratch_t* scratch = nullptr;
905
110
        if (try_hyperscan && hs_prepare(context, re_pattern.c_str(), &database, &scratch).ok()) {
906
            // use hyperscan
907
110
            state->search_state.hs_database.reset(database);
908
110
            state->search_state.hs_scratch.reset(scratch);
909
110
        } else {
910
            // fallback to re2
911
            // reset hs_database to nullptr to indicate not use hyperscan
912
0
            state->search_state.hs_database.reset();
913
0
            state->search_state.hs_scratch.reset();
914
915
0
            RE2::Options opts;
916
0
            opts.set_never_nl(false);
917
0
            opts.set_dot_nl(true);
918
0
            state->search_state.regex = std::make_unique<RE2>(re_pattern, opts);
919
0
            if (!state->search_state.regex->ok()) {
920
0
                return Status::InternalError("Invalid regex expression: {}(origin: {})", re_pattern,
921
0
                                             pattern_str);
922
0
            }
923
0
        }
924
925
110
        state->function = constant_regex_fn;
926
110
        state->scalar_function = constant_regex_fn_scalar;
927
110
    }
928
347
    return Status::OK();
929
347
}
930
931
1.07k
Status FunctionLike::open(FunctionContext* context, FunctionContext::FunctionStateScope scope) {
932
1.07k
    if (scope != FunctionContext::THREAD_LOCAL) {
933
536
        return Status::OK();
934
536
    }
935
536
    std::shared_ptr<LikeState> state = std::make_shared<LikeState>();
936
536
    state->is_like_pattern = true;
937
536
    state->function = like_fn;
938
536
    state->scalar_function = like_fn_scalar;
939
536
    if (context->is_col_constant(2)) {
940
1
        state->has_custom_escape = true;
941
1
        const auto escape_col = context->get_constant_col(2)->column_ptr;
942
1
        const auto& escape = escape_col->get_data_at(0);
943
1
        if (escape.size != 1) {
944
0
            return Status::InternalError("Escape character must be a single character, got: {}",
945
0
                                         escape.to_string());
946
0
        }
947
1
        state->escape_char = escape.data[0];
948
1
    }
949
536
    if (context->is_col_constant(1)) {
950
347
        const auto pattern_col = context->get_constant_col(1)->column_ptr;
951
347
        const auto& pattern = pattern_col->get_data_at(0);
952
347
        RETURN_IF_ERROR(construct_like_const_state(context, pattern, state));
953
347
    }
954
536
    context->set_function_state(scope, state);
955
956
536
    return Status::OK();
957
536
}
958
959
Status FunctionRegexpLike::open(FunctionContext* context,
960
122
                                FunctionContext::FunctionStateScope scope) {
961
122
    if (scope != FunctionContext::THREAD_LOCAL) {
962
61
        return Status::OK();
963
61
    }
964
61
    std::shared_ptr<LikeState> state = std::make_shared<LikeState>();
965
61
    context->set_function_state(scope, state);
966
61
    state->is_like_pattern = false;
967
61
    state->function = regexp_fn;
968
61
    state->scalar_function = regexp_fn_scalar;
969
61
    if (context->is_col_constant(1)) {
970
61
        const auto pattern_col = context->get_constant_col(1)->column_ptr;
971
61
        const auto& pattern = pattern_col->get_data_at(0);
972
973
61
        std::string pattern_str = pattern.to_string();
974
61
        std::string search_string;
975
61
        if (RE2::FullMatch(pattern_str, ALLPASS_RE)) {
976
4
            state->search_state.set_search_string("");
977
4
            state->function = constant_allpass_fn;
978
4
            state->scalar_function = constant_allpass_fn_scalar;
979
57
        } else if (RE2::FullMatch(pattern_str, EQUALS_RE, &search_string)) {
980
7
            state->search_state.set_search_string(search_string);
981
7
            state->function = constant_equals_fn;
982
7
            state->scalar_function = constant_equals_fn_scalar;
983
50
        } else if (RE2::FullMatch(pattern_str, STARTS_WITH_RE, &search_string)) {
984
7
            state->search_state.set_search_string(search_string);
985
7
            state->function = constant_starts_with_fn;
986
7
            state->scalar_function = constant_starts_with_fn_scalar;
987
43
        } else if (RE2::FullMatch(pattern_str, ENDS_WITH_RE, &search_string)) {
988
7
            state->search_state.set_search_string(search_string);
989
7
            state->function = constant_ends_with_fn;
990
7
            state->scalar_function = constant_ends_with_fn_scalar;
991
36
        } else if (RE2::FullMatch(pattern_str, SUBSTRING_RE, &search_string)) {
992
14
            state->search_state.set_search_string(search_string);
993
14
            state->function = constant_substring_fn;
994
14
            state->scalar_function = constant_substring_fn_scalar;
995
22
        } else {
996
22
            hs_database_t* database = nullptr;
997
22
            hs_scratch_t* scratch = nullptr;
998
22
            if (hs_prepare(context, pattern_str.c_str(), &database, &scratch).ok()) {
999
                // use hyperscan
1000
22
                state->search_state.hs_database.reset(database);
1001
22
                state->search_state.hs_scratch.reset(scratch);
1002
22
            } else {
1003
                // fallback to re2
1004
                // reset hs_database to nullptr to indicate not use hyperscan
1005
0
                state->search_state.hs_database.reset();
1006
0
                state->search_state.hs_scratch.reset();
1007
0
                RE2::Options opts;
1008
0
                opts.set_never_nl(false);
1009
0
                opts.set_dot_nl(true);
1010
0
                state->search_state.regex = std::make_unique<RE2>(pattern_str, opts);
1011
0
                if (!state->search_state.regex->ok()) {
1012
0
                    return Status::InternalError("Invalid regex expression: {}", pattern_str);
1013
0
                }
1014
0
            }
1015
22
            state->function = constant_regex_fn;
1016
22
            state->scalar_function = constant_regex_fn_scalar;
1017
22
        }
1018
61
    }
1019
61
    return Status::OK();
1020
61
}
1021
1022
1
void register_function_like(SimpleFunctionFactory& factory) {
1023
1
    factory.register_function<FunctionLike>();
1024
1
}
1025
1026
1
void register_function_regexp(SimpleFunctionFactory& factory) {
1027
1
    factory.register_function<FunctionRegexpLike>();
1028
1
    factory.register_alias(FunctionRegexpLike::name, FunctionRegexpLike::alias);
1029
1
}
1030
#include "common/compile_check_end.h"
1031
} // namespace doris::vectorized