Coverage Report

Created: 2025-09-30 22:18

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