Coverage Report

Created: 2025-04-30 15:20

/root/doris/be/src/vec/functions/like.cpp
Line
Count
Source (jump to first uncovered line)
1
// Licensed to the Apache Software Foundation (ASF) under one
2
// or more contributor license agreements.  See the NOTICE file
3
// distributed with this work for additional information
4
// regarding copyright ownership.  The ASF licenses this file
5
// to you under the Apache License, Version 2.0 (the
6
// "License"); you may not use this file except in compliance
7
// with the License.  You may obtain a copy of the License at
8
//
9
//   http://www.apache.org/licenses/LICENSE-2.0
10
//
11
// Unless required by applicable law or agreed to in writing,
12
// software distributed under the License is distributed on an
13
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14
// KIND, either express or implied.  See the License for the
15
// specific language governing permissions and limitations
16
// under the License.
17
18
#include "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/columns/columns_number.h"
34
#include "vec/common/string_ref.h"
35
#include "vec/core/block.h"
36
#include "vec/core/column_with_type_and_name.h"
37
#include "vec/functions/simple_function_factory.h"
38
39
namespace doris::vectorized {
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
24
    VectorAllpassSearchState() : VectorPatternSearchState(FunctionLikeBase::vector_allpass_fn) {}
65
66
24
    ~VectorAllpassSearchState() override = default;
67
68
2
    void like_pattern_match(const std::string& pattern_str) override {
69
2
        if (!pattern_str.empty() && RE2::FullMatch(pattern_str, LIKE_ALLPASS_RE)) {
70
0
            _search_strings->insert_default();
71
2
        } else {
72
2
            _pattern_matched = false;
73
2
        }
74
2
    }
75
76
22
    void regexp_pattern_match(const std::string& pattern_str) override {
77
22
        if (RE2::FullMatch(pattern_str, ALLPASS_RE)) {
78
0
            _search_strings->insert_default();
79
22
        } else {
80
22
            _pattern_matched = false;
81
22
        }
82
22
    }
83
};
84
85
struct VectorEqualSearchState : public VectorPatternSearchState {
86
24
    VectorEqualSearchState() : VectorPatternSearchState(FunctionLikeBase::vector_equals_fn) {}
87
88
24
    ~VectorEqualSearchState() override = default;
89
90
2
    void like_pattern_match(const std::string& pattern_str) override {
91
2
        _search_string.clear();
92
2
        if (pattern_str.empty() || RE2::FullMatch(pattern_str, LIKE_EQUALS_RE, &_search_string)) {
93
0
            FunctionLike::remove_escape_character(&_search_string);
94
0
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
95
2
        } else {
96
2
            _pattern_matched = false;
97
2
        }
98
2
    }
99
100
22
    void regexp_pattern_match(const std::string& pattern_str) override {
101
22
        _search_string.clear();
102
22
        if (RE2::FullMatch(pattern_str, EQUALS_RE, &_search_string)) {
103
4
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
104
18
        } else {
105
18
            _pattern_matched = false;
106
18
        }
107
22
    }
108
};
109
110
struct VectorSubStringSearchState : public VectorPatternSearchState {
111
    VectorSubStringSearchState()
112
24
            : VectorPatternSearchState(FunctionLikeBase::vector_substring_fn) {}
113
114
24
    ~VectorSubStringSearchState() override = default;
115
116
6
    void like_pattern_match(const std::string& pattern_str) override {
117
6
        _search_string.clear();
118
6
        if (RE2::FullMatch(pattern_str, LIKE_SUBSTRING_RE, &_search_string)) {
119
4
            FunctionLike::remove_escape_character(&_search_string);
120
4
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
121
4
        } else {
122
2
            _pattern_matched = false;
123
2
        }
124
6
    }
125
126
22
    void regexp_pattern_match(const std::string& pattern_str) override {
127
22
        _search_string.clear();
128
22
        if (RE2::FullMatch(pattern_str, SUBSTRING_RE, &_search_string)) {
129
0
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
130
22
        } else {
131
22
            _pattern_matched = false;
132
22
        }
133
22
    }
134
};
135
136
struct VectorStartsWithSearchState : public VectorPatternSearchState {
137
    VectorStartsWithSearchState()
138
24
            : VectorPatternSearchState(FunctionLikeBase::vector_starts_with_fn) {}
139
140
24
    ~VectorStartsWithSearchState() override = default;
141
142
2
    void like_pattern_match(const std::string& pattern_str) override {
143
2
        _search_string.clear();
144
2
        if (RE2::FullMatch(pattern_str, LIKE_STARTS_WITH_RE, &_search_string)) {
145
0
            FunctionLike::remove_escape_character(&_search_string);
146
0
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
147
2
        } else {
148
2
            _pattern_matched = false;
149
2
        }
150
2
    }
151
152
22
    void regexp_pattern_match(const std::string& pattern_str) override {
153
22
        _search_string.clear();
154
22
        if (RE2::FullMatch(pattern_str, STARTS_WITH_RE, &_search_string)) {
155
4
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
156
18
        } else {
157
18
            _pattern_matched = false;
158
18
        }
159
22
    }
160
};
161
162
struct VectorEndsWithSearchState : public VectorPatternSearchState {
163
24
    VectorEndsWithSearchState() : VectorPatternSearchState(FunctionLikeBase::vector_ends_with_fn) {}
164
165
24
    ~VectorEndsWithSearchState() override = default;
166
167
2
    void like_pattern_match(const std::string& pattern_str) override {
168
2
        _search_string.clear();
169
2
        if (RE2::FullMatch(pattern_str, LIKE_ENDS_WITH_RE, &_search_string)) {
170
0
            FunctionLike::remove_escape_character(&_search_string);
171
0
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
172
2
        } else {
173
2
            _pattern_matched = false;
174
2
        }
175
2
    }
176
177
22
    void regexp_pattern_match(const std::string& pattern_str) override {
178
22
        _search_string.clear();
179
22
        if (RE2::FullMatch(pattern_str, ENDS_WITH_RE, &_search_string)) {
180
4
            _search_strings->insert_data(_search_string.c_str(), _search_string.size());
181
18
        } else {
182
18
            _pattern_matched = false;
183
18
        }
184
22
    }
185
};
186
187
0
Status LikeSearchState::clone(LikeSearchState& cloned) {
188
0
    cloned.escape_char = escape_char;
189
0
    cloned.set_search_string(search_string);
190
191
0
    std::string re_pattern;
192
0
    FunctionLike::convert_like_pattern(this, pattern_str, &re_pattern);
193
0
    if (hs_database) { // use hyperscan
194
0
        hs_database_t* database = nullptr;
195
0
        hs_scratch_t* scratch = nullptr;
196
0
        RETURN_IF_ERROR(FunctionLike::hs_prepare(nullptr, re_pattern.c_str(), &database, &scratch));
197
198
0
        cloned.hs_database.reset(database);
199
0
        cloned.hs_scratch.reset(scratch);
200
0
    } else { // fallback to re2
201
0
        cloned.hs_database.reset();
202
0
        cloned.hs_scratch.reset();
203
204
0
        RE2::Options opts;
205
0
        opts.set_never_nl(false);
206
0
        opts.set_dot_nl(true);
207
0
        cloned.regex = std::make_unique<RE2>(re_pattern, opts);
208
0
        if (!cloned.regex->ok()) {
209
0
            return Status::InternalError("Invalid regex expression: {}", re_pattern);
210
0
        }
211
0
    }
212
213
0
    return Status::OK();
214
0
}
215
216
Status FunctionLikeBase::constant_allpass_fn(LikeSearchState* state, const ColumnString& vals,
217
                                             const StringRef& pattern,
218
0
                                             ColumnUInt8::Container& result) {
219
0
    memset(result.data(), 1, vals.size());
220
0
    return Status::OK();
221
0
}
222
223
Status FunctionLikeBase::constant_allpass_fn_scalar(LikeSearchState* state, const StringRef& val,
224
                                                    const StringRef& pattern,
225
0
                                                    unsigned char* result) {
226
0
    *result = 1;
227
0
    return Status::OK();
228
0
}
229
230
Status FunctionLikeBase::vector_allpass_fn(const ColumnString& vals,
231
                                           const ColumnString& search_strings,
232
0
                                           ColumnUInt8::Container& result) {
233
0
    DCHECK(vals.size() == search_strings.size());
234
0
    DCHECK(vals.size() == result.size());
235
0
    memset(result.data(), 1, vals.size());
236
0
    return Status::OK();
237
0
}
238
239
Status FunctionLikeBase::constant_starts_with_fn(LikeSearchState* state, const ColumnString& val,
240
                                                 const StringRef& pattern,
241
0
                                                 ColumnUInt8::Container& result) {
242
0
    auto sz = val.size();
243
0
    for (size_t i = 0; i < sz; i++) {
244
0
        const auto& str_ref = val.get_data_at(i);
245
0
        result[i] = (str_ref.size >= state->search_string_sv.size) &&
246
0
                    str_ref.start_with(state->search_string_sv);
247
0
    }
248
0
    return Status::OK();
249
0
}
250
251
Status FunctionLikeBase::constant_starts_with_fn_scalar(LikeSearchState* state,
252
                                                        const StringRef& val,
253
                                                        const StringRef& pattern,
254
0
                                                        unsigned char* result) {
255
0
    *result = (val.size >= state->search_string_sv.size) &&
256
0
              (state->search_string_sv == val.substring(0, state->search_string_sv.size));
257
0
    return Status::OK();
258
0
}
259
260
Status FunctionLikeBase::vector_starts_with_fn(const ColumnString& vals,
261
                                               const ColumnString& search_strings,
262
4
                                               ColumnUInt8::Container& result) {
263
4
    DCHECK(vals.size() == search_strings.size());
264
4
    DCHECK(vals.size() == result.size());
265
4
    auto sz = vals.size();
266
8
    for (size_t i = 0; i < sz; ++i) {
267
4
        const auto& str_sv = vals.get_data_at(i);
268
4
        const auto& search_string_sv = search_strings.get_data_at(i);
269
4
        result[i] = (str_sv.size >= search_string_sv.size) && str_sv.start_with(search_string_sv);
270
4
    }
271
4
    return Status::OK();
272
4
}
273
274
Status FunctionLikeBase::constant_ends_with_fn(LikeSearchState* state, const ColumnString& val,
275
                                               const StringRef& pattern,
276
0
                                               ColumnUInt8::Container& result) {
277
0
    auto sz = val.size();
278
0
    for (size_t i = 0; i < sz; i++) {
279
0
        const auto& str_ref = val.get_data_at(i);
280
0
        result[i] = (str_ref.size >= state->search_string_sv.size) &&
281
0
                    str_ref.end_with(state->search_string_sv);
282
0
    }
283
0
    return Status::OK();
284
0
}
285
286
Status FunctionLikeBase::constant_ends_with_fn_scalar(LikeSearchState* state, 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
4
                                             ColumnUInt8::Container& result) {
298
4
    DCHECK(vals.size() == search_strings.size());
299
4
    DCHECK(vals.size() == result.size());
300
4
    auto sz = vals.size();
301
8
    for (size_t i = 0; i < sz; ++i) {
302
4
        const auto& str_sv = vals.get_data_at(i);
303
4
        const auto& search_string_sv = search_strings.get_data_at(i);
304
4
        result[i] = (str_sv.size >= search_string_sv.size) && str_sv.end_with(search_string_sv);
305
4
    }
306
4
    return Status::OK();
307
4
}
308
309
Status FunctionLikeBase::constant_equals_fn(LikeSearchState* state, const ColumnString& val,
310
                                            const StringRef& pattern,
311
0
                                            ColumnUInt8::Container& result) {
312
0
    auto sz = val.size();
313
0
    for (size_t i = 0; i < sz; i++) {
314
0
        result[i] = (val.get_data_at(i) == state->search_string_sv);
315
0
    }
316
0
    return Status::OK();
317
0
}
318
319
Status FunctionLikeBase::constant_equals_fn_scalar(LikeSearchState* state, const StringRef& val,
320
                                                   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
4
                                          ColumnUInt8::Container& result) {
329
4
    DCHECK(vals.size() == search_strings.size());
330
4
    DCHECK(vals.size() == result.size());
331
4
    auto sz = vals.size();
332
8
    for (size_t i = 0; i < sz; ++i) {
333
4
        const auto& str_sv = vals.get_data_at(i);
334
4
        const auto& search_string_sv = search_strings.get_data_at(i);
335
4
        result[i] = str_sv == search_string_sv;
336
4
    }
337
4
    return Status::OK();
338
4
}
339
340
Status FunctionLikeBase::constant_substring_fn(LikeSearchState* state, const ColumnString& val,
341
                                               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(LikeSearchState* state, const StringRef& val,
355
                                                      const StringRef& pattern,
356
0
                                                      unsigned char* result) {
357
0
    if (state->search_string_sv.size == 0) {
358
0
        *result = true;
359
0
        return Status::OK();
360
0
    }
361
0
    *result = state->substring_pattern.search(val) != -1;
362
0
    return Status::OK();
363
0
}
364
365
Status FunctionLikeBase::vector_substring_fn(const ColumnString& vals,
366
                                             const ColumnString& search_strings,
367
0
                                             ColumnUInt8::Container& result) {
368
0
    DCHECK(vals.size() == search_strings.size());
369
0
    DCHECK(vals.size() == result.size());
370
0
    auto sz = vals.size();
371
0
    for (size_t i = 0; i < sz; ++i) {
372
0
        const auto& str_sv = vals.get_data_at(i);
373
0
        const auto& search_string_sv = search_strings.get_data_at(i);
374
0
        if (search_string_sv.size == 0) {
375
0
            result[i] = true;
376
0
            continue;
377
0
        }
378
0
        doris::StringSearch substring_search(&search_string_sv);
379
0
        result[i] = substring_search.search(str_sv) != -1;
380
0
    }
381
0
    return Status::OK();
382
0
}
383
384
Status FunctionLikeBase::constant_regex_fn_scalar(LikeSearchState* state, const StringRef& val,
385
10
                                                  const StringRef& pattern, unsigned char* result) {
386
10
    if (state->hs_database) { // use hyperscan
387
10
        auto ret = hs_scan(state->hs_database.get(), val.data, val.size, 0, state->hs_scratch.get(),
388
10
                           doris::vectorized::LikeSearchState::hs_match_handler, (void*)result);
389
10
        if (ret != HS_SUCCESS && ret != HS_SCAN_TERMINATED) {
390
0
            return Status::RuntimeError(fmt::format("hyperscan error: {}", ret));
391
0
        }
392
10
    } else { // fallback to re2
393
0
        *result = RE2::PartialMatch(re2::StringPiece(val.data, val.size), *state->regex);
394
0
    }
395
396
10
    return Status::OK();
397
10
}
398
399
Status FunctionLikeBase::regexp_fn_scalar(LikeSearchState* state, const StringRef& val,
400
60
                                          const StringRef& pattern, unsigned char* result) {
401
60
    RE2::Options opts;
402
60
    opts.set_never_nl(false);
403
60
    opts.set_dot_nl(true);
404
60
    re2::RE2 re(re2::StringPiece(pattern.data, pattern.size), opts);
405
60
    if (re.ok()) {
406
60
        *result = RE2::PartialMatch(re2::StringPiece(val.data, val.size), re);
407
60
    } else {
408
0
        return Status::RuntimeError("Invalid pattern: {}", pattern.debug_string());
409
0
    }
410
411
60
    return Status::OK();
412
60
}
413
414
Status FunctionLikeBase::constant_regex_fn(LikeSearchState* state, const ColumnString& val,
415
                                           const StringRef& pattern,
416
0
                                           ColumnUInt8::Container& result) {
417
0
    auto sz = val.size();
418
0
    if (state->hs_database) { // use hyperscan
419
0
        for (size_t i = 0; i < sz; i++) {
420
0
            const auto& str_ref = val.get_data_at(i);
421
0
            auto ret = hs_scan(state->hs_database.get(), str_ref.data, str_ref.size, 0,
422
0
                               state->hs_scratch.get(),
423
0
                               doris::vectorized::LikeSearchState::hs_match_handler,
424
0
                               (void*)(result.data() + i));
425
0
            if (ret != HS_SUCCESS && ret != HS_SCAN_TERMINATED) {
426
0
                return Status::RuntimeError(fmt::format("hyperscan error: {}", ret));
427
0
            }
428
0
        }
429
0
    } 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
0
    return Status::OK();
438
0
}
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, 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
10
                                    hs_database_t** database, hs_scratch_t** scratch) {
483
10
    hs_compile_error_t* compile_err;
484
10
    auto res = hs_compile(expression, HS_FLAG_DOTALL | HS_FLAG_ALLOWEMPTY | HS_FLAG_UTF8,
485
10
                          HS_MODE_BLOCK, nullptr, database, &compile_err);
486
487
10
    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
10
    hs_free_compile_error(compile_err);
495
496
10
    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
10
    return Status::OK();
505
10
}
506
507
Status FunctionLikeBase::execute_impl(FunctionContext* context, Block& block,
508
                                      const ColumnNumbers& arguments, uint32_t result,
509
32
                                      size_t input_rows_count) const {
510
32
    const auto values_col =
511
32
            block.get_by_position(arguments[0]).column->convert_to_full_column_if_const();
512
32
    const auto* values = check_and_get_column<ColumnString>(values_col.get());
513
514
32
    if (!values) {
515
0
        return Status::InternalError("Not supported input arguments types");
516
0
    }
517
    // result column
518
32
    auto res = ColumnUInt8::create();
519
32
    ColumnUInt8::Container& vec_res = res->get_data();
520
    // set default value to 0, and match functions only need to set 1/true
521
32
    vec_res.resize_fill(input_rows_count);
522
32
    auto* state = reinterpret_cast<LikeState*>(
523
32
            context->get_function_state(FunctionContext::THREAD_LOCAL));
524
    // for constant_substring_fn, use long run length search for performance
525
32
    if (constant_substring_fn ==
526
32
        *(state->function.target<doris::Status (*)(LikeSearchState* state, const ColumnString&,
527
32
                                                   const StringRef&, ColumnUInt8::Container&)>())) {
528
8
        RETURN_IF_ERROR(execute_substring(values->get_chars(), values->get_offsets(), vec_res,
529
8
                                          &state->search_state));
530
24
    } else {
531
24
        const auto pattern_col = block.get_by_position(arguments[1]).column;
532
24
        if (const auto* str_patterns = check_and_get_column<ColumnString>(pattern_col.get())) {
533
24
            RETURN_IF_ERROR(
534
24
                    vector_non_const(*values, *str_patterns, vec_res, state, input_rows_count));
535
24
        } else if (const auto* const_patterns =
536
0
                           check_and_get_column<ColumnConst>(pattern_col.get())) {
537
0
            const auto& pattern_val = const_patterns->get_data_at(0);
538
0
            RETURN_IF_ERROR(vector_const(*values, &pattern_val, vec_res, state->function,
539
0
                                         &state->search_state));
540
0
        } else {
541
0
            return Status::InternalError("Not supported input arguments types");
542
0
        }
543
24
    }
544
32
    block.replace_by_position(result, std::move(res));
545
32
    return Status::OK();
546
32
}
547
548
Status FunctionLikeBase::execute_substring(const ColumnString::Chars& values,
549
                                           const ColumnString::Offsets& value_offsets,
550
                                           ColumnUInt8::Container& result,
551
8
                                           LikeSearchState* search_state) const {
552
    // treat continuous multi string data as a long string data
553
8
    const UInt8* begin = values.data();
554
8
    const UInt8* end = begin + values.size();
555
8
    const UInt8* pos = begin;
556
557
    /// Current index in the array of strings.
558
8
    size_t i = 0;
559
8
    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
14
    while (pos < end) {
563
        // search return matched substring start offset
564
8
        pos = (UInt8*)search_state->substring_pattern.search((char*)pos, end - pos);
565
8
        if (pos >= end) {
566
2
            break;
567
2
        }
568
569
        /// Determine which index it refers to.
570
        /// begin + value_offsets[i] is the start offset of string at i+1
571
6
        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
6
        if (pos + needle_size <= begin + value_offsets[i]) {
577
6
            result[i] = 1;
578
6
        }
579
580
        // move to next string offset
581
6
        pos = begin + value_offsets[i];
582
6
        ++i;
583
6
    }
584
585
8
    return Status::OK();
586
8
}
587
588
Status FunctionLikeBase::vector_const(const ColumnString& values, const StringRef* pattern_val,
589
                                      ColumnUInt8::Container& result, const LikeFn& function,
590
0
                                      LikeSearchState* search_state) const {
591
0
    RETURN_IF_ERROR((function)(search_state, values,
592
0
                               *reinterpret_cast<const StringRef*>(pattern_val), result));
593
0
    return Status::OK();
594
0
}
595
596
template <bool LIKE_PATTERN>
597
24
VPatternSearchStateSPtr FunctionLikeBase::pattern_type_recognition(const ColumnString& patterns) {
598
24
    VPatternSearchStateSPtr allpass_state = std::make_shared<VectorAllpassSearchState>();
599
24
    VPatternSearchStateSPtr equal_state = std::make_shared<VectorEqualSearchState>();
600
24
    VPatternSearchStateSPtr substring_state = std::make_shared<VectorSubStringSearchState>();
601
24
    VPatternSearchStateSPtr starts_with_state = std::make_shared<VectorStartsWithSearchState>();
602
24
    VPatternSearchStateSPtr ends_with_state = std::make_shared<VectorEndsWithSearchState>();
603
24
    size_t size = patterns.size();
604
605
52
    for (size_t i = 0; i < size; ++i) {
606
30
        if (!allpass_state->_pattern_matched && !equal_state->_pattern_matched &&
607
30
            !substring_state->_pattern_matched && !starts_with_state->_pattern_matched &&
608
30
            !ends_with_state->_pattern_matched) {
609
2
            return nullptr;
610
2
        }
611
28
        std::string pattern_str = patterns.get_data_at(i).to_string();
612
28
        if (allpass_state->_pattern_matched) {
613
24
            if constexpr (LIKE_PATTERN) {
614
22
                allpass_state->like_pattern_match(pattern_str);
615
22
            } else {
616
22
                allpass_state->regexp_pattern_match(pattern_str);
617
22
            }
618
24
        }
619
28
        if (equal_state->_pattern_matched) {
620
24
            if constexpr (LIKE_PATTERN) {
621
22
                equal_state->like_pattern_match(pattern_str);
622
22
            } else {
623
22
                equal_state->regexp_pattern_match(pattern_str);
624
22
            }
625
24
        }
626
28
        if (substring_state->_pattern_matched) {
627
28
            if constexpr (LIKE_PATTERN) {
628
22
                substring_state->like_pattern_match(pattern_str);
629
22
            } else {
630
22
                substring_state->regexp_pattern_match(pattern_str);
631
22
            }
632
28
        }
633
28
        if (starts_with_state->_pattern_matched) {
634
24
            if constexpr (LIKE_PATTERN) {
635
22
                starts_with_state->like_pattern_match(pattern_str);
636
22
            } else {
637
22
                starts_with_state->regexp_pattern_match(pattern_str);
638
22
            }
639
24
        }
640
28
        if (ends_with_state->_pattern_matched) {
641
24
            if constexpr (LIKE_PATTERN) {
642
22
                ends_with_state->like_pattern_match(pattern_str);
643
22
            } else {
644
22
                ends_with_state->regexp_pattern_match(pattern_str);
645
22
            }
646
24
        }
647
28
    }
648
649
22
    if (allpass_state->_pattern_matched) {
650
0
        return allpass_state;
651
22
    } else if (equal_state->_pattern_matched) {
652
4
        return equal_state;
653
18
    } else if (substring_state->_pattern_matched) {
654
0
        return substring_state;
655
18
    } else if (starts_with_state->_pattern_matched) {
656
4
        return starts_with_state;
657
14
    } else if (ends_with_state->_pattern_matched) {
658
4
        return ends_with_state;
659
10
    } else {
660
10
        return nullptr;
661
10
    }
662
22
}
_ZN5doris10vectorized16FunctionLikeBase24pattern_type_recognitionILb1EEESt10shared_ptrINS0_24VectorPatternSearchStateEERKNS0_9ColumnStrIjEE
Line
Count
Source
597
2
VPatternSearchStateSPtr FunctionLikeBase::pattern_type_recognition(const ColumnString& patterns) {
598
2
    VPatternSearchStateSPtr allpass_state = std::make_shared<VectorAllpassSearchState>();
599
2
    VPatternSearchStateSPtr equal_state = std::make_shared<VectorEqualSearchState>();
600
2
    VPatternSearchStateSPtr substring_state = std::make_shared<VectorSubStringSearchState>();
601
2
    VPatternSearchStateSPtr starts_with_state = std::make_shared<VectorStartsWithSearchState>();
602
2
    VPatternSearchStateSPtr ends_with_state = std::make_shared<VectorEndsWithSearchState>();
603
2
    size_t size = patterns.size();
604
605
8
    for (size_t i = 0; i < size; ++i) {
606
8
        if (!allpass_state->_pattern_matched && !equal_state->_pattern_matched &&
607
8
            !substring_state->_pattern_matched && !starts_with_state->_pattern_matched &&
608
8
            !ends_with_state->_pattern_matched) {
609
2
            return nullptr;
610
2
        }
611
6
        std::string pattern_str = patterns.get_data_at(i).to_string();
612
6
        if (allpass_state->_pattern_matched) {
613
2
            if constexpr (LIKE_PATTERN) {
614
2
                allpass_state->like_pattern_match(pattern_str);
615
2
            } else {
616
2
                allpass_state->regexp_pattern_match(pattern_str);
617
2
            }
618
2
        }
619
6
        if (equal_state->_pattern_matched) {
620
2
            if constexpr (LIKE_PATTERN) {
621
2
                equal_state->like_pattern_match(pattern_str);
622
2
            } else {
623
2
                equal_state->regexp_pattern_match(pattern_str);
624
2
            }
625
2
        }
626
6
        if (substring_state->_pattern_matched) {
627
6
            if constexpr (LIKE_PATTERN) {
628
6
                substring_state->like_pattern_match(pattern_str);
629
6
            } else {
630
6
                substring_state->regexp_pattern_match(pattern_str);
631
6
            }
632
6
        }
633
6
        if (starts_with_state->_pattern_matched) {
634
2
            if constexpr (LIKE_PATTERN) {
635
2
                starts_with_state->like_pattern_match(pattern_str);
636
2
            } else {
637
2
                starts_with_state->regexp_pattern_match(pattern_str);
638
2
            }
639
2
        }
640
6
        if (ends_with_state->_pattern_matched) {
641
2
            if constexpr (LIKE_PATTERN) {
642
2
                ends_with_state->like_pattern_match(pattern_str);
643
2
            } else {
644
2
                ends_with_state->regexp_pattern_match(pattern_str);
645
2
            }
646
2
        }
647
6
    }
648
649
0
    if (allpass_state->_pattern_matched) {
650
0
        return allpass_state;
651
0
    } else if (equal_state->_pattern_matched) {
652
0
        return equal_state;
653
0
    } else if (substring_state->_pattern_matched) {
654
0
        return substring_state;
655
0
    } else if (starts_with_state->_pattern_matched) {
656
0
        return starts_with_state;
657
0
    } else if (ends_with_state->_pattern_matched) {
658
0
        return ends_with_state;
659
0
    } else {
660
0
        return nullptr;
661
0
    }
662
0
}
_ZN5doris10vectorized16FunctionLikeBase24pattern_type_recognitionILb0EEESt10shared_ptrINS0_24VectorPatternSearchStateEERKNS0_9ColumnStrIjEE
Line
Count
Source
597
22
VPatternSearchStateSPtr FunctionLikeBase::pattern_type_recognition(const ColumnString& patterns) {
598
22
    VPatternSearchStateSPtr allpass_state = std::make_shared<VectorAllpassSearchState>();
599
22
    VPatternSearchStateSPtr equal_state = std::make_shared<VectorEqualSearchState>();
600
22
    VPatternSearchStateSPtr substring_state = std::make_shared<VectorSubStringSearchState>();
601
22
    VPatternSearchStateSPtr starts_with_state = std::make_shared<VectorStartsWithSearchState>();
602
22
    VPatternSearchStateSPtr ends_with_state = std::make_shared<VectorEndsWithSearchState>();
603
22
    size_t size = patterns.size();
604
605
44
    for (size_t i = 0; i < size; ++i) {
606
22
        if (!allpass_state->_pattern_matched && !equal_state->_pattern_matched &&
607
22
            !substring_state->_pattern_matched && !starts_with_state->_pattern_matched &&
608
22
            !ends_with_state->_pattern_matched) {
609
0
            return nullptr;
610
0
        }
611
22
        std::string pattern_str = patterns.get_data_at(i).to_string();
612
22
        if (allpass_state->_pattern_matched) {
613
22
            if constexpr (LIKE_PATTERN) {
614
22
                allpass_state->like_pattern_match(pattern_str);
615
22
            } else {
616
22
                allpass_state->regexp_pattern_match(pattern_str);
617
22
            }
618
22
        }
619
22
        if (equal_state->_pattern_matched) {
620
22
            if constexpr (LIKE_PATTERN) {
621
22
                equal_state->like_pattern_match(pattern_str);
622
22
            } else {
623
22
                equal_state->regexp_pattern_match(pattern_str);
624
22
            }
625
22
        }
626
22
        if (substring_state->_pattern_matched) {
627
22
            if constexpr (LIKE_PATTERN) {
628
22
                substring_state->like_pattern_match(pattern_str);
629
22
            } else {
630
22
                substring_state->regexp_pattern_match(pattern_str);
631
22
            }
632
22
        }
633
22
        if (starts_with_state->_pattern_matched) {
634
22
            if constexpr (LIKE_PATTERN) {
635
22
                starts_with_state->like_pattern_match(pattern_str);
636
22
            } else {
637
22
                starts_with_state->regexp_pattern_match(pattern_str);
638
22
            }
639
22
        }
640
22
        if (ends_with_state->_pattern_matched) {
641
22
            if constexpr (LIKE_PATTERN) {
642
22
                ends_with_state->like_pattern_match(pattern_str);
643
22
            } else {
644
22
                ends_with_state->regexp_pattern_match(pattern_str);
645
22
            }
646
22
        }
647
22
    }
648
649
22
    if (allpass_state->_pattern_matched) {
650
0
        return allpass_state;
651
22
    } else if (equal_state->_pattern_matched) {
652
4
        return equal_state;
653
18
    } else if (substring_state->_pattern_matched) {
654
0
        return substring_state;
655
18
    } else if (starts_with_state->_pattern_matched) {
656
4
        return starts_with_state;
657
14
    } else if (ends_with_state->_pattern_matched) {
658
4
        return ends_with_state;
659
10
    } else {
660
10
        return nullptr;
661
10
    }
662
22
}
663
664
Status FunctionLikeBase::vector_non_const(const ColumnString& values, const ColumnString& patterns,
665
                                          ColumnUInt8::Container& result, LikeState* state,
666
24
                                          size_t input_rows_count) const {
667
24
    VPatternSearchStateSPtr vector_search_state;
668
24
    if (state->is_like_pattern) {
669
2
        vector_search_state = pattern_type_recognition<true>(patterns);
670
22
    } else {
671
22
        vector_search_state = pattern_type_recognition<false>(patterns);
672
22
    }
673
24
    if (vector_search_state == nullptr) {
674
        // pattern type recognition failed, use default case
675
82
        for (int i = 0; i < input_rows_count; ++i) {
676
70
            const auto pattern_val = patterns.get_data_at(i);
677
70
            const auto value_val = values.get_data_at(i);
678
70
            RETURN_IF_ERROR((state->scalar_function)(&state->search_state, value_val, pattern_val,
679
70
                                                     &result[i]));
680
70
        }
681
12
        return Status::OK();
682
12
    }
683
12
    const auto* search_strings =
684
12
            static_cast<const ColumnString*>(vector_search_state->_search_strings.get());
685
12
    return (vector_search_state->_vector_function)(values, *search_strings, result);
686
24
}
687
688
Status FunctionLike::like_fn(LikeSearchState* state, const ColumnString& val,
689
0
                             const StringRef& pattern, ColumnUInt8::Container& result) {
690
0
    std::string re_pattern;
691
0
    convert_like_pattern(state, std::string(pattern.data, pattern.size), &re_pattern);
692
0
    return regexp_fn(state, val, {re_pattern.c_str(), re_pattern.size()}, result);
693
0
}
694
695
Status FunctionLike::like_fn_scalar(LikeSearchState* state, const StringRef& val,
696
60
                                    const StringRef& pattern, unsigned char* result) {
697
60
    std::string re_pattern;
698
60
    convert_like_pattern(state, std::string(pattern.data, pattern.size), &re_pattern);
699
700
60
    return regexp_fn_scalar(state, StringRef(val.data, val.size),
701
60
                            {re_pattern.c_str(), re_pattern.size()}, result);
702
60
}
703
704
void FunctionLike::convert_like_pattern(LikeSearchState* state, const std::string& pattern,
705
60
                                        std::string* re_pattern) {
706
60
    re_pattern->clear();
707
708
60
    if (pattern.empty()) {
709
8
        re_pattern->append("^$");
710
8
        return;
711
8
    }
712
713
    // add ^ to pattern head to match line head
714
52
    if (!pattern.empty() && pattern[0] != '%') {
715
34
        re_pattern->append("^");
716
34
    }
717
718
52
    bool is_escaped = false;
719
    // expect % and _, all chars should keep it literal means.
720
214
    for (char i : pattern) {
721
214
        if (is_escaped) { // last is \, this should be escape
722
8
            if (i == '[' || i == ']' || i == '(' || i == ')' || i == '{' || i == '}' || i == '-' ||
723
8
                i == '*' || i == '+' || i == '\\' || i == '|' || i == '/' || i == ':' || i == '^' ||
724
8
                i == '.' || i == '$' || i == '?') {
725
0
                re_pattern->append(1, '\\');
726
8
            } else if (i != '%' && i != '_') {
727
8
                re_pattern->append(2, '\\');
728
8
            }
729
8
            re_pattern->append(1, i);
730
8
            is_escaped = false;
731
206
        } else {
732
206
            switch (i) {
733
44
            case '%':
734
44
                re_pattern->append(".*");
735
44
                break;
736
42
            case '_':
737
42
                re_pattern->append(".");
738
42
                break;
739
120
            default:
740
120
                is_escaped = i == state->escape_char;
741
120
                if (!is_escaped) {
742
                    // special for hyperscan: [, ], (, ), {, }, -, *, +, \, |, /, :, ^, ., $, ?
743
112
                    if (i == '[' || i == ']' || i == '(' || i == ')' || i == '{' || i == '}' ||
744
112
                        i == '-' || i == '*' || i == '+' || i == '\\' || i == '|' || i == '/' ||
745
112
                        i == ':' || i == '^' || i == '.' || i == '$' || i == '?') {
746
14
                        re_pattern->append(1, '\\');
747
14
                    }
748
112
                    re_pattern->append(1, i);
749
112
                }
750
120
                break;
751
206
            }
752
206
        }
753
214
    }
754
755
    // add $ to pattern tail to match line tail
756
52
    if (!pattern.empty() && re_pattern->back() != '*') {
757
26
        re_pattern->append("$");
758
26
    }
759
52
}
760
761
4
void FunctionLike::remove_escape_character(std::string* search_string) {
762
4
    std::string tmp_search_string;
763
4
    tmp_search_string.swap(*search_string);
764
4
    int len = tmp_search_string.length();
765
10
    for (int i = 0; i < len;) {
766
6
        if (tmp_search_string[i] == '\\' && i + 1 < len &&
767
6
            (tmp_search_string[i + 1] == '%' || tmp_search_string[i + 1] == '_' ||
768
0
             tmp_search_string[i + 1] == '\\')) {
769
0
            search_string->append(1, tmp_search_string[i + 1]);
770
0
            i += 2;
771
6
        } else {
772
6
            search_string->append(1, tmp_search_string[i]);
773
6
            i++;
774
6
        }
775
6
    }
776
4
}
777
778
0
bool re2_full_match(const std::string& str, const RE2& re, std::vector<std::string>& results) {
779
0
    if (!re.ok()) {
780
0
        return false;
781
0
    }
782
783
0
    std::vector<RE2::Arg> arguments;
784
0
    std::vector<RE2::Arg*> arguments_ptrs;
785
0
    std::size_t args_count = re.NumberOfCapturingGroups();
786
0
    arguments.resize(args_count);
787
0
    arguments_ptrs.resize(args_count);
788
0
    results.resize(args_count);
789
0
    for (std::size_t i = 0; i < args_count; ++i) {
790
0
        arguments[i] = &results[i];
791
0
        arguments_ptrs[i] = &arguments[i];
792
0
    }
793
794
0
    return RE2::FullMatchN(str, re, arguments_ptrs.data(), args_count);
795
0
}
796
797
0
void verbose_log_match(const std::string& str, const std::string& pattern_name, const RE2& re) {
798
0
    std::vector<std::string> results;
799
0
    VLOG_DEBUG << "arg str: " << str << ", size: " << str.size() << ", pattern " << pattern_name
800
0
               << ": " << re.pattern() << ", size: " << re.pattern().size();
801
0
    if (re2_full_match(str, re, results)) {
802
0
        for (int i = 0; i < results.size(); ++i) {
803
0
            VLOG_DEBUG << "match " << i << ": " << results[i] << ", size: " << results[i].size();
804
0
        }
805
0
    } else {
806
0
        VLOG_DEBUG << "no match";
807
0
    }
808
0
}
809
810
Status FunctionLike::construct_like_const_state(FunctionContext* context, const StringRef& pattern,
811
                                                std::shared_ptr<LikeState>& state,
812
0
                                                bool try_hyperscan) {
813
0
    std::string pattern_str = pattern.to_string();
814
0
    state->search_state.pattern_str = pattern_str;
815
0
    std::string search_string;
816
817
0
    if (!pattern_str.empty() && RE2::FullMatch(pattern_str, LIKE_ALLPASS_RE)) {
818
0
        state->search_state.set_search_string("");
819
0
        state->function = constant_allpass_fn;
820
0
        state->scalar_function = constant_allpass_fn_scalar;
821
0
    } else if (pattern_str.empty() || RE2::FullMatch(pattern_str, LIKE_EQUALS_RE, &search_string)) {
822
0
        if (VLOG_DEBUG_IS_ON) {
823
0
            verbose_log_match(pattern_str, "LIKE_EQUALS_RE", LIKE_EQUALS_RE);
824
0
            VLOG_DEBUG << "search_string : " << search_string << ", size: " << search_string.size();
825
0
        }
826
0
        remove_escape_character(&search_string);
827
0
        if (VLOG_DEBUG_IS_ON) {
828
0
            VLOG_DEBUG << "search_string escape removed: " << search_string
829
0
                       << ", size: " << search_string.size();
830
0
        }
831
0
        state->search_state.set_search_string(search_string);
832
0
        state->function = constant_equals_fn;
833
0
        state->scalar_function = constant_equals_fn_scalar;
834
0
    } else if (RE2::FullMatch(pattern_str, LIKE_STARTS_WITH_RE, &search_string)) {
835
0
        if (VLOG_DEBUG_IS_ON) {
836
0
            verbose_log_match(pattern_str, "LIKE_STARTS_WITH_RE", LIKE_STARTS_WITH_RE);
837
0
            VLOG_DEBUG << "search_string : " << search_string << ", size: " << search_string.size();
838
0
        }
839
0
        remove_escape_character(&search_string);
840
0
        if (VLOG_DEBUG_IS_ON) {
841
0
            VLOG_DEBUG << "search_string escape removed: " << search_string
842
0
                       << ", size: " << search_string.size();
843
0
        }
844
0
        state->search_state.set_search_string(search_string);
845
0
        state->function = constant_starts_with_fn;
846
0
        state->scalar_function = constant_starts_with_fn_scalar;
847
0
    } else if (RE2::FullMatch(pattern_str, LIKE_ENDS_WITH_RE, &search_string)) {
848
0
        if (VLOG_DEBUG_IS_ON) {
849
0
            verbose_log_match(pattern_str, "LIKE_ENDS_WITH_RE", LIKE_ENDS_WITH_RE);
850
0
            VLOG_DEBUG << "search_string : " << search_string << ", size: " << search_string.size();
851
0
        }
852
0
        remove_escape_character(&search_string);
853
0
        if (VLOG_DEBUG_IS_ON) {
854
0
            VLOG_DEBUG << "search_string escape removed: " << search_string
855
0
                       << ", size: " << search_string.size();
856
0
        }
857
0
        state->search_state.set_search_string(search_string);
858
0
        state->function = constant_ends_with_fn;
859
0
        state->scalar_function = constant_ends_with_fn_scalar;
860
0
    } else if (RE2::FullMatch(pattern_str, LIKE_SUBSTRING_RE, &search_string)) {
861
0
        if (VLOG_DEBUG_IS_ON) {
862
0
            verbose_log_match(pattern_str, "LIKE_SUBSTRING_RE", LIKE_SUBSTRING_RE);
863
0
            VLOG_DEBUG << "search_string : " << search_string << ", size: " << search_string.size();
864
0
        }
865
0
        remove_escape_character(&search_string);
866
0
        if (VLOG_DEBUG_IS_ON) {
867
0
            VLOG_DEBUG << "search_string escape removed: " << search_string
868
0
                       << ", size: " << search_string.size();
869
0
        }
870
0
        state->search_state.set_search_string(search_string);
871
0
        state->function = constant_substring_fn;
872
0
        state->scalar_function = constant_substring_fn_scalar;
873
0
    } else {
874
0
        std::string re_pattern;
875
0
        convert_like_pattern(&state->search_state, pattern_str, &re_pattern);
876
0
        if (VLOG_DEBUG_IS_ON) {
877
0
            VLOG_DEBUG << "hyperscan, pattern str: " << pattern_str
878
0
                       << ", size: " << pattern_str.size() << ", re pattern: " << re_pattern
879
0
                       << ", size: " << re_pattern.size();
880
0
        }
881
882
0
        hs_database_t* database = nullptr;
883
0
        hs_scratch_t* scratch = nullptr;
884
0
        if (try_hyperscan && hs_prepare(context, re_pattern.c_str(), &database, &scratch).ok()) {
885
            // use hyperscan
886
0
            state->search_state.hs_database.reset(database);
887
0
            state->search_state.hs_scratch.reset(scratch);
888
0
        } else {
889
            // fallback to re2
890
            // reset hs_database to nullptr to indicate not use hyperscan
891
0
            state->search_state.hs_database.reset();
892
0
            state->search_state.hs_scratch.reset();
893
894
0
            RE2::Options opts;
895
0
            opts.set_never_nl(false);
896
0
            opts.set_dot_nl(true);
897
0
            state->search_state.regex = std::make_unique<RE2>(re_pattern, opts);
898
0
            if (!state->search_state.regex->ok()) {
899
0
                return Status::InternalError("Invalid regex expression: {}(origin: {})", re_pattern,
900
0
                                             pattern_str);
901
0
            }
902
0
        }
903
904
0
        state->function = constant_regex_fn;
905
0
        state->scalar_function = constant_regex_fn_scalar;
906
0
    }
907
0
    return Status::OK();
908
0
}
909
910
4
Status FunctionLike::open(FunctionContext* context, FunctionContext::FunctionStateScope scope) {
911
4
    if (scope != FunctionContext::THREAD_LOCAL) {
912
2
        return Status::OK();
913
2
    }
914
2
    std::shared_ptr<LikeState> state = std::make_shared<LikeState>();
915
2
    state->is_like_pattern = true;
916
2
    state->function = like_fn;
917
2
    state->scalar_function = like_fn_scalar;
918
2
    if (context->is_col_constant(1)) {
919
0
        const auto pattern_col = context->get_constant_col(1)->column_ptr;
920
0
        const auto& pattern = pattern_col->get_data_at(0);
921
0
        RETURN_IF_ERROR(construct_like_const_state(context, pattern, state));
922
0
    }
923
2
    context->set_function_state(scope, state);
924
925
2
    return Status::OK();
926
2
}
927
928
Status FunctionRegexpLike::open(FunctionContext* context,
929
68
                                FunctionContext::FunctionStateScope scope) {
930
68
    if (scope != FunctionContext::THREAD_LOCAL) {
931
34
        return Status::OK();
932
34
    }
933
34
    std::shared_ptr<LikeState> state = std::make_shared<LikeState>();
934
34
    context->set_function_state(scope, state);
935
34
    state->is_like_pattern = false;
936
34
    state->function = regexp_fn;
937
34
    state->scalar_function = regexp_fn_scalar;
938
34
    if (context->is_col_constant(1)) {
939
34
        const auto pattern_col = context->get_constant_col(1)->column_ptr;
940
34
        const auto& pattern = pattern_col->get_data_at(0);
941
942
34
        std::string pattern_str = pattern.to_string();
943
34
        std::string search_string;
944
34
        if (RE2::FullMatch(pattern_str, ALLPASS_RE)) {
945
0
            state->search_state.set_search_string("");
946
0
            state->function = constant_allpass_fn;
947
0
            state->scalar_function = constant_allpass_fn_scalar;
948
34
        } else if (RE2::FullMatch(pattern_str, EQUALS_RE, &search_string)) {
949
4
            state->search_state.set_search_string(search_string);
950
4
            state->function = constant_equals_fn;
951
4
            state->scalar_function = constant_equals_fn_scalar;
952
30
        } else if (RE2::FullMatch(pattern_str, STARTS_WITH_RE, &search_string)) {
953
4
            state->search_state.set_search_string(search_string);
954
4
            state->function = constant_starts_with_fn;
955
4
            state->scalar_function = constant_starts_with_fn_scalar;
956
26
        } else if (RE2::FullMatch(pattern_str, ENDS_WITH_RE, &search_string)) {
957
4
            state->search_state.set_search_string(search_string);
958
4
            state->function = constant_ends_with_fn;
959
4
            state->scalar_function = constant_ends_with_fn_scalar;
960
22
        } else if (RE2::FullMatch(pattern_str, SUBSTRING_RE, &search_string)) {
961
12
            state->search_state.set_search_string(search_string);
962
12
            state->function = constant_substring_fn;
963
12
            state->scalar_function = constant_substring_fn_scalar;
964
12
        } else {
965
10
            hs_database_t* database = nullptr;
966
10
            hs_scratch_t* scratch = nullptr;
967
10
            if (hs_prepare(context, pattern_str.c_str(), &database, &scratch).ok()) {
968
                // use hyperscan
969
10
                state->search_state.hs_database.reset(database);
970
10
                state->search_state.hs_scratch.reset(scratch);
971
10
            } else {
972
                // fallback to re2
973
                // reset hs_database to nullptr to indicate not use hyperscan
974
0
                state->search_state.hs_database.reset();
975
0
                state->search_state.hs_scratch.reset();
976
0
                RE2::Options opts;
977
0
                opts.set_never_nl(false);
978
0
                opts.set_dot_nl(true);
979
0
                state->search_state.regex = std::make_unique<RE2>(pattern_str, opts);
980
0
                if (!state->search_state.regex->ok()) {
981
0
                    return Status::InternalError("Invalid regex expression: {}", pattern_str);
982
0
                }
983
0
            }
984
10
            state->function = constant_regex_fn;
985
10
            state->scalar_function = constant_regex_fn_scalar;
986
10
        }
987
34
    }
988
34
    return Status::OK();
989
34
}
990
991
2
void register_function_like(SimpleFunctionFactory& factory) {
992
2
    factory.register_function<FunctionLike>();
993
2
}
994
995
2
void register_function_regexp(SimpleFunctionFactory& factory) {
996
2
    factory.register_function<FunctionRegexpLike>();
997
2
    factory.register_alias(FunctionRegexpLike::name, FunctionRegexpLike::alias);
998
2
}
999
1000
} // namespace doris::vectorized