Coverage Report

Created: 2026-06-01 20:29

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
be/src/exprs/function/function_regexp.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 <glog/logging.h>
19
#include <re2/re2.h>
20
#include <re2/stringpiece.h>
21
#include <stddef.h>
22
23
#include <boost/regex.hpp>
24
#include <memory>
25
#include <string>
26
#include <string_view>
27
#include <type_traits>
28
#include <utility>
29
#include <vector>
30
31
#include "common/status.h"
32
#include "core/block/block.h"
33
#include "core/block/column_numbers.h"
34
#include "core/block/column_with_type_and_name.h"
35
#include "core/column/column.h"
36
#include "core/column/column_const.h"
37
#include "core/column/column_execute_util.h"
38
#include "core/column/column_nullable.h"
39
#include "core/column/column_string.h"
40
#include "core/column/column_vector.h"
41
#include "core/data_type/data_type.h"
42
#include "core/data_type/data_type_nullable.h"
43
#include "core/data_type/data_type_number.h"
44
#include "core/data_type/data_type_string.h"
45
#include "core/string_ref.h"
46
#include "core/types.h"
47
#include "exec/common/stringop_substring.h"
48
#include "exprs/aggregate/aggregate_function.h"
49
#include "exprs/function/function.h"
50
#include "exprs/function/simple_function_factory.h"
51
#include "exprs/function_context.h"
52
#include "exprs/string_functions.h"
53
54
namespace doris {
55
56
// Helper structure to hold either RE2 or Boost.Regex
57
struct RegexpExtractEngine {
58
    std::unique_ptr<re2::RE2> re2_regex;
59
    std::unique_ptr<boost::regex> boost_regex;
60
61
0
    bool is_boost() const { return boost_regex != nullptr; }
62
46
    bool is_re2() const { return re2_regex != nullptr; }
63
64
    // Try to compile with RE2 first, fallback to Boost.Regex if RE2 fails
65
    static bool compile(const StringRef& pattern, std::string* error_str,
66
26
                        RegexpExtractEngine& engine, bool enable_extended_regex) {
67
26
        re2::RE2::Options options;
68
26
        options.set_log_errors(false); // avoid RE2 printing to stderr; we handle errors ourselves
69
26
        options.set_dot_nl(true); // make '.' match '\n' by default, consistent with REGEXP/LIKE
70
26
        engine.re2_regex =
71
26
                std::make_unique<re2::RE2>(re2::StringPiece(pattern.data, pattern.size), options);
72
73
26
        if (engine.re2_regex->ok()) {
74
26
            return true;
75
26
        } else if (!enable_extended_regex) {
76
0
            *error_str = fmt::format(
77
0
                    "Invalid regex pattern: {}. Error: {}. If you need advanced regex features, "
78
0
                    "try setting enable_extended_regex=true",
79
0
                    std::string(pattern.data, pattern.size), engine.re2_regex->error());
80
0
            return false;
81
0
        }
82
83
        // RE2 failed, try Boost.Regex for advanced features like zero-width assertions
84
0
        engine.re2_regex.reset();
85
0
        try {
86
0
            boost::regex::flag_type flags = boost::regex::normal;
87
0
            engine.boost_regex = std::make_unique<boost::regex>(pattern.data,
88
0
                                                                pattern.data + pattern.size, flags);
89
0
            return true;
90
0
        } catch (const boost::regex_error& e) {
91
0
            if (error_str) {
92
0
                *error_str = fmt::format("Invalid regex pattern: {}. Error: {}",
93
0
                                         std::string(pattern.data, pattern.size), e.what());
94
0
            }
95
0
            return false;
96
0
        }
97
0
    }
98
99
    // Get number of capturing groups
100
23
    int number_of_capturing_groups() const {
101
23
        if (is_re2()) {
102
23
            return re2_regex->NumberOfCapturingGroups();
103
23
        } else if (is_boost()) {
104
0
            return static_cast<int>(boost_regex->mark_count());
105
0
        }
106
0
        return 0;
107
23
    }
108
109
    // Match function for extraction
110
16
    bool match_and_extract(const char* data, size_t size, int index, std::string& result) const {
111
16
        if (is_re2()) {
112
16
            int max_matches = 1 + re2_regex->NumberOfCapturingGroups();
113
16
            if (index >= max_matches) {
114
0
                return false;
115
0
            }
116
16
            std::vector<re2::StringPiece> matches(max_matches);
117
16
            bool success = re2_regex->Match(re2::StringPiece(data, size), 0, size,
118
16
                                            re2::RE2::UNANCHORED, matches.data(), max_matches);
119
16
            if (success && index < matches.size()) {
120
16
                const re2::StringPiece& match = matches[index];
121
16
                result.assign(match.data(), match.size());
122
16
                return true;
123
16
            }
124
0
            return false;
125
16
        } else if (is_boost()) {
126
0
            boost::cmatch matches;
127
0
            bool success = boost::regex_search(data, data + size, matches, *boost_regex);
128
0
            if (success && index < matches.size()) {
129
0
                result = matches[index].str();
130
0
                return true;
131
0
            }
132
0
            return false;
133
0
        }
134
0
        return false;
135
16
    }
136
137
    // Match all occurrences and extract the first capturing group
138
    void match_all_and_extract(const char* data, size_t size,
139
7
                               std::vector<std::string>& results) const {
140
7
        if (is_re2()) {
141
7
            int max_matches = 1 + re2_regex->NumberOfCapturingGroups();
142
7
            if (max_matches < 2) {
143
0
                return; // No capturing groups
144
0
            }
145
146
7
            size_t pos = 0;
147
19
            while (pos < size) {
148
18
                const char* str_pos = data + pos;
149
18
                size_t str_size = size - pos;
150
18
                std::vector<re2::StringPiece> matches(max_matches);
151
18
                bool success = re2_regex->Match(re2::StringPiece(str_pos, str_size), 0, str_size,
152
18
                                                re2::RE2::UNANCHORED, matches.data(), max_matches);
153
18
                if (!success) {
154
6
                    break;
155
6
                }
156
12
                if (matches[0].empty()) {
157
0
                    pos += 1;
158
0
                    continue;
159
0
                }
160
                // Extract first capturing group
161
12
                if (matches.size() > 1 && !matches[1].empty()) {
162
12
                    results.emplace_back(matches[1].data(), matches[1].size());
163
12
                }
164
                // Move position forward
165
12
                auto offset = std::string(str_pos, str_size)
166
12
                                      .find(std::string(matches[0].data(), matches[0].size()));
167
12
                pos += offset + matches[0].size();
168
12
            }
169
7
        } else if (is_boost()) {
170
0
            const char* search_start = data;
171
0
            const char* search_end = data + size;
172
0
            boost::match_results<const char*> matches;
173
174
0
            while (boost::regex_search(search_start, search_end, matches, *boost_regex)) {
175
0
                if (matches.size() > 1 && matches[1].matched) {
176
0
                    results.emplace_back(matches[1].str());
177
0
                }
178
0
                if (matches[0].length() == 0) {
179
0
                    if (search_start == search_end) {
180
0
                        break;
181
0
                    }
182
0
                    search_start += 1;
183
0
                } else {
184
0
                    search_start = matches[0].second;
185
0
                }
186
0
            }
187
0
        }
188
7
    }
189
};
190
191
struct RegexpCountImpl {
192
    using StringColumnView = ColumnView<TYPE_STRING>;
193
194
    static void execute_impl(FunctionContext* context, ColumnPtr argument_columns[],
195
13
                             size_t input_rows_count, ColumnInt32::Container& result_data) {
196
13
        auto str_col = StringColumnView::create(argument_columns[0]);
197
13
        auto pattern_col = StringColumnView::create(argument_columns[1]);
198
31
        for (size_t i = 0; i < input_rows_count; ++i) {
199
18
            DCHECK(!str_col.is_null_at(i));
200
18
            DCHECK(!pattern_col.is_null_at(i));
201
18
            result_data[i] = _execute_inner_loop(context, str_col, pattern_col, i);
202
18
        }
203
13
    }
204
    static int _execute_inner_loop(FunctionContext* context, const StringColumnView& str_col,
205
18
                                   const StringColumnView& pattern_col, const size_t index_now) {
206
18
        re2::RE2* re = reinterpret_cast<re2::RE2*>(
207
18
                context->get_function_state(FunctionContext::THREAD_LOCAL));
208
18
        std::unique_ptr<re2::RE2> scoped_re;
209
18
        if (re == nullptr) {
210
12
            std::string error_str;
211
12
            const auto pattern = pattern_col.value_at(index_now);
212
12
            bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(), StringRef(),
213
12
                                                     scoped_re);
214
12
            if (!st) {
215
0
                context->add_warning(error_str.c_str());
216
0
                throw Exception(Status::InvalidArgument(error_str));
217
0
                return 0;
218
0
            }
219
12
            re = scoped_re.get();
220
12
        }
221
222
18
        const auto str = str_col.value_at(index_now);
223
18
        int count = 0;
224
18
        size_t pos = 0;
225
70
        while (pos < str.size) {
226
64
            auto str_pos = str.data + pos;
227
64
            auto str_size = str.size - pos;
228
64
            re2::StringPiece str_sp_current = re2::StringPiece(str_pos, str_size);
229
64
            re2::StringPiece match;
230
231
64
            bool success = re->Match(str_sp_current, 0, str_size, re2::RE2::UNANCHORED, &match, 1);
232
64
            if (!success) {
233
12
                break;
234
12
            }
235
52
            if (match.empty()) {
236
20
                pos += 1;
237
20
                continue;
238
20
            }
239
32
            count++;
240
32
            size_t match_start = match.data() - str_sp_current.data();
241
32
            pos += match_start + match.size();
242
32
        }
243
244
18
        return count;
245
18
    }
246
};
247
248
class FunctionRegexpCount : public IFunction {
249
public:
250
    static constexpr auto name = "regexp_count";
251
252
21
    static FunctionPtr create() { return std::make_shared<FunctionRegexpCount>(); }
253
254
1
    String get_name() const override { return name; }
255
256
19
    size_t get_number_of_arguments() const override { return 2; }
257
258
19
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
259
19
        return std::make_shared<DataTypeInt32>();
260
19
    }
261
262
38
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
263
38
        if (scope == FunctionContext::THREAD_LOCAL) {
264
19
            if (context->is_col_constant(1)) {
265
12
                DCHECK(!context->get_function_state(scope));
266
12
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
267
12
                const auto& pattern = pattern_col->get_data_at(0);
268
12
                if (pattern.size == 0) {
269
4
                    return Status::OK();
270
4
                }
271
272
8
                std::string error_str;
273
8
                std::unique_ptr<re2::RE2> scoped_re;
274
8
                bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
275
8
                                                         StringRef(), scoped_re);
276
8
                if (!st) {
277
0
                    context->set_error(error_str.c_str());
278
0
                    return Status::InvalidArgument(error_str);
279
0
                }
280
8
                std::shared_ptr<re2::RE2> re(scoped_re.release());
281
8
                context->set_function_state(scope, re);
282
8
            }
283
19
        }
284
34
        return Status::OK();
285
38
    }
286
287
    Status execute_impl(FunctionContext* context, Block& block, const ColumnNumbers& arguments,
288
13
                        uint32_t result, size_t input_rows_count) const override {
289
13
        auto result_data_column = ColumnInt32::create(input_rows_count);
290
13
        auto& result_data = result_data_column->get_data();
291
292
13
        ColumnPtr argument_columns[2];
293
294
13
        argument_columns[0] = block.get_by_position(arguments[0]).column;
295
13
        argument_columns[1] = block.get_by_position(arguments[1]).column;
296
13
        RegexpCountImpl::execute_impl(context, argument_columns, input_rows_count, result_data);
297
298
13
        block.get_by_position(result).column = std::move(result_data_column);
299
13
        return Status::OK();
300
13
    }
301
};
302
303
struct ThreeParamTypes {
304
2
    static DataTypes get_variadic_argument_types() {
305
2
        return {std::make_shared<DataTypeString>(), std::make_shared<DataTypeString>(),
306
2
                std::make_shared<DataTypeString>()};
307
2
    }
308
};
309
310
struct FourParamTypes {
311
2
    static DataTypes get_variadic_argument_types() {
312
2
        return {std::make_shared<DataTypeString>(), std::make_shared<DataTypeString>(),
313
2
                std::make_shared<DataTypeString>(), std::make_shared<DataTypeString>()};
314
2
    }
315
};
316
317
// template FunctionRegexpFunctionality is used for regexp_replace/regexp_replace_one
318
template <typename Impl, typename ParamTypes>
319
class FunctionRegexpReplace : public IFunction {
320
public:
321
    static constexpr auto name = Impl::name;
322
323
24
    static FunctionPtr create() { return std::make_shared<FunctionRegexpReplace>(); }
_ZN5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE6createEv
Line
Count
Source
323
10
    static FunctionPtr create() { return std::make_shared<FunctionRegexpReplace>(); }
_ZN5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE6createEv
Line
Count
Source
323
2
    static FunctionPtr create() { return std::make_shared<FunctionRegexpReplace>(); }
_ZN5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE6createEv
Line
Count
Source
323
10
    static FunctionPtr create() { return std::make_shared<FunctionRegexpReplace>(); }
_ZN5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE6createEv
Line
Count
Source
323
2
    static FunctionPtr create() { return std::make_shared<FunctionRegexpReplace>(); }
324
325
0
    String get_name() const override { return name; }
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE8get_nameB5cxx11Ev
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE8get_nameB5cxx11Ev
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE8get_nameB5cxx11Ev
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE8get_nameB5cxx11Ev
326
327
0
    size_t get_number_of_arguments() const override {
328
0
        return get_variadic_argument_types_impl().size();
329
0
    }
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE23get_number_of_argumentsEv
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE23get_number_of_argumentsEv
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE23get_number_of_argumentsEv
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE23get_number_of_argumentsEv
330
331
20
    bool is_variadic() const override { return true; }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE11is_variadicEv
Line
Count
Source
331
9
    bool is_variadic() const override { return true; }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE11is_variadicEv
Line
Count
Source
331
1
    bool is_variadic() const override { return true; }
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE11is_variadicEv
Line
Count
Source
331
9
    bool is_variadic() const override { return true; }
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE11is_variadicEv
Line
Count
Source
331
1
    bool is_variadic() const override { return true; }
332
333
16
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
334
16
        return make_nullable(std::make_shared<DataTypeString>());
335
16
    }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
Line
Count
Source
333
8
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
334
8
        return make_nullable(std::make_shared<DataTypeString>());
335
8
    }
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
Line
Count
Source
333
8
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
334
8
        return make_nullable(std::make_shared<DataTypeString>());
335
8
    }
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
336
337
4
    DataTypes get_variadic_argument_types_impl() const override {
338
4
        return ParamTypes::get_variadic_argument_types();
339
4
    }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE32get_variadic_argument_types_implEv
Line
Count
Source
337
1
    DataTypes get_variadic_argument_types_impl() const override {
338
1
        return ParamTypes::get_variadic_argument_types();
339
1
    }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE32get_variadic_argument_types_implEv
Line
Count
Source
337
1
    DataTypes get_variadic_argument_types_impl() const override {
338
1
        return ParamTypes::get_variadic_argument_types();
339
1
    }
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE32get_variadic_argument_types_implEv
Line
Count
Source
337
1
    DataTypes get_variadic_argument_types_impl() const override {
338
1
        return ParamTypes::get_variadic_argument_types();
339
1
    }
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE32get_variadic_argument_types_implEv
Line
Count
Source
337
1
    DataTypes get_variadic_argument_types_impl() const override {
338
1
        return ParamTypes::get_variadic_argument_types();
339
1
    }
340
341
32
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
342
32
        if (scope == FunctionContext::THREAD_LOCAL) {
343
16
            if (context->is_col_constant(1)) {
344
16
                DCHECK(!context->get_function_state(scope));
345
16
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
346
16
                const auto& pattern = pattern_col->get_data_at(0);
347
16
                if (pattern.size == 0) {
348
4
                    return Status::OK();
349
4
                }
350
351
12
                std::string error_str;
352
12
                std::unique_ptr<re2::RE2> scoped_re;
353
12
                StringRef options_value;
354
12
                if constexpr (std::is_same_v<FourParamTypes, ParamTypes>) {
355
0
                    DCHECK_EQ(context->get_num_args(), 4);
356
0
                    DCHECK(context->is_col_constant(3));
357
0
                    const auto options_col = context->get_constant_col(3)->column_ptr;
358
0
                    options_value = options_col->get_data_at(0);
359
0
                }
360
361
12
                bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
362
12
                                                         options_value, scoped_re);
363
12
                if (!st) {
364
0
                    context->set_error(error_str.c_str());
365
0
                    return Status::InvalidArgument(error_str);
366
0
                }
367
12
                std::shared_ptr<re2::RE2> re(scoped_re.release());
368
12
                context->set_function_state(scope, re);
369
12
            }
370
16
        }
371
28
        return Status::OK();
372
32
    }
_ZN5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
Line
Count
Source
341
16
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
342
16
        if (scope == FunctionContext::THREAD_LOCAL) {
343
8
            if (context->is_col_constant(1)) {
344
8
                DCHECK(!context->get_function_state(scope));
345
8
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
346
8
                const auto& pattern = pattern_col->get_data_at(0);
347
8
                if (pattern.size == 0) {
348
2
                    return Status::OK();
349
2
                }
350
351
6
                std::string error_str;
352
6
                std::unique_ptr<re2::RE2> scoped_re;
353
6
                StringRef options_value;
354
                if constexpr (std::is_same_v<FourParamTypes, ParamTypes>) {
355
                    DCHECK_EQ(context->get_num_args(), 4);
356
                    DCHECK(context->is_col_constant(3));
357
                    const auto options_col = context->get_constant_col(3)->column_ptr;
358
                    options_value = options_col->get_data_at(0);
359
                }
360
361
6
                bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
362
6
                                                         options_value, scoped_re);
363
6
                if (!st) {
364
0
                    context->set_error(error_str.c_str());
365
0
                    return Status::InvalidArgument(error_str);
366
0
                }
367
6
                std::shared_ptr<re2::RE2> re(scoped_re.release());
368
6
                context->set_function_state(scope, re);
369
6
            }
370
8
        }
371
14
        return Status::OK();
372
16
    }
Unexecuted instantiation: _ZN5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
_ZN5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
Line
Count
Source
341
16
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
342
16
        if (scope == FunctionContext::THREAD_LOCAL) {
343
8
            if (context->is_col_constant(1)) {
344
8
                DCHECK(!context->get_function_state(scope));
345
8
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
346
8
                const auto& pattern = pattern_col->get_data_at(0);
347
8
                if (pattern.size == 0) {
348
2
                    return Status::OK();
349
2
                }
350
351
6
                std::string error_str;
352
6
                std::unique_ptr<re2::RE2> scoped_re;
353
6
                StringRef options_value;
354
                if constexpr (std::is_same_v<FourParamTypes, ParamTypes>) {
355
                    DCHECK_EQ(context->get_num_args(), 4);
356
                    DCHECK(context->is_col_constant(3));
357
                    const auto options_col = context->get_constant_col(3)->column_ptr;
358
                    options_value = options_col->get_data_at(0);
359
                }
360
361
6
                bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
362
6
                                                         options_value, scoped_re);
363
6
                if (!st) {
364
0
                    context->set_error(error_str.c_str());
365
0
                    return Status::InvalidArgument(error_str);
366
0
                }
367
6
                std::shared_ptr<re2::RE2> re(scoped_re.release());
368
6
                context->set_function_state(scope, re);
369
6
            }
370
8
        }
371
14
        return Status::OK();
372
16
    }
Unexecuted instantiation: _ZN5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
373
374
    Status execute_impl(FunctionContext* context, Block& block, const ColumnNumbers& arguments,
375
12
                        uint32_t result, size_t input_rows_count) const override {
376
12
        size_t argument_size = arguments.size();
377
378
12
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
379
12
        auto result_data_column = ColumnString::create();
380
12
        auto& result_data = result_data_column->get_chars();
381
12
        auto& result_offset = result_data_column->get_offsets();
382
12
        result_offset.resize(input_rows_count);
383
384
12
        bool col_const[3];
385
12
        ColumnPtr argument_columns[3];
386
48
        for (int i = 0; i < 3; ++i) {
387
36
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
388
36
        }
389
12
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
390
0
                                                     *block.get_by_position(arguments[0]).column)
391
0
                                                     .convert_to_full_column()
392
12
                                           : block.get_by_position(arguments[0]).column;
393
394
12
        default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block, arguments);
395
396
12
        StringRef options_value;
397
12
        if (col_const[1] && col_const[2]) {
398
0
            Impl::execute_impl_const_args(context, argument_columns, options_value,
399
0
                                          input_rows_count, result_data, result_offset,
400
0
                                          result_null_map->get_data());
401
12
        } else {
402
            // the options have check in FE, so is always const, and get idx of 0
403
12
            if (argument_size == 4) {
404
0
                options_value = block.get_by_position(arguments[3]).column->get_data_at(0);
405
0
            }
406
12
            Impl::execute_impl(context, argument_columns, options_value, input_rows_count,
407
12
                               result_data, result_offset, result_null_map->get_data());
408
12
        }
409
410
12
        block.get_by_position(result).column =
411
12
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
412
12
        return Status::OK();
413
12
    }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
375
6
                        uint32_t result, size_t input_rows_count) const override {
376
6
        size_t argument_size = arguments.size();
377
378
6
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
379
6
        auto result_data_column = ColumnString::create();
380
6
        auto& result_data = result_data_column->get_chars();
381
6
        auto& result_offset = result_data_column->get_offsets();
382
6
        result_offset.resize(input_rows_count);
383
384
6
        bool col_const[3];
385
6
        ColumnPtr argument_columns[3];
386
24
        for (int i = 0; i < 3; ++i) {
387
18
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
388
18
        }
389
6
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
390
0
                                                     *block.get_by_position(arguments[0]).column)
391
0
                                                     .convert_to_full_column()
392
6
                                           : block.get_by_position(arguments[0]).column;
393
394
6
        default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block, arguments);
395
396
6
        StringRef options_value;
397
6
        if (col_const[1] && col_const[2]) {
398
0
            Impl::execute_impl_const_args(context, argument_columns, options_value,
399
0
                                          input_rows_count, result_data, result_offset,
400
0
                                          result_null_map->get_data());
401
6
        } else {
402
            // the options have check in FE, so is always const, and get idx of 0
403
6
            if (argument_size == 4) {
404
0
                options_value = block.get_by_position(arguments[3]).column->get_data_at(0);
405
0
            }
406
6
            Impl::execute_impl(context, argument_columns, options_value, input_rows_count,
407
6
                               result_data, result_offset, result_null_map->get_data());
408
6
        }
409
410
6
        block.get_by_position(result).column =
411
6
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
412
6
        return Status::OK();
413
6
    }
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
375
6
                        uint32_t result, size_t input_rows_count) const override {
376
6
        size_t argument_size = arguments.size();
377
378
6
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
379
6
        auto result_data_column = ColumnString::create();
380
6
        auto& result_data = result_data_column->get_chars();
381
6
        auto& result_offset = result_data_column->get_offsets();
382
6
        result_offset.resize(input_rows_count);
383
384
6
        bool col_const[3];
385
6
        ColumnPtr argument_columns[3];
386
24
        for (int i = 0; i < 3; ++i) {
387
18
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
388
18
        }
389
6
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
390
0
                                                     *block.get_by_position(arguments[0]).column)
391
0
                                                     .convert_to_full_column()
392
6
                                           : block.get_by_position(arguments[0]).column;
393
394
6
        default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block, arguments);
395
396
6
        StringRef options_value;
397
6
        if (col_const[1] && col_const[2]) {
398
0
            Impl::execute_impl_const_args(context, argument_columns, options_value,
399
0
                                          input_rows_count, result_data, result_offset,
400
0
                                          result_null_map->get_data());
401
6
        } else {
402
            // the options have check in FE, so is always const, and get idx of 0
403
6
            if (argument_size == 4) {
404
0
                options_value = block.get_by_position(arguments[3]).column->get_data_at(0);
405
0
            }
406
6
            Impl::execute_impl(context, argument_columns, options_value, input_rows_count,
407
6
                               result_data, result_offset, result_null_map->get_data());
408
6
        }
409
410
6
        block.get_by_position(result).column =
411
6
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
412
6
        return Status::OK();
413
6
    }
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
414
};
415
416
struct RegexpReplaceImpl {
417
    static constexpr auto name = "regexp_replace";
418
    static void execute_impl(FunctionContext* context, ColumnPtr argument_columns[],
419
                             const StringRef& options_value, size_t input_rows_count,
420
                             ColumnString::Chars& result_data, ColumnString::Offsets& result_offset,
421
6
                             NullMap& null_map) {
422
6
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
423
6
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
424
6
        const auto* replace_col = check_and_get_column<ColumnString>(argument_columns[2].get());
425
426
12
        for (size_t i = 0; i < input_rows_count; ++i) {
427
6
            if (null_map[i]) {
428
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
429
0
                continue;
430
0
            }
431
6
            _execute_inner_loop<false>(context, str_col, pattern_col, replace_col, options_value,
432
6
                                       result_data, result_offset, null_map, i);
433
6
        }
434
6
    }
435
    static void execute_impl_const_args(FunctionContext* context, ColumnPtr argument_columns[],
436
                                        const StringRef& options_value, size_t input_rows_count,
437
                                        ColumnString::Chars& result_data,
438
0
                                        ColumnString::Offsets& result_offset, NullMap& null_map) {
439
0
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
440
0
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
441
0
        const auto* replace_col = check_and_get_column<ColumnString>(argument_columns[2].get());
442
443
0
        for (size_t i = 0; i < input_rows_count; ++i) {
444
0
            if (null_map[i]) {
445
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
446
0
                continue;
447
0
            }
448
0
            _execute_inner_loop<true>(context, str_col, pattern_col, replace_col, options_value,
449
0
                                      result_data, result_offset, null_map, i);
450
0
        }
451
0
    }
452
    template <bool Const>
453
    static void _execute_inner_loop(FunctionContext* context, const ColumnString* str_col,
454
                                    const ColumnString* pattern_col,
455
                                    const ColumnString* replace_col, const StringRef& options_value,
456
                                    ColumnString::Chars& result_data,
457
                                    ColumnString::Offsets& result_offset, NullMap& null_map,
458
6
                                    const size_t index_now) {
459
6
        re2::RE2* re = reinterpret_cast<re2::RE2*>(
460
6
                context->get_function_state(FunctionContext::THREAD_LOCAL));
461
6
        std::unique_ptr<re2::RE2> scoped_re; // destroys re if state->re is nullptr
462
6
        if (re == nullptr) {
463
2
            std::string error_str;
464
2
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
465
2
            bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
466
2
                                                     options_value, scoped_re);
467
2
            if (!st) {
468
0
                context->add_warning(error_str.c_str());
469
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
470
0
                return;
471
0
            }
472
2
            re = scoped_re.get();
473
2
        }
474
475
6
        re2::StringPiece replace_str = re2::StringPiece(
476
6
                replace_col->get_data_at(index_check_const(index_now, Const)).to_string_view());
477
478
6
        std::string result_str(str_col->get_data_at(index_now).to_string());
479
6
        re2::RE2::GlobalReplace(&result_str, *re, replace_str);
480
6
        StringOP::push_value_string(result_str, index_now, result_data, result_offset);
481
6
    }
Unexecuted instantiation: _ZN5doris17RegexpReplaceImpl19_execute_inner_loopILb1EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES7_S7_RKNS_9StringRefERNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSB_IjLm4096ESE_Lm16ELm15EEESG_m
_ZN5doris17RegexpReplaceImpl19_execute_inner_loopILb0EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES7_S7_RKNS_9StringRefERNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSB_IjLm4096ESE_Lm16ELm15EEESG_m
Line
Count
Source
458
6
                                    const size_t index_now) {
459
6
        re2::RE2* re = reinterpret_cast<re2::RE2*>(
460
6
                context->get_function_state(FunctionContext::THREAD_LOCAL));
461
6
        std::unique_ptr<re2::RE2> scoped_re; // destroys re if state->re is nullptr
462
6
        if (re == nullptr) {
463
2
            std::string error_str;
464
2
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
465
2
            bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
466
2
                                                     options_value, scoped_re);
467
2
            if (!st) {
468
0
                context->add_warning(error_str.c_str());
469
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
470
0
                return;
471
0
            }
472
2
            re = scoped_re.get();
473
2
        }
474
475
6
        re2::StringPiece replace_str = re2::StringPiece(
476
6
                replace_col->get_data_at(index_check_const(index_now, Const)).to_string_view());
477
478
6
        std::string result_str(str_col->get_data_at(index_now).to_string());
479
6
        re2::RE2::GlobalReplace(&result_str, *re, replace_str);
480
6
        StringOP::push_value_string(result_str, index_now, result_data, result_offset);
481
6
    }
482
};
483
484
struct RegexpReplaceOneImpl {
485
    static constexpr auto name = "regexp_replace_one";
486
487
    static void execute_impl(FunctionContext* context, ColumnPtr argument_columns[],
488
                             const StringRef& options_value, size_t input_rows_count,
489
                             ColumnString::Chars& result_data, ColumnString::Offsets& result_offset,
490
6
                             NullMap& null_map) {
491
6
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
492
6
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
493
6
        const auto* replace_col = check_and_get_column<ColumnString>(argument_columns[2].get());
494
        // 3 args
495
12
        for (size_t i = 0; i < input_rows_count; ++i) {
496
6
            if (null_map[i]) {
497
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
498
0
                continue;
499
0
            }
500
6
            _execute_inner_loop<false>(context, str_col, pattern_col, replace_col, options_value,
501
6
                                       result_data, result_offset, null_map, i);
502
6
        }
503
6
    }
504
505
    static void execute_impl_const_args(FunctionContext* context, ColumnPtr argument_columns[],
506
                                        const StringRef& options_value, size_t input_rows_count,
507
                                        ColumnString::Chars& result_data,
508
0
                                        ColumnString::Offsets& result_offset, NullMap& null_map) {
509
0
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
510
0
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
511
0
        const auto* replace_col = check_and_get_column<ColumnString>(argument_columns[2].get());
512
        // 3 args
513
0
        for (size_t i = 0; i < input_rows_count; ++i) {
514
0
            if (null_map[i]) {
515
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
516
0
                continue;
517
0
            }
518
0
            _execute_inner_loop<true>(context, str_col, pattern_col, replace_col, options_value,
519
0
                                      result_data, result_offset, null_map, i);
520
0
        }
521
0
    }
522
    template <bool Const>
523
    static void _execute_inner_loop(FunctionContext* context, const ColumnString* str_col,
524
                                    const ColumnString* pattern_col,
525
                                    const ColumnString* replace_col, const StringRef& options_value,
526
                                    ColumnString::Chars& result_data,
527
                                    ColumnString::Offsets& result_offset, NullMap& null_map,
528
6
                                    const size_t index_now) {
529
6
        re2::RE2* re = reinterpret_cast<re2::RE2*>(
530
6
                context->get_function_state(FunctionContext::THREAD_LOCAL));
531
6
        std::unique_ptr<re2::RE2> scoped_re; // destroys re if state->re is nullptr
532
6
        if (re == nullptr) {
533
2
            std::string error_str;
534
2
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
535
2
            bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
536
2
                                                     options_value, scoped_re);
537
2
            if (!st) {
538
0
                context->add_warning(error_str.c_str());
539
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
540
0
                return;
541
0
            }
542
2
            re = scoped_re.get();
543
2
        }
544
545
6
        re2::StringPiece replace_str = re2::StringPiece(
546
6
                replace_col->get_data_at(index_check_const(index_now, Const)).to_string_view());
547
548
6
        std::string result_str(str_col->get_data_at(index_now).to_string());
549
6
        re2::RE2::Replace(&result_str, *re, replace_str);
550
6
        StringOP::push_value_string(result_str, index_now, result_data, result_offset);
551
6
    }
Unexecuted instantiation: _ZN5doris20RegexpReplaceOneImpl19_execute_inner_loopILb1EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES7_S7_RKNS_9StringRefERNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSB_IjLm4096ESE_Lm16ELm15EEESG_m
_ZN5doris20RegexpReplaceOneImpl19_execute_inner_loopILb0EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES7_S7_RKNS_9StringRefERNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSB_IjLm4096ESE_Lm16ELm15EEESG_m
Line
Count
Source
528
6
                                    const size_t index_now) {
529
6
        re2::RE2* re = reinterpret_cast<re2::RE2*>(
530
6
                context->get_function_state(FunctionContext::THREAD_LOCAL));
531
6
        std::unique_ptr<re2::RE2> scoped_re; // destroys re if state->re is nullptr
532
6
        if (re == nullptr) {
533
2
            std::string error_str;
534
2
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
535
2
            bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
536
2
                                                     options_value, scoped_re);
537
2
            if (!st) {
538
0
                context->add_warning(error_str.c_str());
539
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
540
0
                return;
541
0
            }
542
2
            re = scoped_re.get();
543
2
        }
544
545
6
        re2::StringPiece replace_str = re2::StringPiece(
546
6
                replace_col->get_data_at(index_check_const(index_now, Const)).to_string_view());
547
548
6
        std::string result_str(str_col->get_data_at(index_now).to_string());
549
6
        re2::RE2::Replace(&result_str, *re, replace_str);
550
6
        StringOP::push_value_string(result_str, index_now, result_data, result_offset);
551
6
    }
552
};
553
554
template <bool ReturnNull>
555
struct RegexpExtractImpl {
556
    static constexpr auto name = ReturnNull ? "regexp_extract_or_null" : "regexp_extract";
557
    // 3 args
558
    static void execute_impl(FunctionContext* context, ColumnPtr argument_columns[],
559
                             size_t input_rows_count, ColumnString::Chars& result_data,
560
16
                             ColumnString::Offsets& result_offset, NullMap& null_map) {
561
16
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
562
16
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
563
16
        const auto* index_col = check_and_get_column<ColumnInt64>(argument_columns[2].get());
564
32
        for (size_t i = 0; i < input_rows_count; ++i) {
565
16
            if (null_map[i]) {
566
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
567
0
                continue;
568
0
            }
569
16
            const auto& index_data = index_col->get_int(i);
570
16
            if (index_data < 0) {
571
0
                ReturnNull ? StringOP::push_null_string(i, result_data, result_offset, null_map)
572
0
                           : StringOP::push_empty_string(i, result_data, result_offset);
573
0
                continue;
574
0
            }
575
16
            _execute_inner_loop<false>(context, str_col, pattern_col, index_data, result_data,
576
16
                                       result_offset, null_map, i);
577
16
        }
578
16
    }
_ZN5doris17RegexpExtractImplILb1EE12execute_implEPNS_15FunctionContextEPNS_3COWINS_7IColumnEE13immutable_ptrIS5_EEmRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSA_IjLm4096ESD_Lm16ELm15EEESF_
Line
Count
Source
560
8
                             ColumnString::Offsets& result_offset, NullMap& null_map) {
561
8
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
562
8
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
563
8
        const auto* index_col = check_and_get_column<ColumnInt64>(argument_columns[2].get());
564
16
        for (size_t i = 0; i < input_rows_count; ++i) {
565
8
            if (null_map[i]) {
566
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
567
0
                continue;
568
0
            }
569
8
            const auto& index_data = index_col->get_int(i);
570
8
            if (index_data < 0) {
571
0
                ReturnNull ? StringOP::push_null_string(i, result_data, result_offset, null_map)
572
0
                           : StringOP::push_empty_string(i, result_data, result_offset);
573
0
                continue;
574
0
            }
575
8
            _execute_inner_loop<false>(context, str_col, pattern_col, index_data, result_data,
576
8
                                       result_offset, null_map, i);
577
8
        }
578
8
    }
_ZN5doris17RegexpExtractImplILb0EE12execute_implEPNS_15FunctionContextEPNS_3COWINS_7IColumnEE13immutable_ptrIS5_EEmRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSA_IjLm4096ESD_Lm16ELm15EEESF_
Line
Count
Source
560
8
                             ColumnString::Offsets& result_offset, NullMap& null_map) {
561
8
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
562
8
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
563
8
        const auto* index_col = check_and_get_column<ColumnInt64>(argument_columns[2].get());
564
16
        for (size_t i = 0; i < input_rows_count; ++i) {
565
8
            if (null_map[i]) {
566
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
567
0
                continue;
568
0
            }
569
8
            const auto& index_data = index_col->get_int(i);
570
8
            if (index_data < 0) {
571
0
                ReturnNull ? StringOP::push_null_string(i, result_data, result_offset, null_map)
572
0
                           : StringOP::push_empty_string(i, result_data, result_offset);
573
0
                continue;
574
0
            }
575
8
            _execute_inner_loop<false>(context, str_col, pattern_col, index_data, result_data,
576
8
                                       result_offset, null_map, i);
577
8
        }
578
8
    }
579
580
    static void execute_impl_const_args(FunctionContext* context, ColumnPtr argument_columns[],
581
                                        size_t input_rows_count, ColumnString::Chars& result_data,
582
0
                                        ColumnString::Offsets& result_offset, NullMap& null_map) {
583
0
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
584
0
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
585
0
        const auto* index_col = check_and_get_column<ColumnInt64>(argument_columns[2].get());
586
587
0
        const auto& index_data = index_col->get_int(0);
588
0
        if (index_data < 0) {
589
0
            for (size_t i = 0; i < input_rows_count; ++i) {
590
0
                ReturnNull ? StringOP::push_null_string(i, result_data, result_offset, null_map)
591
0
                           : StringOP::push_empty_string(i, result_data, result_offset);
592
0
            }
593
0
            return;
594
0
        }
595
596
0
        for (size_t i = 0; i < input_rows_count; ++i) {
597
0
            if (null_map[i]) {
598
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
599
0
                continue;
600
0
            }
601
602
0
            _execute_inner_loop<true>(context, str_col, pattern_col, index_data, result_data,
603
0
                                      result_offset, null_map, i);
604
0
        }
605
0
    }
Unexecuted instantiation: _ZN5doris17RegexpExtractImplILb1EE23execute_impl_const_argsEPNS_15FunctionContextEPNS_3COWINS_7IColumnEE13immutable_ptrIS5_EEmRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSA_IjLm4096ESD_Lm16ELm15EEESF_
Unexecuted instantiation: _ZN5doris17RegexpExtractImplILb0EE23execute_impl_const_argsEPNS_15FunctionContextEPNS_3COWINS_7IColumnEE13immutable_ptrIS5_EEmRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSA_IjLm4096ESD_Lm16ELm15EEESF_
606
    template <bool Const>
607
    static void _execute_inner_loop(FunctionContext* context, const ColumnString* str_col,
608
                                    const ColumnString* pattern_col, const Int64 index_data,
609
                                    ColumnString::Chars& result_data,
610
                                    ColumnString::Offsets& result_offset, NullMap& null_map,
611
16
                                    const size_t index_now) {
612
16
        auto* engine = reinterpret_cast<RegexpExtractEngine*>(
613
16
                context->get_function_state(FunctionContext::THREAD_LOCAL));
614
16
        std::unique_ptr<RegexpExtractEngine> scoped_engine;
615
616
16
        if (engine == nullptr) {
617
0
            std::string error_str;
618
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
619
0
            scoped_engine = std::make_unique<RegexpExtractEngine>();
620
0
            bool st = RegexpExtractEngine::compile(pattern, &error_str, *scoped_engine,
621
0
                                                   context->state()->enable_extended_regex());
622
0
            if (!st) {
623
0
                context->add_warning(error_str.c_str());
624
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
625
0
                return;
626
0
            }
627
0
            engine = scoped_engine.get();
628
0
        }
629
630
16
        const auto& str = str_col->get_data_at(index_now);
631
632
16
        int max_matches = 1 + engine->number_of_capturing_groups();
633
16
        if (index_data >= max_matches) {
634
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
635
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
636
0
            return;
637
0
        }
638
639
16
        std::string match_result;
640
16
        bool success = engine->match_and_extract(str.data, str.size, static_cast<int>(index_data),
641
16
                                                 match_result);
642
643
16
        if (!success) {
644
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
645
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
646
0
            return;
647
0
        }
648
649
16
        StringOP::push_value_string(std::string_view(match_result.data(), match_result.size()),
650
16
                                    index_now, result_data, result_offset);
651
16
    }
Unexecuted instantiation: _ZN5doris17RegexpExtractImplILb1EE19_execute_inner_loopILb1EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES8_lRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS9_IjLm4096ESC_Lm16ELm15EEESE_m
_ZN5doris17RegexpExtractImplILb1EE19_execute_inner_loopILb0EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES8_lRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS9_IjLm4096ESC_Lm16ELm15EEESE_m
Line
Count
Source
611
8
                                    const size_t index_now) {
612
8
        auto* engine = reinterpret_cast<RegexpExtractEngine*>(
613
8
                context->get_function_state(FunctionContext::THREAD_LOCAL));
614
8
        std::unique_ptr<RegexpExtractEngine> scoped_engine;
615
616
8
        if (engine == nullptr) {
617
0
            std::string error_str;
618
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
619
0
            scoped_engine = std::make_unique<RegexpExtractEngine>();
620
0
            bool st = RegexpExtractEngine::compile(pattern, &error_str, *scoped_engine,
621
0
                                                   context->state()->enable_extended_regex());
622
0
            if (!st) {
623
0
                context->add_warning(error_str.c_str());
624
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
625
0
                return;
626
0
            }
627
0
            engine = scoped_engine.get();
628
0
        }
629
630
8
        const auto& str = str_col->get_data_at(index_now);
631
632
8
        int max_matches = 1 + engine->number_of_capturing_groups();
633
8
        if (index_data >= max_matches) {
634
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
635
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
636
0
            return;
637
0
        }
638
639
8
        std::string match_result;
640
8
        bool success = engine->match_and_extract(str.data, str.size, static_cast<int>(index_data),
641
8
                                                 match_result);
642
643
8
        if (!success) {
644
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
645
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
646
0
            return;
647
0
        }
648
649
8
        StringOP::push_value_string(std::string_view(match_result.data(), match_result.size()),
650
8
                                    index_now, result_data, result_offset);
651
8
    }
Unexecuted instantiation: _ZN5doris17RegexpExtractImplILb0EE19_execute_inner_loopILb1EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES8_lRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS9_IjLm4096ESC_Lm16ELm15EEESE_m
_ZN5doris17RegexpExtractImplILb0EE19_execute_inner_loopILb0EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES8_lRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS9_IjLm4096ESC_Lm16ELm15EEESE_m
Line
Count
Source
611
8
                                    const size_t index_now) {
612
8
        auto* engine = reinterpret_cast<RegexpExtractEngine*>(
613
8
                context->get_function_state(FunctionContext::THREAD_LOCAL));
614
8
        std::unique_ptr<RegexpExtractEngine> scoped_engine;
615
616
8
        if (engine == nullptr) {
617
0
            std::string error_str;
618
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
619
0
            scoped_engine = std::make_unique<RegexpExtractEngine>();
620
0
            bool st = RegexpExtractEngine::compile(pattern, &error_str, *scoped_engine,
621
0
                                                   context->state()->enable_extended_regex());
622
0
            if (!st) {
623
0
                context->add_warning(error_str.c_str());
624
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
625
0
                return;
626
0
            }
627
0
            engine = scoped_engine.get();
628
0
        }
629
630
8
        const auto& str = str_col->get_data_at(index_now);
631
632
8
        int max_matches = 1 + engine->number_of_capturing_groups();
633
8
        if (index_data >= max_matches) {
634
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
635
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
636
0
            return;
637
0
        }
638
639
8
        std::string match_result;
640
8
        bool success = engine->match_and_extract(str.data, str.size, static_cast<int>(index_data),
641
8
                                                 match_result);
642
643
8
        if (!success) {
644
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
645
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
646
0
            return;
647
0
        }
648
649
8
        StringOP::push_value_string(std::string_view(match_result.data(), match_result.size()),
650
8
                                    index_now, result_data, result_offset);
651
8
    }
652
};
653
654
struct RegexpExtractAllImpl {
655
    static constexpr auto name = "regexp_extract_all";
656
657
0
    size_t get_number_of_arguments() const { return 2; }
658
659
    static void execute_impl(FunctionContext* context, ColumnPtr argument_columns[],
660
                             size_t input_rows_count, ColumnString::Chars& result_data,
661
0
                             ColumnString::Offsets& result_offset, NullMap& null_map) {
662
0
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
663
0
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
664
0
        for (int i = 0; i < input_rows_count; ++i) {
665
0
            if (null_map[i]) {
666
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
667
0
                continue;
668
0
            }
669
0
            _execute_inner_loop<false>(context, str_col, pattern_col, result_data, result_offset,
670
0
                                       null_map, i);
671
0
        }
672
0
    }
673
674
    static void execute_impl_const_args(FunctionContext* context, ColumnPtr argument_columns[],
675
                                        size_t input_rows_count, ColumnString::Chars& result_data,
676
7
                                        ColumnString::Offsets& result_offset, NullMap& null_map) {
677
7
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
678
7
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
679
14
        for (int i = 0; i < input_rows_count; ++i) {
680
7
            if (null_map[i]) {
681
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
682
0
                continue;
683
0
            }
684
7
            _execute_inner_loop<true>(context, str_col, pattern_col, result_data, result_offset,
685
7
                                      null_map, i);
686
7
        }
687
7
    }
688
    template <bool Const>
689
    static void _execute_inner_loop(FunctionContext* context, const ColumnString* str_col,
690
                                    const ColumnString* pattern_col,
691
                                    ColumnString::Chars& result_data,
692
                                    ColumnString::Offsets& result_offset, NullMap& null_map,
693
7
                                    const size_t index_now) {
694
7
        auto* engine = reinterpret_cast<RegexpExtractEngine*>(
695
7
                context->get_function_state(FunctionContext::THREAD_LOCAL));
696
7
        std::unique_ptr<RegexpExtractEngine> scoped_engine;
697
698
7
        if (engine == nullptr) {
699
0
            std::string error_str;
700
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
701
0
            scoped_engine = std::make_unique<RegexpExtractEngine>();
702
0
            bool st = RegexpExtractEngine::compile(pattern, &error_str, *scoped_engine,
703
0
                                                   context->state()->enable_extended_regex());
704
0
            if (!st) {
705
0
                context->add_warning(error_str.c_str());
706
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
707
0
                return;
708
0
            }
709
0
            engine = scoped_engine.get();
710
0
        }
711
712
7
        if (engine->number_of_capturing_groups() == 0) {
713
0
            StringOP::push_empty_string(index_now, result_data, result_offset);
714
0
            return;
715
0
        }
716
7
        const auto& str = str_col->get_data_at(index_now);
717
7
        std::vector<std::string> res_matches;
718
7
        engine->match_all_and_extract(str.data, str.size, res_matches);
719
720
7
        if (res_matches.empty()) {
721
0
            StringOP::push_empty_string(index_now, result_data, result_offset);
722
0
            return;
723
0
        }
724
725
7
        std::string res = "[";
726
19
        for (int j = 0; j < res_matches.size(); ++j) {
727
12
            res += "'" + res_matches[j] + "'";
728
12
            if (j < res_matches.size() - 1) {
729
5
                res += ",";
730
5
            }
731
12
        }
732
7
        res += "]";
733
7
        StringOP::push_value_string(std::string_view(res), index_now, result_data, result_offset);
734
7
    }
_ZN5doris20RegexpExtractAllImpl19_execute_inner_loopILb1EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES7_RNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEESD_m
Line
Count
Source
693
7
                                    const size_t index_now) {
694
7
        auto* engine = reinterpret_cast<RegexpExtractEngine*>(
695
7
                context->get_function_state(FunctionContext::THREAD_LOCAL));
696
7
        std::unique_ptr<RegexpExtractEngine> scoped_engine;
697
698
7
        if (engine == nullptr) {
699
0
            std::string error_str;
700
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
701
0
            scoped_engine = std::make_unique<RegexpExtractEngine>();
702
0
            bool st = RegexpExtractEngine::compile(pattern, &error_str, *scoped_engine,
703
0
                                                   context->state()->enable_extended_regex());
704
0
            if (!st) {
705
0
                context->add_warning(error_str.c_str());
706
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
707
0
                return;
708
0
            }
709
0
            engine = scoped_engine.get();
710
0
        }
711
712
7
        if (engine->number_of_capturing_groups() == 0) {
713
0
            StringOP::push_empty_string(index_now, result_data, result_offset);
714
0
            return;
715
0
        }
716
7
        const auto& str = str_col->get_data_at(index_now);
717
7
        std::vector<std::string> res_matches;
718
7
        engine->match_all_and_extract(str.data, str.size, res_matches);
719
720
7
        if (res_matches.empty()) {
721
0
            StringOP::push_empty_string(index_now, result_data, result_offset);
722
0
            return;
723
0
        }
724
725
7
        std::string res = "[";
726
19
        for (int j = 0; j < res_matches.size(); ++j) {
727
12
            res += "'" + res_matches[j] + "'";
728
12
            if (j < res_matches.size() - 1) {
729
5
                res += ",";
730
5
            }
731
12
        }
732
7
        res += "]";
733
7
        StringOP::push_value_string(std::string_view(res), index_now, result_data, result_offset);
734
7
    }
Unexecuted instantiation: _ZN5doris20RegexpExtractAllImpl19_execute_inner_loopILb0EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES7_RNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEESD_m
735
};
736
737
// template FunctionRegexpFunctionality is used for regexp_xxxx series functions, not for regexp match.
738
template <typename Impl>
739
class FunctionRegexpFunctionality : public IFunction {
740
public:
741
    static constexpr auto name = Impl::name;
742
743
35
    static FunctionPtr create() { return std::make_shared<FunctionRegexpFunctionality>(); }
_ZN5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE6createEv
Line
Count
Source
743
12
    static FunctionPtr create() { return std::make_shared<FunctionRegexpFunctionality>(); }
_ZN5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE6createEv
Line
Count
Source
743
12
    static FunctionPtr create() { return std::make_shared<FunctionRegexpFunctionality>(); }
_ZN5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE6createEv
Line
Count
Source
743
11
    static FunctionPtr create() { return std::make_shared<FunctionRegexpFunctionality>(); }
744
745
3
    String get_name() const override { return name; }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE8get_nameB5cxx11Ev
Line
Count
Source
745
1
    String get_name() const override { return name; }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE8get_nameB5cxx11Ev
Line
Count
Source
745
1
    String get_name() const override { return name; }
_ZNK5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE8get_nameB5cxx11Ev
Line
Count
Source
745
1
    String get_name() const override { return name; }
746
747
29
    size_t get_number_of_arguments() const override {
748
29
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
749
9
            return 2;
750
9
        }
751
0
        return 3;
752
29
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE23get_number_of_argumentsEv
Line
Count
Source
747
10
    size_t get_number_of_arguments() const override {
748
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
749
            return 2;
750
        }
751
10
        return 3;
752
10
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE23get_number_of_argumentsEv
Line
Count
Source
747
10
    size_t get_number_of_arguments() const override {
748
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
749
            return 2;
750
        }
751
10
        return 3;
752
10
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE23get_number_of_argumentsEv
Line
Count
Source
747
9
    size_t get_number_of_arguments() const override {
748
9
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
749
9
            return 2;
750
9
        }
751
0
        return 3;
752
9
    }
753
754
29
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
755
29
        return make_nullable(std::make_shared<DataTypeString>());
756
29
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
Line
Count
Source
754
10
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
755
10
        return make_nullable(std::make_shared<DataTypeString>());
756
10
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
Line
Count
Source
754
10
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
755
10
        return make_nullable(std::make_shared<DataTypeString>());
756
10
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS7_EE
Line
Count
Source
754
9
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
755
9
        return make_nullable(std::make_shared<DataTypeString>());
756
9
    }
757
758
58
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
759
58
        if (scope == FunctionContext::THREAD_LOCAL) {
760
29
            if (context->is_col_constant(1)) {
761
29
                DCHECK(!context->get_function_state(scope));
762
29
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
763
29
                const auto& pattern = pattern_col->get_data_at(0);
764
29
                if (pattern.size == 0) {
765
3
                    return Status::OK();
766
3
                }
767
768
26
                std::string error_str;
769
26
                auto engine = std::make_shared<RegexpExtractEngine>();
770
26
                bool st = RegexpExtractEngine::compile(pattern, &error_str, *engine,
771
26
                                                       context->state()->enable_extended_regex());
772
26
                if (!st) {
773
0
                    context->set_error(error_str.c_str());
774
0
                    return Status::InvalidArgument(error_str);
775
0
                }
776
26
                context->set_function_state(scope, engine);
777
26
            }
778
29
        }
779
55
        return Status::OK();
780
58
    }
_ZN5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
Line
Count
Source
758
20
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
759
20
        if (scope == FunctionContext::THREAD_LOCAL) {
760
10
            if (context->is_col_constant(1)) {
761
10
                DCHECK(!context->get_function_state(scope));
762
10
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
763
10
                const auto& pattern = pattern_col->get_data_at(0);
764
10
                if (pattern.size == 0) {
765
1
                    return Status::OK();
766
1
                }
767
768
9
                std::string error_str;
769
9
                auto engine = std::make_shared<RegexpExtractEngine>();
770
9
                bool st = RegexpExtractEngine::compile(pattern, &error_str, *engine,
771
9
                                                       context->state()->enable_extended_regex());
772
9
                if (!st) {
773
0
                    context->set_error(error_str.c_str());
774
0
                    return Status::InvalidArgument(error_str);
775
0
                }
776
9
                context->set_function_state(scope, engine);
777
9
            }
778
10
        }
779
19
        return Status::OK();
780
20
    }
_ZN5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
Line
Count
Source
758
20
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
759
20
        if (scope == FunctionContext::THREAD_LOCAL) {
760
10
            if (context->is_col_constant(1)) {
761
10
                DCHECK(!context->get_function_state(scope));
762
10
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
763
10
                const auto& pattern = pattern_col->get_data_at(0);
764
10
                if (pattern.size == 0) {
765
1
                    return Status::OK();
766
1
                }
767
768
9
                std::string error_str;
769
9
                auto engine = std::make_shared<RegexpExtractEngine>();
770
9
                bool st = RegexpExtractEngine::compile(pattern, &error_str, *engine,
771
9
                                                       context->state()->enable_extended_regex());
772
9
                if (!st) {
773
0
                    context->set_error(error_str.c_str());
774
0
                    return Status::InvalidArgument(error_str);
775
0
                }
776
9
                context->set_function_state(scope, engine);
777
9
            }
778
10
        }
779
19
        return Status::OK();
780
20
    }
_ZN5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE4openEPNS_15FunctionContextENS3_18FunctionStateScopeE
Line
Count
Source
758
18
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
759
18
        if (scope == FunctionContext::THREAD_LOCAL) {
760
9
            if (context->is_col_constant(1)) {
761
9
                DCHECK(!context->get_function_state(scope));
762
9
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
763
9
                const auto& pattern = pattern_col->get_data_at(0);
764
9
                if (pattern.size == 0) {
765
1
                    return Status::OK();
766
1
                }
767
768
8
                std::string error_str;
769
8
                auto engine = std::make_shared<RegexpExtractEngine>();
770
8
                bool st = RegexpExtractEngine::compile(pattern, &error_str, *engine,
771
8
                                                       context->state()->enable_extended_regex());
772
8
                if (!st) {
773
0
                    context->set_error(error_str.c_str());
774
0
                    return Status::InvalidArgument(error_str);
775
0
                }
776
8
                context->set_function_state(scope, engine);
777
8
            }
778
9
        }
779
17
        return Status::OK();
780
18
    }
781
782
    Status execute_impl(FunctionContext* context, Block& block, const ColumnNumbers& arguments,
783
23
                        uint32_t result, size_t input_rows_count) const override {
784
23
        size_t argument_size = arguments.size();
785
786
23
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
787
23
        auto result_data_column = ColumnString::create();
788
23
        auto& result_data = result_data_column->get_chars();
789
23
        auto& result_offset = result_data_column->get_offsets();
790
23
        result_offset.resize(input_rows_count);
791
792
23
        bool col_const[3];
793
23
        ColumnPtr argument_columns[3];
794
85
        for (int i = 0; i < argument_size; ++i) {
795
62
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
796
62
        }
797
23
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
798
0
                                                     *block.get_by_position(arguments[0]).column)
799
0
                                                     .convert_to_full_column()
800
23
                                           : block.get_by_position(arguments[0]).column;
801
23
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
802
7
            default_preprocess_parameter_columns(argument_columns, col_const, {1}, block,
803
7
                                                 arguments);
804
16
        } else {
805
16
            default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block,
806
16
                                                 arguments);
807
16
        }
808
809
23
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
810
7
            if (col_const[1]) {
811
7
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
812
7
                                              result_data, result_offset,
813
7
                                              result_null_map->get_data());
814
7
            } else {
815
0
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
816
0
                                   result_offset, result_null_map->get_data());
817
0
            }
818
16
        } else {
819
16
            if (col_const[1] && col_const[2]) {
820
0
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
821
0
                                              result_data, result_offset,
822
0
                                              result_null_map->get_data());
823
16
            } else {
824
16
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
825
16
                                   result_offset, result_null_map->get_data());
826
16
            }
827
16
        }
828
829
23
        block.get_by_position(result).column =
830
23
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
831
23
        return Status::OK();
832
23
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
783
8
                        uint32_t result, size_t input_rows_count) const override {
784
8
        size_t argument_size = arguments.size();
785
786
8
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
787
8
        auto result_data_column = ColumnString::create();
788
8
        auto& result_data = result_data_column->get_chars();
789
8
        auto& result_offset = result_data_column->get_offsets();
790
8
        result_offset.resize(input_rows_count);
791
792
8
        bool col_const[3];
793
8
        ColumnPtr argument_columns[3];
794
32
        for (int i = 0; i < argument_size; ++i) {
795
24
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
796
24
        }
797
8
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
798
0
                                                     *block.get_by_position(arguments[0]).column)
799
0
                                                     .convert_to_full_column()
800
8
                                           : block.get_by_position(arguments[0]).column;
801
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
802
            default_preprocess_parameter_columns(argument_columns, col_const, {1}, block,
803
                                                 arguments);
804
8
        } else {
805
8
            default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block,
806
8
                                                 arguments);
807
8
        }
808
809
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
810
            if (col_const[1]) {
811
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
812
                                              result_data, result_offset,
813
                                              result_null_map->get_data());
814
            } else {
815
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
816
                                   result_offset, result_null_map->get_data());
817
            }
818
8
        } else {
819
8
            if (col_const[1] && col_const[2]) {
820
0
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
821
0
                                              result_data, result_offset,
822
0
                                              result_null_map->get_data());
823
8
            } else {
824
8
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
825
8
                                   result_offset, result_null_map->get_data());
826
8
            }
827
8
        }
828
829
8
        block.get_by_position(result).column =
830
8
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
831
8
        return Status::OK();
832
8
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
783
8
                        uint32_t result, size_t input_rows_count) const override {
784
8
        size_t argument_size = arguments.size();
785
786
8
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
787
8
        auto result_data_column = ColumnString::create();
788
8
        auto& result_data = result_data_column->get_chars();
789
8
        auto& result_offset = result_data_column->get_offsets();
790
8
        result_offset.resize(input_rows_count);
791
792
8
        bool col_const[3];
793
8
        ColumnPtr argument_columns[3];
794
32
        for (int i = 0; i < argument_size; ++i) {
795
24
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
796
24
        }
797
8
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
798
0
                                                     *block.get_by_position(arguments[0]).column)
799
0
                                                     .convert_to_full_column()
800
8
                                           : block.get_by_position(arguments[0]).column;
801
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
802
            default_preprocess_parameter_columns(argument_columns, col_const, {1}, block,
803
                                                 arguments);
804
8
        } else {
805
8
            default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block,
806
8
                                                 arguments);
807
8
        }
808
809
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
810
            if (col_const[1]) {
811
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
812
                                              result_data, result_offset,
813
                                              result_null_map->get_data());
814
            } else {
815
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
816
                                   result_offset, result_null_map->get_data());
817
            }
818
8
        } else {
819
8
            if (col_const[1] && col_const[2]) {
820
0
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
821
0
                                              result_data, result_offset,
822
0
                                              result_null_map->get_data());
823
8
            } else {
824
8
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
825
8
                                   result_offset, result_null_map->get_data());
826
8
            }
827
8
        }
828
829
8
        block.get_by_position(result).column =
830
8
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
831
8
        return Status::OK();
832
8
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
783
7
                        uint32_t result, size_t input_rows_count) const override {
784
7
        size_t argument_size = arguments.size();
785
786
7
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
787
7
        auto result_data_column = ColumnString::create();
788
7
        auto& result_data = result_data_column->get_chars();
789
7
        auto& result_offset = result_data_column->get_offsets();
790
7
        result_offset.resize(input_rows_count);
791
792
7
        bool col_const[3];
793
7
        ColumnPtr argument_columns[3];
794
21
        for (int i = 0; i < argument_size; ++i) {
795
14
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
796
14
        }
797
7
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
798
0
                                                     *block.get_by_position(arguments[0]).column)
799
0
                                                     .convert_to_full_column()
800
7
                                           : block.get_by_position(arguments[0]).column;
801
7
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
802
7
            default_preprocess_parameter_columns(argument_columns, col_const, {1}, block,
803
7
                                                 arguments);
804
        } else {
805
            default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block,
806
                                                 arguments);
807
        }
808
809
7
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
810
7
            if (col_const[1]) {
811
7
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
812
7
                                              result_data, result_offset,
813
7
                                              result_null_map->get_data());
814
7
            } else {
815
0
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
816
0
                                   result_offset, result_null_map->get_data());
817
0
            }
818
        } else {
819
            if (col_const[1] && col_const[2]) {
820
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
821
                                              result_data, result_offset,
822
                                              result_null_map->get_data());
823
            } else {
824
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
825
                                   result_offset, result_null_map->get_data());
826
            }
827
        }
828
829
7
        block.get_by_position(result).column =
830
7
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
831
7
        return Status::OK();
832
7
    }
833
};
834
835
1
void register_function_regexp_extract(SimpleFunctionFactory& factory) {
836
1
    factory.register_function<FunctionRegexpReplace<RegexpReplaceImpl, ThreeParamTypes>>();
837
1
    factory.register_function<FunctionRegexpReplace<RegexpReplaceImpl, FourParamTypes>>();
838
1
    factory.register_function<FunctionRegexpReplace<RegexpReplaceOneImpl, ThreeParamTypes>>();
839
1
    factory.register_function<FunctionRegexpReplace<RegexpReplaceOneImpl, FourParamTypes>>();
840
1
    factory.register_function<FunctionRegexpFunctionality<RegexpExtractImpl<true>>>();
841
1
    factory.register_function<FunctionRegexpFunctionality<RegexpExtractImpl<false>>>();
842
1
    factory.register_function<FunctionRegexpFunctionality<RegexpExtractAllImpl>>();
843
1
    factory.register_function<FunctionRegexpCount>();
844
1
}
845
846
} // namespace doris