Coverage Report

Created: 2026-03-16 21:05

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_nullable.h"
38
#include "core/column/column_string.h"
39
#include "core/column/column_vector.h"
40
#include "core/data_type/data_type.h"
41
#include "core/data_type/data_type_nullable.h"
42
#include "core/data_type/data_type_number.h"
43
#include "core/data_type/data_type_string.h"
44
#include "core/string_ref.h"
45
#include "core/types.h"
46
#include "exec/common/stringop_substring.h"
47
#include "exprs/aggregate/aggregate_function.h"
48
#include "exprs/function/function.h"
49
#include "exprs/function/simple_function_factory.h"
50
#include "exprs/function_context.h"
51
#include "exprs/string_functions.h"
52
53
namespace doris {
54
#include "common/compile_check_begin.h"
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
    static void execute_impl(FunctionContext* context, ColumnPtr argument_columns[],
193
0
                             size_t input_rows_count, ColumnInt32::Container& result_data) {
194
0
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
195
0
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
196
0
        for (int i = 0; i < input_rows_count; ++i) {
197
0
            result_data[i] = _execute_inner_loop(context, str_col, pattern_col, i);
198
0
        }
199
0
    }
200
    static int _execute_inner_loop(FunctionContext* context, const ColumnString* str_col,
201
0
                                   const ColumnString* pattern_col, const size_t index_now) {
202
0
        re2::RE2* re = reinterpret_cast<re2::RE2*>(
203
0
                context->get_function_state(FunctionContext::THREAD_LOCAL));
204
0
        std::unique_ptr<re2::RE2> scoped_re;
205
0
        if (re == nullptr) {
206
0
            std::string error_str;
207
0
            DCHECK(pattern_col);
208
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, false));
209
0
            bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(), StringRef(),
210
0
                                                     scoped_re);
211
0
            if (!st) {
212
0
                context->add_warning(error_str.c_str());
213
0
                throw Exception(Status::InvalidArgument(error_str));
214
0
                return 0;
215
0
            }
216
0
            re = scoped_re.get();
217
0
        }
218
219
0
        const auto& str = str_col->get_data_at(index_now);
220
0
        int count = 0;
221
0
        size_t pos = 0;
222
0
        while (pos < str.size) {
223
0
            auto str_pos = str.data + pos;
224
0
            auto str_size = str.size - pos;
225
0
            re2::StringPiece str_sp_current = re2::StringPiece(str_pos, str_size);
226
0
            re2::StringPiece match;
227
228
0
            bool success = re->Match(str_sp_current, 0, str_size, re2::RE2::UNANCHORED, &match, 1);
229
0
            if (!success) {
230
0
                break;
231
0
            }
232
0
            if (match.empty()) {
233
0
                pos += 1;
234
0
                continue;
235
0
            }
236
0
            count++;
237
0
            size_t match_start = match.data() - str_sp_current.data();
238
0
            pos += match_start + match.size();
239
0
        }
240
241
0
        return count;
242
0
    }
243
};
244
245
class FunctionRegexpCount : public IFunction {
246
public:
247
    static constexpr auto name = "regexp_count";
248
249
2
    static FunctionPtr create() { return std::make_shared<FunctionRegexpCount>(); }
250
251
1
    String get_name() const override { return name; }
252
253
0
    size_t get_number_of_arguments() const override { return 2; }
254
255
0
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
256
0
        return std::make_shared<DataTypeInt32>();
257
0
    }
258
259
0
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
260
0
        if (scope == FunctionContext::THREAD_LOCAL) {
261
0
            if (context->is_col_constant(1)) {
262
0
                DCHECK(!context->get_function_state(scope));
263
0
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
264
0
                const auto& pattern = pattern_col->get_data_at(0);
265
0
                if (pattern.size == 0) {
266
0
                    return Status::OK();
267
0
                }
268
269
0
                std::string error_str;
270
0
                std::unique_ptr<re2::RE2> scoped_re;
271
0
                bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
272
0
                                                         StringRef(), scoped_re);
273
0
                if (!st) {
274
0
                    context->set_error(error_str.c_str());
275
0
                    return Status::InvalidArgument(error_str);
276
0
                }
277
0
                std::shared_ptr<re2::RE2> re(scoped_re.release());
278
0
                context->set_function_state(scope, re);
279
0
            }
280
0
        }
281
0
        return Status::OK();
282
0
    }
283
284
    Status execute_impl(FunctionContext* context, Block& block, const ColumnNumbers& arguments,
285
0
                        uint32_t result, size_t input_rows_count) const override {
286
0
        auto result_data_column = ColumnInt32::create(input_rows_count);
287
0
        auto& result_data = result_data_column->get_data();
288
289
0
        ColumnPtr argument_columns[2];
290
291
0
        argument_columns[0] = block.get_by_position(arguments[0]).column;
292
0
        argument_columns[1] = block.get_by_position(arguments[1]).column;
293
0
        RegexpCountImpl::execute_impl(context, argument_columns, input_rows_count, result_data);
294
295
0
        block.get_by_position(result).column = std::move(result_data_column);
296
0
        return Status::OK();
297
0
    }
298
};
299
300
struct ThreeParamTypes {
301
2
    static DataTypes get_variadic_argument_types() {
302
2
        return {std::make_shared<DataTypeString>(), std::make_shared<DataTypeString>(),
303
2
                std::make_shared<DataTypeString>()};
304
2
    }
305
};
306
307
struct FourParamTypes {
308
2
    static DataTypes get_variadic_argument_types() {
309
2
        return {std::make_shared<DataTypeString>(), std::make_shared<DataTypeString>(),
310
2
                std::make_shared<DataTypeString>(), std::make_shared<DataTypeString>()};
311
2
    }
312
};
313
314
// template FunctionRegexpFunctionality is used for regexp_replace/regexp_replace_one
315
template <typename Impl, typename ParamTypes>
316
class FunctionRegexpReplace : public IFunction {
317
public:
318
    static constexpr auto name = Impl::name;
319
320
24
    static FunctionPtr create() { return std::make_shared<FunctionRegexpReplace>(); }
_ZN5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE6createEv
Line
Count
Source
320
10
    static FunctionPtr create() { return std::make_shared<FunctionRegexpReplace>(); }
_ZN5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE6createEv
Line
Count
Source
320
2
    static FunctionPtr create() { return std::make_shared<FunctionRegexpReplace>(); }
_ZN5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE6createEv
Line
Count
Source
320
10
    static FunctionPtr create() { return std::make_shared<FunctionRegexpReplace>(); }
_ZN5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE6createEv
Line
Count
Source
320
2
    static FunctionPtr create() { return std::make_shared<FunctionRegexpReplace>(); }
321
322
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
323
324
0
    size_t get_number_of_arguments() const override {
325
0
        return get_variadic_argument_types_impl().size();
326
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
327
328
20
    bool is_variadic() const override { return true; }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE11is_variadicEv
Line
Count
Source
328
9
    bool is_variadic() const override { return true; }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE11is_variadicEv
Line
Count
Source
328
1
    bool is_variadic() const override { return true; }
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE11is_variadicEv
Line
Count
Source
328
9
    bool is_variadic() const override { return true; }
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE11is_variadicEv
Line
Count
Source
328
1
    bool is_variadic() const override { return true; }
329
330
16
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
331
16
        return make_nullable(std::make_shared<DataTypeString>());
332
16
    }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
Line
Count
Source
330
8
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
331
8
        return make_nullable(std::make_shared<DataTypeString>());
332
8
    }
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
Line
Count
Source
330
8
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
331
8
        return make_nullable(std::make_shared<DataTypeString>());
332
8
    }
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
333
334
4
    DataTypes get_variadic_argument_types_impl() const override {
335
4
        return ParamTypes::get_variadic_argument_types();
336
4
    }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE32get_variadic_argument_types_implEv
Line
Count
Source
334
1
    DataTypes get_variadic_argument_types_impl() const override {
335
1
        return ParamTypes::get_variadic_argument_types();
336
1
    }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE32get_variadic_argument_types_implEv
Line
Count
Source
334
1
    DataTypes get_variadic_argument_types_impl() const override {
335
1
        return ParamTypes::get_variadic_argument_types();
336
1
    }
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE32get_variadic_argument_types_implEv
Line
Count
Source
334
1
    DataTypes get_variadic_argument_types_impl() const override {
335
1
        return ParamTypes::get_variadic_argument_types();
336
1
    }
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE32get_variadic_argument_types_implEv
Line
Count
Source
334
1
    DataTypes get_variadic_argument_types_impl() const override {
335
1
        return ParamTypes::get_variadic_argument_types();
336
1
    }
337
338
32
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
339
32
        if (scope == FunctionContext::THREAD_LOCAL) {
340
16
            if (context->is_col_constant(1)) {
341
16
                DCHECK(!context->get_function_state(scope));
342
16
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
343
16
                const auto& pattern = pattern_col->get_data_at(0);
344
16
                if (pattern.size == 0) {
345
4
                    return Status::OK();
346
4
                }
347
348
12
                std::string error_str;
349
12
                std::unique_ptr<re2::RE2> scoped_re;
350
12
                StringRef options_value;
351
12
                if constexpr (std::is_same_v<FourParamTypes, ParamTypes>) {
352
0
                    DCHECK_EQ(context->get_num_args(), 4);
353
0
                    DCHECK(context->is_col_constant(3));
354
0
                    const auto options_col = context->get_constant_col(3)->column_ptr;
355
0
                    options_value = options_col->get_data_at(0);
356
0
                }
357
358
12
                bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
359
12
                                                         options_value, scoped_re);
360
12
                if (!st) {
361
0
                    context->set_error(error_str.c_str());
362
0
                    return Status::InvalidArgument(error_str);
363
0
                }
364
12
                std::shared_ptr<re2::RE2> re(scoped_re.release());
365
12
                context->set_function_state(scope, re);
366
12
            }
367
16
        }
368
28
        return Status::OK();
369
32
    }
_ZN5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
Line
Count
Source
338
16
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
339
16
        if (scope == FunctionContext::THREAD_LOCAL) {
340
8
            if (context->is_col_constant(1)) {
341
8
                DCHECK(!context->get_function_state(scope));
342
8
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
343
8
                const auto& pattern = pattern_col->get_data_at(0);
344
8
                if (pattern.size == 0) {
345
2
                    return Status::OK();
346
2
                }
347
348
6
                std::string error_str;
349
6
                std::unique_ptr<re2::RE2> scoped_re;
350
6
                StringRef options_value;
351
                if constexpr (std::is_same_v<FourParamTypes, ParamTypes>) {
352
                    DCHECK_EQ(context->get_num_args(), 4);
353
                    DCHECK(context->is_col_constant(3));
354
                    const auto options_col = context->get_constant_col(3)->column_ptr;
355
                    options_value = options_col->get_data_at(0);
356
                }
357
358
6
                bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
359
6
                                                         options_value, scoped_re);
360
6
                if (!st) {
361
0
                    context->set_error(error_str.c_str());
362
0
                    return Status::InvalidArgument(error_str);
363
0
                }
364
6
                std::shared_ptr<re2::RE2> re(scoped_re.release());
365
6
                context->set_function_state(scope, re);
366
6
            }
367
8
        }
368
14
        return Status::OK();
369
16
    }
Unexecuted instantiation: _ZN5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
_ZN5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
Line
Count
Source
338
16
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
339
16
        if (scope == FunctionContext::THREAD_LOCAL) {
340
8
            if (context->is_col_constant(1)) {
341
8
                DCHECK(!context->get_function_state(scope));
342
8
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
343
8
                const auto& pattern = pattern_col->get_data_at(0);
344
8
                if (pattern.size == 0) {
345
2
                    return Status::OK();
346
2
                }
347
348
6
                std::string error_str;
349
6
                std::unique_ptr<re2::RE2> scoped_re;
350
6
                StringRef options_value;
351
                if constexpr (std::is_same_v<FourParamTypes, ParamTypes>) {
352
                    DCHECK_EQ(context->get_num_args(), 4);
353
                    DCHECK(context->is_col_constant(3));
354
                    const auto options_col = context->get_constant_col(3)->column_ptr;
355
                    options_value = options_col->get_data_at(0);
356
                }
357
358
6
                bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
359
6
                                                         options_value, scoped_re);
360
6
                if (!st) {
361
0
                    context->set_error(error_str.c_str());
362
0
                    return Status::InvalidArgument(error_str);
363
0
                }
364
6
                std::shared_ptr<re2::RE2> re(scoped_re.release());
365
6
                context->set_function_state(scope, re);
366
6
            }
367
8
        }
368
14
        return Status::OK();
369
16
    }
Unexecuted instantiation: _ZN5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
370
371
    Status execute_impl(FunctionContext* context, Block& block, const ColumnNumbers& arguments,
372
12
                        uint32_t result, size_t input_rows_count) const override {
373
12
        size_t argument_size = arguments.size();
374
375
12
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
376
12
        auto result_data_column = ColumnString::create();
377
12
        auto& result_data = result_data_column->get_chars();
378
12
        auto& result_offset = result_data_column->get_offsets();
379
12
        result_offset.resize(input_rows_count);
380
381
12
        bool col_const[3];
382
12
        ColumnPtr argument_columns[3];
383
48
        for (int i = 0; i < 3; ++i) {
384
36
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
385
36
        }
386
12
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
387
0
                                                     *block.get_by_position(arguments[0]).column)
388
0
                                                     .convert_to_full_column()
389
12
                                           : block.get_by_position(arguments[0]).column;
390
391
12
        default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block, arguments);
392
393
12
        StringRef options_value;
394
12
        if (col_const[1] && col_const[2]) {
395
0
            Impl::execute_impl_const_args(context, argument_columns, options_value,
396
0
                                          input_rows_count, result_data, result_offset,
397
0
                                          result_null_map->get_data());
398
12
        } else {
399
            // the options have check in FE, so is always const, and get idx of 0
400
12
            if (argument_size == 4) {
401
0
                options_value = block.get_by_position(arguments[3]).column->get_data_at(0);
402
0
            }
403
12
            Impl::execute_impl(context, argument_columns, options_value, input_rows_count,
404
12
                               result_data, result_offset, result_null_map->get_data());
405
12
        }
406
407
12
        block.get_by_position(result).column =
408
12
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
409
12
        return Status::OK();
410
12
    }
_ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_15ThreeParamTypesEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
372
6
                        uint32_t result, size_t input_rows_count) const override {
373
6
        size_t argument_size = arguments.size();
374
375
6
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
376
6
        auto result_data_column = ColumnString::create();
377
6
        auto& result_data = result_data_column->get_chars();
378
6
        auto& result_offset = result_data_column->get_offsets();
379
6
        result_offset.resize(input_rows_count);
380
381
6
        bool col_const[3];
382
6
        ColumnPtr argument_columns[3];
383
24
        for (int i = 0; i < 3; ++i) {
384
18
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
385
18
        }
386
6
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
387
0
                                                     *block.get_by_position(arguments[0]).column)
388
0
                                                     .convert_to_full_column()
389
6
                                           : block.get_by_position(arguments[0]).column;
390
391
6
        default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block, arguments);
392
393
6
        StringRef options_value;
394
6
        if (col_const[1] && col_const[2]) {
395
0
            Impl::execute_impl_const_args(context, argument_columns, options_value,
396
0
                                          input_rows_count, result_data, result_offset,
397
0
                                          result_null_map->get_data());
398
6
        } else {
399
            // the options have check in FE, so is always const, and get idx of 0
400
6
            if (argument_size == 4) {
401
0
                options_value = block.get_by_position(arguments[3]).column->get_data_at(0);
402
0
            }
403
6
            Impl::execute_impl(context, argument_columns, options_value, input_rows_count,
404
6
                               result_data, result_offset, result_null_map->get_data());
405
6
        }
406
407
6
        block.get_by_position(result).column =
408
6
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
409
6
        return Status::OK();
410
6
    }
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_17RegexpReplaceImplENS_14FourParamTypesEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
_ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_15ThreeParamTypesEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
372
6
                        uint32_t result, size_t input_rows_count) const override {
373
6
        size_t argument_size = arguments.size();
374
375
6
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
376
6
        auto result_data_column = ColumnString::create();
377
6
        auto& result_data = result_data_column->get_chars();
378
6
        auto& result_offset = result_data_column->get_offsets();
379
6
        result_offset.resize(input_rows_count);
380
381
6
        bool col_const[3];
382
6
        ColumnPtr argument_columns[3];
383
24
        for (int i = 0; i < 3; ++i) {
384
18
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
385
18
        }
386
6
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
387
0
                                                     *block.get_by_position(arguments[0]).column)
388
0
                                                     .convert_to_full_column()
389
6
                                           : block.get_by_position(arguments[0]).column;
390
391
6
        default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block, arguments);
392
393
6
        StringRef options_value;
394
6
        if (col_const[1] && col_const[2]) {
395
0
            Impl::execute_impl_const_args(context, argument_columns, options_value,
396
0
                                          input_rows_count, result_data, result_offset,
397
0
                                          result_null_map->get_data());
398
6
        } else {
399
            // the options have check in FE, so is always const, and get idx of 0
400
6
            if (argument_size == 4) {
401
0
                options_value = block.get_by_position(arguments[3]).column->get_data_at(0);
402
0
            }
403
6
            Impl::execute_impl(context, argument_columns, options_value, input_rows_count,
404
6
                               result_data, result_offset, result_null_map->get_data());
405
6
        }
406
407
6
        block.get_by_position(result).column =
408
6
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
409
6
        return Status::OK();
410
6
    }
Unexecuted instantiation: _ZNK5doris21FunctionRegexpReplaceINS_20RegexpReplaceOneImplENS_14FourParamTypesEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
411
};
412
413
struct RegexpReplaceImpl {
414
    static constexpr auto name = "regexp_replace";
415
    static void execute_impl(FunctionContext* context, ColumnPtr argument_columns[],
416
                             const StringRef& options_value, size_t input_rows_count,
417
                             ColumnString::Chars& result_data, ColumnString::Offsets& result_offset,
418
6
                             NullMap& null_map) {
419
6
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
420
6
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
421
6
        const auto* replace_col = check_and_get_column<ColumnString>(argument_columns[2].get());
422
423
12
        for (size_t i = 0; i < input_rows_count; ++i) {
424
6
            if (null_map[i]) {
425
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
426
0
                continue;
427
0
            }
428
6
            _execute_inner_loop<false>(context, str_col, pattern_col, replace_col, options_value,
429
6
                                       result_data, result_offset, null_map, i);
430
6
        }
431
6
    }
432
    static void execute_impl_const_args(FunctionContext* context, ColumnPtr argument_columns[],
433
                                        const StringRef& options_value, size_t input_rows_count,
434
                                        ColumnString::Chars& result_data,
435
0
                                        ColumnString::Offsets& result_offset, NullMap& null_map) {
436
0
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
437
0
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
438
0
        const auto* replace_col = check_and_get_column<ColumnString>(argument_columns[2].get());
439
440
0
        for (size_t i = 0; i < input_rows_count; ++i) {
441
0
            if (null_map[i]) {
442
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
443
0
                continue;
444
0
            }
445
0
            _execute_inner_loop<true>(context, str_col, pattern_col, replace_col, options_value,
446
0
                                      result_data, result_offset, null_map, i);
447
0
        }
448
0
    }
449
    template <bool Const>
450
    static void _execute_inner_loop(FunctionContext* context, const ColumnString* str_col,
451
                                    const ColumnString* pattern_col,
452
                                    const ColumnString* replace_col, const StringRef& options_value,
453
                                    ColumnString::Chars& result_data,
454
                                    ColumnString::Offsets& result_offset, NullMap& null_map,
455
6
                                    const size_t index_now) {
456
6
        re2::RE2* re = reinterpret_cast<re2::RE2*>(
457
6
                context->get_function_state(FunctionContext::THREAD_LOCAL));
458
6
        std::unique_ptr<re2::RE2> scoped_re; // destroys re if state->re is nullptr
459
6
        if (re == nullptr) {
460
2
            std::string error_str;
461
2
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
462
2
            bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
463
2
                                                     options_value, scoped_re);
464
2
            if (!st) {
465
0
                context->add_warning(error_str.c_str());
466
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
467
0
                return;
468
0
            }
469
2
            re = scoped_re.get();
470
2
        }
471
472
6
        re2::StringPiece replace_str = re2::StringPiece(
473
6
                replace_col->get_data_at(index_check_const(index_now, Const)).to_string_view());
474
475
6
        std::string result_str(str_col->get_data_at(index_now).to_string());
476
6
        re2::RE2::GlobalReplace(&result_str, *re, replace_str);
477
6
        StringOP::push_value_string(result_str, index_now, result_data, result_offset);
478
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
455
6
                                    const size_t index_now) {
456
6
        re2::RE2* re = reinterpret_cast<re2::RE2*>(
457
6
                context->get_function_state(FunctionContext::THREAD_LOCAL));
458
6
        std::unique_ptr<re2::RE2> scoped_re; // destroys re if state->re is nullptr
459
6
        if (re == nullptr) {
460
2
            std::string error_str;
461
2
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
462
2
            bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
463
2
                                                     options_value, scoped_re);
464
2
            if (!st) {
465
0
                context->add_warning(error_str.c_str());
466
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
467
0
                return;
468
0
            }
469
2
            re = scoped_re.get();
470
2
        }
471
472
6
        re2::StringPiece replace_str = re2::StringPiece(
473
6
                replace_col->get_data_at(index_check_const(index_now, Const)).to_string_view());
474
475
6
        std::string result_str(str_col->get_data_at(index_now).to_string());
476
6
        re2::RE2::GlobalReplace(&result_str, *re, replace_str);
477
6
        StringOP::push_value_string(result_str, index_now, result_data, result_offset);
478
6
    }
479
};
480
481
struct RegexpReplaceOneImpl {
482
    static constexpr auto name = "regexp_replace_one";
483
484
    static void execute_impl(FunctionContext* context, ColumnPtr argument_columns[],
485
                             const StringRef& options_value, size_t input_rows_count,
486
                             ColumnString::Chars& result_data, ColumnString::Offsets& result_offset,
487
6
                             NullMap& null_map) {
488
6
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
489
6
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
490
6
        const auto* replace_col = check_and_get_column<ColumnString>(argument_columns[2].get());
491
        // 3 args
492
12
        for (size_t i = 0; i < input_rows_count; ++i) {
493
6
            if (null_map[i]) {
494
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
495
0
                continue;
496
0
            }
497
6
            _execute_inner_loop<false>(context, str_col, pattern_col, replace_col, options_value,
498
6
                                       result_data, result_offset, null_map, i);
499
6
        }
500
6
    }
501
502
    static void execute_impl_const_args(FunctionContext* context, ColumnPtr argument_columns[],
503
                                        const StringRef& options_value, size_t input_rows_count,
504
                                        ColumnString::Chars& result_data,
505
0
                                        ColumnString::Offsets& result_offset, NullMap& null_map) {
506
0
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
507
0
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
508
0
        const auto* replace_col = check_and_get_column<ColumnString>(argument_columns[2].get());
509
        // 3 args
510
0
        for (size_t i = 0; i < input_rows_count; ++i) {
511
0
            if (null_map[i]) {
512
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
513
0
                continue;
514
0
            }
515
0
            _execute_inner_loop<true>(context, str_col, pattern_col, replace_col, options_value,
516
0
                                      result_data, result_offset, null_map, i);
517
0
        }
518
0
    }
519
    template <bool Const>
520
    static void _execute_inner_loop(FunctionContext* context, const ColumnString* str_col,
521
                                    const ColumnString* pattern_col,
522
                                    const ColumnString* replace_col, const StringRef& options_value,
523
                                    ColumnString::Chars& result_data,
524
                                    ColumnString::Offsets& result_offset, NullMap& null_map,
525
6
                                    const size_t index_now) {
526
6
        re2::RE2* re = reinterpret_cast<re2::RE2*>(
527
6
                context->get_function_state(FunctionContext::THREAD_LOCAL));
528
6
        std::unique_ptr<re2::RE2> scoped_re; // destroys re if state->re is nullptr
529
6
        if (re == nullptr) {
530
2
            std::string error_str;
531
2
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
532
2
            bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
533
2
                                                     options_value, scoped_re);
534
2
            if (!st) {
535
0
                context->add_warning(error_str.c_str());
536
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
537
0
                return;
538
0
            }
539
2
            re = scoped_re.get();
540
2
        }
541
542
6
        re2::StringPiece replace_str = re2::StringPiece(
543
6
                replace_col->get_data_at(index_check_const(index_now, Const)).to_string_view());
544
545
6
        std::string result_str(str_col->get_data_at(index_now).to_string());
546
6
        re2::RE2::Replace(&result_str, *re, replace_str);
547
6
        StringOP::push_value_string(result_str, index_now, result_data, result_offset);
548
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
525
6
                                    const size_t index_now) {
526
6
        re2::RE2* re = reinterpret_cast<re2::RE2*>(
527
6
                context->get_function_state(FunctionContext::THREAD_LOCAL));
528
6
        std::unique_ptr<re2::RE2> scoped_re; // destroys re if state->re is nullptr
529
6
        if (re == nullptr) {
530
2
            std::string error_str;
531
2
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
532
2
            bool st = StringFunctions::compile_regex(pattern, &error_str, StringRef(),
533
2
                                                     options_value, scoped_re);
534
2
            if (!st) {
535
0
                context->add_warning(error_str.c_str());
536
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
537
0
                return;
538
0
            }
539
2
            re = scoped_re.get();
540
2
        }
541
542
6
        re2::StringPiece replace_str = re2::StringPiece(
543
6
                replace_col->get_data_at(index_check_const(index_now, Const)).to_string_view());
544
545
6
        std::string result_str(str_col->get_data_at(index_now).to_string());
546
6
        re2::RE2::Replace(&result_str, *re, replace_str);
547
6
        StringOP::push_value_string(result_str, index_now, result_data, result_offset);
548
6
    }
549
};
550
551
template <bool ReturnNull>
552
struct RegexpExtractImpl {
553
    static constexpr auto name = ReturnNull ? "regexp_extract_or_null" : "regexp_extract";
554
    // 3 args
555
    static void execute_impl(FunctionContext* context, ColumnPtr argument_columns[],
556
                             size_t input_rows_count, ColumnString::Chars& result_data,
557
16
                             ColumnString::Offsets& result_offset, NullMap& null_map) {
558
16
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
559
16
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
560
16
        const auto* index_col = check_and_get_column<ColumnInt64>(argument_columns[2].get());
561
32
        for (size_t i = 0; i < input_rows_count; ++i) {
562
16
            if (null_map[i]) {
563
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
564
0
                continue;
565
0
            }
566
16
            const auto& index_data = index_col->get_int(i);
567
16
            if (index_data < 0) {
568
0
                ReturnNull ? StringOP::push_null_string(i, result_data, result_offset, null_map)
569
0
                           : StringOP::push_empty_string(i, result_data, result_offset);
570
0
                continue;
571
0
            }
572
16
            _execute_inner_loop<false>(context, str_col, pattern_col, index_data, result_data,
573
16
                                       result_offset, null_map, i);
574
16
        }
575
16
    }
_ZN5doris17RegexpExtractImplILb1EE12execute_implEPNS_15FunctionContextEPNS_3COWINS_7IColumnEE13immutable_ptrIS5_EEmRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSA_IjLm4096ESD_Lm16ELm15EEESF_
Line
Count
Source
557
8
                             ColumnString::Offsets& result_offset, NullMap& null_map) {
558
8
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
559
8
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
560
8
        const auto* index_col = check_and_get_column<ColumnInt64>(argument_columns[2].get());
561
16
        for (size_t i = 0; i < input_rows_count; ++i) {
562
8
            if (null_map[i]) {
563
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
564
0
                continue;
565
0
            }
566
8
            const auto& index_data = index_col->get_int(i);
567
8
            if (index_data < 0) {
568
0
                ReturnNull ? StringOP::push_null_string(i, result_data, result_offset, null_map)
569
0
                           : StringOP::push_empty_string(i, result_data, result_offset);
570
0
                continue;
571
0
            }
572
8
            _execute_inner_loop<false>(context, str_col, pattern_col, index_data, result_data,
573
8
                                       result_offset, null_map, i);
574
8
        }
575
8
    }
_ZN5doris17RegexpExtractImplILb0EE12execute_implEPNS_15FunctionContextEPNS_3COWINS_7IColumnEE13immutable_ptrIS5_EEmRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSA_IjLm4096ESD_Lm16ELm15EEESF_
Line
Count
Source
557
8
                             ColumnString::Offsets& result_offset, NullMap& null_map) {
558
8
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
559
8
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
560
8
        const auto* index_col = check_and_get_column<ColumnInt64>(argument_columns[2].get());
561
16
        for (size_t i = 0; i < input_rows_count; ++i) {
562
8
            if (null_map[i]) {
563
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
564
0
                continue;
565
0
            }
566
8
            const auto& index_data = index_col->get_int(i);
567
8
            if (index_data < 0) {
568
0
                ReturnNull ? StringOP::push_null_string(i, result_data, result_offset, null_map)
569
0
                           : StringOP::push_empty_string(i, result_data, result_offset);
570
0
                continue;
571
0
            }
572
8
            _execute_inner_loop<false>(context, str_col, pattern_col, index_data, result_data,
573
8
                                       result_offset, null_map, i);
574
8
        }
575
8
    }
576
577
    static void execute_impl_const_args(FunctionContext* context, ColumnPtr argument_columns[],
578
                                        size_t input_rows_count, ColumnString::Chars& result_data,
579
0
                                        ColumnString::Offsets& result_offset, NullMap& null_map) {
580
0
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
581
0
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
582
0
        const auto* index_col = check_and_get_column<ColumnInt64>(argument_columns[2].get());
583
584
0
        const auto& index_data = index_col->get_int(0);
585
0
        if (index_data < 0) {
586
0
            for (size_t i = 0; i < input_rows_count; ++i) {
587
0
                ReturnNull ? StringOP::push_null_string(i, result_data, result_offset, null_map)
588
0
                           : StringOP::push_empty_string(i, result_data, result_offset);
589
0
            }
590
0
            return;
591
0
        }
592
593
0
        for (size_t i = 0; i < input_rows_count; ++i) {
594
0
            if (null_map[i]) {
595
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
596
0
                continue;
597
0
            }
598
599
0
            _execute_inner_loop<true>(context, str_col, pattern_col, index_data, result_data,
600
0
                                      result_offset, null_map, i);
601
0
        }
602
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_
603
    template <bool Const>
604
    static void _execute_inner_loop(FunctionContext* context, const ColumnString* str_col,
605
                                    const ColumnString* pattern_col, const Int64 index_data,
606
                                    ColumnString::Chars& result_data,
607
                                    ColumnString::Offsets& result_offset, NullMap& null_map,
608
16
                                    const size_t index_now) {
609
16
        auto* engine = reinterpret_cast<RegexpExtractEngine*>(
610
16
                context->get_function_state(FunctionContext::THREAD_LOCAL));
611
16
        std::unique_ptr<RegexpExtractEngine> scoped_engine;
612
613
16
        if (engine == nullptr) {
614
0
            std::string error_str;
615
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
616
0
            scoped_engine = std::make_unique<RegexpExtractEngine>();
617
0
            bool st = RegexpExtractEngine::compile(pattern, &error_str, *scoped_engine,
618
0
                                                   context->state()->enable_extended_regex());
619
0
            if (!st) {
620
0
                context->add_warning(error_str.c_str());
621
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
622
0
                return;
623
0
            }
624
0
            engine = scoped_engine.get();
625
0
        }
626
627
16
        const auto& str = str_col->get_data_at(index_now);
628
629
16
        int max_matches = 1 + engine->number_of_capturing_groups();
630
16
        if (index_data >= max_matches) {
631
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
632
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
633
0
            return;
634
0
        }
635
636
16
        std::string match_result;
637
16
        bool success = engine->match_and_extract(str.data, str.size, static_cast<int>(index_data),
638
16
                                                 match_result);
639
640
16
        if (!success) {
641
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
642
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
643
0
            return;
644
0
        }
645
646
16
        StringOP::push_value_string(std::string_view(match_result.data(), match_result.size()),
647
16
                                    index_now, result_data, result_offset);
648
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
608
8
                                    const size_t index_now) {
609
8
        auto* engine = reinterpret_cast<RegexpExtractEngine*>(
610
8
                context->get_function_state(FunctionContext::THREAD_LOCAL));
611
8
        std::unique_ptr<RegexpExtractEngine> scoped_engine;
612
613
8
        if (engine == nullptr) {
614
0
            std::string error_str;
615
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
616
0
            scoped_engine = std::make_unique<RegexpExtractEngine>();
617
0
            bool st = RegexpExtractEngine::compile(pattern, &error_str, *scoped_engine,
618
0
                                                   context->state()->enable_extended_regex());
619
0
            if (!st) {
620
0
                context->add_warning(error_str.c_str());
621
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
622
0
                return;
623
0
            }
624
0
            engine = scoped_engine.get();
625
0
        }
626
627
8
        const auto& str = str_col->get_data_at(index_now);
628
629
8
        int max_matches = 1 + engine->number_of_capturing_groups();
630
8
        if (index_data >= max_matches) {
631
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
632
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
633
0
            return;
634
0
        }
635
636
8
        std::string match_result;
637
8
        bool success = engine->match_and_extract(str.data, str.size, static_cast<int>(index_data),
638
8
                                                 match_result);
639
640
8
        if (!success) {
641
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
642
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
643
0
            return;
644
0
        }
645
646
8
        StringOP::push_value_string(std::string_view(match_result.data(), match_result.size()),
647
8
                                    index_now, result_data, result_offset);
648
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
608
8
                                    const size_t index_now) {
609
8
        auto* engine = reinterpret_cast<RegexpExtractEngine*>(
610
8
                context->get_function_state(FunctionContext::THREAD_LOCAL));
611
8
        std::unique_ptr<RegexpExtractEngine> scoped_engine;
612
613
8
        if (engine == nullptr) {
614
0
            std::string error_str;
615
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
616
0
            scoped_engine = std::make_unique<RegexpExtractEngine>();
617
0
            bool st = RegexpExtractEngine::compile(pattern, &error_str, *scoped_engine,
618
0
                                                   context->state()->enable_extended_regex());
619
0
            if (!st) {
620
0
                context->add_warning(error_str.c_str());
621
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
622
0
                return;
623
0
            }
624
0
            engine = scoped_engine.get();
625
0
        }
626
627
8
        const auto& str = str_col->get_data_at(index_now);
628
629
8
        int max_matches = 1 + engine->number_of_capturing_groups();
630
8
        if (index_data >= max_matches) {
631
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
632
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
633
0
            return;
634
0
        }
635
636
8
        std::string match_result;
637
8
        bool success = engine->match_and_extract(str.data, str.size, static_cast<int>(index_data),
638
8
                                                 match_result);
639
640
8
        if (!success) {
641
0
            ReturnNull ? StringOP::push_null_string(index_now, result_data, result_offset, null_map)
642
0
                       : StringOP::push_empty_string(index_now, result_data, result_offset);
643
0
            return;
644
0
        }
645
646
8
        StringOP::push_value_string(std::string_view(match_result.data(), match_result.size()),
647
8
                                    index_now, result_data, result_offset);
648
8
    }
649
};
650
651
struct RegexpExtractAllImpl {
652
    static constexpr auto name = "regexp_extract_all";
653
654
0
    size_t get_number_of_arguments() const { return 2; }
655
656
    static void execute_impl(FunctionContext* context, ColumnPtr argument_columns[],
657
                             size_t input_rows_count, ColumnString::Chars& result_data,
658
0
                             ColumnString::Offsets& result_offset, NullMap& null_map) {
659
0
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
660
0
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
661
0
        for (int i = 0; i < input_rows_count; ++i) {
662
0
            if (null_map[i]) {
663
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
664
0
                continue;
665
0
            }
666
0
            _execute_inner_loop<false>(context, str_col, pattern_col, result_data, result_offset,
667
0
                                       null_map, i);
668
0
        }
669
0
    }
670
671
    static void execute_impl_const_args(FunctionContext* context, ColumnPtr argument_columns[],
672
                                        size_t input_rows_count, ColumnString::Chars& result_data,
673
7
                                        ColumnString::Offsets& result_offset, NullMap& null_map) {
674
7
        const auto* str_col = check_and_get_column<ColumnString>(argument_columns[0].get());
675
7
        const auto* pattern_col = check_and_get_column<ColumnString>(argument_columns[1].get());
676
14
        for (int i = 0; i < input_rows_count; ++i) {
677
7
            if (null_map[i]) {
678
0
                StringOP::push_null_string(i, result_data, result_offset, null_map);
679
0
                continue;
680
0
            }
681
7
            _execute_inner_loop<true>(context, str_col, pattern_col, result_data, result_offset,
682
7
                                      null_map, i);
683
7
        }
684
7
    }
685
    template <bool Const>
686
    static void _execute_inner_loop(FunctionContext* context, const ColumnString* str_col,
687
                                    const ColumnString* pattern_col,
688
                                    ColumnString::Chars& result_data,
689
                                    ColumnString::Offsets& result_offset, NullMap& null_map,
690
7
                                    const size_t index_now) {
691
7
        auto* engine = reinterpret_cast<RegexpExtractEngine*>(
692
7
                context->get_function_state(FunctionContext::THREAD_LOCAL));
693
7
        std::unique_ptr<RegexpExtractEngine> scoped_engine;
694
695
7
        if (engine == nullptr) {
696
0
            std::string error_str;
697
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
698
0
            scoped_engine = std::make_unique<RegexpExtractEngine>();
699
0
            bool st = RegexpExtractEngine::compile(pattern, &error_str, *scoped_engine,
700
0
                                                   context->state()->enable_extended_regex());
701
0
            if (!st) {
702
0
                context->add_warning(error_str.c_str());
703
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
704
0
                return;
705
0
            }
706
0
            engine = scoped_engine.get();
707
0
        }
708
709
7
        if (engine->number_of_capturing_groups() == 0) {
710
0
            StringOP::push_empty_string(index_now, result_data, result_offset);
711
0
            return;
712
0
        }
713
7
        const auto& str = str_col->get_data_at(index_now);
714
7
        std::vector<std::string> res_matches;
715
7
        engine->match_all_and_extract(str.data, str.size, res_matches);
716
717
7
        if (res_matches.empty()) {
718
0
            StringOP::push_empty_string(index_now, result_data, result_offset);
719
0
            return;
720
0
        }
721
722
7
        std::string res = "[";
723
19
        for (int j = 0; j < res_matches.size(); ++j) {
724
12
            res += "'" + res_matches[j] + "'";
725
12
            if (j < res_matches.size() - 1) {
726
5
                res += ",";
727
5
            }
728
12
        }
729
7
        res += "]";
730
7
        StringOP::push_value_string(std::string_view(res), index_now, result_data, result_offset);
731
7
    }
_ZN5doris20RegexpExtractAllImpl19_execute_inner_loopILb1EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES7_RNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEESD_m
Line
Count
Source
690
7
                                    const size_t index_now) {
691
7
        auto* engine = reinterpret_cast<RegexpExtractEngine*>(
692
7
                context->get_function_state(FunctionContext::THREAD_LOCAL));
693
7
        std::unique_ptr<RegexpExtractEngine> scoped_engine;
694
695
7
        if (engine == nullptr) {
696
0
            std::string error_str;
697
0
            const auto& pattern = pattern_col->get_data_at(index_check_const(index_now, Const));
698
0
            scoped_engine = std::make_unique<RegexpExtractEngine>();
699
0
            bool st = RegexpExtractEngine::compile(pattern, &error_str, *scoped_engine,
700
0
                                                   context->state()->enable_extended_regex());
701
0
            if (!st) {
702
0
                context->add_warning(error_str.c_str());
703
0
                StringOP::push_null_string(index_now, result_data, result_offset, null_map);
704
0
                return;
705
0
            }
706
0
            engine = scoped_engine.get();
707
0
        }
708
709
7
        if (engine->number_of_capturing_groups() == 0) {
710
0
            StringOP::push_empty_string(index_now, result_data, result_offset);
711
0
            return;
712
0
        }
713
7
        const auto& str = str_col->get_data_at(index_now);
714
7
        std::vector<std::string> res_matches;
715
7
        engine->match_all_and_extract(str.data, str.size, res_matches);
716
717
7
        if (res_matches.empty()) {
718
0
            StringOP::push_empty_string(index_now, result_data, result_offset);
719
0
            return;
720
0
        }
721
722
7
        std::string res = "[";
723
19
        for (int j = 0; j < res_matches.size(); ++j) {
724
12
            res += "'" + res_matches[j] + "'";
725
12
            if (j < res_matches.size() - 1) {
726
5
                res += ",";
727
5
            }
728
12
        }
729
7
        res += "]";
730
7
        StringOP::push_value_string(std::string_view(res), index_now, result_data, result_offset);
731
7
    }
Unexecuted instantiation: _ZN5doris20RegexpExtractAllImpl19_execute_inner_loopILb0EEEvPNS_15FunctionContextEPKNS_9ColumnStrIjEES7_RNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEESD_m
732
};
733
734
// template FunctionRegexpFunctionality is used for regexp_xxxx series functions, not for regexp match.
735
template <typename Impl>
736
class FunctionRegexpFunctionality : public IFunction {
737
public:
738
    static constexpr auto name = Impl::name;
739
740
35
    static FunctionPtr create() { return std::make_shared<FunctionRegexpFunctionality>(); }
_ZN5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE6createEv
Line
Count
Source
740
12
    static FunctionPtr create() { return std::make_shared<FunctionRegexpFunctionality>(); }
_ZN5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE6createEv
Line
Count
Source
740
12
    static FunctionPtr create() { return std::make_shared<FunctionRegexpFunctionality>(); }
_ZN5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE6createEv
Line
Count
Source
740
11
    static FunctionPtr create() { return std::make_shared<FunctionRegexpFunctionality>(); }
741
742
3
    String get_name() const override { return name; }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE8get_nameB5cxx11Ev
Line
Count
Source
742
1
    String get_name() const override { return name; }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE8get_nameB5cxx11Ev
Line
Count
Source
742
1
    String get_name() const override { return name; }
_ZNK5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE8get_nameB5cxx11Ev
Line
Count
Source
742
1
    String get_name() const override { return name; }
743
744
29
    size_t get_number_of_arguments() const override {
745
29
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
746
9
            return 2;
747
9
        }
748
0
        return 3;
749
29
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE23get_number_of_argumentsEv
Line
Count
Source
744
10
    size_t get_number_of_arguments() const override {
745
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
746
            return 2;
747
        }
748
10
        return 3;
749
10
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE23get_number_of_argumentsEv
Line
Count
Source
744
10
    size_t get_number_of_arguments() const override {
745
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
746
            return 2;
747
        }
748
10
        return 3;
749
10
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE23get_number_of_argumentsEv
Line
Count
Source
744
9
    size_t get_number_of_arguments() const override {
745
9
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
746
9
            return 2;
747
9
        }
748
0
        return 3;
749
9
    }
750
751
29
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
752
29
        return make_nullable(std::make_shared<DataTypeString>());
753
29
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
Line
Count
Source
751
10
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
752
10
        return make_nullable(std::make_shared<DataTypeString>());
753
10
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS8_EE
Line
Count
Source
751
10
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
752
10
        return make_nullable(std::make_shared<DataTypeString>());
753
10
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS7_EE
Line
Count
Source
751
9
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
752
9
        return make_nullable(std::make_shared<DataTypeString>());
753
9
    }
754
755
58
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
756
58
        if (scope == FunctionContext::THREAD_LOCAL) {
757
29
            if (context->is_col_constant(1)) {
758
29
                DCHECK(!context->get_function_state(scope));
759
29
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
760
29
                const auto& pattern = pattern_col->get_data_at(0);
761
29
                if (pattern.size == 0) {
762
3
                    return Status::OK();
763
3
                }
764
765
26
                std::string error_str;
766
26
                auto engine = std::make_shared<RegexpExtractEngine>();
767
26
                bool st = RegexpExtractEngine::compile(pattern, &error_str, *engine,
768
26
                                                       context->state()->enable_extended_regex());
769
26
                if (!st) {
770
0
                    context->set_error(error_str.c_str());
771
0
                    return Status::InvalidArgument(error_str);
772
0
                }
773
26
                context->set_function_state(scope, engine);
774
26
            }
775
29
        }
776
55
        return Status::OK();
777
58
    }
_ZN5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
Line
Count
Source
755
20
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
756
20
        if (scope == FunctionContext::THREAD_LOCAL) {
757
10
            if (context->is_col_constant(1)) {
758
10
                DCHECK(!context->get_function_state(scope));
759
10
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
760
10
                const auto& pattern = pattern_col->get_data_at(0);
761
10
                if (pattern.size == 0) {
762
1
                    return Status::OK();
763
1
                }
764
765
9
                std::string error_str;
766
9
                auto engine = std::make_shared<RegexpExtractEngine>();
767
9
                bool st = RegexpExtractEngine::compile(pattern, &error_str, *engine,
768
9
                                                       context->state()->enable_extended_regex());
769
9
                if (!st) {
770
0
                    context->set_error(error_str.c_str());
771
0
                    return Status::InvalidArgument(error_str);
772
0
                }
773
9
                context->set_function_state(scope, engine);
774
9
            }
775
10
        }
776
19
        return Status::OK();
777
20
    }
_ZN5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE4openEPNS_15FunctionContextENS4_18FunctionStateScopeE
Line
Count
Source
755
20
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
756
20
        if (scope == FunctionContext::THREAD_LOCAL) {
757
10
            if (context->is_col_constant(1)) {
758
10
                DCHECK(!context->get_function_state(scope));
759
10
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
760
10
                const auto& pattern = pattern_col->get_data_at(0);
761
10
                if (pattern.size == 0) {
762
1
                    return Status::OK();
763
1
                }
764
765
9
                std::string error_str;
766
9
                auto engine = std::make_shared<RegexpExtractEngine>();
767
9
                bool st = RegexpExtractEngine::compile(pattern, &error_str, *engine,
768
9
                                                       context->state()->enable_extended_regex());
769
9
                if (!st) {
770
0
                    context->set_error(error_str.c_str());
771
0
                    return Status::InvalidArgument(error_str);
772
0
                }
773
9
                context->set_function_state(scope, engine);
774
9
            }
775
10
        }
776
19
        return Status::OK();
777
20
    }
_ZN5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE4openEPNS_15FunctionContextENS3_18FunctionStateScopeE
Line
Count
Source
755
18
    Status open(FunctionContext* context, FunctionContext::FunctionStateScope scope) override {
756
18
        if (scope == FunctionContext::THREAD_LOCAL) {
757
9
            if (context->is_col_constant(1)) {
758
9
                DCHECK(!context->get_function_state(scope));
759
9
                const auto pattern_col = context->get_constant_col(1)->column_ptr;
760
9
                const auto& pattern = pattern_col->get_data_at(0);
761
9
                if (pattern.size == 0) {
762
1
                    return Status::OK();
763
1
                }
764
765
8
                std::string error_str;
766
8
                auto engine = std::make_shared<RegexpExtractEngine>();
767
8
                bool st = RegexpExtractEngine::compile(pattern, &error_str, *engine,
768
8
                                                       context->state()->enable_extended_regex());
769
8
                if (!st) {
770
0
                    context->set_error(error_str.c_str());
771
0
                    return Status::InvalidArgument(error_str);
772
0
                }
773
8
                context->set_function_state(scope, engine);
774
8
            }
775
9
        }
776
17
        return Status::OK();
777
18
    }
778
779
    Status execute_impl(FunctionContext* context, Block& block, const ColumnNumbers& arguments,
780
23
                        uint32_t result, size_t input_rows_count) const override {
781
23
        size_t argument_size = arguments.size();
782
783
23
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
784
23
        auto result_data_column = ColumnString::create();
785
23
        auto& result_data = result_data_column->get_chars();
786
23
        auto& result_offset = result_data_column->get_offsets();
787
23
        result_offset.resize(input_rows_count);
788
789
23
        bool col_const[3];
790
23
        ColumnPtr argument_columns[3];
791
85
        for (int i = 0; i < argument_size; ++i) {
792
62
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
793
62
        }
794
23
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
795
0
                                                     *block.get_by_position(arguments[0]).column)
796
0
                                                     .convert_to_full_column()
797
23
                                           : block.get_by_position(arguments[0]).column;
798
23
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
799
7
            default_preprocess_parameter_columns(argument_columns, col_const, {1}, block,
800
7
                                                 arguments);
801
16
        } else {
802
16
            default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block,
803
16
                                                 arguments);
804
16
        }
805
806
23
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
807
7
            if (col_const[1]) {
808
7
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
809
7
                                              result_data, result_offset,
810
7
                                              result_null_map->get_data());
811
7
            } else {
812
0
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
813
0
                                   result_offset, result_null_map->get_data());
814
0
            }
815
16
        } else {
816
16
            if (col_const[1] && col_const[2]) {
817
0
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
818
0
                                              result_data, result_offset,
819
0
                                              result_null_map->get_data());
820
16
            } else {
821
16
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
822
16
                                   result_offset, result_null_map->get_data());
823
16
            }
824
16
        }
825
826
23
        block.get_by_position(result).column =
827
23
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
828
23
        return Status::OK();
829
23
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb1EEEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
780
8
                        uint32_t result, size_t input_rows_count) const override {
781
8
        size_t argument_size = arguments.size();
782
783
8
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
784
8
        auto result_data_column = ColumnString::create();
785
8
        auto& result_data = result_data_column->get_chars();
786
8
        auto& result_offset = result_data_column->get_offsets();
787
8
        result_offset.resize(input_rows_count);
788
789
8
        bool col_const[3];
790
8
        ColumnPtr argument_columns[3];
791
32
        for (int i = 0; i < argument_size; ++i) {
792
24
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
793
24
        }
794
8
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
795
0
                                                     *block.get_by_position(arguments[0]).column)
796
0
                                                     .convert_to_full_column()
797
8
                                           : block.get_by_position(arguments[0]).column;
798
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
799
            default_preprocess_parameter_columns(argument_columns, col_const, {1}, block,
800
                                                 arguments);
801
8
        } else {
802
8
            default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block,
803
8
                                                 arguments);
804
8
        }
805
806
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
807
            if (col_const[1]) {
808
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
809
                                              result_data, result_offset,
810
                                              result_null_map->get_data());
811
            } else {
812
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
813
                                   result_offset, result_null_map->get_data());
814
            }
815
8
        } else {
816
8
            if (col_const[1] && col_const[2]) {
817
0
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
818
0
                                              result_data, result_offset,
819
0
                                              result_null_map->get_data());
820
8
            } else {
821
8
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
822
8
                                   result_offset, result_null_map->get_data());
823
8
            }
824
8
        }
825
826
8
        block.get_by_position(result).column =
827
8
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
828
8
        return Status::OK();
829
8
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_17RegexpExtractImplILb0EEEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
780
8
                        uint32_t result, size_t input_rows_count) const override {
781
8
        size_t argument_size = arguments.size();
782
783
8
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
784
8
        auto result_data_column = ColumnString::create();
785
8
        auto& result_data = result_data_column->get_chars();
786
8
        auto& result_offset = result_data_column->get_offsets();
787
8
        result_offset.resize(input_rows_count);
788
789
8
        bool col_const[3];
790
8
        ColumnPtr argument_columns[3];
791
32
        for (int i = 0; i < argument_size; ++i) {
792
24
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
793
24
        }
794
8
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
795
0
                                                     *block.get_by_position(arguments[0]).column)
796
0
                                                     .convert_to_full_column()
797
8
                                           : block.get_by_position(arguments[0]).column;
798
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
799
            default_preprocess_parameter_columns(argument_columns, col_const, {1}, block,
800
                                                 arguments);
801
8
        } else {
802
8
            default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block,
803
8
                                                 arguments);
804
8
        }
805
806
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
807
            if (col_const[1]) {
808
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
809
                                              result_data, result_offset,
810
                                              result_null_map->get_data());
811
            } else {
812
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
813
                                   result_offset, result_null_map->get_data());
814
            }
815
8
        } else {
816
8
            if (col_const[1] && col_const[2]) {
817
0
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
818
0
                                              result_data, result_offset,
819
0
                                              result_null_map->get_data());
820
8
            } else {
821
8
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
822
8
                                   result_offset, result_null_map->get_data());
823
8
            }
824
8
        }
825
826
8
        block.get_by_position(result).column =
827
8
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
828
8
        return Status::OK();
829
8
    }
_ZNK5doris27FunctionRegexpFunctionalityINS_20RegexpExtractAllImplEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
780
7
                        uint32_t result, size_t input_rows_count) const override {
781
7
        size_t argument_size = arguments.size();
782
783
7
        auto result_null_map = ColumnUInt8::create(input_rows_count, 0);
784
7
        auto result_data_column = ColumnString::create();
785
7
        auto& result_data = result_data_column->get_chars();
786
7
        auto& result_offset = result_data_column->get_offsets();
787
7
        result_offset.resize(input_rows_count);
788
789
7
        bool col_const[3];
790
7
        ColumnPtr argument_columns[3];
791
21
        for (int i = 0; i < argument_size; ++i) {
792
14
            col_const[i] = is_column_const(*block.get_by_position(arguments[i]).column);
793
14
        }
794
7
        argument_columns[0] = col_const[0] ? static_cast<const ColumnConst&>(
795
0
                                                     *block.get_by_position(arguments[0]).column)
796
0
                                                     .convert_to_full_column()
797
7
                                           : block.get_by_position(arguments[0]).column;
798
7
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
799
7
            default_preprocess_parameter_columns(argument_columns, col_const, {1}, block,
800
7
                                                 arguments);
801
        } else {
802
            default_preprocess_parameter_columns(argument_columns, col_const, {1, 2}, block,
803
                                                 arguments);
804
        }
805
806
7
        if constexpr (std::is_same_v<Impl, RegexpExtractAllImpl>) {
807
7
            if (col_const[1]) {
808
7
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
809
7
                                              result_data, result_offset,
810
7
                                              result_null_map->get_data());
811
7
            } else {
812
0
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
813
0
                                   result_offset, result_null_map->get_data());
814
0
            }
815
        } else {
816
            if (col_const[1] && col_const[2]) {
817
                Impl::execute_impl_const_args(context, argument_columns, input_rows_count,
818
                                              result_data, result_offset,
819
                                              result_null_map->get_data());
820
            } else {
821
                Impl::execute_impl(context, argument_columns, input_rows_count, result_data,
822
                                   result_offset, result_null_map->get_data());
823
            }
824
        }
825
826
7
        block.get_by_position(result).column =
827
7
                ColumnNullable::create(std::move(result_data_column), std::move(result_null_map));
828
7
        return Status::OK();
829
7
    }
830
};
831
832
1
void register_function_regexp_extract(SimpleFunctionFactory& factory) {
833
1
    factory.register_function<FunctionRegexpReplace<RegexpReplaceImpl, ThreeParamTypes>>();
834
1
    factory.register_function<FunctionRegexpReplace<RegexpReplaceImpl, FourParamTypes>>();
835
1
    factory.register_function<FunctionRegexpReplace<RegexpReplaceOneImpl, ThreeParamTypes>>();
836
1
    factory.register_function<FunctionRegexpReplace<RegexpReplaceOneImpl, FourParamTypes>>();
837
1
    factory.register_function<FunctionRegexpFunctionality<RegexpExtractImpl<true>>>();
838
1
    factory.register_function<FunctionRegexpFunctionality<RegexpExtractImpl<false>>>();
839
1
    factory.register_function<FunctionRegexpFunctionality<RegexpExtractAllImpl>>();
840
1
    factory.register_function<FunctionRegexpCount>();
841
1
}
842
843
} // namespace doris