Coverage Report

Created: 2026-04-10 04:05

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