Coverage Report

Created: 2026-04-14 03:58

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
be/src/storage/field.h
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
#pragma once
19
20
#include <cstddef>
21
#include <sstream>
22
#include <string>
23
24
#include "core/arena.h"
25
#include "core/value/map_value.h"
26
#include "runtime/collection_value.h"
27
#include "storage/key_coder.h"
28
#include "storage/olap_common.h"
29
#include "storage/olap_define.h"
30
#include "storage/tablet/tablet_schema.h"
31
#include "storage/types.h"
32
#include "storage/utils.h"
33
#include "util/hash_util.hpp"
34
#include "util/json/path_in_data.h"
35
#include "util/slice.h"
36
37
namespace doris {
38
// A Field is used to represent a column in memory format.
39
// User can use this class to access or deal with column data in memory.
40
class StorageField {
41
public:
42
    StorageField(const TabletColumn& column)
43
22.4M
            : _type_info(get_type_info(&column)),
44
22.4M
              _desc(column),
45
22.4M
              _length(column.length()),
46
22.4M
              _key_coder(get_key_coder(column.type())),
47
22.4M
              _name(column.name()),
48
22.4M
              _index_size(column.index_length()),
49
22.4M
              _is_nullable(column.is_nullable()),
50
22.4M
              _unique_id(column.unique_id()),
51
22.4M
              _parent_unique_id(column.parent_unique_id()),
52
22.4M
              _is_extracted_column(column.is_extracted_column()),
53
22.4M
              _path(column.path_info_ptr()) {}
54
55
22.5M
    virtual ~StorageField() = default;
56
57
821k
    size_t size() const { return _type_info->size(); }
58
5
    size_t length() const { return _length; }
59
0
    size_t field_size() const { return size() + 1; }
60
0
    size_t index_size() const { return _index_size; }
61
16.6M
    int32_t unique_id() const { return _unique_id; }
62
36
    int32_t parent_unique_id() const { return _parent_unique_id; }
63
16.1M
    bool is_extracted_column() const { return _is_extracted_column; }
64
24.3M
    const std::string& name() const { return _name; }
65
0
    const PathInDataPtr& path() const { return _path; }
66
67
0
    virtual void set_to_max(char* buf) const { return _type_info->set_to_max(buf); }
68
69
2
    virtual void set_to_min(char* buf) const { return _type_info->set_to_min(buf); }
70
71
50.4k
    virtual StorageField* clone() const {
72
50.4k
        auto* local = new StorageField(_desc);
73
50.4k
        this->clone(local);
74
50.4k
        return local;
75
50.4k
    }
76
77
22.2M
    FieldType type() const { return _type_info->type(); }
78
2.29k
    const TypeInfo* type_info() const { return _type_info.get(); }
79
17.7M
    bool is_nullable() const { return _is_nullable; }
80
81
    // similar to `full_encode_ascending`, but only encode part (the first `index_size` bytes) of the value.
82
    // only applicable to string type
83
21.2k
    void encode_ascending(const void* value, std::string* buf) const {
84
21.2k
        _key_coder->encode_ascending(value, _index_size, buf);
85
21.2k
    }
86
87
    // encode the provided `value` into `buf`.
88
3.73M
    void full_encode_ascending(const void* value, std::string* buf) const {
89
3.73M
        _key_coder->full_encode_ascending(value, buf);
90
3.73M
    }
91
28.5k
    void add_sub_field(std::unique_ptr<StorageField> sub_field) {
92
28.5k
        _sub_fields.emplace_back(std::move(sub_field));
93
28.5k
    }
94
6.88k
    StorageField* get_sub_field(size_t i) const { return _sub_fields[i].get(); }
95
2.73k
    size_t get_sub_field_count() const { return _sub_fields.size(); }
96
97
1.07M
    void set_precision(int32_t precision) { _precision = precision; }
98
1.07M
    void set_scale(int32_t scale) { _scale = scale; }
99
0
    int32_t get_precision() const { return _precision; }
100
4.12M
    int32_t get_scale() const { return _scale; }
101
28.6M
    const TabletColumn& get_desc() const { return _desc; }
102
103
8.08M
    int32_t get_unique_id() const {
104
8.08M
        return is_extracted_column() ? parent_unique_id() : unique_id();
105
8.08M
    }
106
107
protected:
108
    TypeInfoPtr _type_info;
109
    TabletColumn _desc;
110
    // unit : byte
111
    // except for strings, other types have fixed lengths
112
    // Note that, the struct type itself has fixed length, but due to
113
    // its number of subfields is a variable, so the actual length of
114
    // a struct field is not fixed.
115
    size_t _length;
116
117
2.47M
    void clone(StorageField* other) const {
118
2.47M
        other->_type_info = clone_type_info(this->_type_info.get());
119
2.47M
        other->_key_coder = this->_key_coder;
120
2.47M
        other->_name = this->_name;
121
2.47M
        other->_index_size = this->_index_size;
122
2.47M
        other->_is_nullable = this->_is_nullable;
123
2.47M
        other->_sub_fields.clear();
124
2.47M
        other->_precision = this->_precision;
125
2.47M
        other->_scale = this->_scale;
126
2.47M
        other->_unique_id = this->_unique_id;
127
2.47M
        other->_parent_unique_id = this->_parent_unique_id;
128
2.47M
        other->_is_extracted_column = this->_is_extracted_column;
129
2.47M
        for (const auto& f : _sub_fields) {
130
0
            StorageField* item = f->clone();
131
0
            other->add_sub_field(std::unique_ptr<StorageField>(item));
132
0
        }
133
2.47M
    }
134
135
private:
136
    // maximum length of Field, unit : bytes
137
    // usually equal to length, except for variable-length strings
138
    const KeyCoder* _key_coder;
139
    std::string _name;
140
    size_t _index_size;
141
    bool _is_nullable;
142
    std::vector<std::unique_ptr<StorageField>> _sub_fields;
143
    int32_t _precision;
144
    int32_t _scale;
145
    int32_t _unique_id;
146
    int32_t _parent_unique_id;
147
    bool _is_extracted_column = false;
148
    PathInDataPtr _path;
149
};
150
151
class MapField : public StorageField {
152
public:
153
5.01k
    MapField(const TabletColumn& column) : StorageField(column) {}
154
};
155
156
class StructField : public StorageField {
157
public:
158
1.78k
    StructField(const TabletColumn& column) : StorageField(column) {}
159
};
160
161
class ArrayField : public StorageField {
162
public:
163
8.96k
    ArrayField(const TabletColumn& column) : StorageField(column) {}
164
};
165
166
class CharField : public StorageField {
167
public:
168
3.24k
    CharField(const TabletColumn& column) : StorageField(column) {}
169
170
0
    CharField* clone() const override {
171
0
        auto* local = new CharField(_desc);
172
0
        StorageField::clone(local);
173
0
        return local;
174
0
    }
175
176
0
    void set_to_max(char* ch) const override {
177
0
        auto slice = reinterpret_cast<Slice*>(ch);
178
0
        slice->size = _length;
179
0
        memset(slice->data, 0xFF, slice->size);
180
0
    }
181
};
182
183
class VarcharField : public StorageField {
184
public:
185
2.10M
    VarcharField(const TabletColumn& column) : StorageField(column) {}
186
187
0
    VarcharField* clone() const override {
188
0
        auto* local = new VarcharField(_desc);
189
0
        StorageField::clone(local);
190
0
        return local;
191
0
    }
192
193
2
    void set_to_max(char* ch) const override {
194
2
        auto slice = reinterpret_cast<Slice*>(ch);
195
2
        slice->size = _length - OLAP_VARCHAR_MAX_BYTES;
196
2
        memset(slice->data, 0xFF, slice->size);
197
2
    }
198
};
199
class StringField : public StorageField {
200
public:
201
13.8M
    StringField(const TabletColumn& column) : StorageField(column) {}
202
203
2.42M
    StringField* clone() const override {
204
2.42M
        auto* local = new StringField(_desc);
205
2.42M
        StorageField::clone(local);
206
2.42M
        return local;
207
2.42M
    }
208
209
0
    void set_to_max(char* ch) const override {
210
0
        auto slice = reinterpret_cast<Slice*>(ch);
211
0
        memset(slice->data, 0xFF, slice->size);
212
0
    }
213
};
214
215
class BitmapAggField : public StorageField {
216
public:
217
76
    BitmapAggField(const TabletColumn& column) : StorageField(column) {}
218
219
0
    BitmapAggField* clone() const override {
220
0
        auto* local = new BitmapAggField(_desc);
221
0
        StorageField::clone(local);
222
0
        return local;
223
0
    }
224
};
225
226
class QuantileStateAggField : public StorageField {
227
public:
228
48
    QuantileStateAggField(const TabletColumn& column) : StorageField(column) {}
229
230
0
    QuantileStateAggField* clone() const override {
231
0
        auto* local = new QuantileStateAggField(_desc);
232
0
        StorageField::clone(local);
233
0
        return local;
234
0
    }
235
};
236
237
class AggStateField : public StorageField {
238
public:
239
208
    AggStateField(const TabletColumn& column) : StorageField(column) {}
240
241
0
    AggStateField* clone() const override {
242
0
        auto* local = new AggStateField(_desc);
243
0
        StorageField::clone(local);
244
0
        return local;
245
0
    }
246
};
247
248
class HllAggField : public StorageField {
249
public:
250
786
    HllAggField(const TabletColumn& column) : StorageField(column) {}
251
252
0
    HllAggField* clone() const override {
253
0
        auto* local = new HllAggField(_desc);
254
0
        StorageField::clone(local);
255
0
        return local;
256
0
    }
257
};
258
259
class StorageFieldFactory {
260
public:
261
20.1M
    static StorageField* create(const TabletColumn& column) {
262
        // for key column
263
20.1M
        if (column.is_key()) {
264
10.4M
            switch (column.type()) {
265
58
            case FieldType::OLAP_FIELD_TYPE_CHAR:
266
58
                return new CharField(column);
267
10.3M
            case FieldType::OLAP_FIELD_TYPE_VARCHAR:
268
10.3M
            case FieldType::OLAP_FIELD_TYPE_STRING:
269
10.3M
                return new StringField(column);
270
0
            case FieldType::OLAP_FIELD_TYPE_STRUCT: {
271
0
                auto* local = new StructField(column);
272
0
                for (uint32_t i = 0; i < column.get_subtype_count(); i++) {
273
0
                    std::unique_ptr<StorageField> sub_field(
274
0
                            StorageFieldFactory::create(column.get_sub_column(i)));
275
0
                    local->add_sub_field(std::move(sub_field));
276
0
                }
277
0
                return local;
278
10.3M
            }
279
0
            case FieldType::OLAP_FIELD_TYPE_ARRAY: {
280
0
                std::unique_ptr<StorageField> item_field(
281
0
                        StorageFieldFactory::create(column.get_sub_column(0)));
282
0
                auto* local = new ArrayField(column);
283
0
                local->add_sub_field(std::move(item_field));
284
0
                return local;
285
10.3M
            }
286
0
            case FieldType::OLAP_FIELD_TYPE_MAP: {
287
0
                std::unique_ptr<StorageField> key_field(
288
0
                        StorageFieldFactory::create(column.get_sub_column(0)));
289
0
                std::unique_ptr<StorageField> val_field(
290
0
                        StorageFieldFactory::create(column.get_sub_column(1)));
291
0
                auto* local = new MapField(column);
292
0
                local->add_sub_field(std::move(key_field));
293
0
                local->add_sub_field(std::move(val_field));
294
0
                return local;
295
10.3M
            }
296
45
            case FieldType::OLAP_FIELD_TYPE_DECIMAL:
297
45
                [[fallthrough]];
298
121
            case FieldType::OLAP_FIELD_TYPE_DECIMAL32:
299
121
                [[fallthrough]];
300
201
            case FieldType::OLAP_FIELD_TYPE_DECIMAL64:
301
201
                [[fallthrough]];
302
252
            case FieldType::OLAP_FIELD_TYPE_DECIMAL128I:
303
252
                [[fallthrough]];
304
254
            case FieldType::OLAP_FIELD_TYPE_DECIMAL256:
305
254
                [[fallthrough]];
306
255
            case FieldType::OLAP_FIELD_TYPE_TIMESTAMPTZ:
307
255
                [[fallthrough]];
308
3.77k
            case FieldType::OLAP_FIELD_TYPE_DATETIMEV2: {
309
3.77k
                StorageField* field = new StorageField(column);
310
3.77k
                field->set_precision(column.precision());
311
3.77k
                field->set_scale(column.frac());
312
3.77k
                return field;
313
255
            }
314
59.1k
            default:
315
59.1k
                return new StorageField(column);
316
10.4M
            }
317
10.4M
        }
318
319
        // for value column
320
9.68M
        switch (column.aggregation()) {
321
9.67M
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_NONE:
322
9.68M
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_SUM:
323
9.68M
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_MIN:
324
9.68M
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_MAX:
325
9.68M
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_REPLACE:
326
9.68M
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_REPLACE_IF_NOT_NULL:
327
9.68M
            switch (column.type()) {
328
3.18k
            case FieldType::OLAP_FIELD_TYPE_CHAR:
329
3.18k
                return new CharField(column);
330
2.10M
            case FieldType::OLAP_FIELD_TYPE_VARCHAR:
331
2.10M
                return new VarcharField(column);
332
1.10M
            case FieldType::OLAP_FIELD_TYPE_STRING:
333
1.10M
                return new StringField(column);
334
1.78k
            case FieldType::OLAP_FIELD_TYPE_STRUCT: {
335
1.78k
                auto* local = new StructField(column);
336
11.3k
                for (uint32_t i = 0; i < column.get_subtype_count(); i++) {
337
9.53k
                    std::unique_ptr<StorageField> sub_field(
338
9.53k
                            StorageFieldFactory::create(column.get_sub_column(i)));
339
9.53k
                    local->add_sub_field(std::move(sub_field));
340
9.53k
                }
341
1.78k
                return local;
342
0
            }
343
8.96k
            case FieldType::OLAP_FIELD_TYPE_ARRAY: {
344
8.96k
                std::unique_ptr<StorageField> item_field(
345
8.96k
                        StorageFieldFactory::create(column.get_sub_column(0)));
346
8.96k
                auto* local = new ArrayField(column);
347
8.96k
                local->add_sub_field(std::move(item_field));
348
8.96k
                return local;
349
0
            }
350
5.02k
            case FieldType::OLAP_FIELD_TYPE_MAP: {
351
5.02k
                DCHECK(column.get_subtype_count() == 2);
352
5.02k
                auto* local = new MapField(column);
353
5.02k
                std::unique_ptr<StorageField> key_field(
354
5.02k
                        StorageFieldFactory::create(column.get_sub_column(0)));
355
5.02k
                std::unique_ptr<StorageField> value_field(
356
5.02k
                        StorageFieldFactory::create(column.get_sub_column(1)));
357
5.02k
                local->add_sub_field(std::move(key_field));
358
5.02k
                local->add_sub_field(std::move(value_field));
359
5.02k
                return local;
360
0
            }
361
1
            case FieldType::OLAP_FIELD_TYPE_DECIMAL:
362
1
                [[fallthrough]];
363
1.92k
            case FieldType::OLAP_FIELD_TYPE_DECIMAL32:
364
1.92k
                [[fallthrough]];
365
8.12k
            case FieldType::OLAP_FIELD_TYPE_DECIMAL64:
366
8.12k
                [[fallthrough]];
367
11.4k
            case FieldType::OLAP_FIELD_TYPE_DECIMAL128I:
368
11.4k
                [[fallthrough]];
369
11.4k
            case FieldType::OLAP_FIELD_TYPE_DECIMAL256:
370
11.4k
                [[fallthrough]];
371
11.4k
            case FieldType::OLAP_FIELD_TYPE_TIMESTAMPTZ:
372
11.4k
                [[fallthrough]];
373
1.06M
            case FieldType::OLAP_FIELD_TYPE_DATETIMEV2: {
374
1.06M
                StorageField* field = new StorageField(column);
375
1.06M
                field->set_precision(column.precision());
376
1.06M
                field->set_scale(column.frac());
377
1.06M
                return field;
378
11.4k
            }
379
5.39M
            default:
380
5.39M
                return new StorageField(column);
381
9.68M
            }
382
786
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_HLL_UNION:
383
786
            return new HllAggField(column);
384
76
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_BITMAP_UNION:
385
76
            return new BitmapAggField(column);
386
48
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_QUANTILE_UNION:
387
48
            return new QuantileStateAggField(column);
388
208
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_GENERIC:
389
208
            return new AggStateField(column);
390
0
        case FieldAggregationMethod::OLAP_FIELD_AGGREGATION_UNKNOWN:
391
0
            CHECK(false) << ", value column no agg type";
392
0
            return nullptr;
393
9.68M
        }
394
0
        return nullptr;
395
9.68M
    }
396
397
    static StorageField* create_by_type(const FieldType& type) {
398
        TabletColumn column(FieldAggregationMethod::OLAP_FIELD_AGGREGATION_NONE, type);
399
        return create(column);
400
    }
401
};
402
} // namespace doris