Coverage Report

Created: 2026-10-16 08:30

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/root/doris/be/src/runtime/descriptors.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
// This file is copied from
18
// https://github.com/apache/impala/blob/branch-2.9.0/be/src/runtime/descriptors.cc
19
// and modified by Doris
20
21
#include "runtime/descriptors.h"
22
23
#include <fmt/format.h>
24
#include <gen_cpp/Descriptors_types.h>
25
#include <gen_cpp/Types_types.h>
26
#include <gen_cpp/descriptors.pb.h>
27
#include <stddef.h>
28
#include <thrift/protocol/TDebugProtocol.h>
29
30
#include <algorithm>
31
#include <boost/algorithm/string/join.hpp>
32
33
#include "common/exception.h"
34
#include "common/object_pool.h"
35
#include "core/column/column_nothing.h"
36
#include "core/data_type/data_type_array.h"
37
#include "core/data_type/data_type_decimal.h"
38
#include "core/data_type/data_type_factory.hpp"
39
#include "core/data_type/data_type_map.h"
40
#include "core/data_type/data_type_struct.h"
41
#include "core/types.h"
42
#include "exec/common/util.hpp"
43
#include "exprs/aggregate/aggregate_function.h"
44
#include "exprs/function/function_helpers.h"
45
#include "exprs/vexpr.h"
46
#include "util/string_util.h"
47
48
namespace doris {
49
#include "common/compile_check_begin.h"
50
const int RowDescriptor::INVALID_IDX = -1;
51
52
SlotDescriptor::SlotDescriptor(const TSlotDescriptor& tdesc)
53
772k
        : _id(tdesc.id),
54
772k
          _type(DataTypeFactory::instance().create_data_type(tdesc.slotType,
55
772k
                                                             tdesc.nullIndicatorBit != -1)),
56
772k
          _parent(tdesc.parent),
57
772k
          _col_pos(tdesc.columnPos),
58
772k
          _col_name(tdesc.colName),
59
772k
          _col_name_lower_case(to_lower(tdesc.colName)),
60
772k
          _col_unique_id(tdesc.col_unique_id),
61
772k
          _slot_idx(tdesc.slotIdx),
62
772k
          _field_idx(-1),
63
772k
          _is_key(tdesc.is_key),
64
772k
          _column_paths(tdesc.column_paths),
65
772k
          _all_access_paths(tdesc.__isset.all_access_paths ? tdesc.all_access_paths
66
772k
                                                           : TColumnAccessPaths {}),
67
772k
          _predicate_access_paths(tdesc.__isset.predicate_access_paths
68
772k
                                          ? tdesc.predicate_access_paths
69
772k
                                          : TColumnAccessPaths {}),
70
18.4E
          _is_auto_increment(tdesc.__isset.is_auto_increment ? tdesc.is_auto_increment : false),
71
772k
          _col_default_value(tdesc.__isset.col_default_value ? tdesc.col_default_value : "") {
72
772k
    if (tdesc.__isset.virtual_column_expr) {
73
        // Make sure virtual column is valid.
74
104
        if (tdesc.virtual_column_expr.nodes.empty()) {
75
2
            throw doris::Exception(doris::ErrorCode::FATAL_ERROR,
76
2
                                   "Virtual column expr node is empty, col_name: {}, "
77
2
                                   "col_unique_id: {}",
78
2
                                   tdesc.colName, tdesc.col_unique_id);
79
2
        }
80
102
        const auto& node = tdesc.virtual_column_expr.nodes[0];
81
102
        if (node.node_type == TExprNodeType::SLOT_REF) {
82
2
            throw doris::Exception(doris::ErrorCode::FATAL_ERROR,
83
2
                                   "Virtual column expr node is slot ref, col_name: {}, "
84
2
                                   "col_unique_id: {}",
85
2
                                   tdesc.colName, tdesc.col_unique_id);
86
2
        }
87
100
        this->virtual_column_expr = std::make_shared<doris::TExpr>(tdesc.virtual_column_expr);
88
100
    }
89
772k
}
90
91
SlotDescriptor::SlotDescriptor(const PSlotDescriptor& pdesc)
92
95
        : _id(pdesc.id()),
93
95
          _type(DataTypeFactory::instance().create_data_type(pdesc.slot_type(),
94
95
                                                             pdesc.null_indicator_bit() != -1)),
95
95
          _parent(pdesc.parent()),
96
95
          _col_pos(pdesc.column_pos()),
97
95
          _col_name(pdesc.col_name()),
98
95
          _col_name_lower_case(to_lower(pdesc.col_name())),
99
95
          _col_unique_id(pdesc.col_unique_id()),
100
95
          _slot_idx(pdesc.slot_idx()),
101
95
          _field_idx(-1),
102
95
          _is_key(pdesc.is_key()),
103
95
          _column_paths(pdesc.column_paths().begin(), pdesc.column_paths().end()),
104
95
          _is_auto_increment(pdesc.is_auto_increment()) {
105
95
    auto convert_to_thrift_column_access_path = [](const PColumnAccessPath& pb_path) {
106
5
        TColumnAccessPath thrift_path;
107
5
        thrift_path.type = (TAccessPathType::type)pb_path.type();
108
5
        if (pb_path.has_version()) {
109
4
            thrift_path.__set_version(pb_path.version());
110
4
        }
111
5
        if (pb_path.has_data_access_path()) {
112
3
            thrift_path.__isset.data_access_path = true;
113
11
            for (int i = 0; i < pb_path.data_access_path().path_size(); ++i) {
114
8
                thrift_path.data_access_path.path.push_back(pb_path.data_access_path().path(i));
115
8
            }
116
3
        }
117
5
        if (pb_path.has_meta_access_path()) {
118
2
            thrift_path.__isset.meta_access_path = true;
119
8
            for (int i = 0; i < pb_path.meta_access_path().path_size(); ++i) {
120
6
                thrift_path.meta_access_path.path.push_back(pb_path.meta_access_path().path(i));
121
6
            }
122
2
        }
123
5
        return thrift_path;
124
5
    };
125
95
    for (const auto& pb_path : pdesc.all_access_paths()) {
126
4
        _all_access_paths.push_back(convert_to_thrift_column_access_path(pb_path));
127
4
    }
128
95
    for (const auto& pb_path : pdesc.predicate_access_paths()) {
129
1
        _predicate_access_paths.push_back(convert_to_thrift_column_access_path(pb_path));
130
1
    }
131
95
}
132
133
#ifdef BE_TEST
134
SlotDescriptor::SlotDescriptor()
135
492
        : _id(0),
136
492
          _type(nullptr),
137
492
          _parent(0),
138
492
          _col_pos(0),
139
492
          _col_unique_id(0),
140
492
          _slot_idx(0),
141
492
          _field_idx(-1),
142
492
          _is_key(false),
143
492
          _is_auto_increment(false) {}
144
#endif
145
146
119
void SlotDescriptor::to_protobuf(PSlotDescriptor* pslot) const {
147
119
    pslot->set_id(_id);
148
119
    pslot->set_parent(_parent);
149
119
    _type->to_protobuf(pslot->mutable_slot_type());
150
119
    pslot->set_column_pos(_col_pos);
151
119
    pslot->set_byte_offset(0);
152
119
    pslot->set_null_indicator_byte(0);
153
119
    pslot->set_null_indicator_bit(_type->is_nullable() ? 0 : -1);
154
119
    pslot->set_col_name(_col_name);
155
119
    pslot->set_slot_idx(_slot_idx);
156
119
    pslot->set_col_unique_id(_col_unique_id);
157
119
    pslot->set_is_key(_is_key);
158
119
    pslot->set_is_auto_increment(_is_auto_increment);
159
119
    pslot->set_col_type(_type->get_primitive_type());
160
119
    for (const std::string& path : _column_paths) {
161
4
        pslot->add_column_paths(path);
162
4
    }
163
119
    auto convert_to_protobuf_column_access_path = [](const TColumnAccessPath& thrift_path,
164
119
                                                     doris::PColumnAccessPath* pb_path) {
165
5
        pb_path->Clear();
166
5
        pb_path->set_type((PAccessPathType)thrift_path.type); // 使用 reinterpret_cast 进行类型转换
167
5
        if (thrift_path.__isset.version) {
168
4
            pb_path->set_version(thrift_path.version);
169
4
        }
170
5
        if (thrift_path.__isset.data_access_path) {
171
3
            auto* pb_data = pb_path->mutable_data_access_path();
172
3
            pb_data->Clear();
173
8
            for (const auto& s : thrift_path.data_access_path.path) {
174
8
                pb_data->add_path(s);
175
8
            }
176
3
        }
177
5
        if (thrift_path.__isset.meta_access_path) {
178
2
            auto* pb_meta = pb_path->mutable_meta_access_path();
179
2
            pb_meta->Clear();
180
6
            for (const auto& s : thrift_path.meta_access_path.path) {
181
6
                pb_meta->add_path(s);
182
6
            }
183
2
        }
184
5
    };
185
119
    for (const auto& path : _all_access_paths) {
186
4
        auto* pb_path = pslot->add_all_access_paths();
187
4
        convert_to_protobuf_column_access_path(path, pb_path);
188
4
    }
189
119
    for (const auto& path : _predicate_access_paths) {
190
1
        auto* pb_path = pslot->add_predicate_access_paths();
191
1
        convert_to_protobuf_column_access_path(path, pb_path);
192
1
    }
193
119
}
194
195
758k
DataTypePtr SlotDescriptor::get_data_type_ptr() const {
196
758k
    return get_data_type_with_default_argument(type());
197
758k
}
198
199
271k
MutableColumnPtr SlotDescriptor::get_empty_mutable_column() const {
200
271k
    if (this->get_virtual_column_expr() != nullptr) {
201
0
        return ColumnNothing::create(0);
202
0
    }
203
204
271k
    return type()->create_column();
205
271k
}
206
207
389
bool SlotDescriptor::is_nullable() const {
208
389
    return _type->is_nullable();
209
389
}
210
211
863
PrimitiveType SlotDescriptor::col_type() const {
212
863
    return _type->get_primitive_type();
213
863
}
214
215
8
std::string SlotDescriptor::debug_string() const {
216
8
    const bool is_virtual = this->get_virtual_column_expr() != nullptr;
217
8
    return fmt::format(
218
8
            "SlotDescriptor(id={}, type={}, col_name={}, col_unique_id={}, "
219
8
            "is_virtual={})",
220
8
            _id, _type->get_name(), _col_name, _col_unique_id, is_virtual);
221
8
}
222
223
TableDescriptor::TableDescriptor(const TTableDescriptor& tdesc)
224
242
        : _table_type(tdesc.tableType),
225
242
          _name(tdesc.tableName),
226
242
          _database(tdesc.dbName),
227
242
          _table_id(tdesc.id),
228
242
          _num_cols(tdesc.numCols),
229
242
          _num_clustering_cols(tdesc.numClusteringCols) {}
230
231
0
std::string TableDescriptor::debug_string() const {
232
0
    std::stringstream out;
233
0
    out << "#cols=" << _num_cols << " #clustering_cols=" << _num_clustering_cols;
234
0
    return out.str();
235
0
}
236
237
117
OlapTableDescriptor::OlapTableDescriptor(const TTableDescriptor& tdesc) : TableDescriptor(tdesc) {}
238
239
0
std::string OlapTableDescriptor::debug_string() const {
240
0
    std::stringstream out;
241
0
    out << "OlapTable(" << TableDescriptor::debug_string() << ")";
242
0
    return out.str();
243
0
}
244
245
DictionaryTableDescriptor::DictionaryTableDescriptor(const TTableDescriptor& tdesc)
246
0
        : TableDescriptor(tdesc) {}
247
248
0
std::string DictionaryTableDescriptor::debug_string() const {
249
0
    std::stringstream out;
250
0
    out << "Dictionary(" << TableDescriptor::debug_string() << ")";
251
0
    return out.str();
252
0
}
253
254
SchemaTableDescriptor::SchemaTableDescriptor(const TTableDescriptor& tdesc)
255
0
        : TableDescriptor(tdesc), _schema_table_type(tdesc.schemaTable.tableType) {}
256
0
SchemaTableDescriptor::~SchemaTableDescriptor() = default;
257
258
0
std::string SchemaTableDescriptor::debug_string() const {
259
0
    std::stringstream out;
260
0
    out << "SchemaTable(" << TableDescriptor::debug_string() << ")";
261
0
    return out.str();
262
0
}
263
264
BrokerTableDescriptor::BrokerTableDescriptor(const TTableDescriptor& tdesc)
265
0
        : TableDescriptor(tdesc) {}
266
267
0
BrokerTableDescriptor::~BrokerTableDescriptor() = default;
268
269
0
std::string BrokerTableDescriptor::debug_string() const {
270
0
    std::stringstream out;
271
0
    out << "BrokerTable(" << TableDescriptor::debug_string() << ")";
272
0
    return out.str();
273
0
}
274
275
36
HiveTableDescriptor::HiveTableDescriptor(const TTableDescriptor& tdesc) : TableDescriptor(tdesc) {}
276
277
36
HiveTableDescriptor::~HiveTableDescriptor() = default;
278
279
0
std::string HiveTableDescriptor::debug_string() const {
280
0
    std::stringstream out;
281
0
    out << "HiveTable(" << TableDescriptor::debug_string() << ")";
282
0
    return out.str();
283
0
}
284
285
IcebergTableDescriptor::IcebergTableDescriptor(const TTableDescriptor& tdesc)
286
0
        : TableDescriptor(tdesc) {}
287
288
0
IcebergTableDescriptor::~IcebergTableDescriptor() = default;
289
290
0
std::string IcebergTableDescriptor::debug_string() const {
291
0
    std::stringstream out;
292
0
    out << "IcebergTable(" << TableDescriptor::debug_string() << ")";
293
0
    return out.str();
294
0
}
295
296
MaxComputeTableDescriptor::MaxComputeTableDescriptor(const TTableDescriptor& tdesc)
297
0
        : TableDescriptor(tdesc),
298
0
          _region(tdesc.mcTable.region),
299
0
          _project(tdesc.mcTable.project),
300
0
          _table(tdesc.mcTable.table),
301
0
          _odps_url(tdesc.mcTable.odps_url),
302
0
          _tunnel_url(tdesc.mcTable.tunnel_url),
303
0
          _access_key(tdesc.mcTable.access_key),
304
0
          _secret_key(tdesc.mcTable.secret_key),
305
0
          _public_access(tdesc.mcTable.public_access) {
306
0
    if (tdesc.mcTable.__isset.endpoint) {
307
0
        _endpoint = tdesc.mcTable.endpoint;
308
0
    } else {
309
0
        _init_status = Status::InvalidArgument(
310
0
                "fail to init MaxComputeTableDescriptor, missing endpoint.");
311
0
    }
312
313
0
    if (tdesc.mcTable.__isset.quota) {
314
0
        _quota = tdesc.mcTable.quota;
315
0
    } else {
316
0
        _init_status =
317
0
                Status::InvalidArgument("fail to init MaxComputeTableDescriptor, missing quota.");
318
0
    }
319
320
0
    if (tdesc.mcTable.__isset.properties) [[likely]] {
321
0
        _props = tdesc.mcTable.properties;
322
0
    } else {
323
0
        static const std::string MC_ACCESS_KEY = "mc.access_key";
324
0
        static const std::string MC_SECRET_KEY = "mc.secret_key";
325
0
        _props.insert({MC_ACCESS_KEY, _access_key});
326
0
        _props.insert({MC_SECRET_KEY, _secret_key});
327
0
    }
328
0
}
329
330
0
MaxComputeTableDescriptor::~MaxComputeTableDescriptor() = default;
331
332
0
std::string MaxComputeTableDescriptor::debug_string() const {
333
0
    std::stringstream out;
334
0
    out << "MaxComputeTable(" << TableDescriptor::debug_string() << ")";
335
0
    return out.str();
336
0
}
337
338
TrinoConnectorTableDescriptor::TrinoConnectorTableDescriptor(const TTableDescriptor& tdesc)
339
0
        : TableDescriptor(tdesc) {}
340
341
0
TrinoConnectorTableDescriptor::~TrinoConnectorTableDescriptor() = default;
342
343
0
std::string TrinoConnectorTableDescriptor::debug_string() const {
344
0
    std::stringstream out;
345
0
    out << "TrinoConnectorTable(" << TableDescriptor::debug_string() << ")";
346
0
    return out.str();
347
0
}
348
349
0
EsTableDescriptor::EsTableDescriptor(const TTableDescriptor& tdesc) : TableDescriptor(tdesc) {}
350
351
0
EsTableDescriptor::~EsTableDescriptor() = default;
352
353
0
std::string EsTableDescriptor::debug_string() const {
354
0
    std::stringstream out;
355
0
    out << "EsTable(" << TableDescriptor::debug_string() << ")";
356
0
    return out.str();
357
0
}
358
359
MySQLTableDescriptor::MySQLTableDescriptor(const TTableDescriptor& tdesc)
360
89
        : TableDescriptor(tdesc),
361
89
          _mysql_db(tdesc.mysqlTable.db),
362
89
          _mysql_table(tdesc.mysqlTable.table),
363
89
          _host(tdesc.mysqlTable.host),
364
89
          _port(tdesc.mysqlTable.port),
365
89
          _user(tdesc.mysqlTable.user),
366
89
          _passwd(tdesc.mysqlTable.passwd),
367
89
          _charset(tdesc.mysqlTable.charset) {}
368
369
0
std::string MySQLTableDescriptor::debug_string() const {
370
0
    std::stringstream out;
371
0
    out << "MySQLTable(" << TableDescriptor::debug_string() << " _db" << _mysql_db
372
0
        << " table=" << _mysql_table << " host=" << _host << " port=" << _port << " user=" << _user
373
0
        << " passwd=" << _passwd << " charset=" << _charset;
374
0
    return out.str();
375
0
}
376
377
JdbcTableDescriptor::JdbcTableDescriptor(const TTableDescriptor& tdesc)
378
0
        : TableDescriptor(tdesc),
379
0
          _jdbc_catalog_id(tdesc.jdbcTable.catalog_id),
380
0
          _jdbc_resource_name(tdesc.jdbcTable.jdbc_resource_name),
381
0
          _jdbc_driver_url(tdesc.jdbcTable.jdbc_driver_url),
382
0
          _jdbc_driver_class(tdesc.jdbcTable.jdbc_driver_class),
383
0
          _jdbc_driver_checksum(tdesc.jdbcTable.jdbc_driver_checksum),
384
0
          _jdbc_url(tdesc.jdbcTable.jdbc_url),
385
0
          _jdbc_table_name(tdesc.jdbcTable.jdbc_table_name),
386
0
          _jdbc_user(tdesc.jdbcTable.jdbc_user),
387
0
          _jdbc_passwd(tdesc.jdbcTable.jdbc_password),
388
0
          _connection_pool_min_size(tdesc.jdbcTable.connection_pool_min_size),
389
0
          _connection_pool_max_size(tdesc.jdbcTable.connection_pool_max_size),
390
0
          _connection_pool_max_wait_time(tdesc.jdbcTable.connection_pool_max_wait_time),
391
0
          _connection_pool_max_life_time(tdesc.jdbcTable.connection_pool_max_life_time),
392
0
          _connection_pool_keep_alive(tdesc.jdbcTable.connection_pool_keep_alive) {}
393
394
0
std::string JdbcTableDescriptor::debug_string() const {
395
0
    fmt::memory_buffer buf;
396
0
    fmt::format_to(
397
0
            buf,
398
0
            "JDBCTable({} ,_jdbc_catalog_id = {}, _jdbc_resource_name={} ,_jdbc_driver_url={} "
399
0
            ",_jdbc_driver_class={} ,_jdbc_driver_checksum={} ,_jdbc_url={} "
400
0
            ",_jdbc_table_name={} ,_jdbc_user={} ,_jdbc_passwd={} ,_connection_pool_min_size={} "
401
0
            ",_connection_pool_max_size={} ,_connection_pool_max_wait_time={} "
402
0
            ",_connection_pool_max_life_time={} ,_connection_pool_keep_alive={})",
403
0
            TableDescriptor::debug_string(), _jdbc_catalog_id, _jdbc_resource_name,
404
0
            _jdbc_driver_url, _jdbc_driver_class, _jdbc_driver_checksum, _jdbc_url,
405
0
            _jdbc_table_name, _jdbc_user, _jdbc_passwd, _connection_pool_min_size,
406
0
            _connection_pool_max_size, _connection_pool_max_wait_time,
407
0
            _connection_pool_max_life_time, _connection_pool_keep_alive);
408
0
    return fmt::to_string(buf);
409
0
}
410
411
RemoteDorisTableDescriptor::RemoteDorisTableDescriptor(const TTableDescriptor& tdesc)
412
0
        : TableDescriptor(tdesc) {}
413
414
0
RemoteDorisTableDescriptor::~RemoteDorisTableDescriptor() = default;
415
416
0
std::string RemoteDorisTableDescriptor::debug_string() const {
417
0
    std::stringstream out;
418
0
    out << "RemoteDorisTable(" << TableDescriptor::debug_string() << ")";
419
0
    return out.str();
420
0
}
421
422
TupleDescriptor::TupleDescriptor(const TTupleDescriptor& tdesc, bool own_slots)
423
292k
        : _id(tdesc.id),
424
292k
          _num_materialized_slots(0),
425
292k
          _has_varlen_slots(false),
426
292k
          _own_slots(own_slots) {}
427
428
TupleDescriptor::TupleDescriptor(const PTupleDescriptor& pdesc, bool own_slots)
429
22
        : _id(pdesc.id()),
430
22
          _num_materialized_slots(0),
431
22
          _has_varlen_slots(false),
432
22
          _own_slots(own_slots) {}
433
434
768k
void TupleDescriptor::add_slot(SlotDescriptor* slot) {
435
768k
    _slots.push_back(slot);
436
768k
    ++_num_materialized_slots;
437
438
768k
    if (is_complex_type(slot->type()->get_primitive_type()) ||
439
768k
        is_var_len_object(slot->type()->get_primitive_type()) ||
440
768k
        is_string_type(slot->type()->get_primitive_type())) {
441
129k
        _has_varlen_slots = true;
442
129k
    }
443
768k
}
444
445
19
void TupleDescriptor::to_protobuf(PTupleDescriptor* ptuple) const {
446
19
    ptuple->Clear();
447
19
    ptuple->set_id(_id);
448
    // Useless not set
449
19
    ptuple->set_byte_size(0);
450
19
    ptuple->set_table_id(-1);
451
19
    ptuple->set_num_null_bytes(0);
452
19
}
453
454
2
std::string TupleDescriptor::debug_string() const {
455
2
    std::stringstream out;
456
2
    out << "Tuple(id=" << _id;
457
2
    if (_table_desc != nullptr) {
458
        //out << " " << _table_desc->debug_string();
459
0
    }
460
461
2
    out << " slots=[";
462
8
    for (size_t i = 0; i < _slots.size(); ++i) {
463
6
        if (i > 0) {
464
4
            out << ", ";
465
4
        }
466
6
        out << _slots[i]->debug_string();
467
6
    }
468
469
2
    out << "]";
470
2
    out << " has_varlen_slots=" << _has_varlen_slots;
471
2
    out << ")";
472
2
    return out.str();
473
2
}
474
475
RowDescriptor::RowDescriptor(const DescriptorTbl& desc_tbl, const std::vector<TTupleId>& row_tuples,
476
                             const std::vector<bool>& nullable_tuples)
477
360k
        : _tuple_idx_nullable_map(nullable_tuples) {
478
360k
    DCHECK(nullable_tuples.size() == row_tuples.size())
479
0
            << "nullable_tuples size " << nullable_tuples.size() << " != row_tuples size "
480
0
            << row_tuples.size();
481
360k
    DCHECK_GT(row_tuples.size(), 0);
482
360k
    _num_materialized_slots = 0;
483
360k
    _num_slots = 0;
484
485
433k
    for (int row_tuple : row_tuples) {
486
433k
        TupleDescriptor* tupleDesc = desc_tbl.get_tuple_descriptor(row_tuple);
487
433k
        _num_materialized_slots += tupleDesc->num_materialized_slots();
488
433k
        _num_slots += tupleDesc->slots().size();
489
433k
        _tuple_desc_map.push_back(tupleDesc);
490
433k
        DCHECK(_tuple_desc_map.back() != nullptr);
491
433k
    }
492
493
360k
    init_tuple_idx_map();
494
360k
    init_has_varlen_slots();
495
360k
}
496
497
RowDescriptor::RowDescriptor(TupleDescriptor* tuple_desc, bool is_nullable)
498
4.23k
        : _tuple_desc_map(1, tuple_desc), _tuple_idx_nullable_map(1, is_nullable) {
499
4.23k
    init_tuple_idx_map();
500
4.23k
    init_has_varlen_slots();
501
4.23k
    _num_slots = static_cast<int32_t>(tuple_desc->slots().size());
502
4.23k
}
503
504
0
RowDescriptor::RowDescriptor(const RowDescriptor& lhs_row_desc, const RowDescriptor& rhs_row_desc) {
505
0
    _tuple_desc_map.insert(_tuple_desc_map.end(), lhs_row_desc._tuple_desc_map.begin(),
506
0
                           lhs_row_desc._tuple_desc_map.end());
507
0
    _tuple_desc_map.insert(_tuple_desc_map.end(), rhs_row_desc._tuple_desc_map.begin(),
508
0
                           rhs_row_desc._tuple_desc_map.end());
509
0
    _tuple_idx_nullable_map.insert(_tuple_idx_nullable_map.end(),
510
0
                                   lhs_row_desc._tuple_idx_nullable_map.begin(),
511
0
                                   lhs_row_desc._tuple_idx_nullable_map.end());
512
0
    _tuple_idx_nullable_map.insert(_tuple_idx_nullable_map.end(),
513
0
                                   rhs_row_desc._tuple_idx_nullable_map.begin(),
514
0
                                   rhs_row_desc._tuple_idx_nullable_map.end());
515
0
    init_tuple_idx_map();
516
0
    init_has_varlen_slots();
517
518
0
    _num_slots = lhs_row_desc.num_slots() + rhs_row_desc.num_slots();
519
0
}
520
521
365k
void RowDescriptor::init_tuple_idx_map() {
522
    // find max id
523
365k
    TupleId max_id = 0;
524
437k
    for (auto& i : _tuple_desc_map) {
525
437k
        max_id = std::max(i->id(), max_id);
526
437k
    }
527
528
365k
    _tuple_idx_map.resize(max_id + 1, INVALID_IDX);
529
802k
    for (int i = 0; i < _tuple_desc_map.size(); ++i) {
530
437k
        _tuple_idx_map[_tuple_desc_map[i]->id()] = i;
531
437k
    }
532
365k
}
533
534
365k
void RowDescriptor::init_has_varlen_slots() {
535
365k
    _has_varlen_slots = false;
536
415k
    for (auto& i : _tuple_desc_map) {
537
415k
        if (i->has_varlen_slots()) {
538
112k
            _has_varlen_slots = true;
539
112k
            break;
540
112k
        }
541
415k
    }
542
365k
}
543
544
1
int RowDescriptor::get_tuple_idx(TupleId id) const {
545
    // comment CHECK temporarily to make fuzzy test run smoothly
546
    // DCHECK_LT(id, _tuple_idx_map.size()) << "RowDescriptor: " << debug_string();
547
1
    if (_tuple_idx_map.size() <= id) {
548
0
        return RowDescriptor::INVALID_IDX;
549
0
    }
550
1
    return _tuple_idx_map[id];
551
1
}
552
553
0
void RowDescriptor::to_thrift(std::vector<TTupleId>* row_tuple_ids) {
554
0
    row_tuple_ids->clear();
555
556
0
    for (auto& i : _tuple_desc_map) {
557
0
        row_tuple_ids->push_back(i->id());
558
0
    }
559
0
}
560
561
void RowDescriptor::to_protobuf(
562
0
        google::protobuf::RepeatedField<google::protobuf::int32>* row_tuple_ids) const {
563
0
    row_tuple_ids->Clear();
564
0
    for (auto* desc : _tuple_desc_map) {
565
0
        row_tuple_ids->Add(desc->id());
566
0
    }
567
0
}
568
569
0
bool RowDescriptor::is_prefix_of(const RowDescriptor& other_desc) const {
570
0
    if (_tuple_desc_map.size() > other_desc._tuple_desc_map.size()) {
571
0
        return false;
572
0
    }
573
574
0
    for (int i = 0; i < _tuple_desc_map.size(); ++i) {
575
        // pointer comparison okay, descriptors are unique
576
0
        if (_tuple_desc_map[i] != other_desc._tuple_desc_map[i]) {
577
0
            return false;
578
0
        }
579
0
    }
580
581
0
    return true;
582
0
}
583
584
0
bool RowDescriptor::equals(const RowDescriptor& other_desc) const {
585
0
    if (_tuple_desc_map.size() != other_desc._tuple_desc_map.size()) {
586
0
        return false;
587
0
    }
588
589
0
    for (int i = 0; i < _tuple_desc_map.size(); ++i) {
590
        // pointer comparison okay, descriptors are unique
591
0
        if (_tuple_desc_map[i] != other_desc._tuple_desc_map[i]) {
592
0
            return false;
593
0
        }
594
0
    }
595
596
0
    return true;
597
0
}
598
599
0
std::string RowDescriptor::debug_string() const {
600
0
    std::stringstream ss;
601
602
0
    ss << "tuple_desc_map: [";
603
0
    for (int i = 0; i < _tuple_desc_map.size(); ++i) {
604
0
        ss << _tuple_desc_map[i]->debug_string();
605
0
        if (i != _tuple_desc_map.size() - 1) {
606
0
            ss << ", ";
607
0
        }
608
0
    }
609
0
    ss << "] ";
610
611
0
    ss << "tuple_id_map: [";
612
0
    for (int i = 0; i < _tuple_idx_map.size(); ++i) {
613
0
        ss << _tuple_idx_map[i];
614
0
        if (i != _tuple_idx_map.size() - 1) {
615
0
            ss << ", ";
616
0
        }
617
0
    }
618
0
    ss << "] ";
619
620
0
    ss << "tuple_is_nullable: [";
621
0
    for (int i = 0; i < _tuple_idx_nullable_map.size(); ++i) {
622
0
        ss << _tuple_idx_nullable_map[i];
623
0
        if (i != _tuple_idx_nullable_map.size() - 1) {
624
0
            ss << ", ";
625
0
        }
626
0
    }
627
0
    ss << "] ";
628
629
0
    return ss.str();
630
0
}
631
632
258k
int RowDescriptor::get_column_id(int slot_id) const {
633
258k
    int column_id_counter = 0;
634
258k
    for (auto* const tuple_desc : _tuple_desc_map) {
635
420k
        for (auto* const slot : tuple_desc->slots()) {
636
420k
            if (slot->id() == slot_id) {
637
258k
                return column_id_counter;
638
258k
            }
639
162k
            column_id_counter++;
640
162k
        }
641
258k
    }
642
98
    return -1;
643
258k
}
644
645
Status DescriptorTbl::create(ObjectPool* pool, const TDescriptorTable& thrift_tbl,
646
76.6k
                             DescriptorTbl** tbl) {
647
76.6k
    *tbl = pool->add(new DescriptorTbl());
648
649
    // deserialize table descriptors first, they are being referenced by tuple descriptors
650
76.6k
    for (const auto& tdesc : thrift_tbl.tableDescriptors) {
651
242
        TableDescriptor* desc = nullptr;
652
653
242
        switch (tdesc.tableType) {
654
89
        case TTableType::MYSQL_TABLE:
655
89
            desc = pool->add(new MySQLTableDescriptor(tdesc));
656
89
            break;
657
658
117
        case TTableType::OLAP_TABLE:
659
117
            desc = pool->add(new OlapTableDescriptor(tdesc));
660
117
            break;
661
662
0
        case TTableType::SCHEMA_TABLE:
663
0
            desc = pool->add(new SchemaTableDescriptor(tdesc));
664
0
            break;
665
0
        case TTableType::BROKER_TABLE:
666
0
            desc = pool->add(new BrokerTableDescriptor(tdesc));
667
0
            break;
668
0
        case TTableType::ES_TABLE:
669
0
            desc = pool->add(new EsTableDescriptor(tdesc));
670
0
            break;
671
36
        case TTableType::HIVE_TABLE:
672
36
            desc = pool->add(new HiveTableDescriptor(tdesc));
673
36
            break;
674
0
        case TTableType::ICEBERG_TABLE:
675
0
            desc = pool->add(new IcebergTableDescriptor(tdesc));
676
0
            break;
677
0
        case TTableType::JDBC_TABLE:
678
0
            desc = pool->add(new JdbcTableDescriptor(tdesc));
679
0
            break;
680
0
        case TTableType::MAX_COMPUTE_TABLE:
681
0
            desc = pool->add(new MaxComputeTableDescriptor(tdesc));
682
0
            break;
683
0
        case TTableType::TRINO_CONNECTOR_TABLE:
684
0
            desc = pool->add(new TrinoConnectorTableDescriptor(tdesc));
685
0
            break;
686
0
        case TTableType::DICTIONARY_TABLE:
687
0
            desc = pool->add(new DictionaryTableDescriptor(tdesc));
688
0
            break;
689
0
        case TTableType::REMOTE_DORIS_TABLE:
690
0
            desc = pool->add(new RemoteDorisTableDescriptor(tdesc));
691
0
            break;
692
0
        default:
693
0
            DCHECK(false) << "invalid table type: " << tdesc.tableType;
694
242
        }
695
696
242
        (*tbl)->_tbl_desc_map[static_cast<int32_t>(tdesc.id)] = desc;
697
242
    }
698
699
292k
    for (const auto& tdesc : thrift_tbl.tupleDescriptors) {
700
292k
        TupleDescriptor* desc = pool->add(new TupleDescriptor(tdesc));
701
702
        // fix up table pointer
703
292k
        if (tdesc.__isset.tableId) {
704
218
            desc->_table_desc = (*tbl)->get_table_descriptor(static_cast<int32_t>(tdesc.tableId));
705
218
            DCHECK(desc->_table_desc != nullptr);
706
218
        }
707
708
292k
        (*tbl)->_tuple_desc_map[tdesc.id] = desc;
709
292k
        (*tbl)->_row_tuples.emplace_back(tdesc.id);
710
292k
    }
711
712
768k
    for (const auto& tdesc : thrift_tbl.slotDescriptors) {
713
768k
        SlotDescriptor* slot_d = pool->add(new SlotDescriptor(tdesc));
714
768k
        (*tbl)->_slot_desc_map[tdesc.id] = slot_d;
715
716
        // link to parent
717
768k
        auto entry = (*tbl)->_tuple_desc_map.find(tdesc.parent);
718
719
768k
        if (entry == (*tbl)->_tuple_desc_map.end()) {
720
0
            return Status::InternalError("unknown tid in slot descriptor msg");
721
0
        }
722
768k
        entry->second->add_slot(slot_d);
723
768k
    }
724
725
76.6k
    return Status::OK();
726
76.6k
}
727
728
218
TableDescriptor* DescriptorTbl::get_table_descriptor(TableId id) const {
729
    // TODO: is there some boost function to do exactly this?
730
218
    auto i = _tbl_desc_map.find(id);
731
732
218
    if (i == _tbl_desc_map.end()) {
733
0
        return nullptr;
734
218
    } else {
735
218
        return i->second;
736
218
    }
737
218
}
738
739
639k
TupleDescriptor* DescriptorTbl::get_tuple_descriptor(TupleId id) const {
740
    // TODO: is there some boost function to do exactly this?
741
639k
    auto i = _tuple_desc_map.find(id);
742
743
639k
    if (i == _tuple_desc_map.end()) {
744
8
        return nullptr;
745
639k
    } else {
746
639k
        return i->second;
747
639k
    }
748
639k
}
749
750
258k
SlotDescriptor* DescriptorTbl::get_slot_descriptor(SlotId id) const {
751
    // TODO: is there some boost function to do exactly this?
752
258k
    auto i = _slot_desc_map.find(id);
753
754
258k
    if (i == _slot_desc_map.end()) {
755
0
        return nullptr;
756
258k
    } else {
757
258k
        return i->second;
758
258k
    }
759
258k
}
760
761
0
std::string DescriptorTbl::debug_string() const {
762
0
    std::stringstream out;
763
0
    out << "tuples:\n";
764
765
0
    for (auto i : _tuple_desc_map) {
766
0
        out << i.second->debug_string() << '\n';
767
0
    }
768
769
0
    return out.str();
770
0
}
771
#include "common/compile_check_end.h"
772
} // namespace doris