/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 "util/string_util.h" |
36 | | #include "vec/aggregate_functions/aggregate_function.h" |
37 | | #include "vec/columns/column_nothing.h" |
38 | | #include "vec/core/types.h" |
39 | | #include "vec/data_types/data_type_array.h" |
40 | | #include "vec/data_types/data_type_decimal.h" |
41 | | #include "vec/data_types/data_type_factory.hpp" |
42 | | #include "vec/data_types/data_type_map.h" |
43 | | #include "vec/data_types/data_type_struct.h" |
44 | | #include "vec/exprs/vexpr.h" |
45 | | #include "vec/functions/function_helpers.h" |
46 | | #include "vec/utils/util.hpp" |
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 | 767k | : _id(tdesc.id), |
54 | 767k | _type(vectorized::DataTypeFactory::instance().create_data_type( |
55 | 767k | tdesc.slotType, tdesc.nullIndicatorBit != -1)), |
56 | 767k | _parent(tdesc.parent), |
57 | 767k | _col_pos(tdesc.columnPos), |
58 | 767k | _col_name(tdesc.colName), |
59 | 767k | _col_name_lower_case(to_lower(tdesc.colName)), |
60 | 767k | _col_unique_id(tdesc.col_unique_id), |
61 | 767k | _slot_idx(tdesc.slotIdx), |
62 | 767k | _field_idx(-1), |
63 | 767k | _is_key(tdesc.is_key), |
64 | 767k | _column_paths(tdesc.column_paths), |
65 | 767k | _is_auto_increment(tdesc.__isset.is_auto_increment ? tdesc.is_auto_increment : false), |
66 | 767k | _col_default_value(tdesc.__isset.col_default_value ? tdesc.col_default_value : "") { |
67 | 767k | if (tdesc.__isset.virtual_column_expr) { |
68 | | // Make sure virtual column is valid. |
69 | 97 | if (tdesc.virtual_column_expr.nodes.empty()) { |
70 | 2 | throw doris::Exception(doris::ErrorCode::FATAL_ERROR, |
71 | 2 | "Virtual column expr node is empty, col_name: {}, " |
72 | 2 | "col_unique_id: {}", |
73 | 2 | tdesc.colName, tdesc.col_unique_id); |
74 | 2 | } |
75 | 95 | const auto& node = tdesc.virtual_column_expr.nodes[0]; |
76 | 95 | if (node.node_type == TExprNodeType::SLOT_REF) { |
77 | 2 | throw doris::Exception(doris::ErrorCode::FATAL_ERROR, |
78 | 2 | "Virtual column expr node is slot ref, col_name: {}, " |
79 | 2 | "col_unique_id: {}", |
80 | 2 | tdesc.colName, tdesc.col_unique_id); |
81 | 2 | } |
82 | 93 | this->virtual_column_expr = std::make_shared<doris::TExpr>(tdesc.virtual_column_expr); |
83 | 93 | } |
84 | 767k | } |
85 | | |
86 | | SlotDescriptor::SlotDescriptor(const PSlotDescriptor& pdesc) |
87 | 84 | : _id(pdesc.id()), |
88 | 84 | _type(vectorized::DataTypeFactory::instance().create_data_type( |
89 | 84 | pdesc.slot_type(), pdesc.null_indicator_bit() != -1)), |
90 | 84 | _parent(pdesc.parent()), |
91 | 84 | _col_pos(pdesc.column_pos()), |
92 | 84 | _col_name(pdesc.col_name()), |
93 | 84 | _col_name_lower_case(to_lower(pdesc.col_name())), |
94 | 84 | _col_unique_id(pdesc.col_unique_id()), |
95 | 84 | _slot_idx(pdesc.slot_idx()), |
96 | 84 | _field_idx(-1), |
97 | 84 | _is_key(pdesc.is_key()), |
98 | 84 | _column_paths(pdesc.column_paths().begin(), pdesc.column_paths().end()), |
99 | 84 | _is_auto_increment(pdesc.is_auto_increment()) {} |
100 | | |
101 | | #ifdef BE_TEST |
102 | | SlotDescriptor::SlotDescriptor() |
103 | 299 | : _id(0), |
104 | 299 | _type(nullptr), |
105 | 299 | _parent(0), |
106 | 299 | _col_pos(0), |
107 | 299 | _col_unique_id(0), |
108 | 299 | _slot_idx(0), |
109 | 299 | _field_idx(-1), |
110 | 299 | _is_key(false), |
111 | 299 | _is_auto_increment(false) {} |
112 | | #endif |
113 | | |
114 | 96 | void SlotDescriptor::to_protobuf(PSlotDescriptor* pslot) const { |
115 | 96 | pslot->set_id(_id); |
116 | 96 | pslot->set_parent(_parent); |
117 | 96 | _type->to_protobuf(pslot->mutable_slot_type()); |
118 | 96 | pslot->set_column_pos(_col_pos); |
119 | 96 | pslot->set_byte_offset(0); |
120 | 96 | pslot->set_null_indicator_byte(0); |
121 | 96 | pslot->set_null_indicator_bit(_type->is_nullable() ? 0 : -1); |
122 | 96 | pslot->set_col_name(_col_name); |
123 | 96 | pslot->set_slot_idx(_slot_idx); |
124 | 96 | pslot->set_col_unique_id(_col_unique_id); |
125 | 96 | pslot->set_is_key(_is_key); |
126 | 96 | pslot->set_is_auto_increment(_is_auto_increment); |
127 | 96 | pslot->set_col_type(_type->get_primitive_type()); |
128 | 96 | for (const std::string& path : _column_paths) { |
129 | 0 | pslot->add_column_paths(path); |
130 | 0 | } |
131 | 96 | } |
132 | | |
133 | 748k | vectorized::DataTypePtr SlotDescriptor::get_data_type_ptr() const { |
134 | 748k | return vectorized::get_data_type_with_default_argument(type()); |
135 | 748k | } |
136 | | |
137 | 271k | vectorized::MutableColumnPtr SlotDescriptor::get_empty_mutable_column() const { |
138 | 271k | if (this->get_virtual_column_expr() != nullptr) { |
139 | 0 | return vectorized::ColumnNothing::create(0); |
140 | 0 | } |
141 | | |
142 | 271k | return type()->create_column(); |
143 | 271k | } |
144 | | |
145 | 363 | bool SlotDescriptor::is_nullable() const { |
146 | 363 | return _type->is_nullable(); |
147 | 363 | } |
148 | | |
149 | 180 | PrimitiveType SlotDescriptor::col_type() const { |
150 | 180 | return _type->get_primitive_type(); |
151 | 180 | } |
152 | | |
153 | 8 | std::string SlotDescriptor::debug_string() const { |
154 | 8 | const bool is_virtual = this->get_virtual_column_expr() != nullptr; |
155 | 8 | return fmt::format( |
156 | 8 | "SlotDescriptor(id={}, type={}, col_name={}, col_unique_id={}, " |
157 | 8 | "is_virtual={})", |
158 | 8 | _id, _type->get_name(), _col_name, _col_unique_id, is_virtual); |
159 | 8 | } |
160 | | |
161 | | TableDescriptor::TableDescriptor(const TTableDescriptor& tdesc) |
162 | 142 | : _table_type(tdesc.tableType), |
163 | 142 | _name(tdesc.tableName), |
164 | 142 | _database(tdesc.dbName), |
165 | 142 | _table_id(tdesc.id), |
166 | 142 | _num_cols(tdesc.numCols), |
167 | 142 | _num_clustering_cols(tdesc.numClusteringCols) {} |
168 | | |
169 | 0 | std::string TableDescriptor::debug_string() const { |
170 | 0 | std::stringstream out; |
171 | 0 | out << "#cols=" << _num_cols << " #clustering_cols=" << _num_clustering_cols; |
172 | 0 | return out.str(); |
173 | 0 | } |
174 | | |
175 | 58 | OlapTableDescriptor::OlapTableDescriptor(const TTableDescriptor& tdesc) : TableDescriptor(tdesc) {} |
176 | | |
177 | 0 | std::string OlapTableDescriptor::debug_string() const { |
178 | 0 | std::stringstream out; |
179 | 0 | out << "OlapTable(" << TableDescriptor::debug_string() << ")"; |
180 | 0 | return out.str(); |
181 | 0 | } |
182 | | |
183 | | DictionaryTableDescriptor::DictionaryTableDescriptor(const TTableDescriptor& tdesc) |
184 | 0 | : TableDescriptor(tdesc) {} |
185 | | |
186 | 0 | std::string DictionaryTableDescriptor::debug_string() const { |
187 | 0 | std::stringstream out; |
188 | 0 | out << "Dictionary(" << TableDescriptor::debug_string() << ")"; |
189 | 0 | return out.str(); |
190 | 0 | } |
191 | | |
192 | | SchemaTableDescriptor::SchemaTableDescriptor(const TTableDescriptor& tdesc) |
193 | 0 | : TableDescriptor(tdesc), _schema_table_type(tdesc.schemaTable.tableType) {} |
194 | 0 | SchemaTableDescriptor::~SchemaTableDescriptor() = default; |
195 | | |
196 | 0 | std::string SchemaTableDescriptor::debug_string() const { |
197 | 0 | std::stringstream out; |
198 | 0 | out << "SchemaTable(" << TableDescriptor::debug_string() << ")"; |
199 | 0 | return out.str(); |
200 | 0 | } |
201 | | |
202 | | BrokerTableDescriptor::BrokerTableDescriptor(const TTableDescriptor& tdesc) |
203 | 0 | : TableDescriptor(tdesc) {} |
204 | | |
205 | 0 | BrokerTableDescriptor::~BrokerTableDescriptor() = default; |
206 | | |
207 | 0 | std::string BrokerTableDescriptor::debug_string() const { |
208 | 0 | std::stringstream out; |
209 | 0 | out << "BrokerTable(" << TableDescriptor::debug_string() << ")"; |
210 | 0 | return out.str(); |
211 | 0 | } |
212 | | |
213 | 0 | HiveTableDescriptor::HiveTableDescriptor(const TTableDescriptor& tdesc) : TableDescriptor(tdesc) {} |
214 | | |
215 | 0 | HiveTableDescriptor::~HiveTableDescriptor() = default; |
216 | | |
217 | 0 | std::string HiveTableDescriptor::debug_string() const { |
218 | 0 | std::stringstream out; |
219 | 0 | out << "HiveTable(" << TableDescriptor::debug_string() << ")"; |
220 | 0 | return out.str(); |
221 | 0 | } |
222 | | |
223 | | IcebergTableDescriptor::IcebergTableDescriptor(const TTableDescriptor& tdesc) |
224 | 0 | : TableDescriptor(tdesc) {} |
225 | | |
226 | 0 | IcebergTableDescriptor::~IcebergTableDescriptor() = default; |
227 | | |
228 | 0 | std::string IcebergTableDescriptor::debug_string() const { |
229 | 0 | std::stringstream out; |
230 | 0 | out << "IcebergTable(" << TableDescriptor::debug_string() << ")"; |
231 | 0 | return out.str(); |
232 | 0 | } |
233 | | |
234 | | MaxComputeTableDescriptor::MaxComputeTableDescriptor(const TTableDescriptor& tdesc) |
235 | 0 | : TableDescriptor(tdesc), |
236 | 0 | _region(tdesc.mcTable.region), |
237 | 0 | _project(tdesc.mcTable.project), |
238 | 0 | _table(tdesc.mcTable.table), |
239 | 0 | _odps_url(tdesc.mcTable.odps_url), |
240 | 0 | _tunnel_url(tdesc.mcTable.tunnel_url), |
241 | 0 | _access_key(tdesc.mcTable.access_key), |
242 | 0 | _secret_key(tdesc.mcTable.secret_key), |
243 | 0 | _public_access(tdesc.mcTable.public_access) { |
244 | 0 | if (tdesc.mcTable.__isset.endpoint) { |
245 | 0 | _endpoint = tdesc.mcTable.endpoint; |
246 | 0 | } else { |
247 | 0 | _init_status = Status::InvalidArgument( |
248 | 0 | "fail to init MaxComputeTableDescriptor, missing endpoint."); |
249 | 0 | } |
250 | |
|
251 | 0 | if (tdesc.mcTable.__isset.quota) { |
252 | 0 | _quota = tdesc.mcTable.quota; |
253 | 0 | } else { |
254 | 0 | _init_status = |
255 | 0 | Status::InvalidArgument("fail to init MaxComputeTableDescriptor, missing quota."); |
256 | 0 | } |
257 | 0 | } |
258 | | |
259 | 0 | MaxComputeTableDescriptor::~MaxComputeTableDescriptor() = default; |
260 | | |
261 | 0 | std::string MaxComputeTableDescriptor::debug_string() const { |
262 | 0 | std::stringstream out; |
263 | 0 | out << "MaxComputeTable(" << TableDescriptor::debug_string() << ")"; |
264 | 0 | return out.str(); |
265 | 0 | } |
266 | | |
267 | | TrinoConnectorTableDescriptor::TrinoConnectorTableDescriptor(const TTableDescriptor& tdesc) |
268 | 0 | : TableDescriptor(tdesc) {} |
269 | | |
270 | 0 | TrinoConnectorTableDescriptor::~TrinoConnectorTableDescriptor() = default; |
271 | | |
272 | 0 | std::string TrinoConnectorTableDescriptor::debug_string() const { |
273 | 0 | std::stringstream out; |
274 | 0 | out << "TrinoConnectorTable(" << TableDescriptor::debug_string() << ")"; |
275 | 0 | return out.str(); |
276 | 0 | } |
277 | | |
278 | 0 | EsTableDescriptor::EsTableDescriptor(const TTableDescriptor& tdesc) : TableDescriptor(tdesc) {} |
279 | | |
280 | 0 | EsTableDescriptor::~EsTableDescriptor() = default; |
281 | | |
282 | 0 | std::string EsTableDescriptor::debug_string() const { |
283 | 0 | std::stringstream out; |
284 | 0 | out << "EsTable(" << TableDescriptor::debug_string() << ")"; |
285 | 0 | return out.str(); |
286 | 0 | } |
287 | | |
288 | | MySQLTableDescriptor::MySQLTableDescriptor(const TTableDescriptor& tdesc) |
289 | 84 | : TableDescriptor(tdesc), |
290 | 84 | _mysql_db(tdesc.mysqlTable.db), |
291 | 84 | _mysql_table(tdesc.mysqlTable.table), |
292 | 84 | _host(tdesc.mysqlTable.host), |
293 | 84 | _port(tdesc.mysqlTable.port), |
294 | 84 | _user(tdesc.mysqlTable.user), |
295 | 84 | _passwd(tdesc.mysqlTable.passwd), |
296 | 84 | _charset(tdesc.mysqlTable.charset) {} |
297 | | |
298 | 0 | std::string MySQLTableDescriptor::debug_string() const { |
299 | 0 | std::stringstream out; |
300 | 0 | out << "MySQLTable(" << TableDescriptor::debug_string() << " _db" << _mysql_db |
301 | 0 | << " table=" << _mysql_table << " host=" << _host << " port=" << _port << " user=" << _user |
302 | 0 | << " passwd=" << _passwd << " charset=" << _charset; |
303 | 0 | return out.str(); |
304 | 0 | } |
305 | | |
306 | | JdbcTableDescriptor::JdbcTableDescriptor(const TTableDescriptor& tdesc) |
307 | 0 | : TableDescriptor(tdesc), |
308 | 0 | _jdbc_catalog_id(tdesc.jdbcTable.catalog_id), |
309 | 0 | _jdbc_resource_name(tdesc.jdbcTable.jdbc_resource_name), |
310 | 0 | _jdbc_driver_url(tdesc.jdbcTable.jdbc_driver_url), |
311 | 0 | _jdbc_driver_class(tdesc.jdbcTable.jdbc_driver_class), |
312 | 0 | _jdbc_driver_checksum(tdesc.jdbcTable.jdbc_driver_checksum), |
313 | 0 | _jdbc_url(tdesc.jdbcTable.jdbc_url), |
314 | 0 | _jdbc_table_name(tdesc.jdbcTable.jdbc_table_name), |
315 | 0 | _jdbc_user(tdesc.jdbcTable.jdbc_user), |
316 | 0 | _jdbc_passwd(tdesc.jdbcTable.jdbc_password), |
317 | 0 | _connection_pool_min_size(tdesc.jdbcTable.connection_pool_min_size), |
318 | 0 | _connection_pool_max_size(tdesc.jdbcTable.connection_pool_max_size), |
319 | 0 | _connection_pool_max_wait_time(tdesc.jdbcTable.connection_pool_max_wait_time), |
320 | 0 | _connection_pool_max_life_time(tdesc.jdbcTable.connection_pool_max_life_time), |
321 | 0 | _connection_pool_keep_alive(tdesc.jdbcTable.connection_pool_keep_alive) {} |
322 | | |
323 | 0 | std::string JdbcTableDescriptor::debug_string() const { |
324 | 0 | fmt::memory_buffer buf; |
325 | 0 | fmt::format_to( |
326 | 0 | buf, |
327 | 0 | "JDBCTable({} ,_jdbc_catalog_id = {}, _jdbc_resource_name={} ,_jdbc_driver_url={} " |
328 | 0 | ",_jdbc_driver_class={} ,_jdbc_driver_checksum={} ,_jdbc_url={} " |
329 | 0 | ",_jdbc_table_name={} ,_jdbc_user={} ,_jdbc_passwd={} ,_connection_pool_min_size={} " |
330 | 0 | ",_connection_pool_max_size={} ,_connection_pool_max_wait_time={} " |
331 | 0 | ",_connection_pool_max_life_time={} ,_connection_pool_keep_alive={})", |
332 | 0 | TableDescriptor::debug_string(), _jdbc_catalog_id, _jdbc_resource_name, |
333 | 0 | _jdbc_driver_url, _jdbc_driver_class, _jdbc_driver_checksum, _jdbc_url, |
334 | 0 | _jdbc_table_name, _jdbc_user, _jdbc_passwd, _connection_pool_min_size, |
335 | 0 | _connection_pool_max_size, _connection_pool_max_wait_time, |
336 | 0 | _connection_pool_max_life_time, _connection_pool_keep_alive); |
337 | 0 | return fmt::to_string(buf); |
338 | 0 | } |
339 | | |
340 | | RemoteDorisTableDescriptor::RemoteDorisTableDescriptor(const TTableDescriptor& tdesc) |
341 | 0 | : TableDescriptor(tdesc) {} |
342 | | |
343 | 0 | RemoteDorisTableDescriptor::~RemoteDorisTableDescriptor() = default; |
344 | | |
345 | 0 | std::string RemoteDorisTableDescriptor::debug_string() const { |
346 | 0 | std::stringstream out; |
347 | 0 | out << "RemoteDorisTable(" << TableDescriptor::debug_string() << ")"; |
348 | 0 | return out.str(); |
349 | 0 | } |
350 | | |
351 | | TupleDescriptor::TupleDescriptor(const TTupleDescriptor& tdesc, bool own_slots) |
352 | 292k | : _id(tdesc.id), |
353 | 292k | _num_materialized_slots(0), |
354 | 292k | _has_varlen_slots(false), |
355 | 292k | _own_slots(own_slots) {} |
356 | | |
357 | | TupleDescriptor::TupleDescriptor(const PTupleDescriptor& pdesc, bool own_slots) |
358 | 18 | : _id(pdesc.id()), |
359 | 18 | _num_materialized_slots(0), |
360 | 18 | _has_varlen_slots(false), |
361 | 18 | _own_slots(own_slots) {} |
362 | | |
363 | 767k | void TupleDescriptor::add_slot(SlotDescriptor* slot) { |
364 | 767k | _slots.push_back(slot); |
365 | 767k | ++_num_materialized_slots; |
366 | | |
367 | 767k | if (is_complex_type(slot->type()->get_primitive_type()) || |
368 | 767k | is_var_len_object(slot->type()->get_primitive_type()) || |
369 | 767k | is_string_type(slot->type()->get_primitive_type())) { |
370 | 129k | _has_varlen_slots = true; |
371 | 129k | } |
372 | 767k | } |
373 | | |
374 | 16 | void TupleDescriptor::to_protobuf(PTupleDescriptor* ptuple) const { |
375 | 16 | ptuple->Clear(); |
376 | 16 | ptuple->set_id(_id); |
377 | | // Useless not set |
378 | 16 | ptuple->set_byte_size(0); |
379 | 16 | ptuple->set_table_id(-1); |
380 | 16 | ptuple->set_num_null_bytes(0); |
381 | 16 | } |
382 | | |
383 | 2 | std::string TupleDescriptor::debug_string() const { |
384 | 2 | std::stringstream out; |
385 | 2 | out << "Tuple(id=" << _id; |
386 | 2 | if (_table_desc != nullptr) { |
387 | | //out << " " << _table_desc->debug_string(); |
388 | 0 | } |
389 | | |
390 | 2 | out << " slots=["; |
391 | 8 | for (size_t i = 0; i < _slots.size(); ++i) { |
392 | 6 | if (i > 0) { |
393 | 4 | out << ", "; |
394 | 4 | } |
395 | 6 | out << _slots[i]->debug_string(); |
396 | 6 | } |
397 | | |
398 | 2 | out << "]"; |
399 | 2 | out << " has_varlen_slots=" << _has_varlen_slots; |
400 | 2 | out << ")"; |
401 | 2 | return out.str(); |
402 | 2 | } |
403 | | |
404 | | RowDescriptor::RowDescriptor(const DescriptorTbl& desc_tbl, const std::vector<TTupleId>& row_tuples, |
405 | | const std::vector<bool>& nullable_tuples) |
406 | 432k | : _tuple_idx_nullable_map(nullable_tuples) { |
407 | 432k | DCHECK(nullable_tuples.size() == row_tuples.size()) |
408 | 0 | << "nullable_tuples size " << nullable_tuples.size() << " != row_tuples size " |
409 | 0 | << row_tuples.size(); |
410 | 432k | DCHECK_GT(row_tuples.size(), 0); |
411 | 432k | _num_materialized_slots = 0; |
412 | 432k | _num_slots = 0; |
413 | | |
414 | 504k | for (int row_tuple : row_tuples) { |
415 | 504k | TupleDescriptor* tupleDesc = desc_tbl.get_tuple_descriptor(row_tuple); |
416 | 504k | _num_materialized_slots += tupleDesc->num_materialized_slots(); |
417 | 504k | _num_slots += tupleDesc->slots().size(); |
418 | 504k | _tuple_desc_map.push_back(tupleDesc); |
419 | 504k | DCHECK(_tuple_desc_map.back() != nullptr); |
420 | 504k | } |
421 | | |
422 | 432k | init_tuple_idx_map(); |
423 | 432k | init_has_varlen_slots(); |
424 | 432k | } |
425 | | |
426 | | RowDescriptor::RowDescriptor(TupleDescriptor* tuple_desc, bool is_nullable) |
427 | 4.09k | : _tuple_desc_map(1, tuple_desc), _tuple_idx_nullable_map(1, is_nullable) { |
428 | 4.09k | init_tuple_idx_map(); |
429 | 4.09k | init_has_varlen_slots(); |
430 | 4.09k | _num_slots = static_cast<int32_t>(tuple_desc->slots().size()); |
431 | 4.09k | } |
432 | | |
433 | 0 | RowDescriptor::RowDescriptor(const RowDescriptor& lhs_row_desc, const RowDescriptor& rhs_row_desc) { |
434 | 0 | _tuple_desc_map.insert(_tuple_desc_map.end(), lhs_row_desc._tuple_desc_map.begin(), |
435 | 0 | lhs_row_desc._tuple_desc_map.end()); |
436 | 0 | _tuple_desc_map.insert(_tuple_desc_map.end(), rhs_row_desc._tuple_desc_map.begin(), |
437 | 0 | rhs_row_desc._tuple_desc_map.end()); |
438 | 0 | _tuple_idx_nullable_map.insert(_tuple_idx_nullable_map.end(), |
439 | 0 | lhs_row_desc._tuple_idx_nullable_map.begin(), |
440 | 0 | lhs_row_desc._tuple_idx_nullable_map.end()); |
441 | 0 | _tuple_idx_nullable_map.insert(_tuple_idx_nullable_map.end(), |
442 | 0 | rhs_row_desc._tuple_idx_nullable_map.begin(), |
443 | 0 | rhs_row_desc._tuple_idx_nullable_map.end()); |
444 | 0 | init_tuple_idx_map(); |
445 | 0 | init_has_varlen_slots(); |
446 | |
|
447 | 0 | _num_slots = lhs_row_desc.num_slots() + rhs_row_desc.num_slots(); |
448 | 0 | } |
449 | | |
450 | 436k | void RowDescriptor::init_tuple_idx_map() { |
451 | | // find max id |
452 | 436k | TupleId max_id = 0; |
453 | 508k | for (auto& i : _tuple_desc_map) { |
454 | 508k | max_id = std::max(i->id(), max_id); |
455 | 508k | } |
456 | | |
457 | 436k | _tuple_idx_map.resize(max_id + 1, INVALID_IDX); |
458 | 945k | for (int i = 0; i < _tuple_desc_map.size(); ++i) { |
459 | 508k | _tuple_idx_map[_tuple_desc_map[i]->id()] = i; |
460 | 508k | } |
461 | 436k | } |
462 | | |
463 | 436k | void RowDescriptor::init_has_varlen_slots() { |
464 | 436k | _has_varlen_slots = false; |
465 | 487k | for (auto& i : _tuple_desc_map) { |
466 | 487k | if (i->has_varlen_slots()) { |
467 | 133k | _has_varlen_slots = true; |
468 | 133k | break; |
469 | 133k | } |
470 | 487k | } |
471 | 436k | } |
472 | | |
473 | 0 | int RowDescriptor::get_tuple_idx(TupleId id) const { |
474 | | // comment CHECK temporarily to make fuzzy test run smoothly |
475 | | // DCHECK_LT(id, _tuple_idx_map.size()) << "RowDescriptor: " << debug_string(); |
476 | 0 | if (_tuple_idx_map.size() <= id) { |
477 | 0 | return RowDescriptor::INVALID_IDX; |
478 | 0 | } |
479 | 0 | return _tuple_idx_map[id]; |
480 | 0 | } |
481 | | |
482 | 0 | void RowDescriptor::to_thrift(std::vector<TTupleId>* row_tuple_ids) { |
483 | 0 | row_tuple_ids->clear(); |
484 | |
|
485 | 0 | for (auto& i : _tuple_desc_map) { |
486 | 0 | row_tuple_ids->push_back(i->id()); |
487 | 0 | } |
488 | 0 | } |
489 | | |
490 | | void RowDescriptor::to_protobuf( |
491 | 0 | google::protobuf::RepeatedField<google::protobuf::int32>* row_tuple_ids) const { |
492 | 0 | row_tuple_ids->Clear(); |
493 | 0 | for (auto* desc : _tuple_desc_map) { |
494 | 0 | row_tuple_ids->Add(desc->id()); |
495 | 0 | } |
496 | 0 | } |
497 | | |
498 | 0 | bool RowDescriptor::is_prefix_of(const RowDescriptor& other_desc) const { |
499 | 0 | if (_tuple_desc_map.size() > other_desc._tuple_desc_map.size()) { |
500 | 0 | return false; |
501 | 0 | } |
502 | | |
503 | 0 | for (int i = 0; i < _tuple_desc_map.size(); ++i) { |
504 | | // pointer comparison okay, descriptors are unique |
505 | 0 | if (_tuple_desc_map[i] != other_desc._tuple_desc_map[i]) { |
506 | 0 | return false; |
507 | 0 | } |
508 | 0 | } |
509 | | |
510 | 0 | return true; |
511 | 0 | } |
512 | | |
513 | 0 | bool RowDescriptor::equals(const RowDescriptor& other_desc) const { |
514 | 0 | if (_tuple_desc_map.size() != other_desc._tuple_desc_map.size()) { |
515 | 0 | return false; |
516 | 0 | } |
517 | | |
518 | 0 | for (int i = 0; i < _tuple_desc_map.size(); ++i) { |
519 | | // pointer comparison okay, descriptors are unique |
520 | 0 | if (_tuple_desc_map[i] != other_desc._tuple_desc_map[i]) { |
521 | 0 | return false; |
522 | 0 | } |
523 | 0 | } |
524 | | |
525 | 0 | return true; |
526 | 0 | } |
527 | | |
528 | 0 | std::string RowDescriptor::debug_string() const { |
529 | 0 | std::stringstream ss; |
530 | |
|
531 | 0 | ss << "tuple_desc_map: ["; |
532 | 0 | for (int i = 0; i < _tuple_desc_map.size(); ++i) { |
533 | 0 | ss << _tuple_desc_map[i]->debug_string(); |
534 | 0 | if (i != _tuple_desc_map.size() - 1) { |
535 | 0 | ss << ", "; |
536 | 0 | } |
537 | 0 | } |
538 | 0 | ss << "] "; |
539 | |
|
540 | 0 | ss << "tuple_id_map: ["; |
541 | 0 | for (int i = 0; i < _tuple_idx_map.size(); ++i) { |
542 | 0 | ss << _tuple_idx_map[i]; |
543 | 0 | if (i != _tuple_idx_map.size() - 1) { |
544 | 0 | ss << ", "; |
545 | 0 | } |
546 | 0 | } |
547 | 0 | ss << "] "; |
548 | |
|
549 | 0 | ss << "tuple_is_nullable: ["; |
550 | 0 | for (int i = 0; i < _tuple_idx_nullable_map.size(); ++i) { |
551 | 0 | ss << _tuple_idx_nullable_map[i]; |
552 | 0 | if (i != _tuple_idx_nullable_map.size() - 1) { |
553 | 0 | ss << ", "; |
554 | 0 | } |
555 | 0 | } |
556 | 0 | ss << "] "; |
557 | |
|
558 | 0 | return ss.str(); |
559 | 0 | } |
560 | | |
561 | 258k | int RowDescriptor::get_column_id(int slot_id) const { |
562 | 258k | int column_id_counter = 0; |
563 | 258k | for (auto* const tuple_desc : _tuple_desc_map) { |
564 | 420k | for (auto* const slot : tuple_desc->slots()) { |
565 | 420k | if (slot->id() == slot_id) { |
566 | 258k | return column_id_counter; |
567 | 258k | } |
568 | 162k | column_id_counter++; |
569 | 162k | } |
570 | 258k | } |
571 | 0 | return -1; |
572 | 258k | } |
573 | | |
574 | | Status DescriptorTbl::create(ObjectPool* pool, const TDescriptorTable& thrift_tbl, |
575 | 76.3k | DescriptorTbl** tbl) { |
576 | 76.3k | *tbl = pool->add(new DescriptorTbl()); |
577 | | |
578 | | // deserialize table descriptors first, they are being referenced by tuple descriptors |
579 | 76.3k | for (const auto& tdesc : thrift_tbl.tableDescriptors) { |
580 | 142 | TableDescriptor* desc = nullptr; |
581 | | |
582 | 142 | switch (tdesc.tableType) { |
583 | 84 | case TTableType::MYSQL_TABLE: |
584 | 84 | desc = pool->add(new MySQLTableDescriptor(tdesc)); |
585 | 84 | break; |
586 | | |
587 | 58 | case TTableType::OLAP_TABLE: |
588 | 58 | desc = pool->add(new OlapTableDescriptor(tdesc)); |
589 | 58 | break; |
590 | | |
591 | 0 | case TTableType::SCHEMA_TABLE: |
592 | 0 | desc = pool->add(new SchemaTableDescriptor(tdesc)); |
593 | 0 | break; |
594 | 0 | case TTableType::BROKER_TABLE: |
595 | 0 | desc = pool->add(new BrokerTableDescriptor(tdesc)); |
596 | 0 | break; |
597 | 0 | case TTableType::ES_TABLE: |
598 | 0 | desc = pool->add(new EsTableDescriptor(tdesc)); |
599 | 0 | break; |
600 | 0 | case TTableType::HIVE_TABLE: |
601 | 0 | desc = pool->add(new HiveTableDescriptor(tdesc)); |
602 | 0 | break; |
603 | 0 | case TTableType::ICEBERG_TABLE: |
604 | 0 | desc = pool->add(new IcebergTableDescriptor(tdesc)); |
605 | 0 | break; |
606 | 0 | case TTableType::JDBC_TABLE: |
607 | 0 | desc = pool->add(new JdbcTableDescriptor(tdesc)); |
608 | 0 | break; |
609 | 0 | case TTableType::MAX_COMPUTE_TABLE: |
610 | 0 | desc = pool->add(new MaxComputeTableDescriptor(tdesc)); |
611 | 0 | break; |
612 | 0 | case TTableType::TRINO_CONNECTOR_TABLE: |
613 | 0 | desc = pool->add(new TrinoConnectorTableDescriptor(tdesc)); |
614 | 0 | break; |
615 | 0 | case TTableType::DICTIONARY_TABLE: |
616 | 0 | desc = pool->add(new DictionaryTableDescriptor(tdesc)); |
617 | 0 | break; |
618 | 0 | case TTableType::REMOTE_DORIS_TABLE: |
619 | 0 | desc = pool->add(new RemoteDorisTableDescriptor(tdesc)); |
620 | 0 | break; |
621 | 0 | default: |
622 | 0 | DCHECK(false) << "invalid table type: " << tdesc.tableType; |
623 | 142 | } |
624 | | |
625 | 142 | (*tbl)->_tbl_desc_map[static_cast<int32_t>(tdesc.id)] = desc; |
626 | 142 | } |
627 | | |
628 | 292k | for (const auto& tdesc : thrift_tbl.tupleDescriptors) { |
629 | 292k | TupleDescriptor* desc = pool->add(new TupleDescriptor(tdesc)); |
630 | | |
631 | | // fix up table pointer |
632 | 292k | if (tdesc.__isset.tableId) { |
633 | 132 | desc->_table_desc = (*tbl)->get_table_descriptor(static_cast<int32_t>(tdesc.tableId)); |
634 | 132 | DCHECK(desc->_table_desc != nullptr); |
635 | 132 | } |
636 | | |
637 | 292k | (*tbl)->_tuple_desc_map[tdesc.id] = desc; |
638 | 292k | (*tbl)->_row_tuples.emplace_back(tdesc.id); |
639 | 292k | } |
640 | | |
641 | 767k | for (const auto& tdesc : thrift_tbl.slotDescriptors) { |
642 | 767k | SlotDescriptor* slot_d = pool->add(new SlotDescriptor(tdesc)); |
643 | 767k | (*tbl)->_slot_desc_map[tdesc.id] = slot_d; |
644 | | |
645 | | // link to parent |
646 | 767k | auto entry = (*tbl)->_tuple_desc_map.find(tdesc.parent); |
647 | | |
648 | 767k | if (entry == (*tbl)->_tuple_desc_map.end()) { |
649 | 0 | return Status::InternalError("unknown tid in slot descriptor msg"); |
650 | 0 | } |
651 | 767k | entry->second->add_slot(slot_d); |
652 | 767k | } |
653 | | |
654 | 76.3k | return Status::OK(); |
655 | 76.3k | } |
656 | | |
657 | 132 | TableDescriptor* DescriptorTbl::get_table_descriptor(TableId id) const { |
658 | | // TODO: is there some boost function to do exactly this? |
659 | 132 | auto i = _tbl_desc_map.find(id); |
660 | | |
661 | 132 | if (i == _tbl_desc_map.end()) { |
662 | 0 | return nullptr; |
663 | 132 | } else { |
664 | 132 | return i->second; |
665 | 132 | } |
666 | 132 | } |
667 | | |
668 | 710k | TupleDescriptor* DescriptorTbl::get_tuple_descriptor(TupleId id) const { |
669 | | // TODO: is there some boost function to do exactly this? |
670 | 710k | auto i = _tuple_desc_map.find(id); |
671 | | |
672 | 710k | if (i == _tuple_desc_map.end()) { |
673 | 4 | return nullptr; |
674 | 710k | } else { |
675 | 710k | return i->second; |
676 | 710k | } |
677 | 710k | } |
678 | | |
679 | 258k | SlotDescriptor* DescriptorTbl::get_slot_descriptor(SlotId id) const { |
680 | | // TODO: is there some boost function to do exactly this? |
681 | 258k | auto i = _slot_desc_map.find(id); |
682 | | |
683 | 258k | if (i == _slot_desc_map.end()) { |
684 | 0 | return nullptr; |
685 | 258k | } else { |
686 | 258k | return i->second; |
687 | 258k | } |
688 | 258k | } |
689 | | |
690 | 0 | std::string DescriptorTbl::debug_string() const { |
691 | 0 | std::stringstream out; |
692 | 0 | out << "tuples:\n"; |
693 | |
|
694 | 0 | for (auto i : _tuple_desc_map) { |
695 | 0 | out << i.second->debug_string() << '\n'; |
696 | 0 | } |
697 | |
|
698 | 0 | return out.str(); |
699 | 0 | } |
700 | | #include "common/compile_check_end.h" |
701 | | } // namespace doris |