/root/doris/be/src/runtime/descriptors.cpp
Line | Count | Source (jump to first uncovered line) |
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 | | |
29 | | #include <algorithm> |
30 | | #include <boost/algorithm/string/join.hpp> |
31 | | #include <memory> |
32 | | |
33 | | #include "common/object_pool.h" |
34 | | #include "runtime/primitive_type.h" |
35 | | #include "util/string_util.h" |
36 | | #include "vec/aggregate_functions/aggregate_function.h" |
37 | | #include "vec/data_types/data_type_factory.hpp" |
38 | | |
39 | | namespace doris { |
40 | | |
41 | | const int RowDescriptor::INVALID_IDX = -1; |
42 | | |
43 | | SlotDescriptor::SlotDescriptor(const TSlotDescriptor& tdesc) |
44 | | : _id(tdesc.id), |
45 | | _type(TypeDescriptor::from_thrift(tdesc.slotType)), |
46 | | _parent(tdesc.parent), |
47 | | _col_pos(tdesc.columnPos), |
48 | | _is_nullable(tdesc.nullIndicatorBit != -1), |
49 | | _col_name(tdesc.colName), |
50 | | _col_name_lower_case(to_lower(tdesc.colName)), |
51 | | _col_unique_id(tdesc.col_unique_id), |
52 | | _slot_idx(tdesc.slotIdx), |
53 | | _field_idx(-1), |
54 | | _is_materialized(tdesc.isMaterialized && tdesc.need_materialize), |
55 | | _is_key(tdesc.is_key), |
56 | | _column_paths(tdesc.column_paths), |
57 | | _is_auto_increment(tdesc.__isset.is_auto_increment ? tdesc.is_auto_increment : false), |
58 | 467 | _col_default_value(tdesc.__isset.col_default_value ? tdesc.col_default_value : "") {} |
59 | | |
60 | | SlotDescriptor::SlotDescriptor(const PSlotDescriptor& pdesc) |
61 | | : _id(pdesc.id()), |
62 | | _type(TypeDescriptor::from_protobuf(pdesc.slot_type())), |
63 | | _parent(pdesc.parent()), |
64 | | _col_pos(pdesc.column_pos()), |
65 | | _is_nullable(pdesc.null_indicator_bit() != -1), |
66 | | _col_name(pdesc.col_name()), |
67 | | _col_name_lower_case(to_lower(pdesc.col_name())), |
68 | | _col_unique_id(pdesc.col_unique_id()), |
69 | | _slot_idx(pdesc.slot_idx()), |
70 | | _field_idx(-1), |
71 | | _is_materialized(pdesc.is_materialized()), |
72 | | _is_key(pdesc.is_key()), |
73 | | _column_paths(pdesc.column_paths().begin(), pdesc.column_paths().end()), |
74 | 84 | _is_auto_increment(pdesc.is_auto_increment()) {} |
75 | | |
76 | | #ifdef BE_TEST |
77 | | SlotDescriptor::SlotDescriptor() |
78 | | : _id(0), |
79 | | _type(TypeDescriptor {}), |
80 | | _parent(0), |
81 | | _col_pos(0), |
82 | | _is_nullable(false), |
83 | | _col_unique_id(0), |
84 | | _slot_idx(0), |
85 | | _field_idx(-1), |
86 | | _is_materialized(true), |
87 | | _is_key(false), |
88 | 30 | _is_auto_increment(false) {} |
89 | | #endif |
90 | | |
91 | 96 | void SlotDescriptor::to_protobuf(PSlotDescriptor* pslot) const { |
92 | 96 | pslot->set_id(_id); |
93 | 96 | pslot->set_parent(_parent); |
94 | 96 | _type.to_protobuf(pslot->mutable_slot_type()); |
95 | 96 | pslot->set_column_pos(_col_pos); |
96 | 96 | pslot->set_byte_offset(0); |
97 | 96 | pslot->set_null_indicator_byte(0); |
98 | 96 | pslot->set_null_indicator_bit(_is_nullable ? 0 : -1); |
99 | 96 | pslot->set_col_name(_col_name); |
100 | 96 | pslot->set_slot_idx(_slot_idx); |
101 | 96 | pslot->set_is_materialized(_is_materialized); |
102 | 96 | pslot->set_col_unique_id(_col_unique_id); |
103 | 96 | pslot->set_is_key(_is_key); |
104 | 96 | pslot->set_is_auto_increment(_is_auto_increment); |
105 | 96 | pslot->set_col_type(_type.type); |
106 | 96 | for (const std::string& path : _column_paths) { |
107 | 0 | pslot->add_column_paths(path); |
108 | 0 | } |
109 | 96 | } |
110 | | |
111 | 82 | vectorized::MutableColumnPtr SlotDescriptor::get_empty_mutable_column() const { |
112 | 82 | auto data_type = get_data_type_ptr(); |
113 | 82 | if (data_type) { |
114 | 82 | return data_type->create_column(); |
115 | 82 | } |
116 | 0 | return nullptr; |
117 | 82 | } |
118 | | |
119 | 214 | vectorized::DataTypePtr SlotDescriptor::get_data_type_ptr() const { |
120 | 214 | return vectorized::DataTypeFactory::instance().create_data_type(type(), is_nullable()); |
121 | 214 | } |
122 | | |
123 | 0 | std::string SlotDescriptor::debug_string() const { |
124 | 0 | std::stringstream out; |
125 | 0 | out << "Slot(id=" << _id << " type=" << _type << " col=" << _col_pos |
126 | 0 | << ", colname=" << _col_name << ", nullable=" << is_nullable() << ")"; |
127 | 0 | return out.str(); |
128 | 0 | } |
129 | | |
130 | | TableDescriptor::TableDescriptor(const TTableDescriptor& tdesc) |
131 | | : _table_type(tdesc.tableType), |
132 | | _name(tdesc.tableName), |
133 | | _database(tdesc.dbName), |
134 | | _table_id(tdesc.id), |
135 | | _num_cols(tdesc.numCols), |
136 | 17 | _num_clustering_cols(tdesc.numClusteringCols) {} |
137 | | |
138 | 0 | std::string TableDescriptor::debug_string() const { |
139 | 0 | std::stringstream out; |
140 | 0 | out << "#cols=" << _num_cols << " #clustering_cols=" << _num_clustering_cols; |
141 | 0 | return out.str(); |
142 | 0 | } |
143 | | |
144 | 17 | OlapTableDescriptor::OlapTableDescriptor(const TTableDescriptor& tdesc) : TableDescriptor(tdesc) {} |
145 | | |
146 | 0 | std::string OlapTableDescriptor::debug_string() const { |
147 | 0 | std::stringstream out; |
148 | 0 | out << "OlapTable(" << TableDescriptor::debug_string() << ")"; |
149 | 0 | return out.str(); |
150 | 0 | } |
151 | | |
152 | | SchemaTableDescriptor::SchemaTableDescriptor(const TTableDescriptor& tdesc) |
153 | 0 | : TableDescriptor(tdesc), _schema_table_type(tdesc.schemaTable.tableType) {} |
154 | 0 | SchemaTableDescriptor::~SchemaTableDescriptor() = default; |
155 | | |
156 | 0 | std::string SchemaTableDescriptor::debug_string() const { |
157 | 0 | std::stringstream out; |
158 | 0 | out << "SchemaTable(" << TableDescriptor::debug_string() << ")"; |
159 | 0 | return out.str(); |
160 | 0 | } |
161 | | |
162 | | BrokerTableDescriptor::BrokerTableDescriptor(const TTableDescriptor& tdesc) |
163 | 0 | : TableDescriptor(tdesc) {} |
164 | | |
165 | 0 | BrokerTableDescriptor::~BrokerTableDescriptor() = default; |
166 | | |
167 | 0 | std::string BrokerTableDescriptor::debug_string() const { |
168 | 0 | std::stringstream out; |
169 | 0 | out << "BrokerTable(" << TableDescriptor::debug_string() << ")"; |
170 | 0 | return out.str(); |
171 | 0 | } |
172 | | |
173 | 0 | HiveTableDescriptor::HiveTableDescriptor(const TTableDescriptor& tdesc) : TableDescriptor(tdesc) {} |
174 | | |
175 | 0 | HiveTableDescriptor::~HiveTableDescriptor() = default; |
176 | | |
177 | 0 | std::string HiveTableDescriptor::debug_string() const { |
178 | 0 | std::stringstream out; |
179 | 0 | out << "HiveTable(" << TableDescriptor::debug_string() << ")"; |
180 | 0 | return out.str(); |
181 | 0 | } |
182 | | |
183 | | IcebergTableDescriptor::IcebergTableDescriptor(const TTableDescriptor& tdesc) |
184 | 0 | : TableDescriptor(tdesc) {} |
185 | | |
186 | 0 | IcebergTableDescriptor::~IcebergTableDescriptor() = default; |
187 | | |
188 | 0 | std::string IcebergTableDescriptor::debug_string() const { |
189 | 0 | std::stringstream out; |
190 | 0 | out << "IcebergTable(" << TableDescriptor::debug_string() << ")"; |
191 | 0 | return out.str(); |
192 | 0 | } |
193 | | |
194 | | MaxComputeTableDescriptor::MaxComputeTableDescriptor(const TTableDescriptor& tdesc) |
195 | | : TableDescriptor(tdesc), |
196 | | _region(tdesc.mcTable.region), |
197 | | _project(tdesc.mcTable.project), |
198 | | _table(tdesc.mcTable.table), |
199 | | _odps_url(tdesc.mcTable.odps_url), |
200 | | _tunnel_url(tdesc.mcTable.tunnel_url), |
201 | | _access_key(tdesc.mcTable.access_key), |
202 | | _secret_key(tdesc.mcTable.secret_key), |
203 | 0 | _public_access(tdesc.mcTable.public_access) { |
204 | 0 | if (tdesc.mcTable.__isset.endpoint) { |
205 | 0 | _endpoint = tdesc.mcTable.endpoint; |
206 | 0 | } else { |
207 | 0 | _init_status = Status::InvalidArgument( |
208 | 0 | "fail to init MaxComputeTableDescriptor, missing endpoint."); |
209 | 0 | } |
210 | |
|
211 | 0 | if (tdesc.mcTable.__isset.quota) { |
212 | 0 | _quota = tdesc.mcTable.quota; |
213 | 0 | } else { |
214 | 0 | _init_status = |
215 | 0 | Status::InvalidArgument("fail to init MaxComputeTableDescriptor, missing quota."); |
216 | 0 | } |
217 | 0 | } |
218 | | |
219 | 0 | MaxComputeTableDescriptor::~MaxComputeTableDescriptor() = default; |
220 | | |
221 | 0 | std::string MaxComputeTableDescriptor::debug_string() const { |
222 | 0 | std::stringstream out; |
223 | 0 | out << "MaxComputeTable(" << TableDescriptor::debug_string() << ")"; |
224 | 0 | return out.str(); |
225 | 0 | } |
226 | | |
227 | | TrinoConnectorTableDescriptor::TrinoConnectorTableDescriptor(const TTableDescriptor& tdesc) |
228 | 0 | : TableDescriptor(tdesc) {} |
229 | | |
230 | 0 | TrinoConnectorTableDescriptor::~TrinoConnectorTableDescriptor() = default; |
231 | | |
232 | 0 | std::string TrinoConnectorTableDescriptor::debug_string() const { |
233 | 0 | std::stringstream out; |
234 | 0 | out << "TrinoConnectorTable(" << TableDescriptor::debug_string() << ")"; |
235 | 0 | return out.str(); |
236 | 0 | } |
237 | | |
238 | 0 | EsTableDescriptor::EsTableDescriptor(const TTableDescriptor& tdesc) : TableDescriptor(tdesc) {} |
239 | | |
240 | 0 | EsTableDescriptor::~EsTableDescriptor() = default; |
241 | | |
242 | 0 | std::string EsTableDescriptor::debug_string() const { |
243 | 0 | std::stringstream out; |
244 | 0 | out << "EsTable(" << TableDescriptor::debug_string() << ")"; |
245 | 0 | return out.str(); |
246 | 0 | } |
247 | | |
248 | | MySQLTableDescriptor::MySQLTableDescriptor(const TTableDescriptor& tdesc) |
249 | | : TableDescriptor(tdesc), |
250 | | _mysql_db(tdesc.mysqlTable.db), |
251 | | _mysql_table(tdesc.mysqlTable.table), |
252 | | _host(tdesc.mysqlTable.host), |
253 | | _port(tdesc.mysqlTable.port), |
254 | | _user(tdesc.mysqlTable.user), |
255 | | _passwd(tdesc.mysqlTable.passwd), |
256 | 0 | _charset(tdesc.mysqlTable.charset) {} |
257 | | |
258 | 0 | std::string MySQLTableDescriptor::debug_string() const { |
259 | 0 | std::stringstream out; |
260 | 0 | out << "MySQLTable(" << TableDescriptor::debug_string() << " _db" << _mysql_db |
261 | 0 | << " table=" << _mysql_table << " host=" << _host << " port=" << _port << " user=" << _user |
262 | 0 | << " passwd=" << _passwd << " charset=" << _charset; |
263 | 0 | return out.str(); |
264 | 0 | } |
265 | | |
266 | | ODBCTableDescriptor::ODBCTableDescriptor(const TTableDescriptor& tdesc) |
267 | | : TableDescriptor(tdesc), |
268 | | _db(tdesc.odbcTable.db), |
269 | | _table(tdesc.odbcTable.table), |
270 | | _host(tdesc.odbcTable.host), |
271 | | _port(tdesc.odbcTable.port), |
272 | | _user(tdesc.odbcTable.user), |
273 | | _passwd(tdesc.odbcTable.passwd), |
274 | | _driver(tdesc.odbcTable.driver), |
275 | 0 | _type(tdesc.odbcTable.type) {} |
276 | | |
277 | 0 | std::string ODBCTableDescriptor::debug_string() const { |
278 | 0 | std::stringstream out; |
279 | 0 | out << "ODBCTable(" << TableDescriptor::debug_string() << " _db" << _db << " table=" << _table |
280 | 0 | << " host=" << _host << " port=" << _port << " user=" << _user << " passwd=" << _passwd |
281 | 0 | << " driver=" << _driver << " type" << _type; |
282 | 0 | return out.str(); |
283 | 0 | } |
284 | | |
285 | | JdbcTableDescriptor::JdbcTableDescriptor(const TTableDescriptor& tdesc) |
286 | | : TableDescriptor(tdesc), |
287 | | _jdbc_catalog_id(tdesc.jdbcTable.catalog_id), |
288 | | _jdbc_resource_name(tdesc.jdbcTable.jdbc_resource_name), |
289 | | _jdbc_driver_url(tdesc.jdbcTable.jdbc_driver_url), |
290 | | _jdbc_driver_class(tdesc.jdbcTable.jdbc_driver_class), |
291 | | _jdbc_driver_checksum(tdesc.jdbcTable.jdbc_driver_checksum), |
292 | | _jdbc_url(tdesc.jdbcTable.jdbc_url), |
293 | | _jdbc_table_name(tdesc.jdbcTable.jdbc_table_name), |
294 | | _jdbc_user(tdesc.jdbcTable.jdbc_user), |
295 | | _jdbc_passwd(tdesc.jdbcTable.jdbc_password), |
296 | | _connection_pool_min_size(tdesc.jdbcTable.connection_pool_min_size), |
297 | | _connection_pool_max_size(tdesc.jdbcTable.connection_pool_max_size), |
298 | | _connection_pool_max_wait_time(tdesc.jdbcTable.connection_pool_max_wait_time), |
299 | | _connection_pool_max_life_time(tdesc.jdbcTable.connection_pool_max_life_time), |
300 | 0 | _connection_pool_keep_alive(tdesc.jdbcTable.connection_pool_keep_alive) {} |
301 | | |
302 | 0 | std::string JdbcTableDescriptor::debug_string() const { |
303 | 0 | fmt::memory_buffer buf; |
304 | 0 | fmt::format_to( |
305 | 0 | buf, |
306 | 0 | "JDBCTable({} ,_jdbc_catalog_id = {}, _jdbc_resource_name={} ,_jdbc_driver_url={} " |
307 | 0 | ",_jdbc_driver_class={} ,_jdbc_driver_checksum={} ,_jdbc_url={} " |
308 | 0 | ",_jdbc_table_name={} ,_jdbc_user={} ,_jdbc_passwd={} ,_connection_pool_min_size={} " |
309 | 0 | ",_connection_pool_max_size={} ,_connection_pool_max_wait_time={} " |
310 | 0 | ",_connection_pool_max_life_time={} ,_connection_pool_keep_alive={})", |
311 | 0 | TableDescriptor::debug_string(), _jdbc_catalog_id, _jdbc_resource_name, |
312 | 0 | _jdbc_driver_url, _jdbc_driver_class, _jdbc_driver_checksum, _jdbc_url, |
313 | 0 | _jdbc_table_name, _jdbc_user, _jdbc_passwd, _connection_pool_min_size, |
314 | 0 | _connection_pool_max_size, _connection_pool_max_wait_time, |
315 | 0 | _connection_pool_max_life_time, _connection_pool_keep_alive); |
316 | 0 | return fmt::to_string(buf); |
317 | 0 | } |
318 | | |
319 | | TupleDescriptor::TupleDescriptor(const TTupleDescriptor& tdesc, bool own_slots) |
320 | | : _id(tdesc.id), |
321 | | _num_materialized_slots(0), |
322 | | _has_varlen_slots(false), |
323 | 154 | _own_slots(own_slots) {} |
324 | | |
325 | | TupleDescriptor::TupleDescriptor(const PTupleDescriptor& pdesc, bool own_slots) |
326 | | : _id(pdesc.id()), |
327 | | _num_materialized_slots(0), |
328 | | _has_varlen_slots(false), |
329 | 18 | _own_slots(own_slots) {} |
330 | | |
331 | 551 | void TupleDescriptor::add_slot(SlotDescriptor* slot) { |
332 | 551 | _slots.push_back(slot); |
333 | | |
334 | 551 | if (slot->is_materialized()) { |
335 | 516 | ++_num_materialized_slots; |
336 | | |
337 | 516 | if (slot->type().is_string_type() || slot->type().is_collection_type()) { |
338 | 141 | _has_varlen_slots = true; |
339 | 141 | } |
340 | 516 | } |
341 | 551 | } |
342 | | |
343 | 16 | void TupleDescriptor::to_protobuf(PTupleDescriptor* ptuple) const { |
344 | 16 | ptuple->Clear(); |
345 | 16 | ptuple->set_id(_id); |
346 | | // Useless not set |
347 | 16 | ptuple->set_byte_size(0); |
348 | 16 | ptuple->set_table_id(-1); |
349 | 16 | ptuple->set_num_null_bytes(0); |
350 | 16 | } |
351 | | |
352 | 0 | std::string TupleDescriptor::debug_string() const { |
353 | 0 | std::stringstream out; |
354 | 0 | out << "Tuple(id=" << _id; |
355 | 0 | if (_table_desc != nullptr) { |
356 | | //out << " " << _table_desc->debug_string(); |
357 | 0 | } |
358 | |
|
359 | 0 | out << " slots=["; |
360 | 0 | for (size_t i = 0; i < _slots.size(); ++i) { |
361 | 0 | if (i > 0) { |
362 | 0 | out << ", "; |
363 | 0 | } |
364 | 0 | out << _slots[i]->debug_string(); |
365 | 0 | } |
366 | |
|
367 | 0 | out << "]"; |
368 | 0 | out << " has_varlen_slots=" << _has_varlen_slots; |
369 | 0 | out << ")"; |
370 | 0 | return out.str(); |
371 | 0 | } |
372 | | |
373 | | RowDescriptor::RowDescriptor(const DescriptorTbl& desc_tbl, const std::vector<TTupleId>& row_tuples, |
374 | | const std::vector<bool>& nullable_tuples) |
375 | 354 | : _tuple_idx_nullable_map(nullable_tuples) { |
376 | 354 | DCHECK(nullable_tuples.size() == row_tuples.size()) |
377 | 0 | << "nullable_tuples size " << nullable_tuples.size() << " != row_tuples size " |
378 | 0 | << row_tuples.size(); |
379 | 354 | DCHECK_GT(row_tuples.size(), 0); |
380 | 354 | _num_materialized_slots = 0; |
381 | 354 | _num_slots = 0; |
382 | | |
383 | 432 | for (int row_tuple : row_tuples) { |
384 | 432 | TupleDescriptor* tupleDesc = desc_tbl.get_tuple_descriptor(row_tuple); |
385 | 432 | _num_materialized_slots += tupleDesc->num_materialized_slots(); |
386 | 432 | _num_slots += tupleDesc->slots().size(); |
387 | 432 | _tuple_desc_map.push_back(tupleDesc); |
388 | 432 | DCHECK(_tuple_desc_map.back() != nullptr); |
389 | 432 | } |
390 | | |
391 | 354 | init_tuple_idx_map(); |
392 | 354 | init_has_varlen_slots(); |
393 | 354 | } |
394 | | |
395 | | RowDescriptor::RowDescriptor(TupleDescriptor* tuple_desc, bool is_nullable) |
396 | 23 | : _tuple_desc_map(1, tuple_desc), _tuple_idx_nullable_map(1, is_nullable) { |
397 | 23 | init_tuple_idx_map(); |
398 | 23 | init_has_varlen_slots(); |
399 | 23 | _num_slots = tuple_desc->slots().size(); |
400 | 23 | } |
401 | | |
402 | 0 | RowDescriptor::RowDescriptor(const RowDescriptor& lhs_row_desc, const RowDescriptor& rhs_row_desc) { |
403 | 0 | _tuple_desc_map.insert(_tuple_desc_map.end(), lhs_row_desc._tuple_desc_map.begin(), |
404 | 0 | lhs_row_desc._tuple_desc_map.end()); |
405 | 0 | _tuple_desc_map.insert(_tuple_desc_map.end(), rhs_row_desc._tuple_desc_map.begin(), |
406 | 0 | rhs_row_desc._tuple_desc_map.end()); |
407 | 0 | _tuple_idx_nullable_map.insert(_tuple_idx_nullable_map.end(), |
408 | 0 | lhs_row_desc._tuple_idx_nullable_map.begin(), |
409 | 0 | lhs_row_desc._tuple_idx_nullable_map.end()); |
410 | 0 | _tuple_idx_nullable_map.insert(_tuple_idx_nullable_map.end(), |
411 | 0 | rhs_row_desc._tuple_idx_nullable_map.begin(), |
412 | 0 | rhs_row_desc._tuple_idx_nullable_map.end()); |
413 | 0 | init_tuple_idx_map(); |
414 | 0 | init_has_varlen_slots(); |
415 | |
|
416 | 0 | _num_slots = lhs_row_desc.num_slots() + rhs_row_desc.num_slots(); |
417 | 0 | } |
418 | | |
419 | 377 | void RowDescriptor::init_tuple_idx_map() { |
420 | | // find max id |
421 | 377 | TupleId max_id = 0; |
422 | 455 | for (auto& i : _tuple_desc_map) { |
423 | 455 | max_id = std::max(i->id(), max_id); |
424 | 455 | } |
425 | | |
426 | 377 | _tuple_idx_map.resize(max_id + 1, INVALID_IDX); |
427 | 832 | for (int i = 0; i < _tuple_desc_map.size(); ++i) { |
428 | 455 | _tuple_idx_map[_tuple_desc_map[i]->id()] = i; |
429 | 455 | } |
430 | 377 | } |
431 | | |
432 | 377 | void RowDescriptor::init_has_varlen_slots() { |
433 | 377 | _has_varlen_slots = false; |
434 | 455 | for (auto& i : _tuple_desc_map) { |
435 | 455 | if (i->has_varlen_slots()) { |
436 | 22 | _has_varlen_slots = true; |
437 | 22 | break; |
438 | 22 | } |
439 | 455 | } |
440 | 377 | } |
441 | | |
442 | 0 | int RowDescriptor::get_tuple_idx(TupleId id) const { |
443 | | // comment CHECK temporarily to make fuzzy test run smoothly |
444 | | // DCHECK_LT(id, _tuple_idx_map.size()) << "RowDescriptor: " << debug_string(); |
445 | 0 | if (_tuple_idx_map.size() <= id) { |
446 | 0 | return RowDescriptor::INVALID_IDX; |
447 | 0 | } |
448 | 0 | return _tuple_idx_map[id]; |
449 | 0 | } |
450 | | |
451 | 0 | void RowDescriptor::to_thrift(std::vector<TTupleId>* row_tuple_ids) { |
452 | 0 | row_tuple_ids->clear(); |
453 | |
|
454 | 0 | for (auto& i : _tuple_desc_map) { |
455 | 0 | row_tuple_ids->push_back(i->id()); |
456 | 0 | } |
457 | 0 | } |
458 | | |
459 | | void RowDescriptor::to_protobuf( |
460 | 0 | google::protobuf::RepeatedField<google::protobuf::int32>* row_tuple_ids) const { |
461 | 0 | row_tuple_ids->Clear(); |
462 | 0 | for (auto* desc : _tuple_desc_map) { |
463 | 0 | row_tuple_ids->Add(desc->id()); |
464 | 0 | } |
465 | 0 | } |
466 | | |
467 | 0 | bool RowDescriptor::is_prefix_of(const RowDescriptor& other_desc) const { |
468 | 0 | if (_tuple_desc_map.size() > other_desc._tuple_desc_map.size()) { |
469 | 0 | return false; |
470 | 0 | } |
471 | | |
472 | 0 | for (int i = 0; i < _tuple_desc_map.size(); ++i) { |
473 | | // pointer comparison okay, descriptors are unique |
474 | 0 | if (_tuple_desc_map[i] != other_desc._tuple_desc_map[i]) { |
475 | 0 | return false; |
476 | 0 | } |
477 | 0 | } |
478 | | |
479 | 0 | return true; |
480 | 0 | } |
481 | | |
482 | 0 | bool RowDescriptor::equals(const RowDescriptor& other_desc) const { |
483 | 0 | if (_tuple_desc_map.size() != other_desc._tuple_desc_map.size()) { |
484 | 0 | return false; |
485 | 0 | } |
486 | | |
487 | 0 | for (int i = 0; i < _tuple_desc_map.size(); ++i) { |
488 | | // pointer comparison okay, descriptors are unique |
489 | 0 | if (_tuple_desc_map[i] != other_desc._tuple_desc_map[i]) { |
490 | 0 | return false; |
491 | 0 | } |
492 | 0 | } |
493 | | |
494 | 0 | return true; |
495 | 0 | } |
496 | | |
497 | 0 | std::string RowDescriptor::debug_string() const { |
498 | 0 | std::stringstream ss; |
499 | |
|
500 | 0 | ss << "tuple_desc_map: ["; |
501 | 0 | for (int i = 0; i < _tuple_desc_map.size(); ++i) { |
502 | 0 | ss << _tuple_desc_map[i]->debug_string(); |
503 | 0 | if (i != _tuple_desc_map.size() - 1) { |
504 | 0 | ss << ", "; |
505 | 0 | } |
506 | 0 | } |
507 | 0 | ss << "] "; |
508 | |
|
509 | 0 | ss << "tuple_id_map: ["; |
510 | 0 | for (int i = 0; i < _tuple_idx_map.size(); ++i) { |
511 | 0 | ss << _tuple_idx_map[i]; |
512 | 0 | if (i != _tuple_idx_map.size() - 1) { |
513 | 0 | ss << ", "; |
514 | 0 | } |
515 | 0 | } |
516 | 0 | ss << "] "; |
517 | |
|
518 | 0 | ss << "tuple_is_nullable: ["; |
519 | 0 | for (int i = 0; i < _tuple_idx_nullable_map.size(); ++i) { |
520 | 0 | ss << _tuple_idx_nullable_map[i]; |
521 | 0 | if (i != _tuple_idx_nullable_map.size() - 1) { |
522 | 0 | ss << ", "; |
523 | 0 | } |
524 | 0 | } |
525 | 0 | ss << "] "; |
526 | |
|
527 | 0 | return ss.str(); |
528 | 0 | } |
529 | | |
530 | 55 | int RowDescriptor::get_column_id(int slot_id, bool force_materialize_slot) const { |
531 | 55 | int column_id_counter = 0; |
532 | 56 | for (auto* const tuple_desc : _tuple_desc_map) { |
533 | 103 | for (auto* const slot : tuple_desc->slots()) { |
534 | 103 | if (!force_materialize_slot && !slot->is_materialized()) { |
535 | 0 | continue; |
536 | 0 | } |
537 | 103 | if (slot->id() == slot_id) { |
538 | 55 | return column_id_counter; |
539 | 55 | } |
540 | 48 | column_id_counter++; |
541 | 48 | } |
542 | 56 | } |
543 | 0 | return -1; |
544 | 55 | } |
545 | | |
546 | | Status DescriptorTbl::create(ObjectPool* pool, const TDescriptorTable& thrift_tbl, |
547 | 88 | DescriptorTbl** tbl) { |
548 | 88 | *tbl = pool->add(new DescriptorTbl()); |
549 | | |
550 | | // deserialize table descriptors first, they are being referenced by tuple descriptors |
551 | 88 | for (const auto& tdesc : thrift_tbl.tableDescriptors) { |
552 | 17 | TableDescriptor* desc = nullptr; |
553 | | |
554 | 17 | switch (tdesc.tableType) { |
555 | 0 | case TTableType::MYSQL_TABLE: |
556 | 0 | desc = pool->add(new MySQLTableDescriptor(tdesc)); |
557 | 0 | break; |
558 | | |
559 | 0 | case TTableType::ODBC_TABLE: |
560 | 0 | desc = pool->add(new ODBCTableDescriptor(tdesc)); |
561 | 0 | break; |
562 | | |
563 | 17 | case TTableType::OLAP_TABLE: |
564 | 17 | desc = pool->add(new OlapTableDescriptor(tdesc)); |
565 | 17 | break; |
566 | | |
567 | 0 | case TTableType::SCHEMA_TABLE: |
568 | 0 | desc = pool->add(new SchemaTableDescriptor(tdesc)); |
569 | 0 | break; |
570 | 0 | case TTableType::BROKER_TABLE: |
571 | 0 | desc = pool->add(new BrokerTableDescriptor(tdesc)); |
572 | 0 | break; |
573 | 0 | case TTableType::ES_TABLE: |
574 | 0 | desc = pool->add(new EsTableDescriptor(tdesc)); |
575 | 0 | break; |
576 | 0 | case TTableType::HIVE_TABLE: |
577 | 0 | desc = pool->add(new HiveTableDescriptor(tdesc)); |
578 | 0 | break; |
579 | 0 | case TTableType::ICEBERG_TABLE: |
580 | 0 | desc = pool->add(new IcebergTableDescriptor(tdesc)); |
581 | 0 | break; |
582 | 0 | case TTableType::JDBC_TABLE: |
583 | 0 | desc = pool->add(new JdbcTableDescriptor(tdesc)); |
584 | 0 | break; |
585 | 0 | case TTableType::MAX_COMPUTE_TABLE: |
586 | 0 | desc = pool->add(new MaxComputeTableDescriptor(tdesc)); |
587 | 0 | break; |
588 | 0 | case TTableType::TRINO_CONNECTOR_TABLE: |
589 | 0 | desc = pool->add(new TrinoConnectorTableDescriptor(tdesc)); |
590 | 0 | break; |
591 | 0 | default: |
592 | 0 | DCHECK(false) << "invalid table type: " << tdesc.tableType; |
593 | 17 | } |
594 | | |
595 | 17 | (*tbl)->_tbl_desc_map[tdesc.id] = desc; |
596 | 17 | } |
597 | | |
598 | 135 | for (const auto& tdesc : thrift_tbl.tupleDescriptors) { |
599 | 135 | TupleDescriptor* desc = pool->add(new TupleDescriptor(tdesc)); |
600 | | |
601 | | // fix up table pointer |
602 | 135 | if (tdesc.__isset.tableId) { |
603 | 5 | desc->_table_desc = (*tbl)->get_table_descriptor(tdesc.tableId); |
604 | 5 | DCHECK(desc->_table_desc != nullptr); |
605 | 5 | } |
606 | | |
607 | 135 | (*tbl)->_tuple_desc_map[tdesc.id] = desc; |
608 | 135 | (*tbl)->_row_tuples.emplace_back(tdesc.id); |
609 | 135 | } |
610 | | |
611 | 325 | for (const auto& tdesc : thrift_tbl.slotDescriptors) { |
612 | 325 | SlotDescriptor* slot_d = pool->add(new SlotDescriptor(tdesc)); |
613 | 325 | (*tbl)->_slot_desc_map[tdesc.id] = slot_d; |
614 | | |
615 | | // link to parent |
616 | 325 | auto entry = (*tbl)->_tuple_desc_map.find(tdesc.parent); |
617 | | |
618 | 325 | if (entry == (*tbl)->_tuple_desc_map.end()) { |
619 | 0 | return Status::InternalError("unknown tid in slot descriptor msg"); |
620 | 0 | } |
621 | 325 | entry->second->add_slot(slot_d); |
622 | 325 | } |
623 | | |
624 | 88 | return Status::OK(); |
625 | 88 | } |
626 | | |
627 | 5 | TableDescriptor* DescriptorTbl::get_table_descriptor(TableId id) const { |
628 | | // TODO: is there some boost function to do exactly this? |
629 | 5 | auto i = _tbl_desc_map.find(id); |
630 | | |
631 | 5 | if (i == _tbl_desc_map.end()) { |
632 | 0 | return nullptr; |
633 | 5 | } else { |
634 | 5 | return i->second; |
635 | 5 | } |
636 | 5 | } |
637 | | |
638 | 499 | TupleDescriptor* DescriptorTbl::get_tuple_descriptor(TupleId id) const { |
639 | | // TODO: is there some boost function to do exactly this? |
640 | 499 | auto i = _tuple_desc_map.find(id); |
641 | | |
642 | 499 | if (i == _tuple_desc_map.end()) { |
643 | 4 | return nullptr; |
644 | 495 | } else { |
645 | 495 | return i->second; |
646 | 495 | } |
647 | 499 | } |
648 | | |
649 | 54 | SlotDescriptor* DescriptorTbl::get_slot_descriptor(SlotId id) const { |
650 | | // TODO: is there some boost function to do exactly this? |
651 | 54 | auto i = _slot_desc_map.find(id); |
652 | | |
653 | 54 | if (i == _slot_desc_map.end()) { |
654 | 0 | return nullptr; |
655 | 54 | } else { |
656 | 54 | return i->second; |
657 | 54 | } |
658 | 54 | } |
659 | | |
660 | 0 | std::string DescriptorTbl::debug_string() const { |
661 | 0 | std::stringstream out; |
662 | 0 | out << "tuples:\n"; |
663 | |
|
664 | 0 | for (auto i : _tuple_desc_map) { |
665 | 0 | out << i.second->debug_string() << '\n'; |
666 | 0 | } |
667 | |
|
668 | 0 | return out.str(); |
669 | 0 | } |
670 | | |
671 | | } // namespace doris |