Coverage for python/lsst/dax/apdb/apdbSqlSchema.py: 18%

209 statements  

« prev     ^ index     » next       coverage.py v6.5.0, created at 2023-01-19 10:10 +0000

1# This file is part of dax_apdb. 

2# 

3# Developed for the LSST Data Management System. 

4# This product includes software developed by the LSST Project 

5# (http://www.lsst.org). 

6# See the COPYRIGHT file at the top-level directory of this distribution 

7# for details of code ownership. 

8# 

9# This program is free software: you can redistribute it and/or modify 

10# it under the terms of the GNU General Public License as published by 

11# the Free Software Foundation, either version 3 of the License, or 

12# (at your option) any later version. 

13# 

14# This program is distributed in the hope that it will be useful, 

15# but WITHOUT ANY WARRANTY; without even the implied warranty of 

16# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 

17# GNU General Public License for more details. 

18# 

19# You should have received a copy of the GNU General Public License 

20# along with this program. If not, see <http://www.gnu.org/licenses/>. 

21 

22"""Module responsible for APDB schema operations. 

23""" 

24 

25from __future__ import annotations 

26 

27__all__ = ["ApdbSqlSchema", "ExtraTables"] 

28 

29import enum 

30import logging 

31import uuid 

32from typing import Any, Dict, List, Mapping, Optional, Type 

33 

34import felis.types 

35import sqlalchemy 

36from felis import simple 

37from sqlalchemy import ( 

38 DDL, 

39 Column, 

40 ForeignKeyConstraint, 

41 Index, 

42 MetaData, 

43 PrimaryKeyConstraint, 

44 Table, 

45 UniqueConstraint, 

46 event, 

47 inspect, 

48) 

49from sqlalchemy.dialects.postgresql import UUID 

50 

51from .apdbSchema import ApdbSchema, ApdbTables 

52 

53_LOG = logging.getLogger(__name__) 

54 

55 

56# 

57# Copied from daf_butler. 

58# 

59class GUID(sqlalchemy.TypeDecorator): 

60 """Platform-independent GUID type. 

61 

62 Uses PostgreSQL's UUID type, otherwise uses CHAR(32), storing as 

63 stringified hex values. 

64 """ 

65 

66 impl = sqlalchemy.CHAR 

67 

68 cache_ok = True 

69 

70 def load_dialect_impl(self, dialect: sqlalchemy.Dialect) -> sqlalchemy.TypeEngine: 

71 if dialect.name == "postgresql": 

72 return dialect.type_descriptor(UUID()) 

73 else: 

74 return dialect.type_descriptor(sqlalchemy.CHAR(32)) 

75 

76 def process_bind_param(self, value: Any, dialect: sqlalchemy.Dialect) -> Optional[str]: 

77 if value is None: 

78 return value 

79 

80 # Coerce input to UUID type, in general having UUID on input is the 

81 # only thing that we want but there is code right now that uses ints. 

82 if isinstance(value, int): 

83 value = uuid.UUID(int=value) 

84 elif isinstance(value, bytes): 

85 value = uuid.UUID(bytes=value) 

86 elif isinstance(value, str): 

87 # hexstring 

88 value = uuid.UUID(hex=value) 

89 elif not isinstance(value, uuid.UUID): 

90 raise TypeError(f"Unexpected type of a bind value: {type(value)}") 

91 

92 if dialect.name == "postgresql": 

93 return str(value) 

94 else: 

95 return "%.32x" % value.int 

96 

97 def process_result_value(self, value: Optional[str], dialect: sqlalchemy.Dialect) -> Optional[uuid.UUID]: 

98 if value is None: 

99 return value 

100 else: 

101 return uuid.UUID(hex=value) 

102 

103 

104@enum.unique 

105class ExtraTables(enum.Enum): 

106 """Names of the tables used for tracking insert IDs.""" 

107 

108 DiaInsertId = "DiaInsertId" 

109 """Name of the table for insert ID records.""" 

110 

111 DiaObjectInsertId = "DiaObjectInsertId" 

112 """Name of the table for DIAObject insert ID records.""" 

113 

114 DiaSourceInsertId = "DiaSourceInsertId" 

115 """Name of the table for DIASource insert ID records.""" 

116 

117 DiaForcedSourceInsertId = "DiaFSourceInsertId" 

118 """Name of the table for DIAForcedSource insert ID records.""" 

119 

120 def table_name(self, prefix: str = "") -> str: 

121 """Return full table name.""" 

122 return prefix + self.value 

123 

124 @classmethod 

125 def insert_id_tables(cls) -> Mapping[ExtraTables, ApdbTables]: 

126 """Return mapping of tables used for insert ID tracking to their 

127 corresponding regular tables.""" 

128 return { 

129 cls.DiaObjectInsertId: ApdbTables.DiaObject, 

130 cls.DiaSourceInsertId: ApdbTables.DiaSource, 

131 cls.DiaForcedSourceInsertId: ApdbTables.DiaForcedSource, 

132 } 

133 

134 

135class ApdbSqlSchema(ApdbSchema): 

136 """Class for management of APDB schema. 

137 

138 Attributes 

139 ---------- 

140 objects : `sqlalchemy.Table` 

141 DiaObject table instance 

142 objects_last : `sqlalchemy.Table` 

143 DiaObjectLast table instance, may be None 

144 sources : `sqlalchemy.Table` 

145 DiaSource table instance 

146 forcedSources : `sqlalchemy.Table` 

147 DiaForcedSource table instance 

148 has_insert_id : `bool` 

149 If true then schema has tables for insert ID tracking. 

150 

151 Parameters 

152 ---------- 

153 engine : `sqlalchemy.engine.Engine` 

154 SQLAlchemy engine instance 

155 dia_object_index : `str` 

156 Indexing mode for DiaObject table, see `ApdbSqlConfig.dia_object_index` 

157 for details. 

158 htm_index_column : `str` 

159 Name of a HTM index column for DiaObject and DiaSource tables. 

160 schema_file : `str` 

161 Name of the YAML schema file. 

162 schema_name : `str`, optional 

163 Name of the schema in YAML files. 

164 prefix : `str`, optional 

165 Prefix to add to all schema elements. 

166 namespace : `str`, optional 

167 Namespace (or schema name) to use for all APDB tables. 

168 use_insert_id : `bool`, optional 

169 

170 """ 

171 

172 pixel_id_tables = (ApdbTables.DiaObject, ApdbTables.DiaObjectLast, ApdbTables.DiaSource) 

173 """Tables that need pixelId column for spatial indexing.""" 

174 

175 def __init__( 

176 self, 

177 engine: sqlalchemy.engine.Engine, 

178 dia_object_index: str, 

179 htm_index_column: str, 

180 schema_file: str, 

181 schema_name: str = "ApdbSchema", 

182 prefix: str = "", 

183 namespace: Optional[str] = None, 

184 use_insert_id: bool = False, 

185 ): 

186 

187 super().__init__(schema_file, schema_name) 

188 

189 self._engine = engine 

190 self._dia_object_index = dia_object_index 

191 self._htm_index_column = htm_index_column 

192 self._prefix = prefix 

193 self._use_insert_id = use_insert_id 

194 

195 self._metadata = MetaData(schema=namespace) 

196 

197 # map YAML column types to SQLAlchemy 

198 self._type_map = { 

199 felis.types.Double: self._getDoubleType(engine), 

200 felis.types.Float: sqlalchemy.types.Float, 

201 felis.types.Timestamp: sqlalchemy.types.TIMESTAMP, 

202 felis.types.Long: sqlalchemy.types.BigInteger, 

203 felis.types.Int: sqlalchemy.types.Integer, 

204 felis.types.Short: sqlalchemy.types.Integer, 

205 felis.types.Byte: sqlalchemy.types.Integer, 

206 felis.types.Binary: sqlalchemy.types.LargeBinary, 

207 felis.types.Text: sqlalchemy.types.CHAR, 

208 felis.types.String: sqlalchemy.types.CHAR, 

209 felis.types.Char: sqlalchemy.types.CHAR, 

210 felis.types.Unicode: sqlalchemy.types.CHAR, 

211 felis.types.Boolean: sqlalchemy.types.Boolean, 

212 } 

213 

214 # Add pixelId column and index to tables that need it 

215 for table in self.pixel_id_tables: 

216 tableDef = self.tableSchemas.get(table) 

217 if not tableDef: 

218 continue 

219 column = simple.Column( 

220 id=f"#{htm_index_column}", 

221 name=htm_index_column, 

222 datatype=felis.types.Long, 

223 nullable=False, 

224 value=None, 

225 description="Pixelization index column.", 

226 table=tableDef, 

227 ) 

228 tableDef.columns.append(column) 

229 

230 # Adjust index if needed 

231 if table == ApdbTables.DiaObject and self._dia_object_index == "pix_id_iov": 

232 tableDef.primary_key.insert(0, column) 

233 

234 if table is ApdbTables.DiaObjectLast: 

235 # use it as a leading PK column 

236 tableDef.primary_key.insert(0, column) 

237 else: 

238 # make a regular index 

239 name = f"IDX_{tableDef.name}_{htm_index_column}" 

240 index = simple.Index(id=f"#{name}", name=name, columns=[column]) 

241 tableDef.indexes.append(index) 

242 

243 # generate schema for all tables, must be called last 

244 self._apdb_tables = self._make_apdb_tables() 

245 self._extra_tables = self._make_extra_tables(self._apdb_tables) 

246 

247 self._has_insert_id: bool | None = None 

248 

249 def makeSchema(self, drop: bool = False) -> None: 

250 """Create or re-create all tables. 

251 

252 Parameters 

253 ---------- 

254 drop : `bool`, optional 

255 If True then drop tables before creating new ones. 

256 """ 

257 # Create namespace if it does not exist yet, for now this only makes 

258 # sense for postgres. 

259 if self._metadata.schema: 

260 dialect = self._engine.dialect 

261 quoted_schema = dialect.preparer(dialect).quote_schema(self._metadata.schema) 

262 create_schema = DDL( 

263 "CREATE SCHEMA IF NOT EXISTS %(schema)s", context={"schema": quoted_schema} 

264 ).execute_if(dialect="postgresql") 

265 event.listen(self._metadata, "before_create", create_schema) 

266 

267 # create all tables (optionally drop first) 

268 if drop: 

269 _LOG.info("dropping all tables") 

270 self._metadata.drop_all(self._engine) 

271 _LOG.info("creating all tables") 

272 self._metadata.create_all(self._engine) 

273 

274 # Reset possibly cached value. 

275 self._has_insert_id = None 

276 

277 def get_table(self, table_enum: ApdbTables | ExtraTables) -> Table: 

278 """Return SQLAlchemy table instance for a specified table type/enum. 

279 

280 Parameters 

281 ---------- 

282 table_enum : `ApdbTables` or `ExtraTables` 

283 Type of table to return. 

284 

285 Returns 

286 ------- 

287 table : `sqlalchemy.schema.Table` 

288 Table instance. 

289 

290 Raises 

291 ------ 

292 ValueError 

293 Raised if ``table_enum`` is not valid for this database. 

294 """ 

295 try: 

296 if isinstance(table_enum, ApdbTables): 

297 return self._apdb_tables[table_enum] 

298 else: 

299 return self._extra_tables[table_enum] 

300 except LookupError: 

301 raise ValueError(f"Table type {table_enum} does not exist in the schema") from None 

302 

303 def get_apdb_columns(self, table_enum: ApdbTables | ExtraTables) -> list[Column]: 

304 """Return list of columns defined for a table in APDB schema. 

305 

306 Returned list excludes columns that are implementation-specific, e.g. 

307 ``pixelId`` column is not include in the returned list. 

308 

309 Parameters 

310 ---------- 

311 table_enum : `ApdbTables` or `ExtraTables` 

312 Type of table. 

313 

314 Returns 

315 ------- 

316 table : `list` [`sqlalchemy.schema.Column`] 

317 Table instance. 

318 

319 Raises 

320 ------ 

321 ValueError 

322 Raised if ``table_enum`` is not valid for this database. 

323 """ 

324 table = self.get_table(table_enum) 

325 exclude_columns = set() 

326 if table_enum in self.pixel_id_tables: 

327 exclude_columns.add(self._htm_index_column) 

328 return [column for column in table.columns if column.name not in exclude_columns] 

329 

330 @property 

331 def has_insert_id(self) -> bool: 

332 """Whether insert ID tables are to be used (`bool`).""" 

333 if self._has_insert_id is None: 

334 self._has_insert_id = self._use_insert_id and self._check_insert_id() 

335 return self._has_insert_id 

336 

337 def _check_insert_id(self) -> bool: 

338 """Check whether database has tables for tracking insert IDs.""" 

339 inspector = inspect(self._engine) 

340 db_tables = set(inspector.get_table_names(schema=self._metadata.schema)) 

341 return ExtraTables.DiaInsertId.table_name(self._prefix) in db_tables 

342 

343 def _make_apdb_tables(self, mysql_engine: str = "InnoDB") -> Mapping[ApdbTables, Table]: 

344 """Generate schema for regular tables. 

345 

346 Parameters 

347 ---------- 

348 mysql_engine : `str`, optional 

349 MySQL engine type to use for new tables. 

350 """ 

351 tables = {} 

352 for table_enum in ApdbTables: 

353 

354 if table_enum is ApdbTables.DiaObjectLast and self._dia_object_index != "last_object_table": 

355 continue 

356 

357 columns = self._tableColumns(table_enum) 

358 constraints = self._tableIndices(table_enum) 

359 table = Table( 

360 table_enum.table_name(self._prefix), 

361 self._metadata, 

362 *columns, 

363 *constraints, 

364 mysql_engine=mysql_engine, 

365 ) 

366 tables[table_enum] = table 

367 

368 return tables 

369 

370 def _make_extra_tables( 

371 self, apdb_tables: Mapping[ApdbTables, Table], mysql_engine: str = "InnoDB" 

372 ) -> Mapping[ExtraTables, Table]: 

373 """Generate schema for insert ID tables.""" 

374 tables: dict[ExtraTables, Table] = {} 

375 if not self._use_insert_id: 

376 return tables 

377 

378 # Parent table needs to be defined first 

379 column_defs = [ 

380 Column("insert_id", GUID, primary_key=True), 

381 Column("insert_time", sqlalchemy.types.TIMESTAMP, nullable=False), 

382 ] 

383 parent_table = Table( 

384 ExtraTables.DiaInsertId.table_name(self._prefix), 

385 self._metadata, 

386 *column_defs, 

387 mysql_engine=mysql_engine, 

388 ) 

389 tables[ExtraTables.DiaInsertId] = parent_table 

390 

391 for table_enum, apdb_enum in ExtraTables.insert_id_tables().items(): 

392 apdb_table = apdb_tables[apdb_enum] 

393 columns = self._insertIdColumns(table_enum) 

394 constraints = self._insertIdIndices(table_enum, apdb_table, parent_table) 

395 table = Table( 

396 table_enum.table_name(self._prefix), 

397 self._metadata, 

398 *columns, 

399 *constraints, 

400 mysql_engine=mysql_engine, 

401 ) 

402 tables[table_enum] = table 

403 

404 return tables 

405 

406 def _tableColumns(self, table_name: ApdbTables) -> List[Column]: 

407 """Return set of columns in a table 

408 

409 Parameters 

410 ---------- 

411 table_name : `ApdbTables` 

412 Name of the table. 

413 

414 Returns 

415 ------- 

416 column_defs : `list` 

417 List of `Column` objects. 

418 """ 

419 # get the list of columns in primary key, they are treated somewhat 

420 # specially below 

421 table_schema = self.tableSchemas[table_name] 

422 

423 # convert all column dicts into alchemy Columns 

424 column_defs = [] 

425 for column in table_schema.columns: 

426 kwargs: Dict[str, Any] = dict(nullable=column.nullable) 

427 if column.value is not None: 

428 kwargs.update(server_default=str(column.value)) 

429 if column in table_schema.primary_key: 

430 kwargs.update(autoincrement=False) 

431 ctype = self._type_map[column.datatype] 

432 column_defs.append(Column(column.name, ctype, **kwargs)) 

433 

434 return column_defs 

435 

436 def _tableIndices(self, table_name: ApdbTables) -> List[sqlalchemy.schema.Constraint]: 

437 """Return set of constraints/indices in a table 

438 

439 Parameters 

440 ---------- 

441 table_name : `ApdbTables` 

442 Name of the table. 

443 info : `dict` 

444 Additional options passed to SQLAlchemy index constructor. 

445 

446 Returns 

447 ------- 

448 index_defs : `list` 

449 List of SQLAlchemy index/constraint objects. 

450 """ 

451 

452 table_schema = self.tableSchemas[table_name] 

453 

454 # convert all index dicts into alchemy Columns 

455 index_defs: List[sqlalchemy.schema.Constraint] = [] 

456 if table_schema.primary_key: 

457 index_defs.append(PrimaryKeyConstraint(*[column.name for column in table_schema.primary_key])) 

458 for index in table_schema.indexes: 

459 name = self._prefix + index.name if index.name else "" 

460 index_defs.append(Index(name, *[column.name for column in index.columns])) 

461 for constraint in table_schema.constraints: 

462 kwargs = {} 

463 if constraint.name: 

464 kwargs["name"] = self._prefix + constraint.name 

465 if isinstance(constraint, simple.UniqueConstraint): 

466 index_defs.append(UniqueConstraint(*[column.name for column in constraint.columns], **kwargs)) 

467 

468 return index_defs 

469 

470 def _insertIdColumns(self, table_enum: ExtraTables) -> List[Column]: 

471 """Return list of columns for insert ID tables.""" 

472 column_defs: list[Column] = [Column("insert_id", GUID, nullable=False)] 

473 insert_id_tables = ExtraTables.insert_id_tables() 

474 if table_enum in insert_id_tables: 

475 column_defs += self._tablePkColumns(insert_id_tables[table_enum]) 

476 else: 

477 assert False, "Above branches have to cover all enum values" 

478 return column_defs 

479 

480 def _tablePkColumns(self, table_enum: ApdbTables) -> list[Column]: 

481 """Return a list of columns for table PK.""" 

482 table_schema = self.tableSchemas[table_enum] 

483 column_defs = [] 

484 for column in table_schema.primary_key: 

485 ctype = self._type_map[column.datatype] 

486 column_defs.append(Column(column.name, ctype, nullable=False, autoincrement=False)) 

487 return column_defs 

488 

489 def _insertIdIndices( 

490 self, 

491 table_enum: ExtraTables, 

492 apdb_table: sqlalchemy.schema.Table, 

493 parent_table: sqlalchemy.schema.Table, 

494 ) -> List[sqlalchemy.schema.Constraint]: 

495 """Return set of constraints/indices for insert ID tables.""" 

496 index_defs: List[sqlalchemy.schema.Constraint] = [] 

497 

498 # Special case for insert ID tables that are not in felis schema. 

499 insert_id_tables = ExtraTables.insert_id_tables() 

500 if table_enum in insert_id_tables: 

501 # PK is the same as for original table 

502 pk_names = [column.name for column in self._tablePkColumns(insert_id_tables[table_enum])] 

503 index_defs.append(PrimaryKeyConstraint(*pk_names)) 

504 # Non-unique index on insert_id column. 

505 name = self._prefix + table_enum.name + "_idx" 

506 index_defs.append(Index(name, "insert_id")) 

507 # Foreign key to original table 

508 pk_columns = [apdb_table.columns[column] for column in pk_names] 

509 index_defs.append( 

510 ForeignKeyConstraint(pk_names, pk_columns, onupdate="CASCADE", ondelete="CASCADE") 

511 ) 

512 # Foreign key to parent table 

513 index_defs.append( 

514 ForeignKeyConstraint( 

515 ["insert_id"], [parent_table.columns["insert_id"]], onupdate="CASCADE", ondelete="CASCADE" 

516 ) 

517 ) 

518 else: 

519 assert False, "Above branches have to cover all enum values" 

520 return index_defs 

521 

522 @classmethod 

523 def _getDoubleType(cls, engine: sqlalchemy.engine.Engine) -> Type: 

524 """DOUBLE type is database-specific, select one based on dialect. 

525 

526 Parameters 

527 ---------- 

528 engine : `sqlalchemy.engine.Engine` 

529 Database engine. 

530 

531 Returns 

532 ------- 

533 type_object : `object` 

534 Database-specific type definition. 

535 """ 

536 if engine.name == "mysql": 

537 from sqlalchemy.dialects.mysql import DOUBLE 

538 

539 return DOUBLE(asdecimal=False) 

540 elif engine.name == "postgresql": 

541 from sqlalchemy.dialects.postgresql import DOUBLE_PRECISION 

542 

543 return DOUBLE_PRECISION 

544 elif engine.name == "oracle": 

545 from sqlalchemy.dialects.oracle import DOUBLE_PRECISION 

546 

547 return DOUBLE_PRECISION 

548 elif engine.name == "sqlite": 

549 # all floats in sqlite are 8-byte 

550 from sqlalchemy.dialects.sqlite import REAL 

551 

552 return REAL 

553 else: 

554 raise TypeError("cannot determine DOUBLE type, unexpected dialect: " + engine.name)