Coverage for python/lsst/dax/apdb/apdbSqlSchema.py: 20%

212 statements  

« prev     ^ index     » next       coverage.py v7.3.3, created at 2023-12-20 17:15 +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 collections.abc import Mapping 

33from typing import Any 

34 

35import felis.types 

36import sqlalchemy 

37from felis import simple 

38from sqlalchemy import ( 

39 DDL, 

40 Column, 

41 ForeignKeyConstraint, 

42 Index, 

43 MetaData, 

44 PrimaryKeyConstraint, 

45 Table, 

46 UniqueConstraint, 

47 event, 

48 inspect, 

49) 

50from sqlalchemy.dialects.postgresql import UUID 

51 

52from .apdbSchema import ApdbSchema, ApdbTables 

53 

54_LOG = logging.getLogger(__name__) 

55 

56 

57# 

58# Copied from daf_butler. 

59# 

60class GUID(sqlalchemy.TypeDecorator): 

61 """Platform-independent GUID type. 

62 

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

64 stringified hex values. 

65 """ 

66 

67 impl = sqlalchemy.CHAR 

68 

69 cache_ok = True 

70 

71 def load_dialect_impl(self, dialect: sqlalchemy.engine.Dialect) -> sqlalchemy.types.TypeEngine: 

72 if dialect.name == "postgresql": 

73 return dialect.type_descriptor(UUID()) 

74 else: 

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

76 

77 def process_bind_param(self, value: Any, dialect: sqlalchemy.engine.Dialect) -> str | None: 

78 if value is None: 

79 return value 

80 

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

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

83 if isinstance(value, int): 

84 value = uuid.UUID(int=value) 

85 elif isinstance(value, bytes): 

86 value = uuid.UUID(bytes=value) 

87 elif isinstance(value, str): 

88 # hexstring 

89 value = uuid.UUID(hex=value) 

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

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

92 

93 if dialect.name == "postgresql": 

94 return str(value) 

95 else: 

96 return "%.32x" % value.int 

97 

98 def process_result_value( 

99 self, value: str | uuid.UUID | None, dialect: sqlalchemy.engine.Dialect 

100 ) -> uuid.UUID | None: 

101 if value is None: 

102 return value 

103 elif isinstance(value, uuid.UUID): 

104 # sqlalchemy 2 converts to UUID internally 

105 return value 

106 else: 

107 return uuid.UUID(hex=value) 

108 

109 

110@enum.unique 

111class ExtraTables(enum.Enum): 

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

113 

114 DiaInsertId = "DiaInsertId" 

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

116 

117 DiaObjectInsertId = "DiaObjectInsertId" 

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

119 

120 DiaSourceInsertId = "DiaSourceInsertId" 

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

122 

123 DiaForcedSourceInsertId = "DiaFSourceInsertId" 

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

125 

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

127 """Return full table name.""" 

128 return prefix + self.value 

129 

130 @classmethod 

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

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

133 corresponding regular tables. 

134 """ 

135 return { 

136 cls.DiaObjectInsertId: ApdbTables.DiaObject, 

137 cls.DiaSourceInsertId: ApdbTables.DiaSource, 

138 cls.DiaForcedSourceInsertId: ApdbTables.DiaForcedSource, 

139 } 

140 

141 

142class ApdbSqlSchema(ApdbSchema): 

143 """Class for management of APDB schema. 

144 

145 Attributes 

146 ---------- 

147 objects : `sqlalchemy.Table` 

148 DiaObject table instance 

149 objects_last : `sqlalchemy.Table` 

150 DiaObjectLast table instance, may be None 

151 sources : `sqlalchemy.Table` 

152 DiaSource table instance 

153 forcedSources : `sqlalchemy.Table` 

154 DiaForcedSource table instance 

155 has_insert_id : `bool` 

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

157 

158 Parameters 

159 ---------- 

160 engine : `sqlalchemy.engine.Engine` 

161 SQLAlchemy engine instance 

162 dia_object_index : `str` 

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

164 for details. 

165 htm_index_column : `str` 

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

167 schema_file : `str` 

168 Name of the YAML schema file. 

169 schema_name : `str`, optional 

170 Name of the schema in YAML files. 

171 prefix : `str`, optional 

172 Prefix to add to all schema elements. 

173 namespace : `str`, optional 

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

175 use_insert_id : `bool`, optional 

176 

177 """ 

178 

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

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

181 

182 def __init__( 

183 self, 

184 engine: sqlalchemy.engine.Engine, 

185 dia_object_index: str, 

186 htm_index_column: str, 

187 schema_file: str, 

188 schema_name: str = "ApdbSchema", 

189 prefix: str = "", 

190 namespace: str | None = None, 

191 use_insert_id: bool = False, 

192 ): 

193 super().__init__(schema_file, schema_name) 

194 

195 self._engine = engine 

196 self._dia_object_index = dia_object_index 

197 self._htm_index_column = htm_index_column 

198 self._prefix = prefix 

199 self._use_insert_id = use_insert_id 

200 

201 self._metadata = MetaData(schema=namespace) 

202 

203 # map YAML column types to SQLAlchemy 

204 self._type_map = { 

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

206 felis.types.Float: sqlalchemy.types.Float, 

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

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

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

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

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

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

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

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

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

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

217 felis.types.Boolean: sqlalchemy.types.Boolean, 

218 } 

219 

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

221 for table in self.pixel_id_tables: 

222 tableDef = self.tableSchemas.get(table) 

223 if not tableDef: 

224 continue 

225 column = simple.Column( 

226 id=f"#{htm_index_column}", 

227 name=htm_index_column, 

228 datatype=felis.types.Long, 

229 nullable=False, 

230 value=None, 

231 description="Pixelization index column.", 

232 table=tableDef, 

233 ) 

234 tableDef.columns.append(column) 

235 

236 # Adjust index if needed 

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

238 tableDef.primary_key.insert(0, column) 

239 

240 if table is ApdbTables.DiaObjectLast: 

241 # use it as a leading PK column 

242 tableDef.primary_key.insert(0, column) 

243 else: 

244 # make a regular index 

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

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

247 tableDef.indexes.append(index) 

248 

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

250 self._apdb_tables = self._make_apdb_tables() 

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

252 

253 self._has_insert_id: bool | None = None 

254 

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

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

257 

258 Parameters 

259 ---------- 

260 drop : `bool`, optional 

261 If True then drop tables before creating new ones. 

262 """ 

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

264 # sense for postgres. 

265 if self._metadata.schema: 

266 dialect = self._engine.dialect 

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

268 create_schema = DDL( 

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

270 ).execute_if(dialect="postgresql") 

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

272 

273 # create all tables (optionally drop first) 

274 if drop: 

275 _LOG.info("dropping all tables") 

276 self._metadata.drop_all(self._engine) 

277 _LOG.info("creating all tables") 

278 self._metadata.create_all(self._engine) 

279 

280 # Reset possibly cached value. 

281 self._has_insert_id = None 

282 

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

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

285 

286 Parameters 

287 ---------- 

288 table_enum : `ApdbTables` or `ExtraTables` 

289 Type of table to return. 

290 

291 Returns 

292 ------- 

293 table : `sqlalchemy.schema.Table` 

294 Table instance. 

295 

296 Raises 

297 ------ 

298 ValueError 

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

300 """ 

301 try: 

302 if isinstance(table_enum, ApdbTables): 

303 return self._apdb_tables[table_enum] 

304 else: 

305 return self._extra_tables[table_enum] 

306 except LookupError: 

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

308 

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

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

311 

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

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

314 

315 Parameters 

316 ---------- 

317 table_enum : `ApdbTables` or `ExtraTables` 

318 Type of table. 

319 

320 Returns 

321 ------- 

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

323 Table instance. 

324 

325 Raises 

326 ------ 

327 ValueError 

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

329 """ 

330 table = self.get_table(table_enum) 

331 exclude_columns = set() 

332 if table_enum in self.pixel_id_tables: 

333 exclude_columns.add(self._htm_index_column) 

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

335 

336 @property 

337 def has_insert_id(self) -> bool: 

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

339 if self._has_insert_id is None: 

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

341 return self._has_insert_id 

342 

343 def _check_insert_id(self) -> bool: 

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

345 inspector = inspect(self._engine) 

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

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

348 

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

350 """Generate schema for regular tables. 

351 

352 Parameters 

353 ---------- 

354 mysql_engine : `str`, optional 

355 MySQL engine type to use for new tables. 

356 """ 

357 tables = {} 

358 for table_enum in ApdbTables: 

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

360 continue 

361 

362 columns = self._tableColumns(table_enum) 

363 constraints = self._tableIndices(table_enum) 

364 table = Table( 

365 table_enum.table_name(self._prefix), 

366 self._metadata, 

367 *columns, 

368 *constraints, 

369 mysql_engine=mysql_engine, 

370 ) 

371 tables[table_enum] = table 

372 

373 return tables 

374 

375 def _make_extra_tables( 

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

377 ) -> Mapping[ExtraTables, Table]: 

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

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

380 if not self._use_insert_id: 

381 return tables 

382 

383 # Parent table needs to be defined first 

384 column_defs = [ 

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

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

387 ] 

388 parent_table = Table( 

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

390 self._metadata, 

391 *column_defs, 

392 mysql_engine=mysql_engine, 

393 ) 

394 tables[ExtraTables.DiaInsertId] = parent_table 

395 

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

397 apdb_table = apdb_tables[apdb_enum] 

398 columns = self._insertIdColumns(table_enum) 

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

400 table = Table( 

401 table_enum.table_name(self._prefix), 

402 self._metadata, 

403 *columns, 

404 *constraints, 

405 mysql_engine=mysql_engine, 

406 ) 

407 tables[table_enum] = table 

408 

409 return tables 

410 

411 def _tableColumns(self, table_name: ApdbTables) -> list[Column]: 

412 """Return set of columns in a table 

413 

414 Parameters 

415 ---------- 

416 table_name : `ApdbTables` 

417 Name of the table. 

418 

419 Returns 

420 ------- 

421 column_defs : `list` 

422 List of `Column` objects. 

423 """ 

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

425 # specially below 

426 table_schema = self.tableSchemas[table_name] 

427 

428 # convert all column dicts into alchemy Columns 

429 column_defs: list[Column] = [] 

430 for column in table_schema.columns: 

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

432 if column.value is not None: 

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

434 if column in table_schema.primary_key: 

435 kwargs.update(autoincrement=False) 

436 ctype = self._type_map[column.datatype] 

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

438 

439 return column_defs 

440 

441 def _tableIndices(self, table_name: ApdbTables) -> list[sqlalchemy.schema.SchemaItem]: 

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

443 

444 Parameters 

445 ---------- 

446 table_name : `ApdbTables` 

447 Name of the table. 

448 info : `dict` 

449 Additional options passed to SQLAlchemy index constructor. 

450 

451 Returns 

452 ------- 

453 index_defs : `list` 

454 List of SQLAlchemy index/constraint objects. 

455 """ 

456 table_schema = self.tableSchemas[table_name] 

457 

458 # convert all index dicts into alchemy Columns 

459 index_defs: list[sqlalchemy.schema.SchemaItem] = [] 

460 if table_schema.primary_key: 

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

462 for index in table_schema.indexes: 

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

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

465 for constraint in table_schema.constraints: 

466 constr_name: str | None = None 

467 if constraint.name: 

468 constr_name = self._prefix + constraint.name 

469 if isinstance(constraint, simple.UniqueConstraint): 

470 index_defs.append( 

471 UniqueConstraint(*[column.name for column in constraint.columns], name=constr_name) 

472 ) 

473 

474 return index_defs 

475 

476 def _insertIdColumns(self, table_enum: ExtraTables) -> list[Column]: 

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

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

479 insert_id_tables = ExtraTables.insert_id_tables() 

480 if table_enum in insert_id_tables: 

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

482 else: 

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

484 return column_defs 

485 

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

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

488 table_schema = self.tableSchemas[table_enum] 

489 column_defs: list[Column] = [] 

490 for column in table_schema.primary_key: 

491 ctype = self._type_map[column.datatype] 

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

493 return column_defs 

494 

495 def _insertIdIndices( 

496 self, 

497 table_enum: ExtraTables, 

498 apdb_table: sqlalchemy.schema.Table, 

499 parent_table: sqlalchemy.schema.Table, 

500 ) -> list[sqlalchemy.schema.SchemaItem]: 

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

502 index_defs: list[sqlalchemy.schema.SchemaItem] = [] 

503 

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

505 insert_id_tables = ExtraTables.insert_id_tables() 

506 if table_enum in insert_id_tables: 

507 # PK is the same as for original table 

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

509 index_defs.append(PrimaryKeyConstraint(*pk_names)) 

510 # Non-unique index on insert_id column. 

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

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

513 # Foreign key to original table 

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

515 index_defs.append( 

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

517 ) 

518 # Foreign key to parent table 

519 index_defs.append( 

520 ForeignKeyConstraint( 

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

522 ) 

523 ) 

524 else: 

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

526 return index_defs 

527 

528 @classmethod 

529 def _getDoubleType(cls, engine: sqlalchemy.engine.Engine) -> type | sqlalchemy.types.TypeEngine: 

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

531 

532 Parameters 

533 ---------- 

534 engine : `sqlalchemy.engine.Engine` 

535 Database engine. 

536 

537 Returns 

538 ------- 

539 type_object : `object` 

540 Database-specific type definition. 

541 """ 

542 if engine.name == "mysql": 

543 from sqlalchemy.dialects.mysql import DOUBLE 

544 

545 return DOUBLE(asdecimal=False) 

546 elif engine.name == "postgresql": 

547 from sqlalchemy.dialects.postgresql import DOUBLE_PRECISION 

548 

549 return DOUBLE_PRECISION 

550 elif engine.name == "oracle": 

551 from sqlalchemy.dialects.oracle import DOUBLE_PRECISION 

552 

553 return DOUBLE_PRECISION 

554 elif engine.name == "sqlite": 

555 # all floats in sqlite are 8-byte 

556 from sqlalchemy.dialects.sqlite import REAL 

557 

558 return REAL 

559 else: 

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