Coverage for python/lsst/daf/butler/registry/databases/postgresql.py: 31%

198 statements  

« prev     ^ index     » next       coverage.py v7.3.2, created at 2023-12-01 10:59 +0000

1# This file is part of daf_butler. 

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 software is dual licensed under the GNU General Public License and also 

10# under a 3-clause BSD license. Recipients may choose which of these licenses 

11# to use; please see the files gpl-3.0.txt and/or bsd_license.txt, 

12# respectively. If you choose the GPL option then the following text applies 

13# (but note that there is still no warranty even if you opt for BSD instead): 

14# 

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

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

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

18# (at your option) any later version. 

19# 

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

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

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

23# GNU General Public License for more details. 

24# 

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

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

27from __future__ import annotations 

28 

29from ... import ddl, time_utils 

30 

31__all__ = ["PostgresqlDatabase"] 

32 

33from collections.abc import Iterable, Iterator, Mapping 

34from contextlib import closing, contextmanager 

35from typing import Any 

36 

37import psycopg2 

38import sqlalchemy 

39import sqlalchemy.dialects.postgresql 

40from sqlalchemy import sql 

41 

42from ..._named import NamedValueAbstractSet 

43from ..._timespan import Timespan, TimespanDatabaseRepresentation 

44from ..interfaces import Database 

45from ..nameShrinker import NameShrinker 

46 

47 

48class PostgresqlDatabase(Database): 

49 """An implementation of the `Database` interface for PostgreSQL. 

50 

51 Parameters 

52 ---------- 

53 connection : `sqlalchemy.engine.Connection` 

54 An existing connection created by a previous call to `connect`. 

55 origin : `int` 

56 An integer ID that should be used as the default for any datasets, 

57 quanta, or other entities that use a (autoincrement, origin) compound 

58 primary key. 

59 namespace : `str`, optional 

60 The namespace (schema) this database is associated with. If `None`, 

61 the default schema for the connection is used (which may be `None`). 

62 writeable : `bool`, optional 

63 If `True`, allow write operations on the database, including 

64 ``CREATE TABLE``. 

65 

66 Notes 

67 ----- 

68 This currently requires the psycopg2 driver to be used as the backend for 

69 SQLAlchemy. Running the tests for this class requires the 

70 ``testing.postgresql`` be installed, which we assume indicates that a 

71 PostgreSQL server is installed and can be run locally in userspace. 

72 

73 Some functionality provided by this class (and used by `Registry`) requires 

74 the ``btree_gist`` PostgreSQL server extension to be installed an enabled 

75 on the database being connected to; this is checked at connection time. 

76 """ 

77 

78 def __init__( 

79 self, 

80 *, 

81 engine: sqlalchemy.engine.Engine, 

82 origin: int, 

83 namespace: str | None = None, 

84 writeable: bool = True, 

85 ): 

86 super().__init__(origin=origin, engine=engine, namespace=namespace) 

87 with engine.connect() as connection: 

88 # `Any` to make mypy ignore the line below, can't use type: ignore 

89 dbapi: Any = connection.connection 

90 try: 

91 dsn = dbapi.get_dsn_parameters() 

92 except (AttributeError, KeyError) as err: 

93 raise RuntimeError("Only the psycopg2 driver for PostgreSQL is supported.") from err 

94 if namespace is None: 

95 query = sql.select(sql.func.current_schema()) 

96 namespace = connection.execute(query).scalar() 

97 query_text = "SELECT COUNT(*) FROM pg_extension WHERE extname='btree_gist';" 

98 if not connection.execute(sqlalchemy.text(query_text)).scalar(): 

99 raise RuntimeError( 

100 "The Butler PostgreSQL backend requires the btree_gist extension. " 

101 "As extensions are enabled per-database, this may require an administrator to run " 

102 "`CREATE EXTENSION btree_gist;` in a database before a butler client for it is " 

103 " initialized." 

104 ) 

105 self.namespace = namespace 

106 self.dbname = dsn.get("dbname") 

107 self._writeable = writeable 

108 self._shrinker = NameShrinker(self.dialect.max_identifier_length) 

109 

110 @classmethod 

111 def makeEngine( 

112 cls, uri: str | sqlalchemy.engine.URL, *, writeable: bool = True 

113 ) -> sqlalchemy.engine.Engine: 

114 return sqlalchemy.engine.create_engine(uri, pool_size=1) 

115 

116 @classmethod 

117 def fromEngine( 

118 cls, 

119 engine: sqlalchemy.engine.Engine, 

120 *, 

121 origin: int, 

122 namespace: str | None = None, 

123 writeable: bool = True, 

124 ) -> Database: 

125 return cls(engine=engine, origin=origin, namespace=namespace, writeable=writeable) 

126 

127 @contextmanager 

128 def _transaction( 

129 self, 

130 *, 

131 interrupting: bool = False, 

132 savepoint: bool = False, 

133 lock: Iterable[sqlalchemy.schema.Table] = (), 

134 for_temp_tables: bool = False, 

135 ) -> Iterator[tuple[bool, sqlalchemy.engine.Connection]]: 

136 with super()._transaction(interrupting=interrupting, savepoint=savepoint, lock=lock) as ( 

137 is_new, 

138 connection, 

139 ): 

140 if is_new: 

141 # pgbouncer with transaction-level pooling (which we aim to 

142 # support) says that SET cannot be used, except for a list of 

143 # "Startup parameters" that includes "timezone" (see 

144 # https://www.pgbouncer.org/features.html#fnref:0). But I 

145 # don't see "timezone" in PostgreSQL's list of parameters 

146 # passed when creating a new connection 

147 # (https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-PARAMKEYWORDS). 

148 # Given that the pgbouncer docs say, "PgBouncer detects their 

149 # changes and so it can guarantee they remain consistent for 

150 # the client", I assume we can use "SET TIMESPAN" and pgbouncer 

151 # will take care of clients that share connections being set 

152 # consistently. And if that assumption is wrong, we should 

153 # still probably be okay, since all clients should be Butler 

154 # clients, and they'll all be setting the same thing. 

155 # 

156 # The "SET TRANSACTION READ ONLY" should also be safe, because 

157 # it only ever acts on the current transaction; I think it's 

158 # not included in pgbouncer's declaration that SET is 

159 # incompatible with transaction-level pooling because 

160 # PostgreSQL actually considers SET TRANSACTION to be a 

161 # fundamentally different statement from SET (they have their 

162 # own distinct doc pages, at least). 

163 if not (self.isWriteable() or for_temp_tables): 

164 # PostgreSQL permits writing to temporary tables inside 

165 # read-only transactions, but it doesn't permit creating 

166 # them. 

167 with closing(connection.connection.cursor()) as cursor: 

168 cursor.execute("SET TRANSACTION READ ONLY") 

169 cursor.execute("SET TIME ZONE 0") 

170 else: 

171 with closing(connection.connection.cursor()) as cursor: 

172 # Make timestamps UTC, because we didn't use TIMESTAMPZ 

173 # for the column type. When we can tolerate a schema 

174 # change, we should change that type and remove this 

175 # line. 

176 cursor.execute("SET TIME ZONE 0") 

177 yield is_new, connection 

178 

179 @contextmanager 

180 def temporary_table( 

181 self, spec: ddl.TableSpec, name: str | None = None 

182 ) -> Iterator[sqlalchemy.schema.Table]: 

183 # Docstring inherited. 

184 with self.transaction(for_temp_tables=True), super().temporary_table(spec, name) as table: 

185 yield table 

186 

187 def _lockTables( 

188 self, connection: sqlalchemy.engine.Connection, tables: Iterable[sqlalchemy.schema.Table] = () 

189 ) -> None: 

190 # Docstring inherited. 

191 for table in tables: 

192 connection.execute(sqlalchemy.text(f"LOCK TABLE {table.key} IN EXCLUSIVE MODE")) 

193 

194 def isWriteable(self) -> bool: 

195 return self._writeable 

196 

197 def __str__(self) -> str: 

198 return f"PostgreSQL@{self.dbname}:{self.namespace}" 

199 

200 def shrinkDatabaseEntityName(self, original: str) -> str: 

201 return self._shrinker.shrink(original) 

202 

203 def expandDatabaseEntityName(self, shrunk: str) -> str: 

204 return self._shrinker.expand(shrunk) 

205 

206 def _convertExclusionConstraintSpec( 

207 self, 

208 table: str, 

209 spec: tuple[str | type[TimespanDatabaseRepresentation], ...], 

210 metadata: sqlalchemy.MetaData, 

211 ) -> sqlalchemy.schema.Constraint: 

212 # Docstring inherited. 

213 args: list[tuple[sqlalchemy.schema.Column, str]] = [] 

214 names = ["excl"] 

215 for item in spec: 

216 if isinstance(item, str): 

217 args.append((sqlalchemy.schema.Column(item), "=")) 

218 names.append(item) 

219 elif issubclass(item, TimespanDatabaseRepresentation): 

220 assert item is self.getTimespanRepresentation() 

221 args.append((sqlalchemy.schema.Column(TimespanDatabaseRepresentation.NAME), "&&")) 

222 names.append(TimespanDatabaseRepresentation.NAME) 

223 return sqlalchemy.dialects.postgresql.ExcludeConstraint( 

224 *args, 

225 name=self.shrinkDatabaseEntityName("_".join(names)), 

226 ) 

227 

228 def _make_temporary_table( 

229 self, 

230 connection: sqlalchemy.engine.Connection, 

231 spec: ddl.TableSpec, 

232 name: str | None = None, 

233 **kwargs: Any, 

234 ) -> sqlalchemy.schema.Table: 

235 # Docstring inherited 

236 # Adding ON COMMIT DROP here is really quite defensive: we already 

237 # manually drop the table at the end of the temporary_table context 

238 # manager, and that will usually happen first. But this will guarantee 

239 # that we drop the table at the end of the transaction even if the 

240 # connection lasts longer, and that's good citizenship when connections 

241 # may be multiplexed by e.g. pgbouncer. 

242 return super()._make_temporary_table(connection, spec, name, postgresql_on_commit="DROP", **kwargs) 

243 

244 @classmethod 

245 def getTimespanRepresentation(cls) -> type[TimespanDatabaseRepresentation]: 

246 # Docstring inherited. 

247 return _RangeTimespanRepresentation 

248 

249 def replace(self, table: sqlalchemy.schema.Table, *rows: dict) -> None: 

250 self.assertTableWriteable(table, f"Cannot replace into read-only table {table}.") 

251 if not rows: 

252 return 

253 # This uses special support for UPSERT in PostgreSQL backend: 

254 # https://docs.sqlalchemy.org/en/13/dialects/postgresql.html#insert-on-conflict-upsert 

255 query = sqlalchemy.dialects.postgresql.dml.insert(table) 

256 # In the SET clause assign all columns using special `excluded` 

257 # pseudo-table. If some column in the table does not appear in the 

258 # INSERT list this will set it to NULL. 

259 excluded = query.excluded 

260 data = { 

261 column.name: getattr(excluded, column.name) 

262 for column in table.columns 

263 if column.name not in table.primary_key 

264 } 

265 query = query.on_conflict_do_update(constraint=table.primary_key, set_=data) 

266 with self._transaction() as (_, connection): 

267 connection.execute(query, rows) 

268 

269 def ensure(self, table: sqlalchemy.schema.Table, *rows: dict, primary_key_only: bool = False) -> int: 

270 # Docstring inherited. 

271 self.assertTableWriteable(table, f"Cannot ensure into read-only table {table}.") 

272 if not rows: 

273 return 0 

274 # Like `replace`, this uses UPSERT. 

275 base_insert = sqlalchemy.dialects.postgresql.dml.insert(table) 

276 if primary_key_only: 

277 query = base_insert.on_conflict_do_nothing(constraint=table.primary_key) 

278 else: 

279 query = base_insert.on_conflict_do_nothing() 

280 with self._transaction() as (_, connection): 

281 return connection.execute(query, rows).rowcount 

282 

283 def constant_rows( 

284 self, 

285 fields: NamedValueAbstractSet[ddl.FieldSpec], 

286 *rows: dict, 

287 name: str | None = None, 

288 ) -> sqlalchemy.sql.FromClause: 

289 # Docstring inherited. 

290 return super().constant_rows(fields, *rows, name=name) 

291 

292 

293class _RangeTimespanType(sqlalchemy.TypeDecorator): 

294 """A single-column `Timespan` representation usable only with 

295 PostgreSQL. 

296 

297 This type should be able to take advantage of PostgreSQL's built-in 

298 range operators, and the indexing and EXCLUSION table constraints built 

299 off of them. 

300 """ 

301 

302 impl = sqlalchemy.dialects.postgresql.INT8RANGE 

303 

304 cache_ok = True 

305 

306 def process_bind_param( 

307 self, value: Timespan | None, dialect: sqlalchemy.engine.Dialect 

308 ) -> psycopg2.extras.NumericRange | None: 

309 if value is None: 

310 return None 

311 if not isinstance(value, Timespan): 

312 raise TypeError(f"Unsupported type: {type(value)}, expected Timespan.") 

313 if value.isEmpty(): 

314 return psycopg2.extras.NumericRange(empty=True) 

315 else: 

316 converter = time_utils.TimeConverter() 

317 assert value._nsec[0] >= converter.min_nsec, "Guaranteed by Timespan.__init__." 

318 assert value._nsec[1] <= converter.max_nsec, "Guaranteed by Timespan.__init__." 

319 lower = None if value._nsec[0] == converter.min_nsec else value._nsec[0] 

320 upper = None if value._nsec[1] == converter.max_nsec else value._nsec[1] 

321 return psycopg2.extras.NumericRange(lower=lower, upper=upper) 

322 

323 def process_result_value( 

324 self, value: psycopg2.extras.NumericRange | None, dialect: sqlalchemy.engine.Dialect 

325 ) -> Timespan | None: 

326 if value is None: 

327 return None 

328 if value.isempty: 

329 return Timespan.makeEmpty() 

330 converter = time_utils.TimeConverter() 

331 begin_nsec = converter.min_nsec if value.lower is None else value.lower 

332 end_nsec = converter.max_nsec if value.upper is None else value.upper 

333 return Timespan(begin=None, end=None, _nsec=(begin_nsec, end_nsec)) 

334 

335 

336class _RangeTimespanRepresentation(TimespanDatabaseRepresentation): 

337 """An implementation of `TimespanDatabaseRepresentation` that uses 

338 `_RangeTimespanType` to store a timespan in a single 

339 PostgreSQL-specific field. 

340 

341 Parameters 

342 ---------- 

343 column : `sqlalchemy.sql.ColumnElement` 

344 SQLAlchemy object representing the column. 

345 """ 

346 

347 def __init__(self, column: sqlalchemy.sql.ColumnElement, name: str): 

348 self.column = column 

349 self._name = name 

350 

351 __slots__ = ("column", "_name") 

352 

353 @classmethod 

354 def makeFieldSpecs( 

355 cls, nullable: bool, name: str | None = None, **kwargs: Any 

356 ) -> tuple[ddl.FieldSpec, ...]: 

357 # Docstring inherited. 

358 if name is None: 

359 name = cls.NAME 

360 return ( 

361 ddl.FieldSpec( 

362 name, 

363 dtype=_RangeTimespanType, 

364 nullable=nullable, 

365 default=(None if nullable else sqlalchemy.sql.text("'(,)'::int8range")), 

366 **kwargs, 

367 ), 

368 ) 

369 

370 @classmethod 

371 def getFieldNames(cls, name: str | None = None) -> tuple[str, ...]: 

372 # Docstring inherited. 

373 if name is None: 

374 name = cls.NAME 

375 return (name,) 

376 

377 @classmethod 

378 def update( 

379 cls, extent: Timespan | None, name: str | None = None, result: dict[str, Any] | None = None 

380 ) -> dict[str, Any]: 

381 # Docstring inherited. 

382 if name is None: 

383 name = cls.NAME 

384 if result is None: 

385 result = {} 

386 result[name] = extent 

387 return result 

388 

389 @classmethod 

390 def extract(cls, mapping: Mapping[str, Any], name: str | None = None) -> Timespan | None: 

391 # Docstring inherited. 

392 if name is None: 

393 name = cls.NAME 

394 return mapping[name] 

395 

396 @classmethod 

397 def fromLiteral(cls, timespan: Timespan | None) -> _RangeTimespanRepresentation: 

398 # Docstring inherited. 

399 if timespan is None: 

400 return cls(column=sqlalchemy.sql.null(), name=cls.NAME) 

401 return cls( 

402 column=sqlalchemy.sql.cast( 

403 sqlalchemy.sql.literal(timespan, type_=_RangeTimespanType), type_=_RangeTimespanType 

404 ), 

405 name=cls.NAME, 

406 ) 

407 

408 @classmethod 

409 def from_columns( 

410 cls, columns: sqlalchemy.sql.ColumnCollection, name: str | None = None 

411 ) -> _RangeTimespanRepresentation: 

412 # Docstring inherited. 

413 if name is None: 

414 name = cls.NAME 

415 return cls(columns[name], name) 

416 

417 @property 

418 def name(self) -> str: 

419 # Docstring inherited. 

420 return self._name 

421 

422 def isNull(self) -> sqlalchemy.sql.ColumnElement: 

423 # Docstring inherited. 

424 return self.column.is_(None) 

425 

426 def isEmpty(self) -> sqlalchemy.sql.ColumnElement: 

427 # Docstring inherited 

428 return sqlalchemy.sql.func.isempty(self.column) 

429 

430 def __lt__( 

431 self, other: _RangeTimespanRepresentation | sqlalchemy.sql.ColumnElement 

432 ) -> sqlalchemy.sql.ColumnElement: 

433 # Docstring inherited. 

434 if isinstance(other, sqlalchemy.sql.ColumnElement): 

435 return sqlalchemy.sql.and_( 

436 sqlalchemy.sql.not_(sqlalchemy.sql.func.upper_inf(self.column)), 

437 sqlalchemy.sql.not_(sqlalchemy.sql.func.isempty(self.column)), 

438 sqlalchemy.sql.func.upper(self.column) <= other, 

439 ) 

440 else: 

441 return self.column << other.column 

442 

443 def __gt__( 

444 self, other: _RangeTimespanRepresentation | sqlalchemy.sql.ColumnElement 

445 ) -> sqlalchemy.sql.ColumnElement: 

446 # Docstring inherited. 

447 if isinstance(other, sqlalchemy.sql.ColumnElement): 

448 return sqlalchemy.sql.and_( 

449 sqlalchemy.sql.not_(sqlalchemy.sql.func.lower_inf(self.column)), 

450 sqlalchemy.sql.not_(sqlalchemy.sql.func.isempty(self.column)), 

451 sqlalchemy.sql.func.lower(self.column) > other, 

452 ) 

453 else: 

454 return self.column >> other.column 

455 

456 def overlaps( 

457 self, other: _RangeTimespanRepresentation | sqlalchemy.sql.ColumnElement 

458 ) -> sqlalchemy.sql.ColumnElement: 

459 # Docstring inherited. 

460 if not isinstance(other, _RangeTimespanRepresentation): 

461 return self.contains(other) 

462 return self.column.overlaps(other.column) 

463 

464 def contains( 

465 self, other: _RangeTimespanRepresentation | sqlalchemy.sql.ColumnElement 

466 ) -> sqlalchemy.sql.ColumnElement: 

467 # Docstring inherited 

468 if isinstance(other, _RangeTimespanRepresentation): 

469 return self.column.contains(other.column) 

470 else: 

471 return self.column.contains(other) 

472 

473 def lower(self) -> sqlalchemy.sql.ColumnElement: 

474 # Docstring inherited. 

475 return sqlalchemy.sql.functions.coalesce( 

476 sqlalchemy.sql.func.lower(self.column), sqlalchemy.sql.literal(0) 

477 ) 

478 

479 def upper(self) -> sqlalchemy.sql.ColumnElement: 

480 # Docstring inherited. 

481 return sqlalchemy.sql.functions.coalesce( 

482 sqlalchemy.sql.func.upper(self.column), sqlalchemy.sql.literal(0) 

483 ) 

484 

485 def flatten(self, name: str | None = None) -> tuple[sqlalchemy.sql.ColumnElement]: 

486 # Docstring inherited. 

487 if name is None: 

488 return (self.column,) 

489 else: 

490 return (self.column.label(name),)