Coverage for python/lsst/pipe/base/testUtils.py: 12%

144 statements  

« prev     ^ index     » next       coverage.py v6.5.0, created at 2023-03-11 10:14 +0000

1# This file is part of pipe_base. 

2# 

3# Developed for the LSST Data Management System. 

4# This product includes software developed by the LSST Project 

5# (https://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 <https://www.gnu.org/licenses/>. 

21 

22from __future__ import annotations 

23 

24__all__ = [ 

25 "assertValidInitOutput", 

26 "assertValidOutput", 

27 "getInitInputs", 

28 "lintConnections", 

29 "makeQuantum", 

30 "runTestQuantum", 

31] 

32 

33 

34import collections.abc 

35import itertools 

36import unittest.mock 

37from collections import defaultdict 

38from typing import TYPE_CHECKING, AbstractSet, Any, Dict, Mapping, Optional, Sequence, Set, Union 

39 

40from lsst.daf.butler import ( 

41 Butler, 

42 DataCoordinate, 

43 DataId, 

44 DatasetRef, 

45 DatasetType, 

46 Dimension, 

47 DimensionUniverse, 

48 Quantum, 

49 SkyPixDimension, 

50 StorageClassFactory, 

51) 

52from lsst.pipe.base.connectionTypes import BaseConnection, DimensionedConnection 

53 

54from .butlerQuantumContext import ButlerQuantumContext 

55 

56if TYPE_CHECKING: 56 ↛ 57line 56 didn't jump to line 57, because the condition on line 56 was never true

57 from .config import PipelineTaskConfig 

58 from .connections import PipelineTaskConnections 

59 from .pipelineTask import PipelineTask 

60 from .struct import Struct 

61 

62 

63def makeQuantum( 

64 task: PipelineTask, 

65 butler: Butler, 

66 dataId: DataId, 

67 ioDataIds: Mapping[str, Union[DataId, Sequence[DataId]]], 

68) -> Quantum: 

69 """Create a Quantum for a particular data ID(s). 

70 

71 Parameters 

72 ---------- 

73 task : `lsst.pipe.base.PipelineTask` 

74 The task whose processing the quantum represents. 

75 butler : `lsst.daf.butler.Butler` 

76 The collection the quantum refers to. 

77 dataId: any data ID type 

78 The data ID of the quantum. Must have the same dimensions as 

79 ``task``'s connections class. 

80 ioDataIds : `collections.abc.Mapping` [`str`] 

81 A mapping keyed by input/output names. Values must be data IDs for 

82 single connections and sequences of data IDs for multiple connections. 

83 

84 Returns 

85 ------- 

86 quantum : `lsst.daf.butler.Quantum` 

87 A quantum for ``task``, when called with ``dataIds``. 

88 """ 

89 # This is a type ignore, because `connections` is a dynamic class, but 

90 # it for sure will have this property 

91 connections = task.config.ConnectionsClass(config=task.config) # type: ignore 

92 

93 try: 

94 _checkDimensionsMatch(butler.registry.dimensions, connections.dimensions, dataId.keys()) 

95 except ValueError as e: 

96 raise ValueError("Error in quantum dimensions.") from e 

97 

98 inputs = defaultdict(list) 

99 outputs = defaultdict(list) 

100 for name in itertools.chain(connections.inputs, connections.prerequisiteInputs): 

101 try: 

102 connection = connections.__getattribute__(name) 

103 _checkDataIdMultiplicity(name, ioDataIds[name], connection.multiple) 

104 ids = _normalizeDataIds(ioDataIds[name]) 

105 for id in ids: 

106 ref = _refFromConnection(butler, connection, id) 

107 inputs[ref.datasetType].append(ref) 

108 except (ValueError, KeyError) as e: 

109 raise ValueError(f"Error in connection {name}.") from e 

110 for name in connections.outputs: 

111 try: 

112 connection = connections.__getattribute__(name) 

113 _checkDataIdMultiplicity(name, ioDataIds[name], connection.multiple) 

114 ids = _normalizeDataIds(ioDataIds[name]) 

115 for id in ids: 

116 ref = _refFromConnection(butler, connection, id) 

117 outputs[ref.datasetType].append(ref) 

118 except (ValueError, KeyError) as e: 

119 raise ValueError(f"Error in connection {name}.") from e 

120 quantum = Quantum( 

121 taskClass=type(task), 

122 dataId=DataCoordinate.standardize(dataId, universe=butler.registry.dimensions), 

123 inputs=inputs, 

124 outputs=outputs, 

125 ) 

126 return quantum 

127 

128 

129def _checkDimensionsMatch( 

130 universe: DimensionUniverse, 

131 expected: Union[AbstractSet[str], AbstractSet[Dimension]], 

132 actual: Union[AbstractSet[str], AbstractSet[Dimension]], 

133) -> None: 

134 """Test whether two sets of dimensions agree after conversions. 

135 

136 Parameters 

137 ---------- 

138 universe : `lsst.daf.butler.DimensionUniverse` 

139 The set of all known dimensions. 

140 expected : `Set` [`str`] or `Set` [`~lsst.daf.butler.Dimension`] 

141 The dimensions expected from a task specification. 

142 actual : `Set` [`str`] or `Set` [`~lsst.daf.butler.Dimension`] 

143 The dimensions provided by input. 

144 

145 Raises 

146 ------ 

147 ValueError 

148 Raised if ``expected`` and ``actual`` cannot be reconciled. 

149 """ 

150 if _simplify(universe, expected) != _simplify(universe, actual): 

151 raise ValueError(f"Mismatch in dimensions; expected {expected} but got {actual}.") 

152 

153 

154def _simplify( 

155 universe: DimensionUniverse, dimensions: Union[AbstractSet[str], AbstractSet[Dimension]] 

156) -> Set[str]: 

157 """Reduce a set of dimensions to a string-only form. 

158 

159 Parameters 

160 ---------- 

161 universe : `lsst.daf.butler.DimensionUniverse` 

162 The set of all known dimensions. 

163 dimensions : `Set` [`str`] or `Set` [`~lsst.daf.butler.Dimension`] 

164 A set of dimensions to simplify. 

165 

166 Returns 

167 ------- 

168 dimensions : `Set` [`str`] 

169 A copy of ``dimensions`` reduced to string form, with all spatial 

170 dimensions simplified to ``skypix``. 

171 """ 

172 simplified: Set[str] = set() 

173 for dimension in dimensions: 

174 # skypix not a real Dimension, handle it first 

175 if dimension == "skypix": 

176 simplified.add(dimension) # type: ignore 

177 else: 

178 # Need a Dimension to test spatialness 

179 fullDimension = universe[dimension] if isinstance(dimension, str) else dimension 

180 if isinstance(fullDimension, SkyPixDimension): 

181 simplified.add("skypix") 

182 else: 

183 simplified.add(fullDimension.name) 

184 return simplified 

185 

186 

187def _checkDataIdMultiplicity(name: str, dataIds: Union[DataId, Sequence[DataId]], multiple: bool) -> None: 

188 """Test whether data IDs are scalars for scalar connections and sequences 

189 for multiple connections. 

190 

191 Parameters 

192 ---------- 

193 name : `str` 

194 The name of the connection being tested. 

195 dataIds : any data ID type or `~collections.abc.Sequence` [data ID] 

196 The data ID(s) provided for the connection. 

197 multiple : `bool` 

198 The ``multiple`` field of the connection. 

199 

200 Raises 

201 ------ 

202 ValueError 

203 Raised if ``dataIds`` and ``multiple`` do not match. 

204 """ 

205 if multiple: 

206 if not isinstance(dataIds, collections.abc.Sequence): 

207 raise ValueError(f"Expected multiple data IDs for {name}, got {dataIds}.") 

208 else: 

209 # DataCoordinate is a Mapping 

210 if not isinstance(dataIds, collections.abc.Mapping): 

211 raise ValueError(f"Expected single data ID for {name}, got {dataIds}.") 

212 

213 

214def _normalizeDataIds(dataIds: Union[DataId, Sequence[DataId]]) -> Sequence[DataId]: 

215 """Represent both single and multiple data IDs as a list. 

216 

217 Parameters 

218 ---------- 

219 dataIds : any data ID type or `~collections.abc.Sequence` thereof 

220 The data ID(s) provided for a particular input or output connection. 

221 

222 Returns 

223 ------- 

224 normalizedIds : `~collections.abc.Sequence` [data ID] 

225 A sequence equal to ``dataIds`` if it was already a sequence, or 

226 ``[dataIds]`` if it was a single ID. 

227 """ 

228 if isinstance(dataIds, collections.abc.Sequence): 

229 return dataIds 

230 else: 

231 return [dataIds] 

232 

233 

234def _refFromConnection( 

235 butler: Butler, connection: DimensionedConnection, dataId: DataId, **kwargs: Any 

236) -> DatasetRef: 

237 """Create a DatasetRef for a connection in a collection. 

238 

239 Parameters 

240 ---------- 

241 butler : `lsst.daf.butler.Butler` 

242 The collection to point to. 

243 connection : `lsst.pipe.base.connectionTypes.DimensionedConnection` 

244 The connection defining the dataset type to point to. 

245 dataId 

246 The data ID for the dataset to point to. 

247 **kwargs 

248 Additional keyword arguments used to augment or construct 

249 a `~lsst.daf.butler.DataCoordinate`. 

250 

251 Returns 

252 ------- 

253 ref : `lsst.daf.butler.DatasetRef` 

254 A reference to a dataset compatible with ``connection``, with ID 

255 ``dataId``, in the collection pointed to by ``butler``. 

256 """ 

257 universe = butler.registry.dimensions 

258 # DatasetRef only tests if required dimension is missing, but not extras 

259 _checkDimensionsMatch(universe, set(connection.dimensions), dataId.keys()) 

260 dataId = DataCoordinate.standardize(dataId, **kwargs, universe=universe) 

261 

262 datasetType = butler.registry.getDatasetType(connection.name) 

263 

264 try: 

265 butler.registry.getDatasetType(datasetType.name) 

266 except KeyError: 

267 raise ValueError(f"Invalid dataset type {connection.name}.") 

268 try: 

269 ref = DatasetRef(datasetType=datasetType, dataId=dataId) 

270 return ref 

271 except KeyError as e: 

272 raise ValueError(f"Dataset type ({connection.name}) and ID {dataId.byName()} not compatible.") from e 

273 

274 

275def _resolveTestQuantumInputs(butler: Butler, quantum: Quantum) -> None: 

276 """Look up all input datasets a test quantum in the `Registry` to resolve 

277 all `DatasetRef` objects (i.e. ensure they have not-`None` ``id`` and 

278 ``run`` attributes). 

279 

280 Parameters 

281 ---------- 

282 quantum : `~lsst.daf.butler.Quantum` 

283 Single Quantum instance. 

284 butler : `~lsst.daf.butler.Butler` 

285 Data butler. 

286 """ 

287 # TODO (DM-26819): This function is a direct copy of 

288 # `lsst.ctrl.mpexec.SingleQuantumExecutor.updateQuantumInputs`, but the 

289 # `runTestQuantum` function that calls it is essentially duplicating logic 

290 # in that class as well (albeit not verbatim). We should probably move 

291 # `SingleQuantumExecutor` to ``pipe_base`` and see if it is directly usable 

292 # in test code instead of having these classes at all. 

293 for refsForDatasetType in quantum.inputs.values(): 

294 newRefsForDatasetType = [] 

295 for ref in refsForDatasetType: 

296 if ref.id is None: 

297 resolvedRef = butler.registry.findDataset( 

298 ref.datasetType, ref.dataId, collections=butler.collections 

299 ) 

300 if resolvedRef is None: 

301 raise ValueError( 

302 f"Cannot find {ref.datasetType.name} with id {ref.dataId} " 

303 f"in collections {butler.collections}." 

304 ) 

305 newRefsForDatasetType.append(resolvedRef) 

306 else: 

307 newRefsForDatasetType.append(ref) 

308 refsForDatasetType[:] = newRefsForDatasetType 

309 

310 

311def runTestQuantum( 

312 task: PipelineTask, butler: Butler, quantum: Quantum, mockRun: bool = True 

313) -> Optional[unittest.mock.Mock]: 

314 """Run a PipelineTask on a Quantum. 

315 

316 Parameters 

317 ---------- 

318 task : `lsst.pipe.base.PipelineTask` 

319 The task to run on the quantum. 

320 butler : `lsst.daf.butler.Butler` 

321 The collection to run on. 

322 quantum : `lsst.daf.butler.Quantum` 

323 The quantum to run. 

324 mockRun : `bool` 

325 Whether or not to replace ``task``'s ``run`` method. The default of 

326 `True` is recommended unless ``run`` needs to do real work (e.g., 

327 because the test needs real output datasets). 

328 

329 Returns 

330 ------- 

331 run : `unittest.mock.Mock` or `None` 

332 If ``mockRun`` is set, the mock that replaced ``run``. This object can 

333 be queried for the arguments ``runQuantum`` passed to ``run``. 

334 """ 

335 _resolveTestQuantumInputs(butler, quantum) 

336 butlerQc = ButlerQuantumContext.from_full(butler, quantum) 

337 # This is a type ignore, because `connections` is a dynamic class, but 

338 # it for sure will have this property 

339 connections = task.config.ConnectionsClass(config=task.config) # type: ignore 

340 inputRefs, outputRefs = connections.buildDatasetRefs(quantum) 

341 if mockRun: 

342 with unittest.mock.patch.object(task, "run") as mock, unittest.mock.patch( 

343 "lsst.pipe.base.ButlerQuantumContext.put" 

344 ): 

345 task.runQuantum(butlerQc, inputRefs, outputRefs) 

346 return mock 

347 else: 

348 task.runQuantum(butlerQc, inputRefs, outputRefs) 

349 return None 

350 

351 

352def _assertAttributeMatchesConnection(obj: Any, attrName: str, connection: BaseConnection) -> None: 

353 """Test that an attribute on an object matches the specification given in 

354 a connection. 

355 

356 Parameters 

357 ---------- 

358 obj 

359 An object expected to contain the attribute ``attrName``. 

360 attrName : `str` 

361 The name of the attribute to be tested. 

362 connection : `lsst.pipe.base.connectionTypes.BaseConnection` 

363 The connection, usually some type of output, specifying ``attrName``. 

364 

365 Raises 

366 ------ 

367 AssertionError: 

368 Raised if ``obj.attrName`` does not match what's expected 

369 from ``connection``. 

370 """ 

371 # name 

372 try: 

373 attrValue = obj.__getattribute__(attrName) 

374 except AttributeError: 

375 raise AssertionError(f"No such attribute on {obj!r}: {attrName}") 

376 # multiple 

377 if connection.multiple: 

378 if not isinstance(attrValue, collections.abc.Sequence): 

379 raise AssertionError(f"Expected {attrName} to be a sequence, got {attrValue!r} instead.") 

380 else: 

381 # use lazy evaluation to not use StorageClassFactory unless 

382 # necessary 

383 if isinstance(attrValue, collections.abc.Sequence) and not issubclass( 

384 StorageClassFactory().getStorageClass(connection.storageClass).pytype, collections.abc.Sequence 

385 ): 

386 raise AssertionError(f"Expected {attrName} to be a single value, got {attrValue!r} instead.") 

387 # no test for storageClass, as I'm not sure how much persistence 

388 # depends on duck-typing 

389 

390 

391def assertValidOutput(task: PipelineTask, result: Struct) -> None: 

392 """Test that the output of a call to ``run`` conforms to its own 

393 connections. 

394 

395 Parameters 

396 ---------- 

397 task : `lsst.pipe.base.PipelineTask` 

398 The task whose connections need validation. This is a fully-configured 

399 task object to support features such as optional outputs. 

400 result : `lsst.pipe.base.Struct` 

401 A result object produced by calling ``task.run``. 

402 

403 Raises 

404 ------ 

405 AssertionError: 

406 Raised if ``result`` does not match what's expected from ``task's`` 

407 connections. 

408 """ 

409 # This is a type ignore, because `connections` is a dynamic class, but 

410 # it for sure will have this property 

411 connections = task.config.ConnectionsClass(config=task.config) # type: ignore 

412 

413 for name in connections.outputs: 

414 connection = connections.__getattribute__(name) 

415 _assertAttributeMatchesConnection(result, name, connection) 

416 

417 

418def assertValidInitOutput(task: PipelineTask) -> None: 

419 """Test that a constructed task conforms to its own init-connections. 

420 

421 Parameters 

422 ---------- 

423 task : `lsst.pipe.base.PipelineTask` 

424 The task whose connections need validation. 

425 

426 Raises 

427 ------ 

428 AssertionError: 

429 Raised if ``task`` does not have the state expected from ``task's`` 

430 connections. 

431 """ 

432 # This is a type ignore, because `connections` is a dynamic class, but 

433 # it for sure will have this property 

434 connections = task.config.ConnectionsClass(config=task.config) # type: ignore 

435 

436 for name in connections.initOutputs: 

437 connection = connections.__getattribute__(name) 

438 _assertAttributeMatchesConnection(task, name, connection) 

439 

440 

441def getInitInputs(butler: Butler, config: PipelineTaskConfig) -> Dict[str, Any]: 

442 """Return the initInputs object that would have been passed to a 

443 `~lsst.pipe.base.PipelineTask` constructor. 

444 

445 Parameters 

446 ---------- 

447 butler : `lsst.daf.butler.Butler` 

448 The repository to search for input datasets. Must have 

449 pre-configured collections. 

450 config : `lsst.pipe.base.PipelineTaskConfig` 

451 The config for the task to be constructed. 

452 

453 Returns 

454 ------- 

455 initInputs : `dict` [`str`] 

456 A dictionary of objects in the format of the ``initInputs`` parameter 

457 to `lsst.pipe.base.PipelineTask`. 

458 """ 

459 connections = config.connections.ConnectionsClass(config=config) 

460 initInputs = {} 

461 for name in connections.initInputs: 

462 attribute = getattr(connections, name) 

463 # Get full dataset type to check for consistency problems 

464 dsType = DatasetType( 

465 attribute.name, butler.registry.dimensions.extract(set()), attribute.storageClass 

466 ) 

467 # All initInputs have empty data IDs 

468 initInputs[name] = butler.get(dsType) 

469 

470 return initInputs 

471 

472 

473def lintConnections( 

474 connections: PipelineTaskConnections, 

475 *, 

476 checkMissingMultiple: bool = True, 

477 checkUnnecessaryMultiple: bool = True, 

478) -> None: 

479 """Inspect a connections class for common errors. 

480 

481 These tests are designed to detect misuse of connections features in 

482 standard designs. An unusually designed connections class may trigger 

483 alerts despite being correctly written; specific checks can be turned off 

484 using keywords. 

485 

486 Parameters 

487 ---------- 

488 connections : `lsst.pipe.base.PipelineTaskConnections`-type 

489 The connections class to test. 

490 checkMissingMultiple : `bool` 

491 Whether to test for single connections that would match multiple 

492 datasets at run time. 

493 checkUnnecessaryMultiple : `bool` 

494 Whether to test for multiple connections that would only match 

495 one dataset. 

496 

497 Raises 

498 ------ 

499 AssertionError 

500 Raised if any of the selected checks fail for any connection. 

501 """ 

502 # Since all comparisons are inside the class, don't bother 

503 # normalizing skypix. 

504 quantumDimensions = connections.dimensions 

505 

506 errors = "" 

507 # connectionTypes.DimensionedConnection is implementation detail, 

508 # don't use it. 

509 for name in itertools.chain(connections.inputs, connections.prerequisiteInputs, connections.outputs): 

510 connection: DimensionedConnection = connections.allConnections[name] # type: ignore 

511 connDimensions = set(connection.dimensions) 

512 if checkMissingMultiple and not connection.multiple and connDimensions > quantumDimensions: 

513 errors += ( 

514 f"Connection {name} may be called with multiple values of " 

515 f"{connDimensions - quantumDimensions} but has multiple=False.\n" 

516 ) 

517 if checkUnnecessaryMultiple and connection.multiple and connDimensions <= quantumDimensions: 

518 errors += ( 

519 f"Connection {name} has multiple=True but can only be called with one " 

520 f"value of {connDimensions} for each {quantumDimensions}.\n" 

521 ) 

522 if errors: 

523 raise AssertionError(errors)