Hide keyboard shortcuts

Hot-keys on this page

r m x p   toggle line displays

j k   next/prev highlighted chunk

0   (zero) top of page

1   (one) first highlighted chunk

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# (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/>. 

21from __future__ import annotations 

22 

23"""Module defining GraphBuilder class and related methods. 

24""" 

25 

26__all__ = ['GraphBuilder'] 

27 

28# ------------------------------- 

29# Imports of standard modules -- 

30# ------------------------------- 

31import itertools 

32from collections import ChainMap 

33from contextlib import contextmanager 

34from dataclasses import dataclass 

35from typing import Dict, Iterable, Iterator, List, Set 

36import logging 

37 

38 

39# ----------------------------- 

40# Imports for other modules -- 

41# ----------------------------- 

42from .connections import iterConnections 

43from .pipeline import PipelineDatasetTypes, TaskDatasetTypes, TaskDef, Pipeline 

44from .graph import QuantumGraph 

45from lsst.daf.butler import ( 

46 DataCoordinate, 

47 DatasetRef, 

48 DatasetType, 

49 DimensionGraph, 

50 DimensionUniverse, 

51 NamedKeyDict, 

52 Quantum, 

53) 

54from lsst.daf.butler.registry.queries.exprParser import ParseError, ParserYacc, TreeVisitor 

55from lsst.utils import doImport 

56 

57# ---------------------------------- 

58# Local non-exported definitions -- 

59# ---------------------------------- 

60 

61_LOG = logging.getLogger(__name__.partition(".")[2]) 

62 

63 

64class _DatasetDict(NamedKeyDict[DatasetType, Dict[DataCoordinate, DatasetRef]]): 

65 """A custom dictionary that maps `DatasetType` to a nested dictionary of 

66 the known `DatasetRef` instances of that type. 

67 

68 Parameters 

69 ---------- 

70 args 

71 Positional arguments are forwarded to the `dict` constructor. 

72 universe : `DimensionUniverse` 

73 Universe of all possible dimensions. 

74 """ 

75 def __init__(self, *args, universe: DimensionGraph): 

76 super().__init__(*args) 

77 self.universe = universe 

78 

79 @classmethod 

80 def fromDatasetTypes(cls, datasetTypes: Iterable[DatasetType], *, 

81 universe: DimensionUniverse) -> _DatasetDict: 

82 """Construct a dictionary from a flat iterable of `DatasetType` keys. 

83 

84 Parameters 

85 ---------- 

86 datasetTypes : `iterable` of `DatasetType` 

87 DatasetTypes to use as keys for the dict. Values will be empty 

88 dictionaries. 

89 universe : `DimensionUniverse` 

90 Universe of all possible dimensions. 

91 

92 Returns 

93 ------- 

94 dictionary : `_DatasetDict` 

95 A new `_DatasetDict` instance. 

96 """ 

97 return cls({datasetType: {} for datasetType in datasetTypes}, universe=universe) 

98 

99 @classmethod 

100 def fromSubset(cls, datasetTypes: Iterable[DatasetType], first: _DatasetDict, *rest: _DatasetDict 

101 ) -> _DatasetDict: 

102 """Return a new dictionary by extracting items corresponding to the 

103 given keys from one or more existing dictionaries. 

104 

105 Parameters 

106 ---------- 

107 datasetTypes : `iterable` of `DatasetType` 

108 DatasetTypes to use as keys for the dict. Values will be obtained 

109 by lookups against ``first`` and ``rest``. 

110 first : `_DatasetDict` 

111 Another dictionary from which to extract values. 

112 rest 

113 Additional dictionaries from which to extract values. 

114 

115 Returns 

116 ------- 

117 dictionary : `_DatasetDict` 

118 A new dictionary instance. 

119 """ 

120 combined = ChainMap(first, *rest) 

121 return cls({datasetType: combined[datasetType] for datasetType in datasetTypes}, 

122 universe=first.universe) 

123 

124 @property 

125 def dimensions(self) -> DimensionGraph: 

126 """The union of all dimensions used by all dataset types in this 

127 dictionary, including implied dependencies (`DimensionGraph`). 

128 """ 

129 base = self.universe.empty 

130 if len(self) == 0: 

131 return base 

132 return base.union(*[datasetType.dimensions for datasetType in self.keys()]) 

133 

134 def unpackSingleRefs(self) -> NamedKeyDict[DatasetType, DatasetRef]: 

135 """Unpack nested single-element `DatasetRef` dicts into a new 

136 mapping with `DatasetType` keys and `DatasetRef` values. 

137 

138 This method assumes that each nest contains exactly one item, as is the 

139 case for all "init" datasets. 

140 

141 Returns 

142 ------- 

143 dictionary : `NamedKeyDict` 

144 Dictionary mapping `DatasetType` to `DatasetRef`, with both 

145 `DatasetType` instances and string names usable as keys. 

146 """ 

147 def getOne(refs: Dict[DataCoordinate, DatasetRef]) -> DatasetRef: 

148 ref, = refs.values() 

149 return ref 

150 return NamedKeyDict({datasetType: getOne(refs) for datasetType, refs in self.items()}) 

151 

152 def unpackMultiRefs(self) -> NamedKeyDict[DatasetType, DatasetRef]: 

153 """Unpack nested multi-element `DatasetRef` dicts into a new 

154 mapping with `DatasetType` keys and `set` of `DatasetRef` values. 

155 

156 Returns 

157 ------- 

158 dictionary : `NamedKeyDict` 

159 Dictionary mapping `DatasetType` to `DatasetRef`, with both 

160 `DatasetType` instances and string names usable as keys. 

161 """ 

162 return NamedKeyDict({datasetType: list(refs.values()) for datasetType, refs in self.items()}) 

163 

164 def extract(self, datasetType: DatasetType, dataIds: Iterable[DataCoordinate] 

165 ) -> Iterator[DatasetRef]: 

166 """Iterate over the contained `DatasetRef` instances that match the 

167 given `DatasetType` and data IDs. 

168 

169 Parameters 

170 ---------- 

171 datasetType : `DatasetType` 

172 Dataset type to match. 

173 dataIds : `Iterable` [ `DataCoordinate` ] 

174 Data IDs to match. 

175 

176 Returns 

177 ------- 

178 refs : `Iterator` [ `DatasetRef` ] 

179 DatasetRef instances for which ``ref.datasetType == datasetType`` 

180 and ``ref.dataId`` is in ``dataIds``. 

181 """ 

182 refs = self[datasetType] 

183 return (refs[dataId] for dataId in dataIds) 

184 

185 

186class _QuantumScaffolding: 

187 """Helper class aggregating information about a `Quantum`, used when 

188 constructing a `QuantumGraph`. 

189 

190 See `_PipelineScaffolding` for a top-down description of the full 

191 scaffolding data structure. 

192 

193 Parameters 

194 ---------- 

195 task : _TaskScaffolding 

196 Back-reference to the helper object for the `PipelineTask` this quantum 

197 represents an execution of. 

198 dataId : `DataCoordinate` 

199 Data ID for this quantum. 

200 """ 

201 def __init__(self, task: _TaskScaffolding, dataId: DataCoordinate): 

202 self.task = task 

203 self.dataId = dataId 

204 self.inputs = _DatasetDict.fromDatasetTypes(task.inputs.keys(), universe=dataId.universe) 

205 self.outputs = _DatasetDict.fromDatasetTypes(task.outputs.keys(), universe=dataId.universe) 

206 self.prerequisites = _DatasetDict.fromDatasetTypes(task.prerequisites.keys(), 

207 universe=dataId.universe) 

208 

209 __slots__ = ("task", "dataId", "inputs", "outputs", "prerequisites") 

210 

211 def __repr__(self): 

212 return f"_QuantumScaffolding(taskDef={self.task.taskDef}, dataId={self.dataId}, ...)" 

213 

214 task: _TaskScaffolding 

215 """Back-reference to the helper object for the `PipelineTask` this quantum 

216 represents an execution of. 

217 """ 

218 

219 dataId: DataCoordinate 

220 """Data ID for this quantum. 

221 """ 

222 

223 inputs: _DatasetDict 

224 """Nested dictionary containing `DatasetRef` inputs to this quantum. 

225 

226 This is initialized to map each `DatasetType` to an empty dictionary at 

227 construction. Those nested dictionaries are populated (with data IDs as 

228 keys) with unresolved `DatasetRef` instances in 

229 `_PipelineScaffolding.connectDataIds`. 

230 """ 

231 

232 outputs: _DatasetDict 

233 """Nested dictionary containing `DatasetRef` outputs this quantum. 

234 """ 

235 

236 prerequisites: _DatasetDict 

237 """Nested dictionary containing `DatasetRef` prerequisite inputs to this 

238 quantum. 

239 """ 

240 

241 def makeQuantum(self) -> Quantum: 

242 """Transform the scaffolding object into a true `Quantum` instance. 

243 

244 Returns 

245 ------- 

246 quantum : `Quantum` 

247 An actual `Quantum` instance. 

248 """ 

249 allInputs = self.inputs.unpackMultiRefs() 

250 allInputs.update(self.prerequisites.unpackMultiRefs()) 

251 # Give the task's Connections class an opportunity to remove some 

252 # inputs, or complain if they are unacceptable. 

253 # This will raise if one of the check conditions is not met, which is 

254 # the intended behavior 

255 allInputs = self.task.taskDef.connections.adjustQuantum(allInputs) 

256 return Quantum( 

257 taskName=self.task.taskDef.taskName, 

258 taskClass=self.task.taskDef.taskClass, 

259 dataId=self.dataId, 

260 initInputs=self.task.initInputs.unpackSingleRefs(), 

261 inputs=allInputs, 

262 outputs=self.outputs.unpackMultiRefs(), 

263 ) 

264 

265 

266@dataclass 

267class _TaskScaffolding: 

268 """Helper class aggregating information about a `PipelineTask`, used when 

269 constructing a `QuantumGraph`. 

270 

271 See `_PipelineScaffolding` for a top-down description of the full 

272 scaffolding data structure. 

273 

274 Parameters 

275 ---------- 

276 taskDef : `TaskDef` 

277 Data structure that identifies the task class and its config. 

278 parent : `_PipelineScaffolding` 

279 The parent data structure that will hold the instance being 

280 constructed. 

281 datasetTypes : `TaskDatasetTypes` 

282 Data structure that categorizes the dataset types used by this task. 

283 """ 

284 def __init__(self, taskDef: TaskDef, parent: _PipelineScaffolding, datasetTypes: TaskDatasetTypes): 

285 universe = parent.dimensions.universe 

286 self.taskDef = taskDef 

287 self.dimensions = DimensionGraph(universe, names=taskDef.connections.dimensions) 

288 assert self.dimensions.issubset(parent.dimensions) 

289 # Initialize _DatasetDicts as subsets of the one or two 

290 # corresponding dicts in the parent _PipelineScaffolding. 

291 self.initInputs = _DatasetDict.fromSubset(datasetTypes.initInputs, parent.initInputs, 

292 parent.initIntermediates) 

293 self.initOutputs = _DatasetDict.fromSubset(datasetTypes.initOutputs, parent.initIntermediates, 

294 parent.initOutputs) 

295 self.inputs = _DatasetDict.fromSubset(datasetTypes.inputs, parent.inputs, parent.intermediates) 

296 self.outputs = _DatasetDict.fromSubset(datasetTypes.outputs, parent.intermediates, parent.outputs) 

297 self.prerequisites = _DatasetDict.fromSubset(datasetTypes.prerequisites, parent.prerequisites) 

298 self.dataIds = set() 

299 self.quanta = {} 

300 

301 def __repr__(self): 

302 # Default dataclass-injected __repr__ gets caught in an infinite loop 

303 # because of back-references. 

304 return f"_TaskScaffolding(taskDef={self.taskDef}, ...)" 

305 

306 taskDef: TaskDef 

307 """Data structure that identifies the task class and its config 

308 (`TaskDef`). 

309 """ 

310 

311 dimensions: DimensionGraph 

312 """The dimensions of a single `Quantum` of this task (`DimensionGraph`). 

313 """ 

314 

315 initInputs: _DatasetDict 

316 """Dictionary containing information about datasets used to construct this 

317 task (`_DatasetDict`). 

318 """ 

319 

320 initOutputs: _DatasetDict 

321 """Dictionary containing information about datasets produced as a 

322 side-effect of constructing this task (`_DatasetDict`). 

323 """ 

324 

325 inputs: _DatasetDict 

326 """Dictionary containing information about datasets used as regular, 

327 graph-constraining inputs to this task (`_DatasetDict`). 

328 """ 

329 

330 outputs: _DatasetDict 

331 """Dictionary containing information about datasets produced by this task 

332 (`_DatasetDict`). 

333 """ 

334 

335 prerequisites: _DatasetDict 

336 """Dictionary containing information about input datasets that must be 

337 present in the repository before any Pipeline containing this task is run 

338 (`_DatasetDict`). 

339 """ 

340 

341 quanta: Dict[DataCoordinate, _QuantumScaffolding] 

342 """Dictionary mapping data ID to a scaffolding object for the Quantum of 

343 this task with that data ID. 

344 """ 

345 

346 def makeQuantumSet(self) -> Set[Quantum]: 

347 """Create a `set` of `Quantum` from the information in ``self``. 

348 

349 Returns 

350 ------- 

351 nodes : `set` of `Quantum 

352 The `Quantum` elements corresponding to this task. 

353 """ 

354 return set(q.makeQuantum() for q in self.quanta.values()) 

355 

356 

357@dataclass 

358class _PipelineScaffolding: 

359 """A helper data structure that organizes the information involved in 

360 constructing a `QuantumGraph` for a `Pipeline`. 

361 

362 Parameters 

363 ---------- 

364 pipeline : `Pipeline` 

365 Sequence of tasks from which a graph is to be constructed. Must 

366 have nested task classes already imported. 

367 universe : `DimensionUniverse` 

368 Universe of all possible dimensions. 

369 

370 Notes 

371 ----- 

372 The scaffolding data structure contains nested data structures for both 

373 tasks (`_TaskScaffolding`) and datasets (`_DatasetDict`). The dataset 

374 data structures are shared between the pipeline-level structure (which 

375 aggregates all datasets and categorizes them from the perspective of the 

376 complete pipeline) and the individual tasks that use them as inputs and 

377 outputs. 

378 

379 `QuantumGraph` construction proceeds in four steps, with each corresponding 

380 to a different `_PipelineScaffolding` method: 

381 

382 1. When `_PipelineScaffolding` is constructed, we extract and categorize 

383 the DatasetTypes used by the pipeline (delegating to 

384 `PipelineDatasetTypes.fromPipeline`), then use these to construct the 

385 nested `_TaskScaffolding` and `_DatasetDict` objects. 

386 

387 2. In `connectDataIds`, we construct and run the "Big Join Query", which 

388 returns related tuples of all dimensions used to identify any regular 

389 input, output, and intermediate datasets (not prerequisites). We then 

390 iterate over these tuples of related dimensions, identifying the subsets 

391 that correspond to distinct data IDs for each task and dataset type, 

392 and then create `_QuantumScaffolding` objects. 

393 

394 3. In `resolveDatasetRefs`, we run follow-up queries against all of the 

395 dataset data IDs previously identified, transforming unresolved 

396 DatasetRefs into resolved DatasetRefs where appropriate. We then look 

397 up prerequisite datasets for all quanta. 

398 

399 4. In `makeQuantumGraph`, we construct a `QuantumGraph` from the lists of 

400 per-task `_QuantumScaffolding` objects. 

401 """ 

402 def __init__(self, pipeline, *, registry): 

403 _LOG.debug("Initializing data structures for QuantumGraph generation.") 

404 self.tasks = [] 

405 # Aggregate and categorize the DatasetTypes in the Pipeline. 

406 datasetTypes = PipelineDatasetTypes.fromPipeline(pipeline, registry=registry) 

407 # Construct dictionaries that map those DatasetTypes to structures 

408 # that will (later) hold addiitonal information about them. 

409 for attr in ("initInputs", "initIntermediates", "initOutputs", 

410 "inputs", "intermediates", "outputs", "prerequisites"): 

411 setattr(self, attr, _DatasetDict.fromDatasetTypes(getattr(datasetTypes, attr), 

412 universe=registry.dimensions)) 

413 # Aggregate all dimensions for all non-init, non-prerequisite 

414 # DatasetTypes. These are the ones we'll include in the big join 

415 # query. 

416 self.dimensions = self.inputs.dimensions.union(self.intermediates.dimensions, 

417 self.outputs.dimensions) 

418 # Construct scaffolding nodes for each Task, and add backreferences 

419 # to the Task from each DatasetScaffolding node. 

420 # Note that there's only one scaffolding node for each DatasetType, 

421 # shared by _PipelineScaffolding and all _TaskScaffoldings that 

422 # reference it. 

423 if isinstance(pipeline, Pipeline): 

424 pipeline = pipeline.toExpandedPipeline() 

425 self.tasks = [_TaskScaffolding(taskDef=taskDef, parent=self, datasetTypes=taskDatasetTypes) 

426 for taskDef, taskDatasetTypes in zip(pipeline, 

427 datasetTypes.byTask.values())] 

428 

429 def __repr__(self): 

430 # Default dataclass-injected __repr__ gets caught in an infinite loop 

431 # because of back-references. 

432 return f"_PipelineScaffolding(tasks={self.tasks}, ...)" 

433 

434 tasks: List[_TaskScaffolding] 

435 """Scaffolding data structures for each task in the pipeline 

436 (`list` of `_TaskScaffolding`). 

437 """ 

438 

439 initInputs: _DatasetDict 

440 """Datasets consumed but not produced when constructing the tasks in this 

441 pipeline (`_DatasetDict`). 

442 """ 

443 

444 initIntermediates: _DatasetDict 

445 """Datasets that are both consumed and produced when constructing the tasks 

446 in this pipeline (`_DatasetDict`). 

447 """ 

448 

449 initOutputs: _DatasetDict 

450 """Datasets produced but not consumed when constructing the tasks in this 

451 pipeline (`_DatasetDict`). 

452 """ 

453 

454 inputs: _DatasetDict 

455 """Datasets that are consumed but not produced when running this pipeline 

456 (`_DatasetDict`). 

457 """ 

458 

459 intermediates: _DatasetDict 

460 """Datasets that are both produced and consumed when running this pipeline 

461 (`_DatasetDict`). 

462 """ 

463 

464 outputs: _DatasetDict 

465 """Datasets produced but not consumed when when running this pipeline 

466 (`_DatasetDict`). 

467 """ 

468 

469 prerequisites: _DatasetDict 

470 """Datasets that are consumed when running this pipeline and looked up 

471 per-Quantum when generating the graph (`_DatasetDict`). 

472 """ 

473 

474 dimensions: DimensionGraph 

475 """All dimensions used by any regular input, intermediate, or output 

476 (not prerequisite) dataset; the set of dimension used in the "Big Join 

477 Query" (`DimensionGraph`). 

478 

479 This is required to be a superset of all task quantum dimensions. 

480 """ 

481 

482 @contextmanager 

483 def connectDataIds(self, registry, collections, userQuery): 

484 """Query for the data IDs that connect nodes in the `QuantumGraph`. 

485 

486 This method populates `_TaskScaffolding.dataIds` and 

487 `_DatasetScaffolding.dataIds` (except for those in `prerequisites`). 

488 

489 Parameters 

490 ---------- 

491 registry : `lsst.daf.butler.Registry` 

492 Registry for the data repository; used for all data ID queries. 

493 collections : `lsst.daf.butler.CollectionSearch` 

494 Object representing the collections to search for input datasets. 

495 userQuery : `str`, optional 

496 User-provided expression to limit the data IDs processed. 

497 

498 Returns 

499 ------- 

500 commonDataIds : \ 

501 `lsst.daf.butler.registry.queries.DataCoordinateQueryResults` 

502 An interface to a database temporary table containing all data IDs 

503 that will appear in this `QuantumGraph`. Returned inside a 

504 context manager, which will drop the temporary table at the end of 

505 the `with` block in which this method is called. 

506 """ 

507 _LOG.debug("Building query for data IDs.") 

508 # Initialization datasets always have empty data IDs. 

509 emptyDataId = DataCoordinate.makeEmpty(registry.dimensions) 

510 for datasetType, refs in itertools.chain(self.initInputs.items(), 

511 self.initIntermediates.items(), 

512 self.initOutputs.items()): 

513 refs[emptyDataId] = DatasetRef(datasetType, emptyDataId) 

514 # Run one big query for the data IDs for task dimensions and regular 

515 # inputs and outputs. We limit the query to only dimensions that are 

516 # associated with the input dataset types, but don't (yet) try to 

517 # obtain the dataset_ids for those inputs. 

518 _LOG.debug("Submitting data ID query and materializing results.") 

519 with registry.queryDataIds(self.dimensions, 

520 datasets=list(self.inputs), 

521 collections=collections, 

522 where=userQuery, 

523 ).materialize() as commonDataIds: 

524 _LOG.debug("Expanding data IDs.") 

525 commonDataIds = commonDataIds.expanded() 

526 _LOG.debug("Iterating over query results to associate quanta with datasets.") 

527 # Iterate over query results, populating data IDs for datasets and 

528 # quanta and then connecting them to each other. 

529 n = 0 

530 for n, commonDataId in enumerate(commonDataIds): 

531 # Create DatasetRefs for all DatasetTypes from this result row, 

532 # noting that we might have created some already. 

533 # We remember both those that already existed and those that we 

534 # create now. 

535 refsForRow = {} 

536 for datasetType, refs in itertools.chain(self.inputs.items(), self.intermediates.items(), 

537 self.outputs.items()): 

538 datasetDataId = commonDataId.subset(datasetType.dimensions) 

539 ref = refs.get(datasetDataId) 

540 if ref is None: 

541 ref = DatasetRef(datasetType, datasetDataId) 

542 refs[datasetDataId] = ref 

543 refsForRow[datasetType.name] = ref 

544 # Create _QuantumScaffolding objects for all tasks from this 

545 # result row, noting that we might have created some already. 

546 for task in self.tasks: 

547 quantumDataId = commonDataId.subset(task.dimensions) 

548 quantum = task.quanta.get(quantumDataId) 

549 if quantum is None: 

550 quantum = _QuantumScaffolding(task=task, dataId=quantumDataId) 

551 task.quanta[quantumDataId] = quantum 

552 # Whether this is a new quantum or an existing one, we can 

553 # now associate the DatasetRefs for this row with it. The 

554 # fact that a Quantum data ID and a dataset data ID both 

555 # came from the same result row is what tells us they 

556 # should be associated. 

557 # Many of these associates will be duplicates (because 

558 # another query row that differed from this one only in 

559 # irrelevant dimensions already added them), and we use 

560 # sets to skip. 

561 for datasetType in task.inputs: 

562 ref = refsForRow[datasetType.name] 

563 quantum.inputs[datasetType.name][ref.dataId] = ref 

564 for datasetType in task.outputs: 

565 ref = refsForRow[datasetType.name] 

566 quantum.outputs[datasetType.name][ref.dataId] = ref 

567 _LOG.debug("Finished processing %d rows from data ID query.", n) 

568 yield commonDataIds 

569 

570 def resolveDatasetRefs(self, registry, collections, run, commonDataIds, *, skipExisting=True): 

571 """Perform follow up queries for each dataset data ID produced in 

572 `fillDataIds`. 

573 

574 This method populates `_DatasetScaffolding.refs` (except for those in 

575 `prerequisites`). 

576 

577 Parameters 

578 ---------- 

579 registry : `lsst.daf.butler.Registry` 

580 Registry for the data repository; used for all data ID queries. 

581 collections : `lsst.daf.butler.CollectionSearch` 

582 Object representing the collections to search for input datasets. 

583 run : `str`, optional 

584 Name of the `~lsst.daf.butler.CollectionType.RUN` collection for 

585 output datasets, if it already exists. 

586 commonDataIds : \ 

587 `lsst.daf.butler.registry.queries.DataCoordinateQueryResults` 

588 Result of a previous call to `connectDataIds`. 

589 skipExisting : `bool`, optional 

590 If `True` (default), a Quantum is not created if all its outputs 

591 already exist in ``run``. Ignored if ``run`` is `None`. 

592 

593 Raises 

594 ------ 

595 OutputExistsError 

596 Raised if an output dataset already exists in the output run 

597 and ``skipExisting`` is `False`. The case where some but not all 

598 of a quantum's outputs are present and ``skipExisting`` is `True` 

599 cannot be identified at this stage, and is handled by `fillQuanta` 

600 instead. 

601 """ 

602 # Look up [init] intermediate and output datasets in the output 

603 # collection, if there is an output collection. 

604 if run is not None: 

605 for datasetType, refs in itertools.chain(self.initIntermediates.items(), 

606 self.initOutputs.items(), 

607 self.intermediates.items(), 

608 self.outputs.items()): 

609 _LOG.debug("Resolving %d datasets for intermediate and/or output dataset %s.", 

610 len(refs), datasetType.name) 

611 isInit = datasetType in self.initIntermediates or datasetType in self.initOutputs 

612 resolvedRefQueryResults = commonDataIds.subset( 

613 datasetType.dimensions, 

614 unique=True 

615 ).findDatasets( 

616 datasetType, 

617 collections=run, 

618 deduplicate=True 

619 ) 

620 for resolvedRef in resolvedRefQueryResults: 

621 # TODO: we could easily support per-DatasetType 

622 # skipExisting and I could imagine that being useful - it's 

623 # probably required in order to support writing initOutputs 

624 # before QuantumGraph generation. 

625 assert resolvedRef.dataId in refs 

626 if skipExisting or isInit: 

627 refs[resolvedRef.dataId] = resolvedRef 

628 else: 

629 raise OutputExistsError(f"Output dataset {datasetType.name} already exists in " 

630 f"output RUN collection '{run}' with data ID" 

631 f" {resolvedRef.dataId}.") 

632 # Look up input and initInput datasets in the input collection(s). 

633 for datasetType, refs in itertools.chain(self.initInputs.items(), self.inputs.items()): 

634 _LOG.debug("Resolving %d datasets for input dataset %s.", len(refs), datasetType.name) 

635 resolvedRefQueryResults = commonDataIds.subset( 

636 datasetType.dimensions, 

637 unique=True 

638 ).findDatasets( 

639 datasetType, 

640 collections=collections, 

641 deduplicate=True 

642 ) 

643 dataIdsNotFoundYet = set(refs.keys()) 

644 for resolvedRef in resolvedRefQueryResults: 

645 dataIdsNotFoundYet.discard(resolvedRef.dataId) 

646 refs[resolvedRef.dataId] = resolvedRef 

647 if dataIdsNotFoundYet: 

648 raise RuntimeError( 

649 f"{len(dataIdsNotFoundYet)} dataset(s) of type " 

650 f"'{datasetType.name}' was/were present in a previous " 

651 f"query, but could not be found now." 

652 f"This is either a logic bug in QuantumGraph generation " 

653 f"or the input collections have been modified since " 

654 f"QuantumGraph generation began." 

655 ) 

656 # Copy the resolved DatasetRefs to the _QuantumScaffolding objects, 

657 # replacing the unresolved refs there, and then look up prerequisites. 

658 for task in self.tasks: 

659 _LOG.debug( 

660 "Applying resolutions and finding prerequisites for %d quanta of task with label '%s'.", 

661 len(task.quanta), 

662 task.taskDef.label 

663 ) 

664 lookupFunctions = { 

665 c.name: c.lookupFunction 

666 for c in iterConnections(task.taskDef.connections, "prerequisiteInputs") 

667 if c.lookupFunction is not None 

668 } 

669 dataIdsToSkip = [] 

670 for quantum in task.quanta.values(): 

671 # Process outputs datasets only if there is a run to look for 

672 # outputs in and skipExisting is True. Note that if 

673 # skipExisting is False, any output datasets that already exist 

674 # would have already caused an exception to be raised. 

675 # We never update the DatasetRefs in the quantum because those 

676 # should never be resolved. 

677 if run is not None and skipExisting: 

678 resolvedRefs = [] 

679 unresolvedRefs = [] 

680 for datasetType, originalRefs in quantum.outputs.items(): 

681 for ref in task.outputs.extract(datasetType, originalRefs.keys()): 

682 if ref.id is not None: 

683 resolvedRefs.append(ref) 

684 else: 

685 unresolvedRefs.append(ref) 

686 if resolvedRefs: 

687 if unresolvedRefs: 

688 raise OutputExistsError( 

689 f"Quantum {quantum.dataId} of task with label " 

690 f"'{quantum.task.taskDef.label}' has some outputs that exist " 

691 f"({resolvedRefs}) " 

692 f"and others that don't ({unresolvedRefs})." 

693 ) 

694 else: 

695 # All outputs are already present; skip this 

696 # quantum and continue to the next. 

697 dataIdsToSkip.append(quantum.dataId) 

698 continue 

699 # Update the input DatasetRefs to the resolved ones we already 

700 # searched for. 

701 for datasetType, refs in quantum.inputs.items(): 

702 for ref in task.inputs.extract(datasetType, refs.keys()): 

703 refs[ref.dataId] = ref 

704 # Look up prerequisite datasets in the input collection(s). 

705 # These may have dimensions that extend beyond those we queried 

706 # for originally, because we want to permit those data ID 

707 # values to differ across quanta and dataset types. 

708 for datasetType in task.prerequisites: 

709 lookupFunction = lookupFunctions.get(datasetType.name) 

710 if lookupFunction is not None: 

711 # PipelineTask has provided its own function to do the 

712 # lookup. This always takes precedence. 

713 refs = list( 

714 lookupFunction(datasetType, registry, quantum.dataId, collections) 

715 ) 

716 elif (datasetType.isCalibration() 

717 and datasetType.dimensions <= quantum.dataId.graph 

718 and quantum.dataId.graph.temporal): 

719 # This is a master calibration lookup, which we have to 

720 # handle specially because the query system can't do a 

721 # temporal join on a non-dimension-based timespan yet. 

722 timespan = quantum.dataId.timespan 

723 try: 

724 refs = [registry.findDataset(datasetType, quantum.dataId, 

725 collections=collections, 

726 timespan=timespan)] 

727 except KeyError: 

728 # This dataset type is not present in the registry, 

729 # which just means there are no datasets here. 

730 refs = [] 

731 else: 

732 # Most general case. 

733 refs = list(registry.queryDatasets(datasetType, 

734 collections=collections, 

735 dataId=quantum.dataId, 

736 deduplicate=True).expanded()) 

737 quantum.prerequisites[datasetType].update({ref.dataId: ref for ref in refs 

738 if ref is not None}) 

739 # Actually remove any quanta that we decided to skip above. 

740 if dataIdsToSkip: 

741 _LOG.debug("Pruning %d quanta for task with label '%s' because all of their outputs exist.", 

742 len(dataIdsToSkip), task.taskDef.label) 

743 for dataId in dataIdsToSkip: 

744 del task.quanta[dataId] 

745 

746 def makeQuantumGraph(self): 

747 """Create a `QuantumGraph` from the quanta already present in 

748 the scaffolding data structure. 

749 

750 Returns 

751 ------- 

752 graph : `QuantumGraph` 

753 The full `QuantumGraph`. 

754 """ 

755 graph = QuantumGraph({task.taskDef: task.makeQuantumSet() for task in self.tasks}) 

756 return graph 

757 

758 

759class _InstrumentFinder(TreeVisitor): 

760 """Implementation of TreeVisitor which looks for instrument name 

761 

762 Instrument should be specified as a boolean expression 

763 

764 instrument = 'string' 

765 'string' = instrument 

766 

767 so we only need to find a binary operator where operator is "=", 

768 one side is a string literal and other side is an identifier. 

769 All visit methods return tuple of (type, value), non-useful nodes 

770 return None for both type and value. 

771 """ 

772 def __init__(self): 

773 self.instruments = [] 

774 

775 def visitNumericLiteral(self, value, node): 

776 # do not care about numbers 

777 return (None, None) 

778 

779 def visitStringLiteral(self, value, node): 

780 # return type and value 

781 return ("str", value) 

782 

783 def visitTimeLiteral(self, value, node): 

784 # do not care about these 

785 return (None, None) 

786 

787 def visitRangeLiteral(self, start, stop, stride, node): 

788 # do not care about these 

789 return (None, None) 

790 

791 def visitIdentifier(self, name, node): 

792 if name.lower() == "instrument": 

793 return ("id", "instrument") 

794 return (None, None) 

795 

796 def visitUnaryOp(self, operator, operand, node): 

797 # do not care about these 

798 return (None, None) 

799 

800 def visitBinaryOp(self, operator, lhs, rhs, node): 

801 if operator == "=": 

802 if lhs == ("id", "instrument") and rhs[0] == "str": 

803 self.instruments.append(rhs[1]) 

804 elif rhs == ("id", "instrument") and lhs[0] == "str": 

805 self.instruments.append(lhs[1]) 

806 return (None, None) 

807 

808 def visitIsIn(self, lhs, values, not_in, node): 

809 # do not care about these 

810 return (None, None) 

811 

812 def visitParens(self, expression, node): 

813 # do not care about these 

814 return (None, None) 

815 

816 

817def _findInstruments(queryStr): 

818 """Get the names of any instrument named in the query string by searching 

819 for "instrument = <value>" and similar patterns. 

820 

821 Parameters 

822 ---------- 

823 queryStr : `str` or None 

824 The query string to search, or None if there is no query. 

825 

826 Returns 

827 ------- 

828 instruments : `list` [`str`] 

829 The list of instrument names found in the query. 

830 

831 Raises 

832 ------ 

833 ValueError 

834 If the query expression can not be parsed. 

835 """ 

836 if not queryStr: 

837 return [] 

838 parser = ParserYacc() 

839 finder = _InstrumentFinder() 

840 try: 

841 tree = parser.parse(queryStr) 

842 except ParseError as exc: 

843 raise ValueError(f"failed to parse query expression: {queryStr}") from exc 

844 tree.visit(finder) 

845 return finder.instruments 

846 

847 

848# ------------------------ 

849# Exported definitions -- 

850# ------------------------ 

851 

852 

853class GraphBuilderError(Exception): 

854 """Base class for exceptions generated by graph builder. 

855 """ 

856 pass 

857 

858 

859class OutputExistsError(GraphBuilderError): 

860 """Exception generated when output datasets already exist. 

861 """ 

862 pass 

863 

864 

865class PrerequisiteMissingError(GraphBuilderError): 

866 """Exception generated when a prerequisite dataset does not exist. 

867 """ 

868 pass 

869 

870 

871class GraphBuilder(object): 

872 """GraphBuilder class is responsible for building task execution graph from 

873 a Pipeline. 

874 

875 Parameters 

876 ---------- 

877 registry : `~lsst.daf.butler.Registry` 

878 Data butler instance. 

879 skipExisting : `bool`, optional 

880 If `True` (default), a Quantum is not created if all its outputs 

881 already exist. 

882 """ 

883 

884 def __init__(self, registry, skipExisting=True): 

885 self.registry = registry 

886 self.dimensions = registry.dimensions 

887 self.skipExisting = skipExisting 

888 

889 def makeGraph(self, pipeline, collections, run, userQuery): 

890 """Create execution graph for a pipeline. 

891 

892 Parameters 

893 ---------- 

894 pipeline : `Pipeline` 

895 Pipeline definition, task names/classes and their configs. 

896 collections : `lsst.daf.butler.CollectionSearch` 

897 Object representing the collections to search for input datasets. 

898 run : `str`, optional 

899 Name of the `~lsst.daf.butler.CollectionType.RUN` collection for 

900 output datasets, if it already exists. 

901 userQuery : `str` 

902 String which defines user-defined selection for registry, should be 

903 empty or `None` if there is no restrictions on data selection. 

904 

905 Returns 

906 ------- 

907 graph : `QuantumGraph` 

908 

909 Raises 

910 ------ 

911 UserExpressionError 

912 Raised when user expression cannot be parsed. 

913 OutputExistsError 

914 Raised when output datasets already exist. 

915 Exception 

916 Other exceptions types may be raised by underlying registry 

917 classes. 

918 """ 

919 scaffolding = _PipelineScaffolding(pipeline, registry=self.registry) 

920 

921 instrument = pipeline.getInstrument() 

922 if isinstance(instrument, str): 

923 instrument = doImport(instrument) 

924 instrumentName = instrument.getName() if instrument else None 

925 userQuery = self._verifyInstrumentRestriction(instrumentName, userQuery) 

926 

927 with scaffolding.connectDataIds(self.registry, collections, userQuery) as commonDataIds: 

928 scaffolding.resolveDatasetRefs(self.registry, collections, run, commonDataIds, 

929 skipExisting=self.skipExisting) 

930 return scaffolding.makeQuantumGraph() 

931 

932 @staticmethod 

933 def _verifyInstrumentRestriction(instrumentName, query): 

934 """Add an instrument restriction to the query if it does not have one, 

935 and verify that if given an instrument name that there are no other 

936 instrument restrictions in the query. 

937 

938 Parameters 

939 ---------- 

940 instrumentName : `str` 

941 The name of the instrument that should appear in the query. 

942 query : `str` 

943 The query string. 

944 

945 Returns 

946 ------- 

947 query : `str` 

948 The query string with the instrument added to it if needed. 

949 

950 Raises 

951 ------ 

952 RuntimeError 

953 If the pipeline names an instrument and the query contains more 

954 than one instrument or the name of the instrument in the query does 

955 not match the instrument named by the pipeline. 

956 """ 

957 if not instrumentName: 

958 return query 

959 queryInstruments = _findInstruments(query) 

960 if len(queryInstruments) > 1: 

961 raise RuntimeError(f"When the pipeline has an instrument (\"{instrumentName}\") the query must " 

962 "have zero instruments or one instrument that matches the pipeline. " 

963 f"Found these instruments in the query: {queryInstruments}.") 

964 if not queryInstruments: 

965 # There is not an instrument in the query, add it: 

966 restriction = f"instrument = '{instrumentName}'" 

967 _LOG.debug(f"Adding restriction \"{restriction}\" to query.") 

968 query = f"{restriction} AND ({query})" if query else restriction # (there may not be a query) 

969 elif queryInstruments[0] != instrumentName: 

970 # Since there is an instrument in the query, it should match 

971 # the instrument in the pipeline. 

972 raise RuntimeError(f"The instrument named in the query (\"{queryInstruments[0]}\") does not " 

973 f"match the instrument named by the pipeline (\"{instrumentName}\")") 

974 return query