Coverage for python/lsst/pipe/base/pipeline.py: 20%

420 statements  

« prev     ^ index     » next       coverage.py v6.4.4, created at 2022-09-07 04:06 -0700

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 Pipeline class and related methods. 

24""" 

25 

26__all__ = ["Pipeline", "TaskDef", "TaskDatasetTypes", "PipelineDatasetTypes", "LabelSpecifier"] 

27 

28import copy 

29import logging 

30import os 

31import re 

32import urllib.parse 

33import warnings 

34 

35# ------------------------------- 

36# Imports of standard modules -- 

37# ------------------------------- 

38from dataclasses import dataclass 

39from types import MappingProxyType 

40from typing import ( 

41 TYPE_CHECKING, 

42 AbstractSet, 

43 Callable, 

44 ClassVar, 

45 Dict, 

46 Generator, 

47 Iterable, 

48 Iterator, 

49 Mapping, 

50 Optional, 

51 Set, 

52 Tuple, 

53 Type, 

54 Union, 

55 cast, 

56) 

57 

58# ----------------------------- 

59# Imports for other modules -- 

60from lsst.daf.butler import DatasetType, NamedValueSet, Registry, SkyPixDimension 

61from lsst.resources import ResourcePath, ResourcePathExpression 

62from lsst.utils import doImportType 

63from lsst.utils.introspection import get_full_type_name 

64 

65from . import pipelineIR, pipeTools 

66from ._task_metadata import TaskMetadata 

67from .config import PipelineTaskConfig 

68from .configOverrides import ConfigOverrides 

69from .connections import iterConnections 

70from .pipelineTask import PipelineTask 

71from .task import _TASK_METADATA_TYPE 

72 

73if TYPE_CHECKING: # Imports needed only for type annotations; may be circular. 73 ↛ 74line 73 didn't jump to line 74, because the condition on line 73 was never true

74 from lsst.obs.base import Instrument 

75 from lsst.pex.config import Config 

76 

77# ---------------------------------- 

78# Local non-exported definitions -- 

79# ---------------------------------- 

80 

81_LOG = logging.getLogger(__name__) 

82 

83# ------------------------ 

84# Exported definitions -- 

85# ------------------------ 

86 

87 

88@dataclass 

89class LabelSpecifier: 

90 """A structure to specify a subset of labels to load 

91 

92 This structure may contain a set of labels to be used in subsetting a 

93 pipeline, or a beginning and end point. Beginning or end may be empty, 

94 in which case the range will be a half open interval. Unlike python 

95 iteration bounds, end bounds are *INCLUDED*. Note that range based 

96 selection is not well defined for pipelines that are not linear in nature, 

97 and correct behavior is not guaranteed, or may vary from run to run. 

98 """ 

99 

100 labels: Optional[Set[str]] = None 

101 begin: Optional[str] = None 

102 end: Optional[str] = None 

103 

104 def __post_init__(self) -> None: 

105 if self.labels is not None and (self.begin or self.end): 

106 raise ValueError( 

107 "This struct can only be initialized with a labels set or a begin (and/or) end specifier" 

108 ) 

109 

110 

111class TaskDef: 

112 """TaskDef is a collection of information about task needed by Pipeline. 

113 

114 The information includes task name, configuration object and optional 

115 task class. This class is just a collection of attributes and it exposes 

116 all of them so that attributes could potentially be modified in place 

117 (e.g. if configuration needs extra overrides). 

118 

119 Attributes 

120 ---------- 

121 taskName : `str`, optional 

122 The fully-qualified `PipelineTask` class name. If not provided, 

123 ``taskClass`` must be. 

124 config : `lsst.pipe.base.config.PipelineTaskConfig`, optional 

125 Instance of the configuration class corresponding to this task class, 

126 usually with all overrides applied. This config will be frozen. If 

127 not provided, ``taskClass`` must be provided and 

128 ``taskClass.ConfigClass()`` will be used. 

129 taskClass : `type`, optional 

130 `PipelineTask` class object; if provided and ``taskName`` is as well, 

131 the caller guarantees that they are consistent. If not provided, 

132 ``taskName`` is used to import the type. 

133 label : `str`, optional 

134 Task label, usually a short string unique in a pipeline. If not 

135 provided, ``taskClass`` must be, and ``taskClass._DefaultName`` will 

136 be used. 

137 """ 

138 

139 def __init__( 

140 self, 

141 taskName: Optional[str] = None, 

142 config: Optional[PipelineTaskConfig] = None, 

143 taskClass: Optional[Type[PipelineTask]] = None, 

144 label: Optional[str] = None, 

145 ): 

146 if taskName is None: 

147 if taskClass is None: 

148 raise ValueError("At least one of `taskName` and `taskClass` must be provided.") 

149 taskName = get_full_type_name(taskClass) 

150 elif taskClass is None: 

151 taskClass = doImportType(taskName) 

152 if config is None: 

153 if taskClass is None: 

154 raise ValueError("`taskClass` must be provided if `config` is not.") 

155 config = taskClass.ConfigClass() 

156 if label is None: 

157 if taskClass is None: 

158 raise ValueError("`taskClass` must be provided if `label` is not.") 

159 label = taskClass._DefaultName 

160 self.taskName = taskName 

161 try: 

162 config.validate() 

163 except Exception: 

164 _LOG.error("Configuration validation failed for task %s (%s)", label, taskName) 

165 raise 

166 config.freeze() 

167 self.config = config 

168 self.taskClass = taskClass 

169 self.label = label 

170 self.connections = config.connections.ConnectionsClass(config=config) 

171 

172 @property 

173 def configDatasetName(self) -> str: 

174 """Name of a dataset type for configuration of this task (`str`)""" 

175 return self.label + "_config" 

176 

177 @property 

178 def metadataDatasetName(self) -> Optional[str]: 

179 """Name of a dataset type for metadata of this task, `None` if 

180 metadata is not to be saved (`str`) 

181 """ 

182 if self.config.saveMetadata: 

183 return self.makeMetadataDatasetName(self.label) 

184 else: 

185 return None 

186 

187 @classmethod 

188 def makeMetadataDatasetName(cls, label: str) -> str: 

189 """Construct the name of the dataset type for metadata for a task. 

190 

191 Parameters 

192 ---------- 

193 label : `str` 

194 Label for the task within its pipeline. 

195 

196 Returns 

197 ------- 

198 name : `str` 

199 Name of the task's metadata dataset type. 

200 """ 

201 return f"{label}_metadata" 

202 

203 @property 

204 def logOutputDatasetName(self) -> Optional[str]: 

205 """Name of a dataset type for log output from this task, `None` if 

206 logs are not to be saved (`str`) 

207 """ 

208 if cast(PipelineTaskConfig, self.config).saveLogOutput: 

209 return self.label + "_log" 

210 else: 

211 return None 

212 

213 def __str__(self) -> str: 

214 rep = "TaskDef(" + self.taskName 

215 if self.label: 

216 rep += ", label=" + self.label 

217 rep += ")" 

218 return rep 

219 

220 def __eq__(self, other: object) -> bool: 

221 if not isinstance(other, TaskDef): 

222 return False 

223 # This does not consider equality of configs when determining equality 

224 # as config equality is a difficult thing to define. Should be updated 

225 # after DM-27847 

226 return self.taskClass == other.taskClass and self.label == other.label 

227 

228 def __hash__(self) -> int: 

229 return hash((self.taskClass, self.label)) 

230 

231 @classmethod 

232 def _unreduce(cls, taskName: str, config: PipelineTaskConfig, label: str) -> TaskDef: 

233 """Custom callable for unpickling. 

234 

235 All arguments are forwarded directly to the constructor; this 

236 trampoline is only needed because ``__reduce__`` callables can't be 

237 called with keyword arguments. 

238 """ 

239 return cls(taskName=taskName, config=config, label=label) 

240 

241 def __reduce__(self) -> Tuple[Callable[[str, PipelineTaskConfig, str], TaskDef], Tuple[str, Config, str]]: 

242 return (self._unreduce, (self.taskName, self.config, self.label)) 

243 

244 

245class Pipeline: 

246 """A `Pipeline` is a representation of a series of tasks to run, and the 

247 configuration for those tasks. 

248 

249 Parameters 

250 ---------- 

251 description : `str` 

252 A description of that this pipeline does. 

253 """ 

254 

255 def __init__(self, description: str): 

256 pipeline_dict = {"description": description, "tasks": {}} 

257 self._pipelineIR = pipelineIR.PipelineIR(pipeline_dict) 

258 

259 @classmethod 

260 def fromFile(cls, filename: str) -> Pipeline: 

261 """Load a pipeline defined in a pipeline yaml file. 

262 

263 Parameters 

264 ---------- 

265 filename: `str` 

266 A path that points to a pipeline defined in yaml format. This 

267 filename may also supply additional labels to be used in 

268 subsetting the loaded Pipeline. These labels are separated from 

269 the path by a \\#, and may be specified as a comma separated 

270 list, or a range denoted as beginning..end. Beginning or end may 

271 be empty, in which case the range will be a half open interval. 

272 Unlike python iteration bounds, end bounds are *INCLUDED*. Note 

273 that range based selection is not well defined for pipelines that 

274 are not linear in nature, and correct behavior is not guaranteed, 

275 or may vary from run to run. 

276 

277 Returns 

278 ------- 

279 pipeline: `Pipeline` 

280 The pipeline loaded from specified location with appropriate (if 

281 any) subsetting 

282 

283 Notes 

284 ----- 

285 This method attempts to prune any contracts that contain labels which 

286 are not in the declared subset of labels. This pruning is done using a 

287 string based matching due to the nature of contracts and may prune more 

288 than it should. 

289 """ 

290 return cls.from_uri(filename) 

291 

292 @classmethod 

293 def from_uri(cls, uri: ResourcePathExpression) -> Pipeline: 

294 """Load a pipeline defined in a pipeline yaml file at a location 

295 specified by a URI. 

296 

297 Parameters 

298 ---------- 

299 uri: convertible to `ResourcePath` 

300 If a string is supplied this should be a URI path that points to a 

301 pipeline defined in yaml format, either as a direct path to the 

302 yaml file, or as a directory containing a "pipeline.yaml" file (the 

303 form used by `write_to_uri` with ``expand=True``). This uri may 

304 also supply additional labels to be used in subsetting the loaded 

305 Pipeline. These labels are separated from the path by a \\#, and 

306 may be specified as a comma separated list, or a range denoted as 

307 beginning..end. Beginning or end may be empty, in which case the 

308 range will be a half open interval. Unlike python iteration bounds, 

309 end bounds are *INCLUDED*. Note that range based selection is not 

310 well defined for pipelines that are not linear in nature, and 

311 correct behavior is not guaranteed, or may vary from run to run. 

312 The same specifiers can be used with a `ResourcePath` object, by 

313 being the sole contents in the fragments attribute. 

314 

315 Returns 

316 ------- 

317 pipeline: `Pipeline` 

318 The pipeline loaded from specified location with appropriate (if 

319 any) subsetting 

320 

321 Notes 

322 ----- 

323 This method attempts to prune any contracts that contain labels which 

324 are not in the declared subset of labels. This pruning is done using a 

325 string based matching due to the nature of contracts and may prune more 

326 than it should. 

327 """ 

328 # Split up the uri and any labels that were supplied 

329 uri, label_specifier = cls._parse_file_specifier(uri) 

330 pipeline: Pipeline = cls.fromIR(pipelineIR.PipelineIR.from_uri(uri)) 

331 

332 # If there are labels supplied, only keep those 

333 if label_specifier is not None: 

334 pipeline = pipeline.subsetFromLabels(label_specifier) 

335 return pipeline 

336 

337 def subsetFromLabels(self, labelSpecifier: LabelSpecifier) -> Pipeline: 

338 """Subset a pipeline to contain only labels specified in labelSpecifier 

339 

340 Parameters 

341 ---------- 

342 labelSpecifier : `labelSpecifier` 

343 Object containing labels that describes how to subset a pipeline. 

344 

345 Returns 

346 ------- 

347 pipeline : `Pipeline` 

348 A new pipeline object that is a subset of the old pipeline 

349 

350 Raises 

351 ------ 

352 ValueError 

353 Raised if there is an issue with specified labels 

354 

355 Notes 

356 ----- 

357 This method attempts to prune any contracts that contain labels which 

358 are not in the declared subset of labels. This pruning is done using a 

359 string based matching due to the nature of contracts and may prune more 

360 than it should. 

361 """ 

362 # Labels supplied as a set 

363 if labelSpecifier.labels: 

364 labelSet = labelSpecifier.labels 

365 # Labels supplied as a range, first create a list of all the labels 

366 # in the pipeline sorted according to task dependency. Then only 

367 # keep labels that lie between the supplied bounds 

368 else: 

369 # Create a copy of the pipeline to use when assessing the label 

370 # ordering. Use a dict for fast searching while preserving order. 

371 # Remove contracts so they do not fail in the expansion step. This 

372 # is needed because a user may only configure the tasks they intend 

373 # to run, which may cause some contracts to fail if they will later 

374 # be dropped 

375 pipeline = copy.deepcopy(self) 

376 pipeline._pipelineIR.contracts = [] 

377 labels = {taskdef.label: True for taskdef in pipeline.toExpandedPipeline()} 

378 

379 # Verify the bounds are in the labels 

380 if labelSpecifier.begin is not None: 

381 if labelSpecifier.begin not in labels: 

382 raise ValueError( 

383 f"Beginning of range subset, {labelSpecifier.begin}, not found in " 

384 "pipeline definition" 

385 ) 

386 if labelSpecifier.end is not None: 

387 if labelSpecifier.end not in labels: 

388 raise ValueError( 

389 f"End of range subset, {labelSpecifier.end}, not found in pipeline definition" 

390 ) 

391 

392 labelSet = set() 

393 for label in labels: 

394 if labelSpecifier.begin is not None: 

395 if label != labelSpecifier.begin: 

396 continue 

397 else: 

398 labelSpecifier.begin = None 

399 labelSet.add(label) 

400 if labelSpecifier.end is not None and label == labelSpecifier.end: 

401 break 

402 return Pipeline.fromIR(self._pipelineIR.subset_from_labels(labelSet)) 

403 

404 @staticmethod 

405 def _parse_file_specifier(uri: ResourcePathExpression) -> Tuple[ResourcePath, Optional[LabelSpecifier]]: 

406 """Split appart a uri and any possible label subsets""" 

407 if isinstance(uri, str): 

408 # This is to support legacy pipelines during transition 

409 uri, num_replace = re.subn("[:](?!\\/\\/)", "#", uri) 

410 if num_replace: 

411 warnings.warn( 

412 f"The pipeline file {uri} seems to use the legacy : to separate " 

413 "labels, this is deprecated and will be removed after June 2021, please use " 

414 "# instead.", 

415 category=FutureWarning, 

416 ) 

417 if uri.count("#") > 1: 

418 raise ValueError("Only one set of labels is allowed when specifying a pipeline to load") 

419 # Everything else can be converted directly to ResourcePath. 

420 uri = ResourcePath(uri) 

421 label_subset = uri.fragment or None 

422 

423 specifier: Optional[LabelSpecifier] 

424 if label_subset is not None: 

425 label_subset = urllib.parse.unquote(label_subset) 

426 args: Dict[str, Union[Set[str], str, None]] 

427 # labels supplied as a list 

428 if "," in label_subset: 

429 if ".." in label_subset: 

430 raise ValueError( 

431 "Can only specify a list of labels or a rangewhen loading a Pipline not both" 

432 ) 

433 args = {"labels": set(label_subset.split(","))} 

434 # labels supplied as a range 

435 elif ".." in label_subset: 

436 # Try to de-structure the labelSubset, this will fail if more 

437 # than one range is specified 

438 begin, end, *rest = label_subset.split("..") 

439 if rest: 

440 raise ValueError("Only one range can be specified when loading a pipeline") 

441 args = {"begin": begin if begin else None, "end": end if end else None} 

442 # Assume anything else is a single label 

443 else: 

444 args = {"labels": {label_subset}} 

445 

446 # MyPy doesn't like how cavalier kwarg construction is with types. 

447 specifier = LabelSpecifier(**args) # type: ignore 

448 else: 

449 specifier = None 

450 

451 return uri, specifier 

452 

453 @classmethod 

454 def fromString(cls, pipeline_string: str) -> Pipeline: 

455 """Create a pipeline from string formatted as a pipeline document. 

456 

457 Parameters 

458 ---------- 

459 pipeline_string : `str` 

460 A string that is formatted according like a pipeline document 

461 

462 Returns 

463 ------- 

464 pipeline: `Pipeline` 

465 """ 

466 pipeline = cls.fromIR(pipelineIR.PipelineIR.from_string(pipeline_string)) 

467 return pipeline 

468 

469 @classmethod 

470 def fromIR(cls, deserialized_pipeline: pipelineIR.PipelineIR) -> Pipeline: 

471 """Create a pipeline from an already created `PipelineIR` object. 

472 

473 Parameters 

474 ---------- 

475 deserialized_pipeline: `PipelineIR` 

476 An already created pipeline intermediate representation object 

477 

478 Returns 

479 ------- 

480 pipeline: `Pipeline` 

481 """ 

482 pipeline = cls.__new__(cls) 

483 pipeline._pipelineIR = deserialized_pipeline 

484 return pipeline 

485 

486 @classmethod 

487 def fromPipeline(cls, pipeline: Pipeline) -> Pipeline: 

488 """Create a new pipeline by copying an already existing `Pipeline`. 

489 

490 Parameters 

491 ---------- 

492 pipeline: `Pipeline` 

493 An already created pipeline intermediate representation object 

494 

495 Returns 

496 ------- 

497 pipeline: `Pipeline` 

498 """ 

499 return cls.fromIR(copy.deepcopy(pipeline._pipelineIR)) 

500 

501 def __str__(self) -> str: 

502 return str(self._pipelineIR) 

503 

504 def addInstrument(self, instrument: Union[Instrument, str]) -> None: 

505 """Add an instrument to the pipeline, or replace an instrument that is 

506 already defined. 

507 

508 Parameters 

509 ---------- 

510 instrument : `~lsst.daf.butler.instrument.Instrument` or `str` 

511 Either a derived class object of a `lsst.daf.butler.instrument` or 

512 a string corresponding to a fully qualified 

513 `lsst.daf.butler.instrument` name. 

514 """ 

515 if isinstance(instrument, str): 

516 pass 

517 else: 

518 # TODO: assume that this is a subclass of Instrument, no type 

519 # checking 

520 instrument = get_full_type_name(instrument) 

521 self._pipelineIR.instrument = instrument 

522 

523 def getInstrument(self) -> Optional[str]: 

524 """Get the instrument from the pipeline. 

525 

526 Returns 

527 ------- 

528 instrument : `str`, or None 

529 The fully qualified name of a `lsst.obs.base.Instrument` subclass, 

530 name, or None if the pipeline does not have an instrument. 

531 """ 

532 return self._pipelineIR.instrument 

533 

534 def addTask(self, task: Union[Type[PipelineTask], str], label: str) -> None: 

535 """Add a new task to the pipeline, or replace a task that is already 

536 associated with the supplied label. 

537 

538 Parameters 

539 ---------- 

540 task: `PipelineTask` or `str` 

541 Either a derived class object of a `PipelineTask` or a string 

542 corresponding to a fully qualified `PipelineTask` name. 

543 label: `str` 

544 A label that is used to identify the `PipelineTask` being added 

545 """ 

546 if isinstance(task, str): 

547 taskName = task 

548 elif issubclass(task, PipelineTask): 

549 taskName = get_full_type_name(task) 

550 else: 

551 raise ValueError( 

552 "task must be either a child class of PipelineTask or a string containing" 

553 " a fully qualified name to one" 

554 ) 

555 if not label: 

556 # in some cases (with command line-generated pipeline) tasks can 

557 # be defined without label which is not acceptable, use task 

558 # _DefaultName in that case 

559 if isinstance(task, str): 

560 task_class = doImportType(task) 

561 label = task_class._DefaultName 

562 self._pipelineIR.tasks[label] = pipelineIR.TaskIR(label, taskName) 

563 

564 def removeTask(self, label: str) -> None: 

565 """Remove a task from the pipeline. 

566 

567 Parameters 

568 ---------- 

569 label : `str` 

570 The label used to identify the task that is to be removed 

571 

572 Raises 

573 ------ 

574 KeyError 

575 If no task with that label exists in the pipeline 

576 

577 """ 

578 self._pipelineIR.tasks.pop(label) 

579 

580 def addConfigOverride(self, label: str, key: str, value: object) -> None: 

581 """Apply single config override. 

582 

583 Parameters 

584 ---------- 

585 label : `str` 

586 Label of the task. 

587 key: `str` 

588 Fully-qualified field name. 

589 value : object 

590 Value to be given to a field. 

591 """ 

592 self._addConfigImpl(label, pipelineIR.ConfigIR(rest={key: value})) 

593 

594 def addConfigFile(self, label: str, filename: str) -> None: 

595 """Add overrides from a specified file. 

596 

597 Parameters 

598 ---------- 

599 label : `str` 

600 The label used to identify the task associated with config to 

601 modify 

602 filename : `str` 

603 Path to the override file. 

604 """ 

605 self._addConfigImpl(label, pipelineIR.ConfigIR(file=[filename])) 

606 

607 def addConfigPython(self, label: str, pythonString: str) -> None: 

608 """Add Overrides by running a snippet of python code against a config. 

609 

610 Parameters 

611 ---------- 

612 label : `str` 

613 The label used to identity the task associated with config to 

614 modify. 

615 pythonString: `str` 

616 A string which is valid python code to be executed. This is done 

617 with config as the only local accessible value. 

618 """ 

619 self._addConfigImpl(label, pipelineIR.ConfigIR(python=pythonString)) 

620 

621 def _addConfigImpl(self, label: str, newConfig: pipelineIR.ConfigIR) -> None: 

622 if label == "parameters": 

623 if newConfig.rest.keys() - self._pipelineIR.parameters.mapping.keys(): 

624 raise ValueError("Cannot override parameters that are not defined in pipeline") 

625 self._pipelineIR.parameters.mapping.update(newConfig.rest) 

626 if newConfig.file: 

627 raise ValueError("Setting parameters section with config file is not supported") 

628 if newConfig.python: 

629 raise ValueError("Setting parameters section using python block in unsupported") 

630 return 

631 if label not in self._pipelineIR.tasks: 

632 raise LookupError(f"There are no tasks labeled '{label}' in the pipeline") 

633 self._pipelineIR.tasks[label].add_or_update_config(newConfig) 

634 

635 def write_to_uri(self, uri: ResourcePathExpression) -> None: 

636 """Write the pipeline to a file or directory. 

637 

638 Parameters 

639 ---------- 

640 uri : convertible to `ResourcePath` 

641 URI to write to; may have any scheme with `ResourcePath` write 

642 support or no scheme for a local file/directory. Should have a 

643 ``.yaml``. 

644 """ 

645 self._pipelineIR.write_to_uri(uri) 

646 

647 def toExpandedPipeline(self) -> Generator[TaskDef, None, None]: 

648 """Returns a generator of TaskDefs which can be used to create quantum 

649 graphs. 

650 

651 Returns 

652 ------- 

653 generator : generator of `TaskDef` 

654 The generator returned will be the sorted iterator of tasks which 

655 are to be used in constructing a quantum graph. 

656 

657 Raises 

658 ------ 

659 NotImplementedError 

660 If a dataId is supplied in a config block. This is in place for 

661 future use 

662 """ 

663 taskDefs = [] 

664 for label in self._pipelineIR.tasks: 

665 taskDefs.append(self._buildTaskDef(label)) 

666 

667 # lets evaluate the contracts 

668 if self._pipelineIR.contracts is not None: 

669 label_to_config = {x.label: x.config for x in taskDefs} 

670 for contract in self._pipelineIR.contracts: 

671 # execute this in its own line so it can raise a good error 

672 # message if there was problems with the eval 

673 success = eval(contract.contract, None, label_to_config) 

674 if not success: 

675 extra_info = f": {contract.msg}" if contract.msg is not None else "" 

676 raise pipelineIR.ContractError( 

677 f"Contract(s) '{contract.contract}' were not satisfied{extra_info}" 

678 ) 

679 

680 taskDefs = sorted(taskDefs, key=lambda x: x.label) 

681 yield from pipeTools.orderPipeline(taskDefs) 

682 

683 def _buildTaskDef(self, label: str) -> TaskDef: 

684 if (taskIR := self._pipelineIR.tasks.get(label)) is None: 

685 raise NameError(f"Label {label} does not appear in this pipeline") 

686 taskClass: Type[PipelineTask] = doImportType(taskIR.klass) 

687 taskName = get_full_type_name(taskClass) 

688 config = taskClass.ConfigClass() 

689 overrides = ConfigOverrides() 

690 if self._pipelineIR.instrument is not None: 

691 overrides.addInstrumentOverride(self._pipelineIR.instrument, taskClass._DefaultName) 

692 if taskIR.config is not None: 

693 for configIR in (configIr.formatted(self._pipelineIR.parameters) for configIr in taskIR.config): 

694 if configIR.dataId is not None: 

695 raise NotImplementedError( 

696 "Specializing a config on a partial data id is not yet " 

697 "supported in Pipeline definition" 

698 ) 

699 # only apply override if it applies to everything 

700 if configIR.dataId is None: 

701 if configIR.file: 

702 for configFile in configIR.file: 

703 overrides.addFileOverride(os.path.expandvars(configFile)) 

704 if configIR.python is not None: 

705 overrides.addPythonOverride(configIR.python) 

706 for key, value in configIR.rest.items(): 

707 overrides.addValueOverride(key, value) 

708 overrides.applyTo(config) 

709 return TaskDef(taskName=taskName, config=config, taskClass=taskClass, label=label) 

710 

711 def __iter__(self) -> Generator[TaskDef, None, None]: 

712 return self.toExpandedPipeline() 

713 

714 def __getitem__(self, item: str) -> TaskDef: 

715 return self._buildTaskDef(item) 

716 

717 def __len__(self) -> int: 

718 return len(self._pipelineIR.tasks) 

719 

720 def __eq__(self, other: object) -> bool: 

721 if not isinstance(other, Pipeline): 

722 return False 

723 elif self._pipelineIR == other._pipelineIR: 

724 # Shortcut: if the IR is the same, the expanded pipeline must be 

725 # the same as well. But the converse is not true. 

726 return True 

727 else: 

728 self_expanded = {td.label: (td.taskClass,) for td in self} 

729 other_expanded = {td.label: (td.taskClass,) for td in other} 

730 if self_expanded != other_expanded: 

731 return False 

732 # After DM-27847, we should compare configuration here, or better, 

733 # delegated to TaskDef.__eq__ after making that compare configurations. 

734 raise NotImplementedError( 

735 "Pipelines cannot be compared because config instances cannot be compared; see DM-27847." 

736 ) 

737 

738 

739@dataclass(frozen=True) 

740class TaskDatasetTypes: 

741 """An immutable struct that extracts and classifies the dataset types used 

742 by a `PipelineTask` 

743 """ 

744 

745 initInputs: NamedValueSet[DatasetType] 

746 """Dataset types that are needed as inputs in order to construct this Task. 

747 

748 Task-level `initInputs` may be classified as either 

749 `~PipelineDatasetTypes.initInputs` or 

750 `~PipelineDatasetTypes.initIntermediates` at the Pipeline level. 

751 """ 

752 

753 initOutputs: NamedValueSet[DatasetType] 

754 """Dataset types that may be written after constructing this Task. 

755 

756 Task-level `initOutputs` may be classified as either 

757 `~PipelineDatasetTypes.initOutputs` or 

758 `~PipelineDatasetTypes.initIntermediates` at the Pipeline level. 

759 """ 

760 

761 inputs: NamedValueSet[DatasetType] 

762 """Dataset types that are regular inputs to this Task. 

763 

764 If an input dataset needed for a Quantum cannot be found in the input 

765 collection(s) or produced by another Task in the Pipeline, that Quantum 

766 (and all dependent Quanta) will not be produced. 

767 

768 Task-level `inputs` may be classified as either 

769 `~PipelineDatasetTypes.inputs` or `~PipelineDatasetTypes.intermediates` 

770 at the Pipeline level. 

771 """ 

772 

773 prerequisites: NamedValueSet[DatasetType] 

774 """Dataset types that are prerequisite inputs to this Task. 

775 

776 Prerequisite inputs must exist in the input collection(s) before the 

777 pipeline is run, but do not constrain the graph - if a prerequisite is 

778 missing for a Quantum, `PrerequisiteMissingError` is raised. 

779 

780 Prerequisite inputs are not resolved until the second stage of 

781 QuantumGraph generation. 

782 """ 

783 

784 outputs: NamedValueSet[DatasetType] 

785 """Dataset types that are produced by this Task. 

786 

787 Task-level `outputs` may be classified as either 

788 `~PipelineDatasetTypes.outputs` or `~PipelineDatasetTypes.intermediates` 

789 at the Pipeline level. 

790 """ 

791 

792 @classmethod 

793 def fromTaskDef( 

794 cls, 

795 taskDef: TaskDef, 

796 *, 

797 registry: Registry, 

798 include_configs: bool = True, 

799 storage_class_mapping: Optional[Mapping[str, str]] = None, 

800 ) -> TaskDatasetTypes: 

801 """Extract and classify the dataset types from a single `PipelineTask`. 

802 

803 Parameters 

804 ---------- 

805 taskDef: `TaskDef` 

806 An instance of a `TaskDef` class for a particular `PipelineTask`. 

807 registry: `Registry` 

808 Registry used to construct normalized `DatasetType` objects and 

809 retrieve those that are incomplete. 

810 include_configs : `bool`, optional 

811 If `True` (default) include config dataset types as 

812 ``initOutputs``. 

813 storage_class_mapping : `Mapping` of `str` to `StorageClass`, optional 

814 If a taskdef contains a component dataset type that is unknown 

815 to the registry, its parent StorageClass will be looked up in this 

816 mapping if it is supplied. If the mapping does not contain the 

817 composite dataset type, or the mapping is not supplied an exception 

818 will be raised. 

819 

820 Returns 

821 ------- 

822 types: `TaskDatasetTypes` 

823 The dataset types used by this task. 

824 

825 Raises 

826 ------ 

827 ValueError 

828 Raised if dataset type connection definition differs from 

829 registry definition. 

830 LookupError 

831 Raised if component parent StorageClass could not be determined 

832 and storage_class_mapping does not contain the composite type, or 

833 is set to None. 

834 """ 

835 

836 def makeDatasetTypesSet( 

837 connectionType: str, 

838 is_input: bool, 

839 freeze: bool = True, 

840 ) -> NamedValueSet[DatasetType]: 

841 """Constructs a set of true `DatasetType` objects 

842 

843 Parameters 

844 ---------- 

845 connectionType : `str` 

846 Name of the connection type to produce a set for, corresponds 

847 to an attribute of type `list` on the connection class instance 

848 is_input : `bool` 

849 These are input dataset types, else they are output dataset 

850 types. 

851 freeze : `bool`, optional 

852 If `True`, call `NamedValueSet.freeze` on the object returned. 

853 

854 Returns 

855 ------- 

856 datasetTypes : `NamedValueSet` 

857 A set of all datasetTypes which correspond to the input 

858 connection type specified in the connection class of this 

859 `PipelineTask` 

860 

861 Raises 

862 ------ 

863 ValueError 

864 Raised if dataset type connection definition differs from 

865 registry definition. 

866 LookupError 

867 Raised if component parent StorageClass could not be determined 

868 and storage_class_mapping does not contain the composite type, 

869 or is set to None. 

870 

871 Notes 

872 ----- 

873 This function is a closure over the variables ``registry`` and 

874 ``taskDef``, and ``storage_class_mapping``. 

875 """ 

876 datasetTypes = NamedValueSet[DatasetType]() 

877 for c in iterConnections(taskDef.connections, connectionType): 

878 dimensions = set(getattr(c, "dimensions", set())) 

879 if "skypix" in dimensions: 

880 try: 

881 datasetType = registry.getDatasetType(c.name) 

882 except LookupError as err: 

883 raise LookupError( 

884 f"DatasetType '{c.name}' referenced by " 

885 f"{type(taskDef.connections).__name__} uses 'skypix' as a dimension " 

886 f"placeholder, but does not already exist in the registry. " 

887 f"Note that reference catalog names are now used as the dataset " 

888 f"type name instead of 'ref_cat'." 

889 ) from err 

890 rest1 = set(registry.dimensions.extract(dimensions - set(["skypix"])).names) 

891 rest2 = set( 

892 dim.name for dim in datasetType.dimensions if not isinstance(dim, SkyPixDimension) 

893 ) 

894 if rest1 != rest2: 

895 raise ValueError( 

896 f"Non-skypix dimensions for dataset type {c.name} declared in " 

897 f"connections ({rest1}) are inconsistent with those in " 

898 f"registry's version of this dataset ({rest2})." 

899 ) 

900 else: 

901 # Component dataset types are not explicitly in the 

902 # registry. This complicates consistency checks with 

903 # registry and requires we work out the composite storage 

904 # class. 

905 registryDatasetType = None 

906 try: 

907 registryDatasetType = registry.getDatasetType(c.name) 

908 except KeyError: 

909 compositeName, componentName = DatasetType.splitDatasetTypeName(c.name) 

910 if componentName: 

911 if storage_class_mapping is None or compositeName not in storage_class_mapping: 

912 raise LookupError( 

913 "Component parent class cannot be determined, and " 

914 "composite name was not in storage class mapping, or no " 

915 "storage_class_mapping was supplied" 

916 ) 

917 else: 

918 parentStorageClass = storage_class_mapping[compositeName] 

919 else: 

920 parentStorageClass = None 

921 datasetType = c.makeDatasetType( 

922 registry.dimensions, parentStorageClass=parentStorageClass 

923 ) 

924 registryDatasetType = datasetType 

925 else: 

926 datasetType = c.makeDatasetType( 

927 registry.dimensions, parentStorageClass=registryDatasetType.parentStorageClass 

928 ) 

929 

930 if registryDatasetType and datasetType != registryDatasetType: 

931 # The dataset types differ but first check to see if 

932 # they are compatible before raising. 

933 if is_input: 

934 # This DatasetType must be compatible on get. 

935 is_compatible = datasetType.is_compatible_with(registryDatasetType) 

936 else: 

937 # Has to be able to be converted to expect type 

938 # on put. 

939 is_compatible = registryDatasetType.is_compatible_with(datasetType) 

940 if is_compatible: 

941 # For inputs we want the pipeline to use the 

942 # pipeline definition, for outputs it should use 

943 # the registry definition. 

944 if not is_input: 

945 datasetType = registryDatasetType 

946 _LOG.debug( 

947 "Dataset types differ (task %s != registry %s) but are compatible" 

948 " for %s in %s.", 

949 datasetType, 

950 registryDatasetType, 

951 "input" if is_input else "output", 

952 taskDef.label, 

953 ) 

954 else: 

955 try: 

956 # Explicitly check for storage class just to 

957 # make more specific message. 

958 _ = datasetType.storageClass 

959 except KeyError: 

960 raise ValueError( 

961 "Storage class does not exist for supplied dataset type " 

962 f"{datasetType} for {taskDef.label}." 

963 ) from None 

964 raise ValueError( 

965 f"Supplied dataset type ({datasetType}) inconsistent with " 

966 f"registry definition ({registryDatasetType}) " 

967 f"for {taskDef.label}." 

968 ) 

969 datasetTypes.add(datasetType) 

970 if freeze: 

971 datasetTypes.freeze() 

972 return datasetTypes 

973 

974 # optionally add initOutput dataset for config 

975 initOutputs = makeDatasetTypesSet("initOutputs", is_input=False, freeze=False) 

976 if include_configs: 

977 initOutputs.add( 

978 DatasetType( 

979 taskDef.configDatasetName, 

980 registry.dimensions.empty, 

981 storageClass="Config", 

982 ) 

983 ) 

984 initOutputs.freeze() 

985 

986 # optionally add output dataset for metadata 

987 outputs = makeDatasetTypesSet("outputs", is_input=False, freeze=False) 

988 if taskDef.metadataDatasetName is not None: 

989 # Metadata is supposed to be of the TaskMetadata type, its 

990 # dimensions correspond to a task quantum. 

991 dimensions = registry.dimensions.extract(taskDef.connections.dimensions) 

992 

993 # Allow the storage class definition to be read from the existing 

994 # dataset type definition if present. 

995 try: 

996 current = registry.getDatasetType(taskDef.metadataDatasetName) 

997 except KeyError: 

998 # No previous definition so use the default. 

999 storageClass = "TaskMetadata" if _TASK_METADATA_TYPE is TaskMetadata else "PropertySet" 

1000 else: 

1001 storageClass = current.storageClass.name 

1002 

1003 outputs.update({DatasetType(taskDef.metadataDatasetName, dimensions, storageClass)}) 

1004 if taskDef.logOutputDatasetName is not None: 

1005 # Log output dimensions correspond to a task quantum. 

1006 dimensions = registry.dimensions.extract(taskDef.connections.dimensions) 

1007 outputs.update({DatasetType(taskDef.logOutputDatasetName, dimensions, "ButlerLogRecords")}) 

1008 

1009 outputs.freeze() 

1010 

1011 return cls( 

1012 initInputs=makeDatasetTypesSet("initInputs", is_input=True), 

1013 initOutputs=initOutputs, 

1014 inputs=makeDatasetTypesSet("inputs", is_input=True), 

1015 prerequisites=makeDatasetTypesSet("prerequisiteInputs", is_input=True), 

1016 outputs=outputs, 

1017 ) 

1018 

1019 

1020@dataclass(frozen=True) 

1021class PipelineDatasetTypes: 

1022 """An immutable struct that classifies the dataset types used in a 

1023 `Pipeline`. 

1024 """ 

1025 

1026 packagesDatasetName: ClassVar[str] = "packages" 

1027 """Name of a dataset type used to save package versions. 

1028 """ 

1029 

1030 initInputs: NamedValueSet[DatasetType] 

1031 """Dataset types that are needed as inputs in order to construct the Tasks 

1032 in this Pipeline. 

1033 

1034 This does not include dataset types that are produced when constructing 

1035 other Tasks in the Pipeline (these are classified as `initIntermediates`). 

1036 """ 

1037 

1038 initOutputs: NamedValueSet[DatasetType] 

1039 """Dataset types that may be written after constructing the Tasks in this 

1040 Pipeline. 

1041 

1042 This does not include dataset types that are also used as inputs when 

1043 constructing other Tasks in the Pipeline (these are classified as 

1044 `initIntermediates`). 

1045 """ 

1046 

1047 initIntermediates: NamedValueSet[DatasetType] 

1048 """Dataset types that are both used when constructing one or more Tasks 

1049 in the Pipeline and produced as a side-effect of constructing another 

1050 Task in the Pipeline. 

1051 """ 

1052 

1053 inputs: NamedValueSet[DatasetType] 

1054 """Dataset types that are regular inputs for the full pipeline. 

1055 

1056 If an input dataset needed for a Quantum cannot be found in the input 

1057 collection(s), that Quantum (and all dependent Quanta) will not be 

1058 produced. 

1059 """ 

1060 

1061 prerequisites: NamedValueSet[DatasetType] 

1062 """Dataset types that are prerequisite inputs for the full Pipeline. 

1063 

1064 Prerequisite inputs must exist in the input collection(s) before the 

1065 pipeline is run, but do not constrain the graph - if a prerequisite is 

1066 missing for a Quantum, `PrerequisiteMissingError` is raised. 

1067 

1068 Prerequisite inputs are not resolved until the second stage of 

1069 QuantumGraph generation. 

1070 """ 

1071 

1072 intermediates: NamedValueSet[DatasetType] 

1073 """Dataset types that are output by one Task in the Pipeline and consumed 

1074 as inputs by one or more other Tasks in the Pipeline. 

1075 """ 

1076 

1077 outputs: NamedValueSet[DatasetType] 

1078 """Dataset types that are output by a Task in the Pipeline and not consumed 

1079 by any other Task in the Pipeline. 

1080 """ 

1081 

1082 byTask: Mapping[str, TaskDatasetTypes] 

1083 """Per-Task dataset types, keyed by label in the `Pipeline`. 

1084 

1085 This is guaranteed to be zip-iterable with the `Pipeline` itself (assuming 

1086 neither has been modified since the dataset types were extracted, of 

1087 course). 

1088 """ 

1089 

1090 @classmethod 

1091 def fromPipeline( 

1092 cls, 

1093 pipeline: Union[Pipeline, Iterable[TaskDef]], 

1094 *, 

1095 registry: Registry, 

1096 include_configs: bool = True, 

1097 include_packages: bool = True, 

1098 ) -> PipelineDatasetTypes: 

1099 """Extract and classify the dataset types from all tasks in a 

1100 `Pipeline`. 

1101 

1102 Parameters 

1103 ---------- 

1104 pipeline: `Pipeline` or `Iterable` [ `TaskDef` ] 

1105 A collection of tasks that can be run together. 

1106 registry: `Registry` 

1107 Registry used to construct normalized `DatasetType` objects and 

1108 retrieve those that are incomplete. 

1109 include_configs : `bool`, optional 

1110 If `True` (default) include config dataset types as 

1111 ``initOutputs``. 

1112 include_packages : `bool`, optional 

1113 If `True` (default) include the dataset type for software package 

1114 versions in ``initOutputs``. 

1115 

1116 Returns 

1117 ------- 

1118 types: `PipelineDatasetTypes` 

1119 The dataset types used by this `Pipeline`. 

1120 

1121 Raises 

1122 ------ 

1123 ValueError 

1124 Raised if Tasks are inconsistent about which datasets are marked 

1125 prerequisite. This indicates that the Tasks cannot be run as part 

1126 of the same `Pipeline`. 

1127 """ 

1128 allInputs = NamedValueSet[DatasetType]() 

1129 allOutputs = NamedValueSet[DatasetType]() 

1130 allInitInputs = NamedValueSet[DatasetType]() 

1131 allInitOutputs = NamedValueSet[DatasetType]() 

1132 prerequisites = NamedValueSet[DatasetType]() 

1133 byTask = dict() 

1134 if include_packages: 

1135 allInitOutputs.add( 

1136 DatasetType( 

1137 cls.packagesDatasetName, 

1138 registry.dimensions.empty, 

1139 storageClass="Packages", 

1140 ) 

1141 ) 

1142 # create a list of TaskDefs in case the input is a generator 

1143 pipeline = list(pipeline) 

1144 

1145 # collect all the output dataset types 

1146 typeStorageclassMap: Dict[str, str] = {} 

1147 for taskDef in pipeline: 

1148 for outConnection in iterConnections(taskDef.connections, "outputs"): 

1149 typeStorageclassMap[outConnection.name] = outConnection.storageClass 

1150 

1151 for taskDef in pipeline: 

1152 thisTask = TaskDatasetTypes.fromTaskDef( 

1153 taskDef, 

1154 registry=registry, 

1155 include_configs=include_configs, 

1156 storage_class_mapping=typeStorageclassMap, 

1157 ) 

1158 allInitInputs.update(thisTask.initInputs) 

1159 allInitOutputs.update(thisTask.initOutputs) 

1160 allInputs.update(thisTask.inputs) 

1161 prerequisites.update(thisTask.prerequisites) 

1162 allOutputs.update(thisTask.outputs) 

1163 byTask[taskDef.label] = thisTask 

1164 if not prerequisites.isdisjoint(allInputs): 

1165 raise ValueError( 

1166 "{} marked as both prerequisites and regular inputs".format( 

1167 {dt.name for dt in allInputs & prerequisites} 

1168 ) 

1169 ) 

1170 if not prerequisites.isdisjoint(allOutputs): 

1171 raise ValueError( 

1172 "{} marked as both prerequisites and outputs".format( 

1173 {dt.name for dt in allOutputs & prerequisites} 

1174 ) 

1175 ) 

1176 # Make sure that components which are marked as inputs get treated as 

1177 # intermediates if there is an output which produces the composite 

1178 # containing the component 

1179 intermediateComponents = NamedValueSet[DatasetType]() 

1180 intermediateComposites = NamedValueSet[DatasetType]() 

1181 outputNameMapping = {dsType.name: dsType for dsType in allOutputs} 

1182 for dsType in allInputs: 

1183 # get the name of a possible component 

1184 name, component = dsType.nameAndComponent() 

1185 # if there is a component name, that means this is a component 

1186 # DatasetType, if there is an output which produces the parent of 

1187 # this component, treat this input as an intermediate 

1188 if component is not None: 

1189 # This needs to be in this if block, because someone might have 

1190 # a composite that is a pure input from existing data 

1191 if name in outputNameMapping: 

1192 intermediateComponents.add(dsType) 

1193 intermediateComposites.add(outputNameMapping[name]) 

1194 

1195 def checkConsistency(a: NamedValueSet, b: NamedValueSet) -> None: 

1196 common = a.names & b.names 

1197 for name in common: 

1198 # Any compatibility is allowed. This function does not know 

1199 # if a dataset type is to be used for input or output. 

1200 if not (a[name].is_compatible_with(b[name]) or b[name].is_compatible_with(a[name])): 

1201 raise ValueError(f"Conflicting definitions for dataset type: {a[name]} != {b[name]}.") 

1202 

1203 checkConsistency(allInitInputs, allInitOutputs) 

1204 checkConsistency(allInputs, allOutputs) 

1205 checkConsistency(allInputs, intermediateComposites) 

1206 checkConsistency(allOutputs, intermediateComposites) 

1207 

1208 def frozen(s: AbstractSet[DatasetType]) -> NamedValueSet[DatasetType]: 

1209 assert isinstance(s, NamedValueSet) 

1210 s.freeze() 

1211 return s 

1212 

1213 return cls( 

1214 initInputs=frozen(allInitInputs - allInitOutputs), 

1215 initIntermediates=frozen(allInitInputs & allInitOutputs), 

1216 initOutputs=frozen(allInitOutputs - allInitInputs), 

1217 inputs=frozen(allInputs - allOutputs - intermediateComponents), 

1218 # If there are storage class differences in inputs and outputs 

1219 # the intermediates have to choose priority. Here choose that 

1220 # inputs to tasks much match the requested storage class by 

1221 # applying the inputs over the top of the outputs. 

1222 intermediates=frozen(allOutputs & allInputs | intermediateComponents), 

1223 outputs=frozen(allOutputs - allInputs - intermediateComposites), 

1224 prerequisites=frozen(prerequisites), 

1225 byTask=MappingProxyType(byTask), # MappingProxyType -> frozen view of dict for immutability 

1226 ) 

1227 

1228 @classmethod 

1229 def initOutputNames( 

1230 cls, 

1231 pipeline: Union[Pipeline, Iterable[TaskDef]], 

1232 *, 

1233 include_configs: bool = True, 

1234 include_packages: bool = True, 

1235 ) -> Iterator[str]: 

1236 """Return the names of dataset types ot task initOutputs, Configs, 

1237 and package versions for a pipeline. 

1238 

1239 Parameters 

1240 ---------- 

1241 pipeline: `Pipeline` or `Iterable` [ `TaskDef` ] 

1242 A `Pipeline` instance or collection of `TaskDef` instances. 

1243 include_configs : `bool`, optional 

1244 If `True` (default) include config dataset types. 

1245 include_packages : `bool`, optional 

1246 If `True` (default) include the dataset type for package versions. 

1247 

1248 Yields 

1249 ------ 

1250 datasetTypeName : `str` 

1251 Name of the dataset type. 

1252 """ 

1253 if include_packages: 

1254 # Package versions dataset type 

1255 yield cls.packagesDatasetName 

1256 

1257 if isinstance(pipeline, Pipeline): 

1258 pipeline = pipeline.toExpandedPipeline() 

1259 

1260 for taskDef in pipeline: 

1261 

1262 # all task InitOutputs 

1263 for name in taskDef.connections.initOutputs: 

1264 attribute = getattr(taskDef.connections, name) 

1265 yield attribute.name 

1266 

1267 # config dataset name 

1268 if include_configs: 

1269 yield taskDef.configDatasetName