Coverage for python/lsst/daf/butler/datastores/fileDatastore.py: 85%

921 statements  

« prev     ^ index     » next       coverage.py v7.5.0, created at 2024-04-24 23:49 -0700

1# This file is part of daf_butler. 

2# 

3# Developed for the LSST Data Management System. 

4# This product includes software developed by the LSST Project 

5# (http://www.lsst.org). 

6# See the COPYRIGHT file at the top-level directory of this distribution 

7# for details of code ownership. 

8# 

9# This 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"""Generic file-based datastore code.""" 

24 

25__all__ = ("FileDatastore",) 

26 

27import hashlib 

28import logging 

29from collections import defaultdict 

30from dataclasses import dataclass 

31from typing import ( 

32 TYPE_CHECKING, 

33 Any, 

34 ClassVar, 

35 Dict, 

36 Iterable, 

37 List, 

38 Mapping, 

39 Optional, 

40 Sequence, 

41 Set, 

42 Tuple, 

43 Type, 

44 Union, 

45) 

46 

47from lsst.daf.butler import ( 

48 CompositesMap, 

49 Config, 

50 DatasetId, 

51 DatasetRef, 

52 DatasetRefURIs, 

53 DatasetType, 

54 DatasetTypeNotSupportedError, 

55 Datastore, 

56 DatastoreCacheManager, 

57 DatastoreConfig, 

58 DatastoreDisabledCacheManager, 

59 DatastoreRecordData, 

60 DatastoreValidationError, 

61 FileDataset, 

62 FileDescriptor, 

63 FileTemplates, 

64 FileTemplateValidationError, 

65 Formatter, 

66 FormatterFactory, 

67 Location, 

68 LocationFactory, 

69 Progress, 

70 StorageClass, 

71 StoredDatastoreItemInfo, 

72 StoredFileInfo, 

73 ddl, 

74) 

75from lsst.daf.butler.core.repoRelocation import replaceRoot 

76from lsst.daf.butler.core.utils import transactional 

77from lsst.daf.butler.registry.interfaces import DatastoreRegistryBridge, ReadOnlyDatabaseError 

78from lsst.resources import ResourcePath, ResourcePathExpression 

79from lsst.utils.introspection import get_class_of, get_instance_of 

80from lsst.utils.iteration import chunk_iterable 

81 

82# For VERBOSE logging usage. 

83from lsst.utils.logging import VERBOSE, getLogger 

84from lsst.utils.timer import time_this 

85from sqlalchemy import BigInteger, String 

86 

87from ..registry.interfaces import FakeDatasetRef 

88from .genericDatastore import GenericBaseDatastore 

89 

90if TYPE_CHECKING: 90 ↛ 91line 90 didn't jump to line 91, because the condition on line 90 was never true

91 from lsst.daf.butler import AbstractDatastoreCacheManager, LookupKey 

92 from lsst.daf.butler.registry.interfaces import DatasetIdRef, DatastoreRegistryBridgeManager 

93 

94log = getLogger(__name__) 

95 

96 

97class _IngestPrepData(Datastore.IngestPrepData): 

98 """Helper class for FileDatastore ingest implementation. 

99 

100 Parameters 

101 ---------- 

102 datasets : `list` of `FileDataset` 

103 Files to be ingested by this datastore. 

104 """ 

105 

106 def __init__(self, datasets: List[FileDataset]): 

107 super().__init__(ref for dataset in datasets for ref in dataset.refs) 

108 self.datasets = datasets 

109 

110 

111@dataclass(frozen=True) 

112class DatastoreFileGetInformation: 

113 """Collection of useful parameters needed to retrieve a file from 

114 a Datastore. 

115 """ 

116 

117 location: Location 

118 """The location from which to read the dataset.""" 

119 

120 formatter: Formatter 

121 """The `Formatter` to use to deserialize the dataset.""" 

122 

123 info: StoredFileInfo 

124 """Stored information about this file and its formatter.""" 

125 

126 assemblerParams: Mapping[str, Any] 

127 """Parameters to use for post-processing the retrieved dataset.""" 

128 

129 formatterParams: Mapping[str, Any] 

130 """Parameters that were understood by the associated formatter.""" 

131 

132 component: Optional[str] 

133 """The component to be retrieved (can be `None`).""" 

134 

135 readStorageClass: StorageClass 

136 """The `StorageClass` of the dataset being read.""" 

137 

138 

139class FileDatastore(GenericBaseDatastore): 

140 """Generic Datastore for file-based implementations. 

141 

142 Should always be sub-classed since key abstract methods are missing. 

143 

144 Parameters 

145 ---------- 

146 config : `DatastoreConfig` or `str` 

147 Configuration as either a `Config` object or URI to file. 

148 bridgeManager : `DatastoreRegistryBridgeManager` 

149 Object that manages the interface between `Registry` and datastores. 

150 butlerRoot : `str`, optional 

151 New datastore root to use to override the configuration value. 

152 

153 Raises 

154 ------ 

155 ValueError 

156 If root location does not exist and ``create`` is `False` in the 

157 configuration. 

158 """ 

159 

160 defaultConfigFile: ClassVar[Optional[str]] = None 

161 """Path to configuration defaults. Accessed within the ``config`` resource 

162 or relative to a search path. Can be None if no defaults specified. 

163 """ 

164 

165 root: ResourcePath 

166 """Root directory URI of this `Datastore`.""" 

167 

168 locationFactory: LocationFactory 

169 """Factory for creating locations relative to the datastore root.""" 

170 

171 formatterFactory: FormatterFactory 

172 """Factory for creating instances of formatters.""" 

173 

174 templates: FileTemplates 

175 """File templates that can be used by this `Datastore`.""" 

176 

177 composites: CompositesMap 

178 """Determines whether a dataset should be disassembled on put.""" 

179 

180 defaultConfigFile = "datastores/fileDatastore.yaml" 

181 """Path to configuration defaults. Accessed within the ``config`` resource 

182 or relative to a search path. Can be None if no defaults specified. 

183 """ 

184 

185 @classmethod 

186 def setConfigRoot(cls, root: str, config: Config, full: Config, overwrite: bool = True) -> None: 

187 """Set any filesystem-dependent config options for this Datastore to 

188 be appropriate for a new empty repository with the given root. 

189 

190 Parameters 

191 ---------- 

192 root : `str` 

193 URI to the root of the data repository. 

194 config : `Config` 

195 A `Config` to update. Only the subset understood by 

196 this component will be updated. Will not expand 

197 defaults. 

198 full : `Config` 

199 A complete config with all defaults expanded that can be 

200 converted to a `DatastoreConfig`. Read-only and will not be 

201 modified by this method. 

202 Repository-specific options that should not be obtained 

203 from defaults when Butler instances are constructed 

204 should be copied from ``full`` to ``config``. 

205 overwrite : `bool`, optional 

206 If `False`, do not modify a value in ``config`` if the value 

207 already exists. Default is always to overwrite with the provided 

208 ``root``. 

209 

210 Notes 

211 ----- 

212 If a keyword is explicitly defined in the supplied ``config`` it 

213 will not be overridden by this method if ``overwrite`` is `False`. 

214 This allows explicit values set in external configs to be retained. 

215 """ 

216 Config.updateParameters( 

217 DatastoreConfig, 

218 config, 

219 full, 

220 toUpdate={"root": root}, 

221 toCopy=("cls", ("records", "table")), 

222 overwrite=overwrite, 

223 ) 

224 

225 @classmethod 

226 def makeTableSpec(cls, datasetIdColumnType: type) -> ddl.TableSpec: 

227 return ddl.TableSpec( 

228 fields=[ 

229 ddl.FieldSpec(name="dataset_id", dtype=datasetIdColumnType, primaryKey=True), 

230 ddl.FieldSpec(name="path", dtype=String, length=256, nullable=False), 

231 ddl.FieldSpec(name="formatter", dtype=String, length=128, nullable=False), 

232 ddl.FieldSpec(name="storage_class", dtype=String, length=64, nullable=False), 

233 # Use empty string to indicate no component 

234 ddl.FieldSpec(name="component", dtype=String, length=32, primaryKey=True), 

235 # TODO: should checksum be Base64Bytes instead? 

236 ddl.FieldSpec(name="checksum", dtype=String, length=128, nullable=True), 

237 ddl.FieldSpec(name="file_size", dtype=BigInteger, nullable=True), 

238 ], 

239 unique=frozenset(), 

240 indexes=[tuple(["path"])], 

241 ) 

242 

243 def __init__( 

244 self, 

245 config: Union[DatastoreConfig, str], 

246 bridgeManager: DatastoreRegistryBridgeManager, 

247 butlerRoot: str = None, 

248 ): 

249 super().__init__(config, bridgeManager) 

250 if "root" not in self.config: 250 ↛ 251line 250 didn't jump to line 251, because the condition on line 250 was never true

251 raise ValueError("No root directory specified in configuration") 

252 

253 self._bridgeManager = bridgeManager 

254 

255 # Name ourselves either using an explicit name or a name 

256 # derived from the (unexpanded) root 

257 if "name" in self.config: 

258 self.name = self.config["name"] 

259 else: 

260 # We use the unexpanded root in the name to indicate that this 

261 # datastore can be moved without having to update registry. 

262 self.name = "{}@{}".format(type(self).__name__, self.config["root"]) 

263 

264 # Support repository relocation in config 

265 # Existence of self.root is checked in subclass 

266 self.root = ResourcePath( 

267 replaceRoot(self.config["root"], butlerRoot), forceDirectory=True, forceAbsolute=True 

268 ) 

269 

270 self.locationFactory = LocationFactory(self.root) 

271 self.formatterFactory = FormatterFactory() 

272 

273 # Now associate formatters with storage classes 

274 self.formatterFactory.registerFormatters(self.config["formatters"], universe=bridgeManager.universe) 

275 

276 # Read the file naming templates 

277 self.templates = FileTemplates(self.config["templates"], universe=bridgeManager.universe) 

278 

279 # See if composites should be disassembled 

280 self.composites = CompositesMap(self.config["composites"], universe=bridgeManager.universe) 

281 

282 tableName = self.config["records", "table"] 

283 try: 

284 # Storage of paths and formatters, keyed by dataset_id 

285 self._table = bridgeManager.opaque.register( 

286 tableName, self.makeTableSpec(bridgeManager.datasetIdColumnType) 

287 ) 

288 # Interface to Registry. 

289 self._bridge = bridgeManager.register(self.name) 

290 except ReadOnlyDatabaseError: 

291 # If the database is read only and we just tried and failed to 

292 # create a table, it means someone is trying to create a read-only 

293 # butler client for an empty repo. That should be okay, as long 

294 # as they then try to get any datasets before some other client 

295 # creates the table. Chances are they'rejust validating 

296 # configuration. 

297 pass 

298 

299 # Determine whether checksums should be used - default to False 

300 self.useChecksum = self.config.get("checksum", False) 

301 

302 # Determine whether we can fall back to configuration if a 

303 # requested dataset is not known to registry 

304 self.trustGetRequest = self.config.get("trust_get_request", False) 

305 

306 # Create a cache manager 

307 self.cacheManager: AbstractDatastoreCacheManager 

308 if "cached" in self.config: 308 ↛ 311line 308 didn't jump to line 311, because the condition on line 308 was never false

309 self.cacheManager = DatastoreCacheManager(self.config["cached"], universe=bridgeManager.universe) 

310 else: 

311 self.cacheManager = DatastoreDisabledCacheManager("", universe=bridgeManager.universe) 

312 

313 # Check existence and create directory structure if necessary 

314 if not self.root.exists(): 

315 if "create" not in self.config or not self.config["create"]: 315 ↛ 316line 315 didn't jump to line 316, because the condition on line 315 was never true

316 raise ValueError(f"No valid root and not allowed to create one at: {self.root}") 

317 try: 

318 self.root.mkdir() 

319 except Exception as e: 

320 raise ValueError( 

321 f"Can not create datastore root '{self.root}', check permissions. Got error: {e}" 

322 ) from e 

323 

324 def __str__(self) -> str: 

325 return str(self.root) 

326 

327 @property 

328 def bridge(self) -> DatastoreRegistryBridge: 

329 return self._bridge 

330 

331 def _artifact_exists(self, location: Location) -> bool: 

332 """Check that an artifact exists in this datastore at the specified 

333 location. 

334 

335 Parameters 

336 ---------- 

337 location : `Location` 

338 Expected location of the artifact associated with this datastore. 

339 

340 Returns 

341 ------- 

342 exists : `bool` 

343 True if the location can be found, false otherwise. 

344 """ 

345 log.debug("Checking if resource exists: %s", location.uri) 

346 return location.uri.exists() 

347 

348 def _delete_artifact(self, location: Location) -> None: 

349 """Delete the artifact from the datastore. 

350 

351 Parameters 

352 ---------- 

353 location : `Location` 

354 Location of the artifact associated with this datastore. 

355 """ 

356 if location.pathInStore.isabs(): 356 ↛ 357line 356 didn't jump to line 357, because the condition on line 356 was never true

357 raise RuntimeError(f"Cannot delete artifact with absolute uri {location.uri}.") 

358 

359 try: 

360 location.uri.remove() 

361 except FileNotFoundError: 

362 log.debug("File %s did not exist and so could not be deleted.", location.uri) 

363 raise 

364 except Exception as e: 

365 log.critical("Failed to delete file: %s (%s)", location.uri, e) 

366 raise 

367 log.debug("Successfully deleted file: %s", location.uri) 

368 

369 def addStoredItemInfo(self, refs: Iterable[DatasetRef], infos: Iterable[StoredFileInfo]) -> None: 

370 # Docstring inherited from GenericBaseDatastore 

371 records = [info.rebase(ref).to_record() for ref, info in zip(refs, infos)] 

372 self._table.insert(*records) 

373 

374 def getStoredItemsInfo(self, ref: DatasetIdRef) -> List[StoredFileInfo]: 

375 # Docstring inherited from GenericBaseDatastore 

376 

377 # Look for the dataset_id -- there might be multiple matches 

378 # if we have disassembled the dataset. 

379 records = self._table.fetch(dataset_id=ref.id) 

380 return [StoredFileInfo.from_record(record) for record in records] 

381 

382 def _get_stored_records_associated_with_refs( 

383 self, refs: Iterable[DatasetIdRef] 

384 ) -> Dict[DatasetId, List[StoredFileInfo]]: 

385 """Retrieve all records associated with the provided refs. 

386 

387 Parameters 

388 ---------- 

389 refs : iterable of `DatasetIdRef` 

390 The refs for which records are to be retrieved. 

391 

392 Returns 

393 ------- 

394 records : `dict` of [`DatasetId`, `list` of `StoredFileInfo`] 

395 The matching records indexed by the ref ID. The number of entries 

396 in the dict can be smaller than the number of requested refs. 

397 """ 

398 records = self._table.fetch(dataset_id=[ref.id for ref in refs]) 

399 

400 # Uniqueness is dataset_id + component so can have multiple records 

401 # per ref. 

402 records_by_ref = defaultdict(list) 

403 for record in records: 

404 records_by_ref[record["dataset_id"]].append(StoredFileInfo.from_record(record)) 

405 return records_by_ref 

406 

407 def _refs_associated_with_artifacts( 

408 self, paths: List[Union[str, ResourcePath]] 

409 ) -> Dict[str, Set[DatasetId]]: 

410 """Return paths and associated dataset refs. 

411 

412 Parameters 

413 ---------- 

414 paths : `list` of `str` or `lsst.resources.ResourcePath` 

415 All the paths to include in search. 

416 

417 Returns 

418 ------- 

419 mapping : `dict` of [`str`, `set` [`DatasetId`]] 

420 Mapping of each path to a set of associated database IDs. 

421 """ 

422 records = self._table.fetch(path=[str(path) for path in paths]) 

423 result = defaultdict(set) 

424 for row in records: 

425 result[row["path"]].add(row["dataset_id"]) 

426 return result 

427 

428 def _registered_refs_per_artifact(self, pathInStore: ResourcePath) -> Set[DatasetId]: 

429 """Return all dataset refs associated with the supplied path. 

430 

431 Parameters 

432 ---------- 

433 pathInStore : `lsst.resources.ResourcePath` 

434 Path of interest in the data store. 

435 

436 Returns 

437 ------- 

438 ids : `set` of `int` 

439 All `DatasetRef` IDs associated with this path. 

440 """ 

441 records = list(self._table.fetch(path=str(pathInStore))) 

442 ids = {r["dataset_id"] for r in records} 

443 return ids 

444 

445 def removeStoredItemInfo(self, ref: DatasetIdRef) -> None: 

446 # Docstring inherited from GenericBaseDatastore 

447 self._table.delete(["dataset_id"], {"dataset_id": ref.id}) 

448 

449 def _get_dataset_locations_info(self, ref: DatasetIdRef) -> List[Tuple[Location, StoredFileInfo]]: 

450 r"""Find all the `Location`\ s of the requested dataset in the 

451 `Datastore` and the associated stored file information. 

452 

453 Parameters 

454 ---------- 

455 ref : `DatasetRef` 

456 Reference to the required `Dataset`. 

457 

458 Returns 

459 ------- 

460 results : `list` [`tuple` [`Location`, `StoredFileInfo` ]] 

461 Location of the dataset within the datastore and 

462 stored information about each file and its formatter. 

463 """ 

464 # Get the file information (this will fail if no file) 

465 records = self.getStoredItemsInfo(ref) 

466 

467 # Use the path to determine the location -- we need to take 

468 # into account absolute URIs in the datastore record 

469 return [(r.file_location(self.locationFactory), r) for r in records] 

470 

471 def _can_remove_dataset_artifact(self, ref: DatasetIdRef, location: Location) -> bool: 

472 """Check that there is only one dataset associated with the 

473 specified artifact. 

474 

475 Parameters 

476 ---------- 

477 ref : `DatasetRef` or `FakeDatasetRef` 

478 Dataset to be removed. 

479 location : `Location` 

480 The location of the artifact to be removed. 

481 

482 Returns 

483 ------- 

484 can_remove : `Bool` 

485 True if the artifact can be safely removed. 

486 """ 

487 # Can't ever delete absolute URIs. 

488 if location.pathInStore.isabs(): 

489 return False 

490 

491 # Get all entries associated with this path 

492 allRefs = self._registered_refs_per_artifact(location.pathInStore) 

493 if not allRefs: 

494 raise RuntimeError(f"Datastore inconsistency error. {location.pathInStore} not in registry") 

495 

496 # Remove these refs from all the refs and if there is nothing left 

497 # then we can delete 

498 remainingRefs = allRefs - {ref.id} 

499 

500 if remainingRefs: 

501 return False 

502 return True 

503 

504 def _get_expected_dataset_locations_info(self, ref: DatasetRef) -> List[Tuple[Location, StoredFileInfo]]: 

505 """Predict the location and related file information of the requested 

506 dataset in this datastore. 

507 

508 Parameters 

509 ---------- 

510 ref : `DatasetRef` 

511 Reference to the required `Dataset`. 

512 

513 Returns 

514 ------- 

515 results : `list` [`tuple` [`Location`, `StoredFileInfo` ]] 

516 Expected Location of the dataset within the datastore and 

517 placeholder information about each file and its formatter. 

518 

519 Notes 

520 ----- 

521 Uses the current configuration to determine how we would expect the 

522 datastore files to have been written if we couldn't ask registry. 

523 This is safe so long as there has been no change to datastore 

524 configuration between writing the dataset and wanting to read it. 

525 Will not work for files that have been ingested without using the 

526 standard file template or default formatter. 

527 """ 

528 

529 # If we have a component ref we always need to ask the questions 

530 # of the composite. If the composite is disassembled this routine 

531 # should return all components. If the composite was not 

532 # disassembled the composite is what is stored regardless of 

533 # component request. Note that if the caller has disassembled 

534 # a composite there is no way for this guess to know that 

535 # without trying both the composite and component ref and seeing 

536 # if there is something at the component Location even without 

537 # disassembly being enabled. 

538 if ref.datasetType.isComponent(): 

539 ref = ref.makeCompositeRef() 

540 

541 # See if the ref is a composite that should be disassembled 

542 doDisassembly = self.composites.shouldBeDisassembled(ref) 

543 

544 all_info: List[Tuple[Location, Formatter, StorageClass, Optional[str]]] = [] 

545 

546 if doDisassembly: 

547 for component, componentStorage in ref.datasetType.storageClass.components.items(): 

548 compRef = ref.makeComponentRef(component) 

549 location, formatter = self._determine_put_formatter_location(compRef) 

550 all_info.append((location, formatter, componentStorage, component)) 

551 

552 else: 

553 # Always use the composite ref if no disassembly 

554 location, formatter = self._determine_put_formatter_location(ref) 

555 all_info.append((location, formatter, ref.datasetType.storageClass, None)) 

556 

557 # Convert the list of tuples to have StoredFileInfo as second element 

558 return [ 

559 ( 

560 location, 

561 StoredFileInfo( 

562 formatter=formatter, 

563 path=location.pathInStore.path, 

564 storageClass=storageClass, 

565 component=component, 

566 checksum=None, 

567 file_size=-1, 

568 dataset_id=ref.getCheckedId(), 

569 ), 

570 ) 

571 for location, formatter, storageClass, component in all_info 

572 ] 

573 

574 def _prepare_for_get( 

575 self, ref: DatasetRef, parameters: Optional[Mapping[str, Any]] = None 

576 ) -> List[DatastoreFileGetInformation]: 

577 """Check parameters for ``get`` and obtain formatter and 

578 location. 

579 

580 Parameters 

581 ---------- 

582 ref : `DatasetRef` 

583 Reference to the required Dataset. 

584 parameters : `dict` 

585 `StorageClass`-specific parameters that specify, for example, 

586 a slice of the dataset to be loaded. 

587 

588 Returns 

589 ------- 

590 getInfo : `list` [`DatastoreFileGetInformation`] 

591 Parameters needed to retrieve each file. 

592 """ 

593 log.debug("Retrieve %s from %s with parameters %s", ref, self.name, parameters) 

594 

595 # Get file metadata and internal metadata 

596 fileLocations = self._get_dataset_locations_info(ref) 

597 if not fileLocations: 

598 if not self.trustGetRequest: 

599 raise FileNotFoundError(f"Could not retrieve dataset {ref}.") 

600 # Assume the dataset is where we think it should be 

601 fileLocations = self._get_expected_dataset_locations_info(ref) 

602 

603 # The storage class we want to use eventually 

604 refStorageClass = ref.datasetType.storageClass 

605 

606 if len(fileLocations) > 1: 

607 disassembled = True 

608 

609 # If trust is involved it is possible that there will be 

610 # components listed here that do not exist in the datastore. 

611 # Explicitly check for file artifact existence and filter out any 

612 # that are missing. 

613 if self.trustGetRequest: 

614 fileLocations = [loc for loc in fileLocations if loc[0].uri.exists()] 

615 

616 # For now complain only if we have no components at all. One 

617 # component is probably a problem but we can punt that to the 

618 # assembler. 

619 if not fileLocations: 619 ↛ 620line 619 didn't jump to line 620, because the condition on line 619 was never true

620 raise FileNotFoundError(f"None of the component files for dataset {ref} exist.") 

621 

622 else: 

623 disassembled = False 

624 

625 # Is this a component request? 

626 refComponent = ref.datasetType.component() 

627 

628 fileGetInfo = [] 

629 for location, storedFileInfo in fileLocations: 

630 # The storage class used to write the file 

631 writeStorageClass = storedFileInfo.storageClass 

632 

633 # If this has been disassembled we need read to match the write 

634 if disassembled: 

635 readStorageClass = writeStorageClass 

636 else: 

637 readStorageClass = refStorageClass 

638 

639 formatter = get_instance_of( 

640 storedFileInfo.formatter, 

641 FileDescriptor( 

642 location, 

643 readStorageClass=readStorageClass, 

644 storageClass=writeStorageClass, 

645 parameters=parameters, 

646 ), 

647 ref.dataId, 

648 ) 

649 

650 formatterParams, notFormatterParams = formatter.segregateParameters() 

651 

652 # Of the remaining parameters, extract the ones supported by 

653 # this StorageClass (for components not all will be handled) 

654 assemblerParams = readStorageClass.filterParameters(notFormatterParams) 

655 

656 # The ref itself could be a component if the dataset was 

657 # disassembled by butler, or we disassembled in datastore and 

658 # components came from the datastore records 

659 component = storedFileInfo.component if storedFileInfo.component else refComponent 

660 

661 fileGetInfo.append( 

662 DatastoreFileGetInformation( 

663 location, 

664 formatter, 

665 storedFileInfo, 

666 assemblerParams, 

667 formatterParams, 

668 component, 

669 readStorageClass, 

670 ) 

671 ) 

672 

673 return fileGetInfo 

674 

675 def _prepare_for_put(self, inMemoryDataset: Any, ref: DatasetRef) -> Tuple[Location, Formatter]: 

676 """Check the arguments for ``put`` and obtain formatter and 

677 location. 

678 

679 Parameters 

680 ---------- 

681 inMemoryDataset : `object` 

682 The dataset to store. 

683 ref : `DatasetRef` 

684 Reference to the associated Dataset. 

685 

686 Returns 

687 ------- 

688 location : `Location` 

689 The location to write the dataset. 

690 formatter : `Formatter` 

691 The `Formatter` to use to write the dataset. 

692 

693 Raises 

694 ------ 

695 TypeError 

696 Supplied object and storage class are inconsistent. 

697 DatasetTypeNotSupportedError 

698 The associated `DatasetType` is not handled by this datastore. 

699 """ 

700 self._validate_put_parameters(inMemoryDataset, ref) 

701 return self._determine_put_formatter_location(ref) 

702 

703 def _determine_put_formatter_location(self, ref: DatasetRef) -> Tuple[Location, Formatter]: 

704 """Calculate the formatter and output location to use for put. 

705 

706 Parameters 

707 ---------- 

708 ref : `DatasetRef` 

709 Reference to the associated Dataset. 

710 

711 Returns 

712 ------- 

713 location : `Location` 

714 The location to write the dataset. 

715 formatter : `Formatter` 

716 The `Formatter` to use to write the dataset. 

717 """ 

718 # Work out output file name 

719 try: 

720 template = self.templates.getTemplate(ref) 

721 except KeyError as e: 

722 raise DatasetTypeNotSupportedError(f"Unable to find template for {ref}") from e 

723 

724 # Validate the template to protect against filenames from different 

725 # dataIds returning the same and causing overwrite confusion. 

726 template.validateTemplate(ref) 

727 

728 location = self.locationFactory.fromPath(template.format(ref)) 

729 

730 # Get the formatter based on the storage class 

731 storageClass = ref.datasetType.storageClass 

732 try: 

733 formatter = self.formatterFactory.getFormatter( 

734 ref, FileDescriptor(location, storageClass=storageClass), ref.dataId 

735 ) 

736 except KeyError as e: 

737 raise DatasetTypeNotSupportedError( 

738 f"Unable to find formatter for {ref} in datastore {self.name}" 

739 ) from e 

740 

741 # Now that we know the formatter, update the location 

742 location = formatter.makeUpdatedLocation(location) 

743 

744 return location, formatter 

745 

746 def _overrideTransferMode(self, *datasets: FileDataset, transfer: Optional[str] = None) -> Optional[str]: 

747 # Docstring inherited from base class 

748 if transfer != "auto": 

749 return transfer 

750 

751 # See if the paths are within the datastore or not 

752 inside = [self._pathInStore(d.path) is not None for d in datasets] 

753 

754 if all(inside): 

755 transfer = None 

756 elif not any(inside): 756 ↛ 765line 756 didn't jump to line 765, because the condition on line 756 was never false

757 # Allow ResourcePath to use its own knowledge 

758 transfer = "auto" 

759 else: 

760 # This can happen when importing from a datastore that 

761 # has had some datasets ingested using "direct" mode. 

762 # Also allow ResourcePath to sort it out but warn about it. 

763 # This can happen if you are importing from a datastore 

764 # that had some direct transfer datasets. 

765 log.warning( 

766 "Some datasets are inside the datastore and some are outside. Using 'split' " 

767 "transfer mode. This assumes that the files outside the datastore are " 

768 "still accessible to the new butler since they will not be copied into " 

769 "the target datastore." 

770 ) 

771 transfer = "split" 

772 

773 return transfer 

774 

775 def _pathInStore(self, path: ResourcePathExpression) -> Optional[str]: 

776 """Return path relative to datastore root 

777 

778 Parameters 

779 ---------- 

780 path : `lsst.resources.ResourcePathExpression` 

781 Path to dataset. Can be absolute URI. If relative assumed to 

782 be relative to the datastore. Returns path in datastore 

783 or raises an exception if the path it outside. 

784 

785 Returns 

786 ------- 

787 inStore : `str` 

788 Path relative to datastore root. Returns `None` if the file is 

789 outside the root. 

790 """ 

791 # Relative path will always be relative to datastore 

792 pathUri = ResourcePath(path, forceAbsolute=False) 

793 return pathUri.relative_to(self.root) 

794 

795 def _standardizeIngestPath( 

796 self, path: Union[str, ResourcePath], *, transfer: Optional[str] = None 

797 ) -> Union[str, ResourcePath]: 

798 """Standardize the path of a to-be-ingested file. 

799 

800 Parameters 

801 ---------- 

802 path : `str` or `lsst.resources.ResourcePath` 

803 Path of a file to be ingested. This parameter is not expected 

804 to be all the types that can be used to construct a 

805 `~lsst.resources.ResourcePath`. 

806 transfer : `str`, optional 

807 How (and whether) the dataset should be added to the datastore. 

808 See `ingest` for details of transfer modes. 

809 This implementation is provided only so 

810 `NotImplementedError` can be raised if the mode is not supported; 

811 actual transfers are deferred to `_extractIngestInfo`. 

812 

813 Returns 

814 ------- 

815 path : `str` or `lsst.resources.ResourcePath` 

816 New path in what the datastore considers standard form. If an 

817 absolute URI was given that will be returned unchanged. 

818 

819 Notes 

820 ----- 

821 Subclasses of `FileDatastore` can implement this method instead 

822 of `_prepIngest`. It should not modify the data repository or given 

823 file in any way. 

824 

825 Raises 

826 ------ 

827 NotImplementedError 

828 Raised if the datastore does not support the given transfer mode 

829 (including the case where ingest is not supported at all). 

830 FileNotFoundError 

831 Raised if one of the given files does not exist. 

832 """ 

833 if transfer not in (None, "direct", "split") + self.root.transferModes: 833 ↛ 834line 833 didn't jump to line 834, because the condition on line 833 was never true

834 raise NotImplementedError(f"Transfer mode {transfer} not supported.") 

835 

836 # A relative URI indicates relative to datastore root 

837 srcUri = ResourcePath(path, forceAbsolute=False) 

838 if not srcUri.isabs(): 

839 srcUri = self.root.join(path) 

840 

841 if not srcUri.exists(): 

842 raise FileNotFoundError( 

843 f"Resource at {srcUri} does not exist; note that paths to ingest " 

844 f"are assumed to be relative to {self.root} unless they are absolute." 

845 ) 

846 

847 if transfer is None: 

848 relpath = srcUri.relative_to(self.root) 

849 if not relpath: 

850 raise RuntimeError( 

851 f"Transfer is none but source file ({srcUri}) is not within datastore ({self.root})" 

852 ) 

853 

854 # Return the relative path within the datastore for internal 

855 # transfer 

856 path = relpath 

857 

858 return path 

859 

860 def _extractIngestInfo( 

861 self, 

862 path: ResourcePathExpression, 

863 ref: DatasetRef, 

864 *, 

865 formatter: Union[Formatter, Type[Formatter]], 

866 transfer: Optional[str] = None, 

867 record_validation_info: bool = True, 

868 ) -> StoredFileInfo: 

869 """Relocate (if necessary) and extract `StoredFileInfo` from a 

870 to-be-ingested file. 

871 

872 Parameters 

873 ---------- 

874 path : `lsst.resources.ResourcePathExpression` 

875 URI or path of a file to be ingested. 

876 ref : `DatasetRef` 

877 Reference for the dataset being ingested. Guaranteed to have 

878 ``dataset_id not None`. 

879 formatter : `type` or `Formatter` 

880 `Formatter` subclass to use for this dataset or an instance. 

881 transfer : `str`, optional 

882 How (and whether) the dataset should be added to the datastore. 

883 See `ingest` for details of transfer modes. 

884 record_validation_info : `bool`, optional 

885 If `True`, the default, the datastore can record validation 

886 information associated with the file. If `False` the datastore 

887 will not attempt to track any information such as checksums 

888 or file sizes. This can be useful if such information is tracked 

889 in an external system or if the file is to be compressed in place. 

890 It is up to the datastore whether this parameter is relevant. 

891 

892 Returns 

893 ------- 

894 info : `StoredFileInfo` 

895 Internal datastore record for this file. This will be inserted by 

896 the caller; the `_extractIngestInfo` is only responsible for 

897 creating and populating the struct. 

898 

899 Raises 

900 ------ 

901 FileNotFoundError 

902 Raised if one of the given files does not exist. 

903 FileExistsError 

904 Raised if transfer is not `None` but the (internal) location the 

905 file would be moved to is already occupied. 

906 """ 

907 if self._transaction is None: 907 ↛ 908line 907 didn't jump to line 908, because the condition on line 907 was never true

908 raise RuntimeError("Ingest called without transaction enabled") 

909 

910 # Create URI of the source path, do not need to force a relative 

911 # path to absolute. 

912 srcUri = ResourcePath(path, forceAbsolute=False) 

913 

914 # Track whether we have read the size of the source yet 

915 have_sized = False 

916 

917 tgtLocation: Optional[Location] 

918 if transfer is None or transfer == "split": 

919 # A relative path is assumed to be relative to the datastore 

920 # in this context 

921 if not srcUri.isabs(): 

922 tgtLocation = self.locationFactory.fromPath(srcUri.ospath) 

923 else: 

924 # Work out the path in the datastore from an absolute URI 

925 # This is required to be within the datastore. 

926 pathInStore = srcUri.relative_to(self.root) 

927 if pathInStore is None and transfer is None: 927 ↛ 928line 927 didn't jump to line 928, because the condition on line 927 was never true

928 raise RuntimeError( 

929 f"Unexpectedly learned that {srcUri} is not within datastore {self.root}" 

930 ) 

931 if pathInStore: 931 ↛ 933line 931 didn't jump to line 933, because the condition on line 931 was never false

932 tgtLocation = self.locationFactory.fromPath(pathInStore) 

933 elif transfer == "split": 

934 # Outside the datastore but treat that as a direct ingest 

935 # instead. 

936 tgtLocation = None 

937 else: 

938 raise RuntimeError(f"Unexpected transfer mode encountered: {transfer} for URI {srcUri}") 

939 elif transfer == "direct": 939 ↛ 944line 939 didn't jump to line 944, because the condition on line 939 was never true

940 # Want to store the full URI to the resource directly in 

941 # datastore. This is useful for referring to permanent archive 

942 # storage for raw data. 

943 # Trust that people know what they are doing. 

944 tgtLocation = None 

945 else: 

946 # Work out the name we want this ingested file to have 

947 # inside the datastore 

948 tgtLocation = self._calculate_ingested_datastore_name(srcUri, ref, formatter) 

949 if not tgtLocation.uri.dirname().exists(): 

950 log.debug("Folder %s does not exist yet.", tgtLocation.uri.dirname()) 

951 tgtLocation.uri.dirname().mkdir() 

952 

953 # if we are transferring from a local file to a remote location 

954 # it may be more efficient to get the size and checksum of the 

955 # local file rather than the transferred one 

956 if record_validation_info and srcUri.isLocal: 

957 size = srcUri.size() 

958 checksum = self.computeChecksum(srcUri) if self.useChecksum else None 

959 have_sized = True 

960 

961 # Transfer the resource to the destination. 

962 # Allow overwrite of an existing file. This matches the behavior 

963 # of datastore.put() in that it trusts that registry would not 

964 # be asking to overwrite unless registry thought that the 

965 # overwrite was allowed. 

966 tgtLocation.uri.transfer_from( 

967 srcUri, transfer=transfer, transaction=self._transaction, overwrite=True 

968 ) 

969 

970 if tgtLocation is None: 970 ↛ 972line 970 didn't jump to line 972, because the condition on line 970 was never true

971 # This means we are using direct mode 

972 targetUri = srcUri 

973 targetPath = str(srcUri) 

974 else: 

975 targetUri = tgtLocation.uri 

976 targetPath = tgtLocation.pathInStore.path 

977 

978 # the file should exist in the datastore now 

979 if record_validation_info: 

980 if not have_sized: 

981 size = targetUri.size() 

982 checksum = self.computeChecksum(targetUri) if self.useChecksum else None 

983 else: 

984 # Not recording any file information. 

985 size = -1 

986 checksum = None 

987 

988 return StoredFileInfo( 

989 formatter=formatter, 

990 path=targetPath, 

991 storageClass=ref.datasetType.storageClass, 

992 component=ref.datasetType.component(), 

993 file_size=size, 

994 checksum=checksum, 

995 dataset_id=ref.getCheckedId(), 

996 ) 

997 

998 def _prepIngest(self, *datasets: FileDataset, transfer: Optional[str] = None) -> _IngestPrepData: 

999 # Docstring inherited from Datastore._prepIngest. 

1000 filtered = [] 

1001 for dataset in datasets: 

1002 acceptable = [ref for ref in dataset.refs if self.constraints.isAcceptable(ref)] 

1003 if not acceptable: 

1004 continue 

1005 else: 

1006 dataset.refs = acceptable 

1007 if dataset.formatter is None: 

1008 dataset.formatter = self.formatterFactory.getFormatterClass(dataset.refs[0]) 

1009 else: 

1010 assert isinstance(dataset.formatter, (type, str)) 

1011 formatter_class = get_class_of(dataset.formatter) 

1012 if not issubclass(formatter_class, Formatter): 1012 ↛ 1013line 1012 didn't jump to line 1013, because the condition on line 1012 was never true

1013 raise TypeError(f"Requested formatter {dataset.formatter} is not a Formatter class.") 

1014 dataset.formatter = formatter_class 

1015 dataset.path = self._standardizeIngestPath(dataset.path, transfer=transfer) 

1016 filtered.append(dataset) 

1017 return _IngestPrepData(filtered) 

1018 

1019 @transactional 

1020 def _finishIngest( 

1021 self, 

1022 prepData: Datastore.IngestPrepData, 

1023 *, 

1024 transfer: Optional[str] = None, 

1025 record_validation_info: bool = True, 

1026 ) -> None: 

1027 # Docstring inherited from Datastore._finishIngest. 

1028 refsAndInfos = [] 

1029 progress = Progress("lsst.daf.butler.datastores.FileDatastore.ingest", level=logging.DEBUG) 

1030 for dataset in progress.wrap(prepData.datasets, desc="Ingesting dataset files"): 

1031 # Do ingest as if the first dataset ref is associated with the file 

1032 info = self._extractIngestInfo( 

1033 dataset.path, 

1034 dataset.refs[0], 

1035 formatter=dataset.formatter, 

1036 transfer=transfer, 

1037 record_validation_info=record_validation_info, 

1038 ) 

1039 refsAndInfos.extend([(ref, info) for ref in dataset.refs]) 

1040 self._register_datasets(refsAndInfos) 

1041 

1042 def _calculate_ingested_datastore_name( 

1043 self, srcUri: ResourcePath, ref: DatasetRef, formatter: Union[Formatter, Type[Formatter]] 

1044 ) -> Location: 

1045 """Given a source URI and a DatasetRef, determine the name the 

1046 dataset will have inside datastore. 

1047 

1048 Parameters 

1049 ---------- 

1050 srcUri : `lsst.resources.ResourcePath` 

1051 URI to the source dataset file. 

1052 ref : `DatasetRef` 

1053 Ref associated with the newly-ingested dataset artifact. This 

1054 is used to determine the name within the datastore. 

1055 formatter : `Formatter` or Formatter class. 

1056 Formatter to use for validation. Can be a class or an instance. 

1057 

1058 Returns 

1059 ------- 

1060 location : `Location` 

1061 Target location for the newly-ingested dataset. 

1062 """ 

1063 # Ingesting a file from outside the datastore. 

1064 # This involves a new name. 

1065 template = self.templates.getTemplate(ref) 

1066 location = self.locationFactory.fromPath(template.format(ref)) 

1067 

1068 # Get the extension 

1069 ext = srcUri.getExtension() 

1070 

1071 # Update the destination to include that extension 

1072 location.updateExtension(ext) 

1073 

1074 # Ask the formatter to validate this extension 

1075 formatter.validateExtension(location) 

1076 

1077 return location 

1078 

1079 def _write_in_memory_to_artifact(self, inMemoryDataset: Any, ref: DatasetRef) -> StoredFileInfo: 

1080 """Write out in memory dataset to datastore. 

1081 

1082 Parameters 

1083 ---------- 

1084 inMemoryDataset : `object` 

1085 Dataset to write to datastore. 

1086 ref : `DatasetRef` 

1087 Registry information associated with this dataset. 

1088 

1089 Returns 

1090 ------- 

1091 info : `StoredFileInfo` 

1092 Information describing the artifact written to the datastore. 

1093 """ 

1094 # May need to coerce the in memory dataset to the correct 

1095 # python type. 

1096 inMemoryDataset = ref.datasetType.storageClass.coerce_type(inMemoryDataset) 

1097 

1098 location, formatter = self._prepare_for_put(inMemoryDataset, ref) 

1099 uri = location.uri 

1100 

1101 if not uri.dirname().exists(): 

1102 log.debug("Folder %s does not exist yet so creating it.", uri.dirname()) 

1103 uri.dirname().mkdir() 

1104 

1105 if self._transaction is None: 1105 ↛ 1106line 1105 didn't jump to line 1106, because the condition on line 1105 was never true

1106 raise RuntimeError("Attempting to write artifact without transaction enabled") 

1107 

1108 def _removeFileExists(uri: ResourcePath) -> None: 

1109 """Remove a file and do not complain if it is not there. 

1110 

1111 This is important since a formatter might fail before the file 

1112 is written and we should not confuse people by writing spurious 

1113 error messages to the log. 

1114 """ 

1115 try: 

1116 uri.remove() 

1117 except FileNotFoundError: 

1118 pass 

1119 

1120 # Register a callback to try to delete the uploaded data if 

1121 # something fails below 

1122 self._transaction.registerUndo("artifactWrite", _removeFileExists, uri) 

1123 

1124 data_written = False 

1125 if not uri.isLocal: 

1126 # This is a remote URI. Some datasets can be serialized directly 

1127 # to bytes and sent to the remote datastore without writing a 

1128 # file. If the dataset is intended to be saved to the cache 

1129 # a file is always written and direct write to the remote 

1130 # datastore is bypassed. 

1131 if not self.cacheManager.should_be_cached(ref): 

1132 try: 

1133 serializedDataset = formatter.toBytes(inMemoryDataset) 

1134 except NotImplementedError: 

1135 # Fallback to the file writing option. 

1136 pass 

1137 except Exception as e: 

1138 raise RuntimeError( 

1139 f"Failed to serialize dataset {ref} of type {type(inMemoryDataset)} to bytes." 

1140 ) from e 

1141 else: 

1142 log.debug("Writing bytes directly to %s", uri) 

1143 uri.write(serializedDataset, overwrite=True) 

1144 log.debug("Successfully wrote bytes directly to %s", uri) 

1145 data_written = True 

1146 

1147 if not data_written: 

1148 # Did not write the bytes directly to object store so instead 

1149 # write to temporary file. Always write to a temporary even if 

1150 # using a local file system -- that gives us atomic writes. 

1151 # If a process is killed as the file is being written we do not 

1152 # want it to remain in the correct place but in corrupt state. 

1153 # For local files write to the output directory not temporary dir. 

1154 prefix = uri.dirname() if uri.isLocal else None 

1155 with ResourcePath.temporary_uri(suffix=uri.getExtension(), prefix=prefix) as temporary_uri: 

1156 # Need to configure the formatter to write to a different 

1157 # location and that needs us to overwrite internals 

1158 log.debug("Writing dataset to temporary location at %s", temporary_uri) 

1159 with formatter._updateLocation(Location(None, temporary_uri)): 

1160 try: 

1161 formatter.write(inMemoryDataset) 

1162 except Exception as e: 

1163 raise RuntimeError( 

1164 f"Failed to serialize dataset {ref} of type" 

1165 f" {type(inMemoryDataset)} to " 

1166 f"temporary location {temporary_uri}" 

1167 ) from e 

1168 

1169 # Use move for a local file since that becomes an efficient 

1170 # os.rename. For remote resources we use copy to allow the 

1171 # file to be cached afterwards. 

1172 transfer = "move" if uri.isLocal else "copy" 

1173 

1174 uri.transfer_from(temporary_uri, transfer=transfer, overwrite=True) 

1175 

1176 if transfer == "copy": 

1177 # Cache if required 

1178 self.cacheManager.move_to_cache(temporary_uri, ref) 

1179 

1180 log.debug("Successfully wrote dataset to %s via a temporary file.", uri) 

1181 

1182 # URI is needed to resolve what ingest case are we dealing with 

1183 return self._extractIngestInfo(uri, ref, formatter=formatter) 

1184 

1185 def _read_artifact_into_memory( 

1186 self, 

1187 getInfo: DatastoreFileGetInformation, 

1188 ref: DatasetRef, 

1189 isComponent: bool = False, 

1190 cache_ref: Optional[DatasetRef] = None, 

1191 ) -> Any: 

1192 """Read the artifact from datastore into in memory object. 

1193 

1194 Parameters 

1195 ---------- 

1196 getInfo : `DatastoreFileGetInformation` 

1197 Information about the artifact within the datastore. 

1198 ref : `DatasetRef` 

1199 The registry information associated with this artifact. 

1200 isComponent : `bool` 

1201 Flag to indicate if a component is being read from this artifact. 

1202 cache_ref : `DatasetRef`, optional 

1203 The DatasetRef to use when looking up the file in the cache. 

1204 This ref must have the same ID as the supplied ref but can 

1205 be a parent ref or component ref to indicate to the cache whether 

1206 a composite file is being requested from the cache or a component 

1207 file. Without this the cache will default to the supplied ref but 

1208 it can get confused with read-only derived components for 

1209 disassembled composites. 

1210 

1211 Returns 

1212 ------- 

1213 inMemoryDataset : `object` 

1214 The artifact as a python object. 

1215 """ 

1216 location = getInfo.location 

1217 uri = location.uri 

1218 log.debug("Accessing data from %s", uri) 

1219 

1220 if cache_ref is None: 

1221 cache_ref = ref 

1222 if cache_ref.id != ref.id: 1222 ↛ 1223line 1222 didn't jump to line 1223, because the condition on line 1222 was never true

1223 raise ValueError( 

1224 "The supplied cache dataset ref refers to a different dataset than expected:" 

1225 f" {ref.id} != {cache_ref.id}" 

1226 ) 

1227 

1228 # Cannot recalculate checksum but can compare size as a quick check 

1229 # Do not do this if the size is negative since that indicates 

1230 # we do not know. 

1231 recorded_size = getInfo.info.file_size 

1232 resource_size = uri.size() 

1233 if recorded_size >= 0 and resource_size != recorded_size: 1233 ↛ 1234line 1233 didn't jump to line 1234, because the condition on line 1233 was never true

1234 raise RuntimeError( 

1235 "Integrity failure in Datastore. " 

1236 f"Size of file {uri} ({resource_size}) " 

1237 f"does not match size recorded in registry of {recorded_size}" 

1238 ) 

1239 

1240 # For the general case we have choices for how to proceed. 

1241 # 1. Always use a local file (downloading the remote resource to a 

1242 # temporary file if needed). 

1243 # 2. Use a threshold size and read into memory and use bytes. 

1244 # Use both for now with an arbitrary hand off size. 

1245 # This allows small datasets to be downloaded from remote object 

1246 # stores without requiring a temporary file. 

1247 

1248 formatter = getInfo.formatter 

1249 nbytes_max = 10_000_000 # Arbitrary number that we can tune 

1250 if resource_size <= nbytes_max and formatter.can_read_bytes(): 

1251 with self.cacheManager.find_in_cache(cache_ref, uri.getExtension()) as cached_file: 

1252 if cached_file is not None: 

1253 desired_uri = cached_file 

1254 msg = f" (cached version of {uri})" 

1255 else: 

1256 desired_uri = uri 

1257 msg = "" 

1258 with time_this(log, msg="Reading bytes from %s%s", args=(desired_uri, msg)): 

1259 serializedDataset = desired_uri.read() 

1260 log.debug( 

1261 "Deserializing %s from %d bytes from location %s with formatter %s", 

1262 f"component {getInfo.component}" if isComponent else "", 

1263 len(serializedDataset), 

1264 uri, 

1265 formatter.name(), 

1266 ) 

1267 try: 

1268 result = formatter.fromBytes( 

1269 serializedDataset, component=getInfo.component if isComponent else None 

1270 ) 

1271 except Exception as e: 

1272 raise ValueError( 

1273 f"Failure from formatter '{formatter.name()}' for dataset {ref.id}" 

1274 f" ({ref.datasetType.name} from {uri}): {e}" 

1275 ) from e 

1276 else: 

1277 # Read from file. 

1278 

1279 # Have to update the Location associated with the formatter 

1280 # because formatter.read does not allow an override. 

1281 # This could be improved. 

1282 location_updated = False 

1283 msg = "" 

1284 

1285 # First check in cache for local version. 

1286 # The cache will only be relevant for remote resources but 

1287 # no harm in always asking. Context manager ensures that cache 

1288 # file is not deleted during cache expiration. 

1289 with self.cacheManager.find_in_cache(cache_ref, uri.getExtension()) as cached_file: 

1290 if cached_file is not None: 

1291 msg = f"(via cache read of remote file {uri})" 

1292 uri = cached_file 

1293 location_updated = True 

1294 

1295 with uri.as_local() as local_uri: 

1296 can_be_cached = False 

1297 if uri != local_uri: 1297 ↛ 1299line 1297 didn't jump to line 1299, because the condition on line 1297 was never true

1298 # URI was remote and file was downloaded 

1299 cache_msg = "" 

1300 location_updated = True 

1301 

1302 if self.cacheManager.should_be_cached(cache_ref): 

1303 # In this scenario we want to ask if the downloaded 

1304 # file should be cached but we should not cache 

1305 # it until after we've used it (to ensure it can't 

1306 # be expired whilst we are using it). 

1307 can_be_cached = True 

1308 

1309 # Say that it is "likely" to be cached because 

1310 # if the formatter read fails we will not be 

1311 # caching this file. 

1312 cache_msg = " and likely cached" 

1313 

1314 msg = f"(via download to local file{cache_msg})" 

1315 

1316 # Calculate the (possibly) new location for the formatter 

1317 # to use. 

1318 newLocation = Location(*local_uri.split()) if location_updated else None 

1319 

1320 log.debug( 

1321 "Reading%s from location %s %s with formatter %s", 

1322 f" component {getInfo.component}" if isComponent else "", 

1323 uri, 

1324 msg, 

1325 formatter.name(), 

1326 ) 

1327 try: 

1328 with formatter._updateLocation(newLocation): 

1329 with time_this( 

1330 log, 

1331 msg="Reading%s from location %s %s with formatter %s", 

1332 args=( 

1333 f" component {getInfo.component}" if isComponent else "", 

1334 uri, 

1335 msg, 

1336 formatter.name(), 

1337 ), 

1338 ): 

1339 result = formatter.read(component=getInfo.component if isComponent else None) 

1340 except Exception as e: 

1341 raise ValueError( 

1342 f"Failure from formatter '{formatter.name()}' for dataset {ref.id}" 

1343 f" ({ref.datasetType.name} from {uri}): {e}" 

1344 ) from e 

1345 

1346 # File was read successfully so can move to cache 

1347 if can_be_cached: 1347 ↛ 1348line 1347 didn't jump to line 1348, because the condition on line 1347 was never true

1348 self.cacheManager.move_to_cache(local_uri, cache_ref) 

1349 

1350 return self._post_process_get( 

1351 result, getInfo.readStorageClass, getInfo.assemblerParams, isComponent=isComponent 

1352 ) 

1353 

1354 def knows(self, ref: DatasetRef) -> bool: 

1355 """Check if the dataset is known to the datastore. 

1356 

1357 Does not check for existence of any artifact. 

1358 

1359 Parameters 

1360 ---------- 

1361 ref : `DatasetRef` 

1362 Reference to the required dataset. 

1363 

1364 Returns 

1365 ------- 

1366 exists : `bool` 

1367 `True` if the dataset is known to the datastore. 

1368 """ 

1369 fileLocations = self._get_dataset_locations_info(ref) 

1370 if fileLocations: 

1371 return True 

1372 return False 

1373 

1374 def _process_mexists_records( 

1375 self, 

1376 id_to_ref: Dict[DatasetId, DatasetRef], 

1377 records: Dict[DatasetId, List[StoredFileInfo]], 

1378 all_required: bool, 

1379 artifact_existence: Optional[Dict[ResourcePath, bool]] = None, 

1380 ) -> Dict[DatasetRef, bool]: 

1381 """Helper function for mexists that checks the given records. 

1382 

1383 Parameters 

1384 ---------- 

1385 id_to_ref : `dict` of [`DatasetId`, `DatasetRef`] 

1386 Mapping of the dataset ID to the dataset ref itself. 

1387 records : `dict` of [`DatasetId`, `list` of `StoredFileInfo`] 

1388 Records as generally returned by 

1389 ``_get_stored_records_associated_with_refs``. 

1390 all_required : `bool` 

1391 Flag to indicate whether existence requires all artifacts 

1392 associated with a dataset ID to exist or not for existence. 

1393 artifact_existence : `dict` [`lsst.resources.ResourcePath`, `bool`] 

1394 Optional mapping of datastore artifact to existence. Updated by 

1395 this method with details of all artifacts tested. Can be `None` 

1396 if the caller is not interested. 

1397 

1398 Returns 

1399 ------- 

1400 existence : `dict` of [`DatasetRef`, `bool`] 

1401 Mapping from dataset to boolean indicating existence. 

1402 """ 

1403 # The URIs to be checked and a mapping of those URIs to 

1404 # the dataset ID. 

1405 uris_to_check: List[ResourcePath] = [] 

1406 location_map: Dict[ResourcePath, DatasetId] = {} 

1407 

1408 location_factory = self.locationFactory 

1409 

1410 uri_existence: Dict[ResourcePath, bool] = {} 

1411 for ref_id, infos in records.items(): 

1412 # Key is the dataset Id, value is list of StoredItemInfo 

1413 uris = [info.file_location(location_factory).uri for info in infos] 

1414 location_map.update({uri: ref_id for uri in uris}) 

1415 

1416 # Check the local cache directly for a dataset corresponding 

1417 # to the remote URI. 

1418 if self.cacheManager.file_count > 0: 

1419 ref = id_to_ref[ref_id] 

1420 for uri, storedFileInfo in zip(uris, infos): 

1421 check_ref = ref 

1422 if not ref.datasetType.isComponent() and (component := storedFileInfo.component): 1422 ↛ 1423line 1422 didn't jump to line 1423, because the condition on line 1422 was never true

1423 check_ref = ref.makeComponentRef(component) 

1424 if self.cacheManager.known_to_cache(check_ref, uri.getExtension()): 

1425 # Proxy for URI existence. 

1426 uri_existence[uri] = True 

1427 else: 

1428 uris_to_check.append(uri) 

1429 else: 

1430 # Check all of them. 

1431 uris_to_check.extend(uris) 

1432 

1433 if artifact_existence is not None: 

1434 # If a URI has already been checked remove it from the list 

1435 # and immediately add the status to the output dict. 

1436 filtered_uris_to_check = [] 

1437 for uri in uris_to_check: 

1438 if uri in artifact_existence: 

1439 uri_existence[uri] = artifact_existence[uri] 

1440 else: 

1441 filtered_uris_to_check.append(uri) 

1442 uris_to_check = filtered_uris_to_check 

1443 

1444 # Results. 

1445 dataset_existence: Dict[DatasetRef, bool] = {} 

1446 

1447 uri_existence.update(ResourcePath.mexists(uris_to_check)) 

1448 for uri, exists in uri_existence.items(): 

1449 dataset_id = location_map[uri] 

1450 ref = id_to_ref[dataset_id] 

1451 

1452 # Disassembled composite needs to check all locations. 

1453 # all_required indicates whether all need to exist or not. 

1454 if ref in dataset_existence: 

1455 if all_required: 

1456 exists = dataset_existence[ref] and exists 

1457 else: 

1458 exists = dataset_existence[ref] or exists 

1459 dataset_existence[ref] = exists 

1460 

1461 if artifact_existence is not None: 

1462 artifact_existence.update(uri_existence) 

1463 

1464 return dataset_existence 

1465 

1466 def mexists( 

1467 self, refs: Iterable[DatasetRef], artifact_existence: Optional[Dict[ResourcePath, bool]] = None 

1468 ) -> Dict[DatasetRef, bool]: 

1469 """Check the existence of multiple datasets at once. 

1470 

1471 Parameters 

1472 ---------- 

1473 refs : iterable of `DatasetRef` 

1474 The datasets to be checked. 

1475 artifact_existence : `dict` [`lsst.resources.ResourcePath`, `bool`] 

1476 Optional mapping of datastore artifact to existence. Updated by 

1477 this method with details of all artifacts tested. Can be `None` 

1478 if the caller is not interested. 

1479 

1480 Returns 

1481 ------- 

1482 existence : `dict` of [`DatasetRef`, `bool`] 

1483 Mapping from dataset to boolean indicating existence. 

1484 

1485 Notes 

1486 ----- 

1487 To minimize potentially costly remote existence checks, the local 

1488 cache is checked as a proxy for existence. If a file for this 

1489 `DatasetRef` does exist no check is done for the actual URI. This 

1490 could result in possibly unexpected behavior if the dataset itself 

1491 has been removed from the datastore by another process whilst it is 

1492 still in the cache. 

1493 """ 

1494 chunk_size = 10_000 

1495 dataset_existence: Dict[DatasetRef, bool] = {} 

1496 log.debug("Checking for the existence of multiple artifacts in datastore in chunks of %d", chunk_size) 

1497 n_found_total = 0 

1498 n_checked = 0 

1499 n_chunks = 0 

1500 for chunk in chunk_iterable(refs, chunk_size=chunk_size): 

1501 chunk_result = self._mexists(chunk, artifact_existence) 

1502 if log.isEnabledFor(VERBOSE): 

1503 n_results = len(chunk_result) 

1504 n_checked += n_results 

1505 # Can treat the booleans as 0, 1 integers and sum them. 

1506 n_found = sum(chunk_result.values()) 

1507 n_found_total += n_found 

1508 log.verbose( 

1509 "Number of datasets found in datastore for chunk %d = %d/%d (running total: %d/%d)", 

1510 n_chunks, 

1511 n_found, 

1512 n_results, 

1513 n_found_total, 

1514 n_checked, 

1515 ) 

1516 dataset_existence.update(chunk_result) 

1517 n_chunks += 1 

1518 

1519 return dataset_existence 

1520 

1521 def _mexists( 

1522 self, refs: Iterable[DatasetRef], artifact_existence: Optional[Dict[ResourcePath, bool]] = None 

1523 ) -> Dict[DatasetRef, bool]: 

1524 """Check the existence of multiple datasets at once. 

1525 

1526 Parameters 

1527 ---------- 

1528 refs : iterable of `DatasetRef` 

1529 The datasets to be checked. 

1530 

1531 Returns 

1532 ------- 

1533 existence : `dict` of [`DatasetRef`, `bool`] 

1534 Mapping from dataset to boolean indicating existence. 

1535 """ 

1536 # Need a mapping of dataset_id to dataset ref since the API 

1537 # works with dataset_id 

1538 id_to_ref = {ref.getCheckedId(): ref for ref in refs} 

1539 

1540 # Set of all IDs we are checking for. 

1541 requested_ids = set(id_to_ref.keys()) 

1542 

1543 # The records themselves. Could be missing some entries. 

1544 records = self._get_stored_records_associated_with_refs(refs) 

1545 

1546 dataset_existence = self._process_mexists_records( 

1547 id_to_ref, records, True, artifact_existence=artifact_existence 

1548 ) 

1549 

1550 # Set of IDs that have been handled. 

1551 handled_ids = {ref.id for ref in dataset_existence.keys()} 

1552 

1553 missing_ids = requested_ids - handled_ids 

1554 if missing_ids: 

1555 if not self.trustGetRequest: 

1556 # Must assume these do not exist 

1557 for missing in missing_ids: 

1558 dataset_existence[id_to_ref[missing]] = False 

1559 else: 

1560 log.debug( 

1561 "%d out of %d datasets were not known to datastore during initial existence check.", 

1562 len(missing_ids), 

1563 len(requested_ids), 

1564 ) 

1565 

1566 # Construct data structure identical to that returned 

1567 # by _get_stored_records_associated_with_refs() but using 

1568 # guessed names. 

1569 records = {} 

1570 for missing in missing_ids: 

1571 expected = self._get_expected_dataset_locations_info(id_to_ref[missing]) 

1572 records[missing] = [info for _, info in expected] 

1573 

1574 dataset_existence.update( 

1575 self._process_mexists_records( 

1576 id_to_ref, records, False, artifact_existence=artifact_existence 

1577 ) 

1578 ) 

1579 

1580 return dataset_existence 

1581 

1582 def exists(self, ref: DatasetRef) -> bool: 

1583 """Check if the dataset exists in the datastore. 

1584 

1585 Parameters 

1586 ---------- 

1587 ref : `DatasetRef` 

1588 Reference to the required dataset. 

1589 

1590 Returns 

1591 ------- 

1592 exists : `bool` 

1593 `True` if the entity exists in the `Datastore`. 

1594 

1595 Notes 

1596 ----- 

1597 The local cache is checked as a proxy for existence in the remote 

1598 object store. It is possible that another process on a different 

1599 compute node could remove the file from the object store even 

1600 though it is present in the local cache. 

1601 """ 

1602 fileLocations = self._get_dataset_locations_info(ref) 

1603 

1604 # if we are being asked to trust that registry might not be correct 

1605 # we ask for the expected locations and check them explicitly 

1606 if not fileLocations: 

1607 if not self.trustGetRequest: 

1608 return False 

1609 

1610 # First check the cache. If it is not found we must check 

1611 # the datastore itself. Assume that any component in the cache 

1612 # means that the dataset does exist somewhere. 

1613 if self.cacheManager.known_to_cache(ref): 1613 ↛ 1614line 1613 didn't jump to line 1614, because the condition on line 1613 was never true

1614 return True 

1615 

1616 # When we are guessing a dataset location we can not check 

1617 # for the existence of every component since we can not 

1618 # know if every component was written. Instead we check 

1619 # for the existence of any of the expected locations. 

1620 for location, _ in self._get_expected_dataset_locations_info(ref): 

1621 if self._artifact_exists(location): 

1622 return True 

1623 return False 

1624 

1625 # All listed artifacts must exist. 

1626 for location, storedFileInfo in fileLocations: 

1627 # Checking in cache needs the component ref. 

1628 check_ref = ref 

1629 if not ref.datasetType.isComponent() and (component := storedFileInfo.component): 

1630 check_ref = ref.makeComponentRef(component) 

1631 if self.cacheManager.known_to_cache(check_ref, location.getExtension()): 

1632 continue 

1633 

1634 if not self._artifact_exists(location): 

1635 return False 

1636 

1637 return True 

1638 

1639 def getURIs(self, ref: DatasetRef, predict: bool = False) -> DatasetRefURIs: 

1640 """Return URIs associated with dataset. 

1641 

1642 Parameters 

1643 ---------- 

1644 ref : `DatasetRef` 

1645 Reference to the required dataset. 

1646 predict : `bool`, optional 

1647 If the datastore does not know about the dataset, should it 

1648 return a predicted URI or not? 

1649 

1650 Returns 

1651 ------- 

1652 uris : `DatasetRefURIs` 

1653 The URI to the primary artifact associated with this dataset (if 

1654 the dataset was disassembled within the datastore this may be 

1655 `None`), and the URIs to any components associated with the dataset 

1656 artifact. (can be empty if there are no components). 

1657 """ 

1658 # if this has never been written then we have to guess 

1659 if not self.exists(ref): 

1660 if not predict: 

1661 raise FileNotFoundError("Dataset {} not in this datastore".format(ref)) 

1662 

1663 return self._predict_URIs(ref) 

1664 

1665 # If this is a ref that we have written we can get the path. 

1666 # Get file metadata and internal metadata 

1667 fileLocations = self._get_dataset_locations_info(ref) 

1668 

1669 return self._locations_to_URI(ref, fileLocations) 

1670 

1671 def getURI(self, ref: DatasetRef, predict: bool = False) -> ResourcePath: 

1672 """URI to the Dataset. 

1673 

1674 Parameters 

1675 ---------- 

1676 ref : `DatasetRef` 

1677 Reference to the required Dataset. 

1678 predict : `bool` 

1679 If `True`, allow URIs to be returned of datasets that have not 

1680 been written. 

1681 

1682 Returns 

1683 ------- 

1684 uri : `str` 

1685 URI pointing to the dataset within the datastore. If the 

1686 dataset does not exist in the datastore, and if ``predict`` is 

1687 `True`, the URI will be a prediction and will include a URI 

1688 fragment "#predicted". 

1689 If the datastore does not have entities that relate well 

1690 to the concept of a URI the returned URI will be 

1691 descriptive. The returned URI is not guaranteed to be obtainable. 

1692 

1693 Raises 

1694 ------ 

1695 FileNotFoundError 

1696 Raised if a URI has been requested for a dataset that does not 

1697 exist and guessing is not allowed. 

1698 RuntimeError 

1699 Raised if a request is made for a single URI but multiple URIs 

1700 are associated with this dataset. 

1701 

1702 Notes 

1703 ----- 

1704 When a predicted URI is requested an attempt will be made to form 

1705 a reasonable URI based on file templates and the expected formatter. 

1706 """ 

1707 primary, components = self.getURIs(ref, predict) 

1708 if primary is None or components: 1708 ↛ 1709line 1708 didn't jump to line 1709, because the condition on line 1708 was never true

1709 raise RuntimeError( 

1710 f"Dataset ({ref}) includes distinct URIs for components. Use Datastore.getURIs() instead." 

1711 ) 

1712 return primary 

1713 

1714 def _predict_URIs( 

1715 self, 

1716 ref: DatasetRef, 

1717 ) -> DatasetRefURIs: 

1718 """Predict the URIs of a dataset ref. 

1719 

1720 Parameters 

1721 ---------- 

1722 ref : `DatasetRef` 

1723 Reference to the required Dataset. 

1724 

1725 Returns 

1726 ------- 

1727 URI : DatasetRefUris 

1728 Primary and component URIs. URIs will contain a URI fragment 

1729 "#predicted". 

1730 """ 

1731 uris = DatasetRefURIs() 

1732 

1733 if self.composites.shouldBeDisassembled(ref): 

1734 for component, _ in ref.datasetType.storageClass.components.items(): 

1735 comp_ref = ref.makeComponentRef(component) 

1736 comp_location, _ = self._determine_put_formatter_location(comp_ref) 

1737 

1738 # Add the "#predicted" URI fragment to indicate this is a 

1739 # guess 

1740 uris.componentURIs[component] = ResourcePath(comp_location.uri.geturl() + "#predicted") 

1741 

1742 else: 

1743 location, _ = self._determine_put_formatter_location(ref) 

1744 

1745 # Add the "#predicted" URI fragment to indicate this is a guess 

1746 uris.primaryURI = ResourcePath(location.uri.geturl() + "#predicted") 

1747 

1748 return uris 

1749 

1750 def getManyURIs( 

1751 self, 

1752 refs: Iterable[DatasetRef], 

1753 predict: bool = False, 

1754 allow_missing: bool = False, 

1755 ) -> Dict[DatasetRef, DatasetRefURIs]: 

1756 # Docstring inherited 

1757 

1758 uris: Dict[DatasetRef, DatasetRefURIs] = {} 

1759 

1760 records = self._get_stored_records_associated_with_refs(refs) 

1761 records_keys = records.keys() 

1762 

1763 existing_refs = (ref for ref in refs if ref.id in records_keys) 

1764 missing_refs = (ref for ref in refs if ref.id not in records_keys) 

1765 

1766 for ref in missing_refs: 

1767 # if this has never been written then we have to guess 

1768 if not predict: 

1769 if not allow_missing: 

1770 raise FileNotFoundError("Dataset {} not in this datastore.".format(ref)) 

1771 else: 

1772 uris[ref] = self._predict_URIs(ref) 

1773 

1774 for ref in existing_refs: 

1775 file_infos = records[ref.getCheckedId()] 

1776 file_locations = [(i.file_location(self.locationFactory), i) for i in file_infos] 

1777 uris[ref] = self._locations_to_URI(ref, file_locations) 

1778 

1779 return uris 

1780 

1781 def _locations_to_URI( 

1782 self, 

1783 ref: DatasetRef, 

1784 file_locations: Sequence[Tuple[Location, StoredFileInfo]], 

1785 ) -> DatasetRefURIs: 

1786 """Convert one or more file locations associated with a DatasetRef 

1787 to a DatasetRefURIs. 

1788 

1789 Parameters 

1790 ---------- 

1791 ref : `DatasetRef` 

1792 Reference to the dataset. 

1793 file_locations : Sequence[Tuple[Location, StoredFileInfo]] 

1794 Each item in the sequence is the location of the dataset within the 

1795 datastore and stored information about the file and its formatter. 

1796 If there is only one item in the sequence then it is treated as the 

1797 primary URI. If there is more than one item then they are treated 

1798 as component URIs. If there are no items then an error is raised 

1799 unless ``self.trustGetRequest`` is `True`. 

1800 

1801 Returns 

1802 ------- 

1803 uris: DatasetRefURIs 

1804 Represents the primary URI or component URIs described by the 

1805 inputs. 

1806 

1807 Raises 

1808 ------ 

1809 RuntimeError 

1810 If no file locations are passed in and ``self.trustGetRequest`` is 

1811 `False`. 

1812 FileNotFoundError 

1813 If the a passed-in URI does not exist, and ``self.trustGetRequest`` 

1814 is `False`. 

1815 RuntimeError 

1816 If a passed in `StoredFileInfo`'s ``component`` is `None` (this is 

1817 unexpected). 

1818 """ 

1819 

1820 guessing = False 

1821 uris = DatasetRefURIs() 

1822 

1823 if not file_locations: 

1824 if not self.trustGetRequest: 1824 ↛ 1825line 1824 didn't jump to line 1825, because the condition on line 1824 was never true

1825 raise RuntimeError(f"Unexpectedly got no artifacts for dataset {ref}") 

1826 file_locations = self._get_expected_dataset_locations_info(ref) 

1827 guessing = True 

1828 

1829 if len(file_locations) == 1: 

1830 # No disassembly so this is the primary URI 

1831 uris.primaryURI = file_locations[0][0].uri 

1832 if guessing and not uris.primaryURI.exists(): 1832 ↛ 1833line 1832 didn't jump to line 1833, because the condition on line 1832 was never true

1833 raise FileNotFoundError(f"Expected URI ({uris.primaryURI}) does not exist") 

1834 else: 

1835 for location, file_info in file_locations: 

1836 if file_info.component is None: 1836 ↛ 1837line 1836 didn't jump to line 1837, because the condition on line 1836 was never true

1837 raise RuntimeError(f"Unexpectedly got no component name for a component at {location}") 

1838 if guessing and not location.uri.exists(): 1838 ↛ 1842line 1838 didn't jump to line 1842, because the condition on line 1838 was never true

1839 # If we are trusting then it is entirely possible for 

1840 # some components to be missing. In that case we skip 

1841 # to the next component. 

1842 if self.trustGetRequest: 

1843 continue 

1844 raise FileNotFoundError(f"Expected URI ({location.uri}) does not exist") 

1845 uris.componentURIs[file_info.component] = location.uri 

1846 

1847 return uris 

1848 

1849 def retrieveArtifacts( 

1850 self, 

1851 refs: Iterable[DatasetRef], 

1852 destination: ResourcePath, 

1853 transfer: str = "auto", 

1854 preserve_path: bool = True, 

1855 overwrite: bool = False, 

1856 ) -> List[ResourcePath]: 

1857 """Retrieve the file artifacts associated with the supplied refs. 

1858 

1859 Parameters 

1860 ---------- 

1861 refs : iterable of `DatasetRef` 

1862 The datasets for which file artifacts are to be retrieved. 

1863 A single ref can result in multiple files. The refs must 

1864 be resolved. 

1865 destination : `lsst.resources.ResourcePath` 

1866 Location to write the file artifacts. 

1867 transfer : `str`, optional 

1868 Method to use to transfer the artifacts. Must be one of the options 

1869 supported by `lsst.resources.ResourcePath.transfer_from()`. 

1870 "move" is not allowed. 

1871 preserve_path : `bool`, optional 

1872 If `True` the full path of the file artifact within the datastore 

1873 is preserved. If `False` the final file component of the path 

1874 is used. 

1875 overwrite : `bool`, optional 

1876 If `True` allow transfers to overwrite existing files at the 

1877 destination. 

1878 

1879 Returns 

1880 ------- 

1881 targets : `list` of `lsst.resources.ResourcePath` 

1882 URIs of file artifacts in destination location. Order is not 

1883 preserved. 

1884 """ 

1885 if not destination.isdir(): 1885 ↛ 1886line 1885 didn't jump to line 1886, because the condition on line 1885 was never true

1886 raise ValueError(f"Destination location must refer to a directory. Given {destination}") 

1887 

1888 if transfer == "move": 

1889 raise ValueError("Can not move artifacts out of datastore. Use copy instead.") 

1890 

1891 # Source -> Destination 

1892 # This also helps filter out duplicate DatasetRef in the request 

1893 # that will map to the same underlying file transfer. 

1894 to_transfer: Dict[ResourcePath, ResourcePath] = {} 

1895 

1896 for ref in refs: 

1897 locations = self._get_dataset_locations_info(ref) 

1898 for location, _ in locations: 

1899 source_uri = location.uri 

1900 target_path: ResourcePathExpression 

1901 if preserve_path: 

1902 target_path = location.pathInStore 

1903 if target_path.isabs(): 1903 ↛ 1906line 1903 didn't jump to line 1906, because the condition on line 1903 was never true

1904 # This is an absolute path to an external file. 

1905 # Use the full path. 

1906 target_path = target_path.relativeToPathRoot 

1907 else: 

1908 target_path = source_uri.basename() 

1909 target_uri = destination.join(target_path) 

1910 to_transfer[source_uri] = target_uri 

1911 

1912 # In theory can now parallelize the transfer 

1913 log.debug("Number of artifacts to transfer to %s: %d", str(destination), len(to_transfer)) 

1914 for source_uri, target_uri in to_transfer.items(): 

1915 target_uri.transfer_from(source_uri, transfer=transfer, overwrite=overwrite) 

1916 

1917 return list(to_transfer.values()) 

1918 

1919 def get(self, ref: DatasetRef, parameters: Optional[Mapping[str, Any]] = None) -> Any: 

1920 """Load an InMemoryDataset from the store. 

1921 

1922 Parameters 

1923 ---------- 

1924 ref : `DatasetRef` 

1925 Reference to the required Dataset. 

1926 parameters : `dict` 

1927 `StorageClass`-specific parameters that specify, for example, 

1928 a slice of the dataset to be loaded. 

1929 

1930 Returns 

1931 ------- 

1932 inMemoryDataset : `object` 

1933 Requested dataset or slice thereof as an InMemoryDataset. 

1934 

1935 Raises 

1936 ------ 

1937 FileNotFoundError 

1938 Requested dataset can not be retrieved. 

1939 TypeError 

1940 Return value from formatter has unexpected type. 

1941 ValueError 

1942 Formatter failed to process the dataset. 

1943 """ 

1944 allGetInfo = self._prepare_for_get(ref, parameters) 

1945 refComponent = ref.datasetType.component() 

1946 

1947 # Supplied storage class for the component being read 

1948 refStorageClass = ref.datasetType.storageClass 

1949 

1950 # Create mapping from component name to related info 

1951 allComponents = {i.component: i for i in allGetInfo} 

1952 

1953 # By definition the dataset is disassembled if we have more 

1954 # than one record for it. 

1955 isDisassembled = len(allGetInfo) > 1 

1956 

1957 # Look for the special case where we are disassembled but the 

1958 # component is a derived component that was not written during 

1959 # disassembly. For this scenario we need to check that the 

1960 # component requested is listed as a derived component for the 

1961 # composite storage class 

1962 isDisassembledReadOnlyComponent = False 

1963 if isDisassembled and refComponent: 

1964 # The composite storage class should be accessible through 

1965 # the component dataset type 

1966 compositeStorageClass = ref.datasetType.parentStorageClass 

1967 

1968 # In the unlikely scenario where the composite storage 

1969 # class is not known, we can only assume that this is a 

1970 # normal component. If that assumption is wrong then the 

1971 # branch below that reads a persisted component will fail 

1972 # so there is no need to complain here. 

1973 if compositeStorageClass is not None: 1973 ↛ 1976line 1973 didn't jump to line 1976, because the condition on line 1973 was never false

1974 isDisassembledReadOnlyComponent = refComponent in compositeStorageClass.derivedComponents 

1975 

1976 if isDisassembled and not refComponent: 

1977 # This was a disassembled dataset spread over multiple files 

1978 # and we need to put them all back together again. 

1979 # Read into memory and then assemble 

1980 

1981 # Check that the supplied parameters are suitable for the type read 

1982 refStorageClass.validateParameters(parameters) 

1983 

1984 # We want to keep track of all the parameters that were not used 

1985 # by formatters. We assume that if any of the component formatters 

1986 # use a parameter that we do not need to apply it again in the 

1987 # assembler. 

1988 usedParams = set() 

1989 

1990 components: Dict[str, Any] = {} 

1991 for getInfo in allGetInfo: 

1992 # assemblerParams are parameters not understood by the 

1993 # associated formatter. 

1994 usedParams.update(set(getInfo.formatterParams)) 

1995 

1996 component = getInfo.component 

1997 

1998 if component is None: 1998 ↛ 1999line 1998 didn't jump to line 1999, because the condition on line 1998 was never true

1999 raise RuntimeError(f"Internal error in datastore assembly of {ref}") 

2000 

2001 # We do not want the formatter to think it's reading 

2002 # a component though because it is really reading a 

2003 # standalone dataset -- always tell reader it is not a 

2004 # component. 

2005 components[component] = self._read_artifact_into_memory( 

2006 getInfo, ref.makeComponentRef(component), isComponent=False 

2007 ) 

2008 

2009 inMemoryDataset = ref.datasetType.storageClass.delegate().assemble(components) 

2010 

2011 # Any unused parameters will have to be passed to the assembler 

2012 if parameters: 

2013 unusedParams = {k: v for k, v in parameters.items() if k not in usedParams} 

2014 else: 

2015 unusedParams = {} 

2016 

2017 # Process parameters 

2018 return ref.datasetType.storageClass.delegate().handleParameters( 

2019 inMemoryDataset, parameters=unusedParams 

2020 ) 

2021 

2022 elif isDisassembledReadOnlyComponent: 

2023 compositeStorageClass = ref.datasetType.parentStorageClass 

2024 if compositeStorageClass is None: 2024 ↛ 2025line 2024 didn't jump to line 2025, because the condition on line 2024 was never true

2025 raise RuntimeError( 

2026 f"Unable to retrieve derived component '{refComponent}' since" 

2027 "no composite storage class is available." 

2028 ) 

2029 

2030 if refComponent is None: 2030 ↛ 2032line 2030 didn't jump to line 2032, because the condition on line 2030 was never true

2031 # Mainly for mypy 

2032 raise RuntimeError(f"Internal error in datastore {self.name}: component can not be None here") 

2033 

2034 # Assume that every derived component can be calculated by 

2035 # forwarding the request to a single read/write component. 

2036 # Rather than guessing which rw component is the right one by 

2037 # scanning each for a derived component of the same name, 

2038 # we ask the storage class delegate directly which one is best to 

2039 # use. 

2040 compositeDelegate = compositeStorageClass.delegate() 

2041 forwardedComponent = compositeDelegate.selectResponsibleComponent( 

2042 refComponent, set(allComponents) 

2043 ) 

2044 

2045 # Select the relevant component 

2046 rwInfo = allComponents[forwardedComponent] 

2047 

2048 # For now assume that read parameters are validated against 

2049 # the real component and not the requested component 

2050 forwardedStorageClass = rwInfo.formatter.fileDescriptor.readStorageClass 

2051 forwardedStorageClass.validateParameters(parameters) 

2052 

2053 # The reference to use for the caching must refer to the forwarded 

2054 # component and not the derived component. 

2055 cache_ref = ref.makeCompositeRef().makeComponentRef(forwardedComponent) 

2056 

2057 # Unfortunately the FileDescriptor inside the formatter will have 

2058 # the wrong write storage class so we need to create a new one 

2059 # given the immutability constraint. 

2060 writeStorageClass = rwInfo.info.storageClass 

2061 

2062 # We may need to put some thought into parameters for read 

2063 # components but for now forward them on as is 

2064 readFormatter = type(rwInfo.formatter)( 

2065 FileDescriptor( 

2066 rwInfo.location, 

2067 readStorageClass=refStorageClass, 

2068 storageClass=writeStorageClass, 

2069 parameters=parameters, 

2070 ), 

2071 ref.dataId, 

2072 ) 

2073 

2074 # The assembler can not receive any parameter requests for a 

2075 # derived component at this time since the assembler will 

2076 # see the storage class of the derived component and those 

2077 # parameters will have to be handled by the formatter on the 

2078 # forwarded storage class. 

2079 assemblerParams: Dict[str, Any] = {} 

2080 

2081 # Need to created a new info that specifies the derived 

2082 # component and associated storage class 

2083 readInfo = DatastoreFileGetInformation( 

2084 rwInfo.location, 

2085 readFormatter, 

2086 rwInfo.info, 

2087 assemblerParams, 

2088 {}, 

2089 refComponent, 

2090 refStorageClass, 

2091 ) 

2092 

2093 return self._read_artifact_into_memory(readInfo, ref, isComponent=True, cache_ref=cache_ref) 

2094 

2095 else: 

2096 # Single file request or component from that composite file 

2097 for lookup in (refComponent, None): 2097 ↛ 2102line 2097 didn't jump to line 2102, because the loop on line 2097 didn't complete

2098 if lookup in allComponents: 2098 ↛ 2097line 2098 didn't jump to line 2097, because the condition on line 2098 was never false

2099 getInfo = allComponents[lookup] 

2100 break 

2101 else: 

2102 raise FileNotFoundError( 

2103 f"Component {refComponent} not found for ref {ref} in datastore {self.name}" 

2104 ) 

2105 

2106 # Do not need the component itself if already disassembled 

2107 if isDisassembled: 

2108 isComponent = False 

2109 else: 

2110 isComponent = getInfo.component is not None 

2111 

2112 # For a component read of a composite we want the cache to 

2113 # be looking at the composite ref itself. 

2114 cache_ref = ref.makeCompositeRef() if isComponent else ref 

2115 

2116 # For a disassembled component we can validate parametersagainst 

2117 # the component storage class directly 

2118 if isDisassembled: 

2119 refStorageClass.validateParameters(parameters) 

2120 else: 

2121 # For an assembled composite this could be a derived 

2122 # component derived from a real component. The validity 

2123 # of the parameters is not clear. For now validate against 

2124 # the composite storage class 

2125 getInfo.formatter.fileDescriptor.storageClass.validateParameters(parameters) 

2126 

2127 return self._read_artifact_into_memory(getInfo, ref, isComponent=isComponent, cache_ref=cache_ref) 

2128 

2129 @transactional 

2130 def put(self, inMemoryDataset: Any, ref: DatasetRef) -> None: 

2131 """Write a InMemoryDataset with a given `DatasetRef` to the store. 

2132 

2133 Parameters 

2134 ---------- 

2135 inMemoryDataset : `object` 

2136 The dataset to store. 

2137 ref : `DatasetRef` 

2138 Reference to the associated Dataset. 

2139 

2140 Raises 

2141 ------ 

2142 TypeError 

2143 Supplied object and storage class are inconsistent. 

2144 DatasetTypeNotSupportedError 

2145 The associated `DatasetType` is not handled by this datastore. 

2146 

2147 Notes 

2148 ----- 

2149 If the datastore is configured to reject certain dataset types it 

2150 is possible that the put will fail and raise a 

2151 `DatasetTypeNotSupportedError`. The main use case for this is to 

2152 allow `ChainedDatastore` to put to multiple datastores without 

2153 requiring that every datastore accepts the dataset. 

2154 """ 

2155 

2156 doDisassembly = self.composites.shouldBeDisassembled(ref) 

2157 # doDisassembly = True 

2158 

2159 artifacts = [] 

2160 if doDisassembly: 

2161 components = ref.datasetType.storageClass.delegate().disassemble(inMemoryDataset) 

2162 if components is None: 2162 ↛ 2163line 2162 didn't jump to line 2163, because the condition on line 2162 was never true

2163 raise RuntimeError( 

2164 f"Inconsistent configuration: dataset type {ref.datasetType.name} " 

2165 f"with storage class {ref.datasetType.storageClass.name} " 

2166 "is configured to be disassembled, but cannot be." 

2167 ) 

2168 for component, componentInfo in components.items(): 

2169 # Don't recurse because we want to take advantage of 

2170 # bulk insert -- need a new DatasetRef that refers to the 

2171 # same dataset_id but has the component DatasetType 

2172 # DatasetType does not refer to the types of components 

2173 # So we construct one ourselves. 

2174 compRef = ref.makeComponentRef(component) 

2175 storedInfo = self._write_in_memory_to_artifact(componentInfo.component, compRef) 

2176 artifacts.append((compRef, storedInfo)) 

2177 else: 

2178 # Write the entire thing out 

2179 storedInfo = self._write_in_memory_to_artifact(inMemoryDataset, ref) 

2180 artifacts.append((ref, storedInfo)) 

2181 

2182 self._register_datasets(artifacts) 

2183 

2184 @transactional 

2185 def trash(self, ref: Union[DatasetRef, Iterable[DatasetRef]], ignore_errors: bool = True) -> None: 

2186 # At this point can safely remove these datasets from the cache 

2187 # to avoid confusion later on. If they are not trashed later 

2188 # the cache will simply be refilled. 

2189 self.cacheManager.remove_from_cache(ref) 

2190 

2191 # If we are in trust mode there will be nothing to move to 

2192 # the trash table and we will have to try to delete the file 

2193 # immediately. 

2194 if self.trustGetRequest: 

2195 # Try to keep the logic below for a single file trash. 

2196 if isinstance(ref, DatasetRef): 

2197 refs = {ref} 

2198 else: 

2199 # Will recreate ref at the end of this branch. 

2200 refs = set(ref) 

2201 

2202 # Determine which datasets are known to datastore directly. 

2203 id_to_ref = {ref.getCheckedId(): ref for ref in refs} 

2204 existing_ids = self._get_stored_records_associated_with_refs(refs) 

2205 existing_refs = {id_to_ref[ref_id] for ref_id in existing_ids} 

2206 

2207 missing = refs - existing_refs 

2208 if missing: 

2209 # Do an explicit existence check on these refs. 

2210 # We only care about the artifacts at this point and not 

2211 # the dataset existence. 

2212 artifact_existence: Dict[ResourcePath, bool] = {} 

2213 _ = self.mexists(missing, artifact_existence) 

2214 uris = [uri for uri, exists in artifact_existence.items() if exists] 

2215 

2216 # FUTURE UPGRADE: Implement a parallelized bulk remove. 

2217 log.debug("Removing %d artifacts from datastore that are unknown to datastore", len(uris)) 

2218 for uri in uris: 

2219 try: 

2220 uri.remove() 

2221 except Exception as e: 

2222 if ignore_errors: 

2223 log.debug("Artifact %s could not be removed: %s", uri, e) 

2224 continue 

2225 raise 

2226 

2227 # There is no point asking the code below to remove refs we 

2228 # know are missing so update it with the list of existing 

2229 # records. Try to retain one vs many logic. 

2230 if not existing_refs: 

2231 # Nothing more to do since none of the datasets were 

2232 # known to the datastore record table. 

2233 return 

2234 ref = list(existing_refs) 

2235 if len(ref) == 1: 

2236 ref = ref[0] 

2237 

2238 # Get file metadata and internal metadata 

2239 if not isinstance(ref, DatasetRef): 

2240 log.debug("Doing multi-dataset trash in datastore %s", self.name) 

2241 # Assumed to be an iterable of refs so bulk mode enabled. 

2242 try: 

2243 self.bridge.moveToTrash(ref) 

2244 except Exception as e: 

2245 if ignore_errors: 

2246 log.warning("Unexpected issue moving multiple datasets to trash: %s", e) 

2247 else: 

2248 raise 

2249 return 

2250 

2251 log.debug("Trashing dataset %s in datastore %s", ref, self.name) 

2252 

2253 fileLocations = self._get_dataset_locations_info(ref) 

2254 

2255 if not fileLocations: 

2256 err_msg = f"Requested dataset to trash ({ref}) is not known to datastore {self.name}" 

2257 if ignore_errors: 

2258 log.warning(err_msg) 

2259 return 

2260 else: 

2261 raise FileNotFoundError(err_msg) 

2262 

2263 for location, storedFileInfo in fileLocations: 

2264 if not self._artifact_exists(location): 2264 ↛ 2265line 2264 didn't jump to line 2265

2265 err_msg = ( 

2266 f"Dataset is known to datastore {self.name} but " 

2267 f"associated artifact ({location.uri}) is missing" 

2268 ) 

2269 if ignore_errors: 

2270 log.warning(err_msg) 

2271 return 

2272 else: 

2273 raise FileNotFoundError(err_msg) 

2274 

2275 # Mark dataset as trashed 

2276 try: 

2277 self.bridge.moveToTrash([ref]) 

2278 except Exception as e: 

2279 if ignore_errors: 

2280 log.warning( 

2281 "Attempted to mark dataset (%s) to be trashed in datastore %s " 

2282 "but encountered an error: %s", 

2283 ref, 

2284 self.name, 

2285 e, 

2286 ) 

2287 pass 

2288 else: 

2289 raise 

2290 

2291 @transactional 

2292 def emptyTrash(self, ignore_errors: bool = True) -> None: 

2293 """Remove all datasets from the trash. 

2294 

2295 Parameters 

2296 ---------- 

2297 ignore_errors : `bool` 

2298 If `True` return without error even if something went wrong. 

2299 Problems could occur if another process is simultaneously trying 

2300 to delete. 

2301 """ 

2302 log.debug("Emptying trash in datastore %s", self.name) 

2303 

2304 # Context manager will empty trash iff we finish it without raising. 

2305 # It will also automatically delete the relevant rows from the 

2306 # trash table and the records table. 

2307 with self.bridge.emptyTrash( 

2308 self._table, record_class=StoredFileInfo, record_column="path" 

2309 ) as trash_data: 

2310 # Removing the artifacts themselves requires that the files are 

2311 # not also associated with refs that are not to be trashed. 

2312 # Therefore need to do a query with the file paths themselves 

2313 # and return all the refs associated with them. Can only delete 

2314 # a file if the refs to be trashed are the only refs associated 

2315 # with the file. 

2316 # This requires multiple copies of the trashed items 

2317 trashed, artifacts_to_keep = trash_data 

2318 

2319 if artifacts_to_keep is None: 

2320 # The bridge is not helping us so have to work it out 

2321 # ourselves. This is not going to be as efficient. 

2322 trashed = list(trashed) 

2323 

2324 # The instance check is for mypy since up to this point it 

2325 # does not know the type of info. 

2326 path_map = self._refs_associated_with_artifacts( 

2327 [info.path for _, info in trashed if isinstance(info, StoredFileInfo)] 

2328 ) 

2329 

2330 for ref, info in trashed: 

2331 # Mypy needs to know this is not the base class 

2332 assert isinstance(info, StoredFileInfo), f"Unexpectedly got info of class {type(info)}" 

2333 

2334 # Check for mypy 

2335 assert ref.id is not None, f"Internal logic error in emptyTrash with ref {ref}/{info}" 

2336 

2337 path_map[info.path].remove(ref.id) 

2338 if not path_map[info.path]: 2338 ↛ 2330line 2338 didn't jump to line 2330, because the condition on line 2338 was never false

2339 del path_map[info.path] 

2340 

2341 artifacts_to_keep = set(path_map) 

2342 

2343 for ref, info in trashed: 

2344 # Should not happen for this implementation but need 

2345 # to keep mypy happy. 

2346 assert info is not None, f"Internal logic error in emptyTrash with ref {ref}." 

2347 

2348 # Mypy needs to know this is not the base class 

2349 assert isinstance(info, StoredFileInfo), f"Unexpectedly got info of class {type(info)}" 

2350 

2351 # Check for mypy 

2352 assert ref.id is not None, f"Internal logic error in emptyTrash with ref {ref}/{info}" 

2353 

2354 if info.path in artifacts_to_keep: 

2355 # This is a multi-dataset artifact and we are not 

2356 # removing all associated refs. 

2357 continue 

2358 

2359 # Only trashed refs still known to datastore will be returned. 

2360 location = info.file_location(self.locationFactory) 

2361 

2362 # Point of no return for this artifact 

2363 log.debug("Removing artifact %s from datastore %s", location.uri, self.name) 

2364 try: 

2365 self._delete_artifact(location) 

2366 except FileNotFoundError: 

2367 # If the file itself has been deleted there is nothing 

2368 # we can do about it. It is possible that trash has 

2369 # been run in parallel in another process or someone 

2370 # decided to delete the file. It is unlikely to come 

2371 # back and so we should still continue with the removal 

2372 # of the entry from the trash table. It is also possible 

2373 # we removed it in a previous iteration if it was 

2374 # a multi-dataset artifact. The delete artifact method 

2375 # will log a debug message in this scenario. 

2376 # Distinguishing file missing before trash started and 

2377 # file already removed previously as part of this trash 

2378 # is not worth the distinction with regards to potential 

2379 # memory cost. 

2380 pass 

2381 except Exception as e: 

2382 if ignore_errors: 

2383 # Use a debug message here even though it's not 

2384 # a good situation. In some cases this can be 

2385 # caused by a race between user A and user B 

2386 # and neither of them has permissions for the 

2387 # other's files. Butler does not know about users 

2388 # and trash has no idea what collections these 

2389 # files were in (without guessing from a path). 

2390 log.debug( 

2391 "Encountered error removing artifact %s from datastore %s: %s", 

2392 location.uri, 

2393 self.name, 

2394 e, 

2395 ) 

2396 else: 

2397 raise 

2398 

2399 @transactional 

2400 def transfer_from( 

2401 self, 

2402 source_datastore: Datastore, 

2403 refs: Iterable[DatasetRef], 

2404 local_refs: Optional[Iterable[DatasetRef]] = None, 

2405 transfer: str = "auto", 

2406 artifact_existence: Optional[Dict[ResourcePath, bool]] = None, 

2407 ) -> None: 

2408 # Docstring inherited 

2409 if type(self) is not type(source_datastore): 

2410 raise TypeError( 

2411 f"Datastore mismatch between this datastore ({type(self)}) and the " 

2412 f"source datastore ({type(source_datastore)})." 

2413 ) 

2414 

2415 # Be explicit for mypy 

2416 if not isinstance(source_datastore, FileDatastore): 2416 ↛ 2417line 2416 didn't jump to line 2417, because the condition on line 2416 was never true

2417 raise TypeError( 

2418 "Can only transfer to a FileDatastore from another FileDatastore, not" 

2419 f" {type(source_datastore)}" 

2420 ) 

2421 

2422 # Stop early if "direct" transfer mode is requested. That would 

2423 # require that the URI inside the source datastore should be stored 

2424 # directly in the target datastore, which seems unlikely to be useful 

2425 # since at any moment the source datastore could delete the file. 

2426 if transfer in ("direct", "split"): 

2427 raise ValueError( 

2428 f"Can not transfer from a source datastore using {transfer} mode since" 

2429 " those files are controlled by the other datastore." 

2430 ) 

2431 

2432 # Empty existence lookup if none given. 

2433 if artifact_existence is None: 

2434 artifact_existence = {} 

2435 

2436 # We will go through the list multiple times so must convert 

2437 # generators to lists. 

2438 refs = list(refs) 

2439 

2440 if local_refs is None: 

2441 local_refs = refs 

2442 else: 

2443 local_refs = list(local_refs) 

2444 

2445 # In order to handle disassembled composites the code works 

2446 # at the records level since it can assume that internal APIs 

2447 # can be used. 

2448 # - If the record already exists in the destination this is assumed 

2449 # to be okay. 

2450 # - If there is no record but the source and destination URIs are 

2451 # identical no transfer is done but the record is added. 

2452 # - If the source record refers to an absolute URI currently assume 

2453 # that that URI should remain absolute and will be visible to the 

2454 # destination butler. May need to have a flag to indicate whether 

2455 # the dataset should be transferred. This will only happen if 

2456 # the detached Butler has had a local ingest. 

2457 

2458 # What we really want is all the records in the source datastore 

2459 # associated with these refs. Or derived ones if they don't exist 

2460 # in the source. 

2461 source_records = source_datastore._get_stored_records_associated_with_refs(refs) 

2462 

2463 # The source dataset_ids are the keys in these records 

2464 source_ids = set(source_records) 

2465 log.debug("Number of datastore records found in source: %d", len(source_ids)) 

2466 

2467 # The not None check is to appease mypy 

2468 requested_ids = set(ref.id for ref in refs if ref.id is not None) 

2469 missing_ids = requested_ids - source_ids 

2470 

2471 # Missing IDs can be okay if that datastore has allowed 

2472 # gets based on file existence. Should we transfer what we can 

2473 # or complain about it and warn? 

2474 if missing_ids and not source_datastore.trustGetRequest: 2474 ↛ 2475line 2474 didn't jump to line 2475, because the condition on line 2474 was never true

2475 raise ValueError( 

2476 f"Some datasets are missing from source datastore {source_datastore}: {missing_ids}" 

2477 ) 

2478 

2479 # Need to map these missing IDs to a DatasetRef so we can guess 

2480 # the details. 

2481 if missing_ids: 

2482 log.info( 

2483 "Number of expected datasets missing from source datastore records: %d out of %d", 

2484 len(missing_ids), 

2485 len(requested_ids), 

2486 ) 

2487 id_to_ref = {ref.id: ref for ref in refs if ref.id in missing_ids} 

2488 

2489 # This should be chunked in case we end up having to check 

2490 # the file store since we need some log output to show 

2491 # progress. 

2492 for missing_ids_chunk in chunk_iterable(missing_ids, chunk_size=10_000): 

2493 records = {} 

2494 for missing in missing_ids_chunk: 

2495 # Ask the source datastore where the missing artifacts 

2496 # should be. An execution butler might not know about the 

2497 # artifacts even if they are there. 

2498 expected = source_datastore._get_expected_dataset_locations_info(id_to_ref[missing]) 

2499 records[missing] = [info for _, info in expected] 

2500 

2501 # Call the mexist helper method in case we have not already 

2502 # checked these artifacts such that artifact_existence is 

2503 # empty. This allows us to benefit from parallelism. 

2504 # datastore.mexists() itself does not give us access to the 

2505 # derived datastore record. 

2506 log.verbose("Checking existence of %d datasets unknown to datastore", len(records)) 

2507 ref_exists = source_datastore._process_mexists_records( 

2508 id_to_ref, records, False, artifact_existence=artifact_existence 

2509 ) 

2510 

2511 # Now go through the records and propagate the ones that exist. 

2512 location_factory = source_datastore.locationFactory 

2513 for missing, record_list in records.items(): 

2514 # Skip completely if the ref does not exist. 

2515 ref = id_to_ref[missing] 

2516 if not ref_exists[ref]: 

2517 log.warning("Asked to transfer dataset %s but no file artifacts exist for it.", ref) 

2518 continue 

2519 # Check for file artifact to decide which parts of a 

2520 # disassembled composite do exist. If there is only a 

2521 # single record we don't even need to look because it can't 

2522 # be a composite and must exist. 

2523 if len(record_list) == 1: 

2524 dataset_records = record_list 

2525 else: 

2526 dataset_records = [ 

2527 record 

2528 for record in record_list 

2529 if artifact_existence[record.file_location(location_factory).uri] 

2530 ] 

2531 assert len(dataset_records) > 0, "Disassembled composite should have had some files." 

2532 

2533 # Rely on source_records being a defaultdict. 

2534 source_records[missing].extend(dataset_records) 

2535 

2536 # See if we already have these records 

2537 target_records = self._get_stored_records_associated_with_refs(local_refs) 

2538 

2539 # The artifacts to register 

2540 artifacts = [] 

2541 

2542 # Refs that already exist 

2543 already_present = [] 

2544 

2545 # Now can transfer the artifacts 

2546 for source_ref, target_ref in zip(refs, local_refs): 

2547 if target_ref.id in target_records: 

2548 # Already have an artifact for this. 

2549 already_present.append(target_ref) 

2550 continue 

2551 

2552 # mypy needs to know these are always resolved refs 

2553 for info in source_records[source_ref.getCheckedId()]: 

2554 source_location = info.file_location(source_datastore.locationFactory) 

2555 target_location = info.file_location(self.locationFactory) 

2556 if source_location == target_location: 2556 ↛ 2560line 2556 didn't jump to line 2560, because the condition on line 2556 was never true

2557 # Either the dataset is already in the target datastore 

2558 # (which is how execution butler currently runs) or 

2559 # it is an absolute URI. 

2560 if source_location.pathInStore.isabs(): 

2561 # Just because we can see the artifact when running 

2562 # the transfer doesn't mean it will be generally 

2563 # accessible to a user of this butler. For now warn 

2564 # but assume it will be accessible. 

2565 log.warning( 

2566 "Transfer request for an outside-datastore artifact has been found at %s", 

2567 source_location, 

2568 ) 

2569 else: 

2570 # Need to transfer it to the new location. 

2571 # Assume we should always overwrite. If the artifact 

2572 # is there this might indicate that a previous transfer 

2573 # was interrupted but was not able to be rolled back 

2574 # completely (eg pre-emption) so follow Datastore default 

2575 # and overwrite. 

2576 target_location.uri.transfer_from( 

2577 source_location.uri, transfer=transfer, overwrite=True, transaction=self._transaction 

2578 ) 

2579 

2580 artifacts.append((target_ref, info)) 

2581 

2582 self._register_datasets(artifacts) 

2583 

2584 if already_present: 

2585 n_skipped = len(already_present) 

2586 log.info( 

2587 "Skipped transfer of %d dataset%s already present in datastore", 

2588 n_skipped, 

2589 "" if n_skipped == 1 else "s", 

2590 ) 

2591 

2592 @transactional 

2593 def forget(self, refs: Iterable[DatasetRef]) -> None: 

2594 # Docstring inherited. 

2595 refs = list(refs) 

2596 self.bridge.forget(refs) 

2597 self._table.delete(["dataset_id"], *[{"dataset_id": ref.getCheckedId()} for ref in refs]) 

2598 

2599 def validateConfiguration( 

2600 self, entities: Iterable[Union[DatasetRef, DatasetType, StorageClass]], logFailures: bool = False 

2601 ) -> None: 

2602 """Validate some of the configuration for this datastore. 

2603 

2604 Parameters 

2605 ---------- 

2606 entities : iterable of `DatasetRef`, `DatasetType`, or `StorageClass` 

2607 Entities to test against this configuration. Can be differing 

2608 types. 

2609 logFailures : `bool`, optional 

2610 If `True`, output a log message for every validation error 

2611 detected. 

2612 

2613 Raises 

2614 ------ 

2615 DatastoreValidationError 

2616 Raised if there is a validation problem with a configuration. 

2617 All the problems are reported in a single exception. 

2618 

2619 Notes 

2620 ----- 

2621 This method checks that all the supplied entities have valid file 

2622 templates and also have formatters defined. 

2623 """ 

2624 

2625 templateFailed = None 

2626 try: 

2627 self.templates.validateTemplates(entities, logFailures=logFailures) 

2628 except FileTemplateValidationError as e: 

2629 templateFailed = str(e) 

2630 

2631 formatterFailed = [] 

2632 for entity in entities: 

2633 try: 

2634 self.formatterFactory.getFormatterClass(entity) 

2635 except KeyError as e: 

2636 formatterFailed.append(str(e)) 

2637 if logFailures: 2637 ↛ 2632line 2637 didn't jump to line 2632, because the condition on line 2637 was never false

2638 log.critical("Formatter failure: %s", e) 

2639 

2640 if templateFailed or formatterFailed: 

2641 messages = [] 

2642 if templateFailed: 2642 ↛ 2643line 2642 didn't jump to line 2643, because the condition on line 2642 was never true

2643 messages.append(templateFailed) 

2644 if formatterFailed: 2644 ↛ 2646line 2644 didn't jump to line 2646, because the condition on line 2644 was never false

2645 messages.append(",".join(formatterFailed)) 

2646 msg = ";\n".join(messages) 

2647 raise DatastoreValidationError(msg) 

2648 

2649 def getLookupKeys(self) -> Set[LookupKey]: 

2650 # Docstring is inherited from base class 

2651 return ( 

2652 self.templates.getLookupKeys() 

2653 | self.formatterFactory.getLookupKeys() 

2654 | self.constraints.getLookupKeys() 

2655 ) 

2656 

2657 def validateKey(self, lookupKey: LookupKey, entity: Union[DatasetRef, DatasetType, StorageClass]) -> None: 

2658 # Docstring is inherited from base class 

2659 # The key can be valid in either formatters or templates so we can 

2660 # only check the template if it exists 

2661 if lookupKey in self.templates: 

2662 try: 

2663 self.templates[lookupKey].validateTemplate(entity) 

2664 except FileTemplateValidationError as e: 

2665 raise DatastoreValidationError(e) from e 

2666 

2667 def export( 

2668 self, 

2669 refs: Iterable[DatasetRef], 

2670 *, 

2671 directory: Optional[ResourcePathExpression] = None, 

2672 transfer: Optional[str] = "auto", 

2673 ) -> Iterable[FileDataset]: 

2674 # Docstring inherited from Datastore.export. 

2675 if transfer is not None and directory is None: 2675 ↛ 2676line 2675 didn't jump to line 2676, because the condition on line 2675 was never true

2676 raise RuntimeError(f"Cannot export using transfer mode {transfer} with no export directory given") 

2677 

2678 if transfer == "move": 2678 ↛ 2679line 2678 didn't jump to line 2679, because the condition on line 2678 was never true

2679 raise RuntimeError("Can not export by moving files out of datastore.") 

2680 elif transfer == "direct": 2680 ↛ 2684line 2680 didn't jump to line 2684, because the condition on line 2680 was never true

2681 # For an export, treat this as equivalent to None. We do not 

2682 # want an import to risk using absolute URIs to datasets owned 

2683 # by another datastore. 

2684 log.info("Treating 'direct' transfer mode as in-place export.") 

2685 transfer = None 

2686 

2687 # Force the directory to be a URI object 

2688 directoryUri: Optional[ResourcePath] = None 

2689 if directory is not None: 2689 ↛ 2692line 2689 didn't jump to line 2692, because the condition on line 2689 was never false

2690 directoryUri = ResourcePath(directory, forceDirectory=True) 

2691 

2692 if transfer is not None and directoryUri is not None: 2692 ↛ 2697line 2692 didn't jump to line 2697, because the condition on line 2692 was never false

2693 # mypy needs the second test 

2694 if not directoryUri.exists(): 2694 ↛ 2695line 2694 didn't jump to line 2695, because the condition on line 2694 was never true

2695 raise FileNotFoundError(f"Export location {directory} does not exist") 

2696 

2697 progress = Progress("lsst.daf.butler.datastores.FileDatastore.export", level=logging.DEBUG) 

2698 for ref in progress.wrap(refs, "Exporting dataset files"): 

2699 fileLocations = self._get_dataset_locations_info(ref) 

2700 if not fileLocations: 2700 ↛ 2701line 2700 didn't jump to line 2701, because the condition on line 2700 was never true

2701 raise FileNotFoundError(f"Could not retrieve dataset {ref}.") 

2702 # For now we can not export disassembled datasets 

2703 if len(fileLocations) > 1: 

2704 raise NotImplementedError(f"Can not export disassembled datasets such as {ref}") 

2705 location, storedFileInfo = fileLocations[0] 

2706 

2707 pathInStore = location.pathInStore.path 

2708 if transfer is None: 2708 ↛ 2712line 2708 didn't jump to line 2712, because the condition on line 2708 was never true

2709 # TODO: do we also need to return the readStorageClass somehow? 

2710 # We will use the path in store directly. If this is an 

2711 # absolute URI, preserve it. 

2712 if location.pathInStore.isabs(): 

2713 pathInStore = str(location.uri) 

2714 elif transfer == "direct": 2714 ↛ 2716line 2714 didn't jump to line 2716, because the condition on line 2714 was never true

2715 # Use full URIs to the remote store in the export 

2716 pathInStore = str(location.uri) 

2717 else: 

2718 # mypy needs help 

2719 assert directoryUri is not None, "directoryUri must be defined to get here" 

2720 storeUri = ResourcePath(location.uri) 

2721 

2722 # if the datastore has an absolute URI to a resource, we 

2723 # have two options: 

2724 # 1. Keep the absolute URI in the exported YAML 

2725 # 2. Allocate a new name in the local datastore and transfer 

2726 # it. 

2727 # For now go with option 2 

2728 if location.pathInStore.isabs(): 2728 ↛ 2729line 2728 didn't jump to line 2729, because the condition on line 2728 was never true

2729 template = self.templates.getTemplate(ref) 

2730 newURI = ResourcePath(template.format(ref), forceAbsolute=False) 

2731 pathInStore = str(newURI.updatedExtension(location.pathInStore.getExtension())) 

2732 

2733 exportUri = directoryUri.join(pathInStore) 

2734 exportUri.transfer_from(storeUri, transfer=transfer) 

2735 

2736 yield FileDataset(refs=[ref], path=pathInStore, formatter=storedFileInfo.formatter) 

2737 

2738 @staticmethod 

2739 def computeChecksum( 

2740 uri: ResourcePath, algorithm: str = "blake2b", block_size: int = 8192 

2741 ) -> Optional[str]: 

2742 """Compute the checksum of the supplied file. 

2743 

2744 Parameters 

2745 ---------- 

2746 uri : `lsst.resources.ResourcePath` 

2747 Name of resource to calculate checksum from. 

2748 algorithm : `str`, optional 

2749 Name of algorithm to use. Must be one of the algorithms supported 

2750 by :py:class`hashlib`. 

2751 block_size : `int` 

2752 Number of bytes to read from file at one time. 

2753 

2754 Returns 

2755 ------- 

2756 hexdigest : `str` 

2757 Hex digest of the file. 

2758 

2759 Notes 

2760 ----- 

2761 Currently returns None if the URI is for a remote resource. 

2762 """ 

2763 if algorithm not in hashlib.algorithms_guaranteed: 2763 ↛ 2764line 2763 didn't jump to line 2764, because the condition on line 2763 was never true

2764 raise NameError("The specified algorithm '{}' is not supported by hashlib".format(algorithm)) 

2765 

2766 if not uri.isLocal: 2766 ↛ 2767line 2766 didn't jump to line 2767, because the condition on line 2766 was never true

2767 return None 

2768 

2769 hasher = hashlib.new(algorithm) 

2770 

2771 with uri.as_local() as local_uri: 

2772 with open(local_uri.ospath, "rb") as f: 

2773 for chunk in iter(lambda: f.read(block_size), b""): 

2774 hasher.update(chunk) 

2775 

2776 return hasher.hexdigest() 

2777 

2778 def needs_expanded_data_ids( 

2779 self, 

2780 transfer: Optional[str], 

2781 entity: Optional[Union[DatasetRef, DatasetType, StorageClass]] = None, 

2782 ) -> bool: 

2783 # Docstring inherited. 

2784 # This _could_ also use entity to inspect whether the filename template 

2785 # involves placeholders other than the required dimensions for its 

2786 # dataset type, but that's not necessary for correctness; it just 

2787 # enables more optimizations (perhaps only in theory). 

2788 return transfer not in ("direct", None) 

2789 

2790 def import_records(self, data: Mapping[str, DatastoreRecordData]) -> None: 

2791 # Docstring inherited from the base class. 

2792 record_data = data.get(self.name) 

2793 if not record_data: 2793 ↛ 2794line 2793 didn't jump to line 2794, because the condition on line 2793 was never true

2794 return 

2795 

2796 self._bridge.insert(FakeDatasetRef(dataset_id) for dataset_id in record_data.records.keys()) 

2797 

2798 # TODO: Verify that there are no unexpected table names in the dict? 

2799 unpacked_records = [] 

2800 for dataset_data in record_data.records.values(): 

2801 records = dataset_data.get(self._table.name) 

2802 if records: 2802 ↛ 2800line 2802 didn't jump to line 2800, because the condition on line 2802 was never false

2803 for info in records: 

2804 assert isinstance(info, StoredFileInfo), "Expecting StoredFileInfo records" 

2805 unpacked_records.append(info.to_record()) 

2806 if unpacked_records: 

2807 self._table.insert(*unpacked_records) 

2808 

2809 def export_records(self, refs: Iterable[DatasetIdRef]) -> Mapping[str, DatastoreRecordData]: 

2810 # Docstring inherited from the base class. 

2811 exported_refs = list(self._bridge.check(refs)) 

2812 ids = {ref.getCheckedId() for ref in exported_refs} 

2813 records: defaultdict[DatasetId, defaultdict[str, List[StoredDatastoreItemInfo]]] = defaultdict( 

2814 lambda: defaultdict(list), {id: defaultdict(list) for id in ids} 

2815 ) 

2816 for row in self._table.fetch(dataset_id=ids): 

2817 info: StoredDatastoreItemInfo = StoredFileInfo.from_record(row) 

2818 records[info.dataset_id][self._table.name].append(info) 

2819 

2820 record_data = DatastoreRecordData(records=records) 

2821 return {self.name: record_data}