Coverage for python/lsst/daf/butler/datastores/fileDatastore.py : 78%

Hot-keys on this page
r m x p toggle line displays
j k next/prev highlighted chunk
0 (zero) top of page
1 (one) first highlighted chunk
1# This file is part of 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
23"""Generic file-based datastore code."""
25__all__ = ("FileDatastore", )
27import hashlib
28import logging
29import os
30import tempfile
32from sqlalchemy import BigInteger, String
34from collections import defaultdict
35from dataclasses import dataclass
36from typing import (
37 TYPE_CHECKING,
38 Any,
39 ClassVar,
40 Dict,
41 Iterable,
42 List,
43 Mapping,
44 Optional,
45 Set,
46 Tuple,
47 Type,
48 Union,
49)
51from lsst.daf.butler import (
52 ButlerURI,
53 CompositesMap,
54 Config,
55 FileDataset,
56 DatasetId,
57 DatasetRef,
58 DatasetType,
59 DatasetTypeNotSupportedError,
60 Datastore,
61 DatastoreCacheManager,
62 DatastoreDisabledCacheManager,
63 DatastoreConfig,
64 DatastoreValidationError,
65 FileDescriptor,
66 FileTemplates,
67 FileTemplateValidationError,
68 Formatter,
69 FormatterFactory,
70 Location,
71 LocationFactory,
72 Progress,
73 StorageClass,
74 StoredFileInfo,
75)
77from lsst.daf.butler import ddl
78from lsst.daf.butler.registry.interfaces import (
79 ReadOnlyDatabaseError,
80 DatastoreRegistryBridge,
81)
83from lsst.daf.butler.core.repoRelocation import replaceRoot
84from lsst.daf.butler.core.utils import getInstanceOf, getClassOf, transactional, time_this
85from .genericDatastore import GenericBaseDatastore
87if TYPE_CHECKING: 87 ↛ 88line 87 didn't jump to line 88, because the condition on line 87 was never true
88 from lsst.daf.butler import LookupKey, AbstractDatastoreCacheManager
89 from lsst.daf.butler.registry.interfaces import DatasetIdRef, DatastoreRegistryBridgeManager
91log = logging.getLogger(__name__)
94class _IngestPrepData(Datastore.IngestPrepData):
95 """Helper class for FileDatastore ingest implementation.
97 Parameters
98 ----------
99 datasets : `list` of `FileDataset`
100 Files to be ingested by this datastore.
101 """
102 def __init__(self, datasets: List[FileDataset]):
103 super().__init__(ref for dataset in datasets for ref in dataset.refs)
104 self.datasets = datasets
107@dataclass(frozen=True)
108class DatastoreFileGetInformation:
109 """Collection of useful parameters needed to retrieve a file from
110 a Datastore.
111 """
113 location: Location
114 """The location from which to read the dataset."""
116 formatter: Formatter
117 """The `Formatter` to use to deserialize the dataset."""
119 info: StoredFileInfo
120 """Stored information about this file and its formatter."""
122 assemblerParams: Dict[str, Any]
123 """Parameters to use for post-processing the retrieved dataset."""
125 formatterParams: Dict[str, Any]
126 """Parameters that were understood by the associated formatter."""
128 component: Optional[str]
129 """The component to be retrieved (can be `None`)."""
131 readStorageClass: StorageClass
132 """The `StorageClass` of the dataset being read."""
135class FileDatastore(GenericBaseDatastore):
136 """Generic Datastore for file-based implementations.
138 Should always be sub-classed since key abstract methods are missing.
140 Parameters
141 ----------
142 config : `DatastoreConfig` or `str`
143 Configuration as either a `Config` object or URI to file.
144 bridgeManager : `DatastoreRegistryBridgeManager`
145 Object that manages the interface between `Registry` and datastores.
146 butlerRoot : `str`, optional
147 New datastore root to use to override the configuration value.
149 Raises
150 ------
151 ValueError
152 If root location does not exist and ``create`` is `False` in the
153 configuration.
154 """
156 defaultConfigFile: ClassVar[Optional[str]] = None
157 """Path to configuration defaults. Accessed within the ``config`` resource
158 or relative to a search path. Can be None if no defaults specified.
159 """
161 root: ButlerURI
162 """Root directory URI of this `Datastore`."""
164 locationFactory: LocationFactory
165 """Factory for creating locations relative to the datastore root."""
167 formatterFactory: FormatterFactory
168 """Factory for creating instances of formatters."""
170 templates: FileTemplates
171 """File templates that can be used by this `Datastore`."""
173 composites: CompositesMap
174 """Determines whether a dataset should be disassembled on put."""
176 defaultConfigFile = "datastores/fileDatastore.yaml"
177 """Path to configuration defaults. Accessed within the ``config`` resource
178 or relative to a search path. Can be None if no defaults specified.
179 """
181 @classmethod
182 def setConfigRoot(cls, root: str, config: Config, full: Config, overwrite: bool = True) -> None:
183 """Set any filesystem-dependent config options for this Datastore to
184 be appropriate for a new empty repository with the given root.
186 Parameters
187 ----------
188 root : `str`
189 URI to the root of the data repository.
190 config : `Config`
191 A `Config` to update. Only the subset understood by
192 this component will be updated. Will not expand
193 defaults.
194 full : `Config`
195 A complete config with all defaults expanded that can be
196 converted to a `DatastoreConfig`. Read-only and will not be
197 modified by this method.
198 Repository-specific options that should not be obtained
199 from defaults when Butler instances are constructed
200 should be copied from ``full`` to ``config``.
201 overwrite : `bool`, optional
202 If `False`, do not modify a value in ``config`` if the value
203 already exists. Default is always to overwrite with the provided
204 ``root``.
206 Notes
207 -----
208 If a keyword is explicitly defined in the supplied ``config`` it
209 will not be overridden by this method if ``overwrite`` is `False`.
210 This allows explicit values set in external configs to be retained.
211 """
212 Config.updateParameters(DatastoreConfig, config, full,
213 toUpdate={"root": root},
214 toCopy=("cls", ("records", "table")), overwrite=overwrite)
216 @classmethod
217 def makeTableSpec(cls, datasetIdColumnType: type) -> ddl.TableSpec:
218 return ddl.TableSpec(
219 fields=[
220 ddl.FieldSpec(name="dataset_id", dtype=datasetIdColumnType, primaryKey=True),
221 ddl.FieldSpec(name="path", dtype=String, length=256, nullable=False),
222 ddl.FieldSpec(name="formatter", dtype=String, length=128, nullable=False),
223 ddl.FieldSpec(name="storage_class", dtype=String, length=64, nullable=False),
224 # Use empty string to indicate no component
225 ddl.FieldSpec(name="component", dtype=String, length=32, primaryKey=True),
226 # TODO: should checksum be Base64Bytes instead?
227 ddl.FieldSpec(name="checksum", dtype=String, length=128, nullable=True),
228 ddl.FieldSpec(name="file_size", dtype=BigInteger, nullable=True),
229 ],
230 unique=frozenset(),
231 indexes=[tuple(["path"])],
232 )
234 def __init__(self, config: Union[DatastoreConfig, str],
235 bridgeManager: DatastoreRegistryBridgeManager, butlerRoot: str = None):
236 super().__init__(config, bridgeManager)
237 if "root" not in self.config: 237 ↛ 238line 237 didn't jump to line 238, because the condition on line 237 was never true
238 raise ValueError("No root directory specified in configuration")
240 # Name ourselves either using an explicit name or a name
241 # derived from the (unexpanded) root
242 if "name" in self.config:
243 self.name = self.config["name"]
244 else:
245 # We use the unexpanded root in the name to indicate that this
246 # datastore can be moved without having to update registry.
247 self.name = "{}@{}".format(type(self).__name__,
248 self.config["root"])
250 # Support repository relocation in config
251 # Existence of self.root is checked in subclass
252 self.root = ButlerURI(replaceRoot(self.config["root"], butlerRoot),
253 forceDirectory=True, forceAbsolute=True)
255 self.locationFactory = LocationFactory(self.root)
256 self.formatterFactory = FormatterFactory()
258 # Now associate formatters with storage classes
259 self.formatterFactory.registerFormatters(self.config["formatters"],
260 universe=bridgeManager.universe)
262 # Read the file naming templates
263 self.templates = FileTemplates(self.config["templates"],
264 universe=bridgeManager.universe)
266 # See if composites should be disassembled
267 self.composites = CompositesMap(self.config["composites"],
268 universe=bridgeManager.universe)
270 tableName = self.config["records", "table"]
271 try:
272 # Storage of paths and formatters, keyed by dataset_id
273 self._table = bridgeManager.opaque.register(
274 tableName, self.makeTableSpec(bridgeManager.datasetIdColumnType))
275 # Interface to Registry.
276 self._bridge = bridgeManager.register(self.name)
277 except ReadOnlyDatabaseError:
278 # If the database is read only and we just tried and failed to
279 # create a table, it means someone is trying to create a read-only
280 # butler client for an empty repo. That should be okay, as long
281 # as they then try to get any datasets before some other client
282 # creates the table. Chances are they'rejust validating
283 # configuration.
284 pass
286 # Determine whether checksums should be used - default to False
287 self.useChecksum = self.config.get("checksum", False)
289 # Determine whether we can fall back to configuration if a
290 # requested dataset is not known to registry
291 self.trustGetRequest = self.config.get("trust_get_request", False)
293 # Create a cache manager
294 self.cacheManager: AbstractDatastoreCacheManager
295 if "cached" in self.config: 295 ↛ 299line 295 didn't jump to line 299, because the condition on line 295 was never false
296 self.cacheManager = DatastoreCacheManager(self.config["cached"],
297 universe=bridgeManager.universe)
298 else:
299 self.cacheManager = DatastoreDisabledCacheManager("",
300 universe=bridgeManager.universe)
302 # Check existence and create directory structure if necessary
303 if not self.root.exists():
304 if "create" not in self.config or not self.config["create"]: 304 ↛ 305line 304 didn't jump to line 305, because the condition on line 304 was never true
305 raise ValueError(f"No valid root and not allowed to create one at: {self.root}")
306 try:
307 self.root.mkdir()
308 except Exception as e:
309 raise ValueError(f"Can not create datastore root '{self.root}', check permissions."
310 f" Got error: {e}") from e
312 def __str__(self) -> str:
313 return str(self.root)
315 @property
316 def bridge(self) -> DatastoreRegistryBridge:
317 return self._bridge
319 def _artifact_exists(self, location: Location) -> bool:
320 """Check that an artifact exists in this datastore at the specified
321 location.
323 Parameters
324 ----------
325 location : `Location`
326 Expected location of the artifact associated with this datastore.
328 Returns
329 -------
330 exists : `bool`
331 True if the location can be found, false otherwise.
332 """
333 log.debug("Checking if resource exists: %s", location.uri)
334 return location.uri.exists()
336 def _delete_artifact(self, location: Location) -> None:
337 """Delete the artifact from the datastore.
339 Parameters
340 ----------
341 location : `Location`
342 Location of the artifact associated with this datastore.
343 """
344 if location.pathInStore.isabs(): 344 ↛ 345line 344 didn't jump to line 345, because the condition on line 344 was never true
345 raise RuntimeError(f"Cannot delete artifact with absolute uri {location.uri}.")
347 try:
348 location.uri.remove()
349 except FileNotFoundError:
350 log.debug("File %s did not exist and so could not be deleted.", location.uri)
351 raise
352 except Exception as e:
353 log.critical("Failed to delete file: %s (%s)", location.uri, e)
354 raise
355 log.debug("Successfully deleted file: %s", location.uri)
357 def addStoredItemInfo(self, refs: Iterable[DatasetRef], infos: Iterable[StoredFileInfo]) -> None:
358 # Docstring inherited from GenericBaseDatastore
359 records = [info.to_record(ref) for ref, info in zip(refs, infos)]
360 self._table.insert(*records)
362 def getStoredItemsInfo(self, ref: DatasetIdRef) -> List[StoredFileInfo]:
363 # Docstring inherited from GenericBaseDatastore
365 # Look for the dataset_id -- there might be multiple matches
366 # if we have disassembled the dataset.
367 records = self._table.fetch(dataset_id=ref.id)
368 return [StoredFileInfo.from_record(record) for record in records]
370 def _get_stored_records_associated_with_refs(self,
371 refs: Iterable[DatasetIdRef]
372 ) -> Dict[DatasetId, List[StoredFileInfo]]:
373 """Retrieve all records associated with the provided refs.
375 Parameters
376 ----------
377 refs : iterable of `DatasetIdRef`
378 The refs for which records are to be retrieved.
380 Returns
381 -------
382 records : `dict` of [`DatasetId`, `list` of `StoredFileInfo`]
383 The matching records indexed by the ref ID. The number of entries
384 in the dict can be smaller than the number of requested refs.
385 """
386 records = self._table.fetch(dataset_id=[ref.id for ref in refs])
388 # Uniqueness is dataset_id + component so can have multiple records
389 # per ref.
390 records_by_ref = defaultdict(list)
391 for record in records:
392 records_by_ref[record["dataset_id"]].append(StoredFileInfo.from_record(record))
393 return records_by_ref
395 def _refs_associated_with_artifacts(self, paths: List[Union[str, ButlerURI]]) -> Dict[str,
396 Set[DatasetId]]:
397 """Return paths and associated dataset refs.
399 Parameters
400 ----------
401 paths : `list` of `str` or `ButlerURI`
402 All the paths to include in search.
404 Returns
405 -------
406 mapping : `dict` of [`str`, `set` [`DatasetId`]]
407 Mapping of each path to a set of associated database IDs.
408 """
409 records = self._table.fetch(path=[str(path) for path in paths])
410 result = defaultdict(set)
411 for row in records:
412 result[row["path"]].add(row["dataset_id"])
413 return result
415 def _registered_refs_per_artifact(self, pathInStore: ButlerURI) -> Set[DatasetId]:
416 """Return all dataset refs associated with the supplied path.
418 Parameters
419 ----------
420 pathInStore : `ButlerURI`
421 Path of interest in the data store.
423 Returns
424 -------
425 ids : `set` of `int`
426 All `DatasetRef` IDs associated with this path.
427 """
428 records = list(self._table.fetch(path=str(pathInStore)))
429 ids = {r["dataset_id"] for r in records}
430 return ids
432 def removeStoredItemInfo(self, ref: DatasetIdRef) -> None:
433 # Docstring inherited from GenericBaseDatastore
434 self._table.delete(["dataset_id"], {"dataset_id": ref.id})
436 def _get_dataset_locations_info(self, ref: DatasetIdRef) -> List[Tuple[Location, StoredFileInfo]]:
437 r"""Find all the `Location`\ s of the requested dataset in the
438 `Datastore` and the associated stored file information.
440 Parameters
441 ----------
442 ref : `DatasetRef`
443 Reference to the required `Dataset`.
445 Returns
446 -------
447 results : `list` [`tuple` [`Location`, `StoredFileInfo` ]]
448 Location of the dataset within the datastore and
449 stored information about each file and its formatter.
450 """
451 # Get the file information (this will fail if no file)
452 records = self.getStoredItemsInfo(ref)
454 # Use the path to determine the location -- we need to take
455 # into account absolute URIs in the datastore record
456 return [(r.file_location(self.locationFactory), r) for r in records]
458 def _can_remove_dataset_artifact(self, ref: DatasetIdRef, location: Location) -> bool:
459 """Check that there is only one dataset associated with the
460 specified artifact.
462 Parameters
463 ----------
464 ref : `DatasetRef` or `FakeDatasetRef`
465 Dataset to be removed.
466 location : `Location`
467 The location of the artifact to be removed.
469 Returns
470 -------
471 can_remove : `Bool`
472 True if the artifact can be safely removed.
473 """
474 # Can't ever delete absolute URIs.
475 if location.pathInStore.isabs():
476 return False
478 # Get all entries associated with this path
479 allRefs = self._registered_refs_per_artifact(location.pathInStore)
480 if not allRefs:
481 raise RuntimeError(f"Datastore inconsistency error. {location.pathInStore} not in registry")
483 # Remove these refs from all the refs and if there is nothing left
484 # then we can delete
485 remainingRefs = allRefs - {ref.id}
487 if remainingRefs:
488 return False
489 return True
491 def _get_expected_dataset_locations_info(self, ref: DatasetRef) -> List[Tuple[Location,
492 StoredFileInfo]]:
493 """Predict the location and related file information of the requested
494 dataset in this datastore.
496 Parameters
497 ----------
498 ref : `DatasetRef`
499 Reference to the required `Dataset`.
501 Returns
502 -------
503 results : `list` [`tuple` [`Location`, `StoredFileInfo` ]]
504 Expected Location of the dataset within the datastore and
505 placeholder information about each file and its formatter.
507 Notes
508 -----
509 Uses the current configuration to determine how we would expect the
510 datastore files to have been written if we couldn't ask registry.
511 This is safe so long as there has been no change to datastore
512 configuration between writing the dataset and wanting to read it.
513 Will not work for files that have been ingested without using the
514 standard file template or default formatter.
515 """
517 # If we have a component ref we always need to ask the questions
518 # of the composite. If the composite is disassembled this routine
519 # should return all components. If the composite was not
520 # disassembled the composite is what is stored regardless of
521 # component request. Note that if the caller has disassembled
522 # a composite there is no way for this guess to know that
523 # without trying both the composite and component ref and seeing
524 # if there is something at the component Location even without
525 # disassembly being enabled.
526 if ref.datasetType.isComponent():
527 ref = ref.makeCompositeRef()
529 # See if the ref is a composite that should be disassembled
530 doDisassembly = self.composites.shouldBeDisassembled(ref)
532 all_info: List[Tuple[Location, Formatter, StorageClass, Optional[str]]] = []
534 if doDisassembly:
535 for component, componentStorage in ref.datasetType.storageClass.components.items():
536 compRef = ref.makeComponentRef(component)
537 location, formatter = self._determine_put_formatter_location(compRef)
538 all_info.append((location, formatter, componentStorage, component))
540 else:
541 # Always use the composite ref if no disassembly
542 location, formatter = self._determine_put_formatter_location(ref)
543 all_info.append((location, formatter, ref.datasetType.storageClass, None))
545 # Convert the list of tuples to have StoredFileInfo as second element
546 return [(location, StoredFileInfo(formatter=formatter,
547 path=location.pathInStore.path,
548 storageClass=storageClass,
549 component=component,
550 checksum=None,
551 file_size=-1))
552 for location, formatter, storageClass, component in all_info]
554 def _prepare_for_get(self, ref: DatasetRef,
555 parameters: Optional[Mapping[str, Any]] = None) -> List[DatastoreFileGetInformation]:
556 """Check parameters for ``get`` and obtain formatter and
557 location.
559 Parameters
560 ----------
561 ref : `DatasetRef`
562 Reference to the required Dataset.
563 parameters : `dict`
564 `StorageClass`-specific parameters that specify, for example,
565 a slice of the dataset to be loaded.
567 Returns
568 -------
569 getInfo : `list` [`DatastoreFileGetInformation`]
570 Parameters needed to retrieve each file.
571 """
572 log.debug("Retrieve %s from %s with parameters %s", ref, self.name, parameters)
574 # Get file metadata and internal metadata
575 fileLocations = self._get_dataset_locations_info(ref)
576 if not fileLocations:
577 if not self.trustGetRequest:
578 raise FileNotFoundError(f"Could not retrieve dataset {ref}.")
579 # Assume the dataset is where we think it should be
580 fileLocations = self._get_expected_dataset_locations_info(ref)
582 # The storage class we want to use eventually
583 refStorageClass = ref.datasetType.storageClass
585 if len(fileLocations) > 1:
586 disassembled = True
588 # If trust is involved it is possible that there will be
589 # components listed here that do not exist in the datastore.
590 # Explicitly check for file artifact existence and filter out any
591 # that are missing.
592 if self.trustGetRequest:
593 fileLocations = [loc for loc in fileLocations if loc[0].uri.exists()]
595 # For now complain only if we have no components at all. One
596 # component is probably a problem but we can punt that to the
597 # assembler.
598 if not fileLocations: 598 ↛ 599line 598 didn't jump to line 599, because the condition on line 598 was never true
599 raise FileNotFoundError(f"None of the component files for dataset {ref} exist.")
601 else:
602 disassembled = False
604 # Is this a component request?
605 refComponent = ref.datasetType.component()
607 fileGetInfo = []
608 for location, storedFileInfo in fileLocations:
610 # The storage class used to write the file
611 writeStorageClass = storedFileInfo.storageClass
613 # If this has been disassembled we need read to match the write
614 if disassembled:
615 readStorageClass = writeStorageClass
616 else:
617 readStorageClass = refStorageClass
619 formatter = getInstanceOf(storedFileInfo.formatter,
620 FileDescriptor(location, readStorageClass=readStorageClass,
621 storageClass=writeStorageClass, parameters=parameters),
622 ref.dataId)
624 formatterParams, notFormatterParams = formatter.segregateParameters()
626 # Of the remaining parameters, extract the ones supported by
627 # this StorageClass (for components not all will be handled)
628 assemblerParams = readStorageClass.filterParameters(notFormatterParams)
630 # The ref itself could be a component if the dataset was
631 # disassembled by butler, or we disassembled in datastore and
632 # components came from the datastore records
633 component = storedFileInfo.component if storedFileInfo.component else refComponent
635 fileGetInfo.append(DatastoreFileGetInformation(location, formatter, storedFileInfo,
636 assemblerParams, formatterParams,
637 component, readStorageClass))
639 return fileGetInfo
641 def _prepare_for_put(self, inMemoryDataset: Any, ref: DatasetRef) -> Tuple[Location, Formatter]:
642 """Check the arguments for ``put`` and obtain formatter and
643 location.
645 Parameters
646 ----------
647 inMemoryDataset : `object`
648 The dataset to store.
649 ref : `DatasetRef`
650 Reference to the associated Dataset.
652 Returns
653 -------
654 location : `Location`
655 The location to write the dataset.
656 formatter : `Formatter`
657 The `Formatter` to use to write the dataset.
659 Raises
660 ------
661 TypeError
662 Supplied object and storage class are inconsistent.
663 DatasetTypeNotSupportedError
664 The associated `DatasetType` is not handled by this datastore.
665 """
666 self._validate_put_parameters(inMemoryDataset, ref)
667 return self._determine_put_formatter_location(ref)
669 def _determine_put_formatter_location(self, ref: DatasetRef) -> Tuple[Location, Formatter]:
670 """Calculate the formatter and output location to use for put.
672 Parameters
673 ----------
674 ref : `DatasetRef`
675 Reference to the associated Dataset.
677 Returns
678 -------
679 location : `Location`
680 The location to write the dataset.
681 formatter : `Formatter`
682 The `Formatter` to use to write the dataset.
683 """
684 # Work out output file name
685 try:
686 template = self.templates.getTemplate(ref)
687 except KeyError as e:
688 raise DatasetTypeNotSupportedError(f"Unable to find template for {ref}") from e
690 # Validate the template to protect against filenames from different
691 # dataIds returning the same and causing overwrite confusion.
692 template.validateTemplate(ref)
694 location = self.locationFactory.fromPath(template.format(ref))
696 # Get the formatter based on the storage class
697 storageClass = ref.datasetType.storageClass
698 try:
699 formatter = self.formatterFactory.getFormatter(ref,
700 FileDescriptor(location,
701 storageClass=storageClass),
702 ref.dataId)
703 except KeyError as e:
704 raise DatasetTypeNotSupportedError(f"Unable to find formatter for {ref} in datastore "
705 f"{self.name}") from e
707 # Now that we know the formatter, update the location
708 location = formatter.makeUpdatedLocation(location)
710 return location, formatter
712 def _overrideTransferMode(self, *datasets: FileDataset, transfer: Optional[str] = None) -> Optional[str]:
713 # Docstring inherited from base class
714 if transfer != "auto":
715 return transfer
717 # See if the paths are within the datastore or not
718 inside = [self._pathInStore(d.path) is not None for d in datasets]
720 if all(inside):
721 transfer = None
722 elif not any(inside): 722 ↛ 731line 722 didn't jump to line 731, because the condition on line 722 was never false
723 # Allow ButlerURI to use its own knowledge
724 transfer = "auto"
725 else:
726 # This can happen when importing from a datastore that
727 # has had some datasets ingested using "direct" mode.
728 # Also allow ButlerURI to sort it out but warn about it.
729 # This can happen if you are importing from a datastore
730 # that had some direct transfer datasets.
731 log.warning("Some datasets are inside the datastore and some are outside. Using 'split' "
732 "transfer mode. This assumes that the files outside the datastore are "
733 "still accessible to the new butler since they will not be copied into "
734 "the target datastore.")
735 transfer = "split"
737 return transfer
739 def _pathInStore(self, path: Union[str, ButlerURI]) -> Optional[str]:
740 """Return path relative to datastore root
742 Parameters
743 ----------
744 path : `str` or `ButlerURI`
745 Path to dataset. Can be absolute URI. If relative assumed to
746 be relative to the datastore. Returns path in datastore
747 or raises an exception if the path it outside.
749 Returns
750 -------
751 inStore : `str`
752 Path relative to datastore root. Returns `None` if the file is
753 outside the root.
754 """
755 # Relative path will always be relative to datastore
756 pathUri = ButlerURI(path, forceAbsolute=False)
757 return pathUri.relative_to(self.root)
759 def _standardizeIngestPath(self, path: Union[str, ButlerURI], *,
760 transfer: Optional[str] = None) -> Union[str, ButlerURI]:
761 """Standardize the path of a to-be-ingested file.
763 Parameters
764 ----------
765 path : `str` or `ButlerURI`
766 Path of a file to be ingested.
767 transfer : `str`, optional
768 How (and whether) the dataset should be added to the datastore.
769 See `ingest` for details of transfer modes.
770 This implementation is provided only so
771 `NotImplementedError` can be raised if the mode is not supported;
772 actual transfers are deferred to `_extractIngestInfo`.
774 Returns
775 -------
776 path : `str` or `ButlerURI`
777 New path in what the datastore considers standard form. If an
778 absolute URI was given that will be returned unchanged.
780 Notes
781 -----
782 Subclasses of `FileDatastore` can implement this method instead
783 of `_prepIngest`. It should not modify the data repository or given
784 file in any way.
786 Raises
787 ------
788 NotImplementedError
789 Raised if the datastore does not support the given transfer mode
790 (including the case where ingest is not supported at all).
791 FileNotFoundError
792 Raised if one of the given files does not exist.
793 """
794 if transfer not in (None, "direct", "split") + self.root.transferModes: 794 ↛ 795line 794 didn't jump to line 795, because the condition on line 794 was never true
795 raise NotImplementedError(f"Transfer mode {transfer} not supported.")
797 # A relative URI indicates relative to datastore root
798 srcUri = ButlerURI(path, forceAbsolute=False)
799 if not srcUri.isabs():
800 srcUri = self.root.join(path)
802 if not srcUri.exists():
803 raise FileNotFoundError(f"Resource at {srcUri} does not exist; note that paths to ingest "
804 f"are assumed to be relative to {self.root} unless they are absolute.")
806 if transfer is None:
807 relpath = srcUri.relative_to(self.root)
808 if not relpath:
809 raise RuntimeError(f"Transfer is none but source file ({srcUri}) is not "
810 f"within datastore ({self.root})")
812 # Return the relative path within the datastore for internal
813 # transfer
814 path = relpath
816 return path
818 def _extractIngestInfo(self, path: Union[str, ButlerURI], ref: DatasetRef, *,
819 formatter: Union[Formatter, Type[Formatter]],
820 transfer: Optional[str] = None) -> StoredFileInfo:
821 """Relocate (if necessary) and extract `StoredFileInfo` from a
822 to-be-ingested file.
824 Parameters
825 ----------
826 path : `str` or `ButlerURI`
827 URI or path of a file to be ingested.
828 ref : `DatasetRef`
829 Reference for the dataset being ingested. Guaranteed to have
830 ``dataset_id not None`.
831 formatter : `type` or `Formatter`
832 `Formatter` subclass to use for this dataset or an instance.
833 transfer : `str`, optional
834 How (and whether) the dataset should be added to the datastore.
835 See `ingest` for details of transfer modes.
837 Returns
838 -------
839 info : `StoredFileInfo`
840 Internal datastore record for this file. This will be inserted by
841 the caller; the `_extractIngestInfo` is only resposible for
842 creating and populating the struct.
844 Raises
845 ------
846 FileNotFoundError
847 Raised if one of the given files does not exist.
848 FileExistsError
849 Raised if transfer is not `None` but the (internal) location the
850 file would be moved to is already occupied.
851 """
852 if self._transaction is None: 852 ↛ 853line 852 didn't jump to line 853, because the condition on line 852 was never true
853 raise RuntimeError("Ingest called without transaction enabled")
855 # Create URI of the source path, do not need to force a relative
856 # path to absolute.
857 srcUri = ButlerURI(path, forceAbsolute=False)
859 # Track whether we have read the size of the source yet
860 have_sized = False
862 tgtLocation: Optional[Location]
863 if transfer is None or transfer == "split":
864 # A relative path is assumed to be relative to the datastore
865 # in this context
866 if not srcUri.isabs():
867 tgtLocation = self.locationFactory.fromPath(srcUri.ospath)
868 else:
869 # Work out the path in the datastore from an absolute URI
870 # This is required to be within the datastore.
871 pathInStore = srcUri.relative_to(self.root)
872 if pathInStore is None and transfer is None: 872 ↛ 873line 872 didn't jump to line 873, because the condition on line 872 was never true
873 raise RuntimeError(f"Unexpectedly learned that {srcUri} is "
874 f"not within datastore {self.root}")
875 if pathInStore: 875 ↛ 877line 875 didn't jump to line 877, because the condition on line 875 was never false
876 tgtLocation = self.locationFactory.fromPath(pathInStore)
877 elif transfer == "split":
878 # Outside the datastore but treat that as a direct ingest
879 # instead.
880 tgtLocation = None
881 else:
882 raise RuntimeError(f"Unexpected transfer mode encountered: {transfer} for"
883 f" URI {srcUri}")
884 elif transfer == "direct": 884 ↛ 889line 884 didn't jump to line 889, because the condition on line 884 was never true
885 # Want to store the full URI to the resource directly in
886 # datastore. This is useful for referring to permanent archive
887 # storage for raw data.
888 # Trust that people know what they are doing.
889 tgtLocation = None
890 else:
891 # Work out the name we want this ingested file to have
892 # inside the datastore
893 tgtLocation = self._calculate_ingested_datastore_name(srcUri, ref, formatter)
894 if not tgtLocation.uri.dirname().exists():
895 log.debug("Folder %s does not exist yet.", tgtLocation.uri.dirname())
896 tgtLocation.uri.dirname().mkdir()
898 # if we are transferring from a local file to a remote location
899 # it may be more efficient to get the size and checksum of the
900 # local file rather than the transferred one
901 if not srcUri.scheme or srcUri.scheme == "file": 901 ↛ 907line 901 didn't jump to line 907, because the condition on line 901 was never false
902 size = srcUri.size()
903 checksum = self.computeChecksum(srcUri) if self.useChecksum else None
904 have_sized = True
906 # transfer the resource to the destination
907 tgtLocation.uri.transfer_from(srcUri, transfer=transfer, transaction=self._transaction)
909 if tgtLocation is None: 909 ↛ 911line 909 didn't jump to line 911, because the condition on line 909 was never true
910 # This means we are using direct mode
911 targetUri = srcUri
912 targetPath = str(srcUri)
913 else:
914 targetUri = tgtLocation.uri
915 targetPath = tgtLocation.pathInStore.path
917 # the file should exist in the datastore now
918 if not have_sized:
919 size = targetUri.size()
920 checksum = self.computeChecksum(targetUri) if self.useChecksum else None
922 return StoredFileInfo(formatter=formatter, path=targetPath,
923 storageClass=ref.datasetType.storageClass,
924 component=ref.datasetType.component(),
925 file_size=size, checksum=checksum)
927 def _prepIngest(self, *datasets: FileDataset, transfer: Optional[str] = None) -> _IngestPrepData:
928 # Docstring inherited from Datastore._prepIngest.
929 filtered = []
930 for dataset in datasets:
931 acceptable = [ref for ref in dataset.refs if self.constraints.isAcceptable(ref)]
932 if not acceptable:
933 continue
934 else:
935 dataset.refs = acceptable
936 if dataset.formatter is None:
937 dataset.formatter = self.formatterFactory.getFormatterClass(dataset.refs[0])
938 else:
939 assert isinstance(dataset.formatter, (type, str))
940 dataset.formatter = getClassOf(dataset.formatter)
941 dataset.path = self._standardizeIngestPath(dataset.path, transfer=transfer)
942 filtered.append(dataset)
943 return _IngestPrepData(filtered)
945 @transactional
946 def _finishIngest(self, prepData: Datastore.IngestPrepData, *, transfer: Optional[str] = None) -> None:
947 # Docstring inherited from Datastore._finishIngest.
948 refsAndInfos = []
949 progress = Progress("lsst.daf.butler.datastores.FileDatastore.ingest", level=logging.DEBUG)
950 for dataset in progress.wrap(prepData.datasets, desc="Ingesting dataset files"):
951 # Do ingest as if the first dataset ref is associated with the file
952 info = self._extractIngestInfo(dataset.path, dataset.refs[0], formatter=dataset.formatter,
953 transfer=transfer)
954 refsAndInfos.extend([(ref, info) for ref in dataset.refs])
955 self._register_datasets(refsAndInfos)
957 def _calculate_ingested_datastore_name(self, srcUri: ButlerURI, ref: DatasetRef,
958 formatter: Union[Formatter, Type[Formatter]]) -> Location:
959 """Given a source URI and a DatasetRef, determine the name the
960 dataset will have inside datastore.
962 Parameters
963 ----------
964 srcUri : `ButlerURI`
965 URI to the source dataset file.
966 ref : `DatasetRef`
967 Ref associated with the newly-ingested dataset artifact. This
968 is used to determine the name within the datastore.
969 formatter : `Formatter` or Formatter class.
970 Formatter to use for validation. Can be a class or an instance.
972 Returns
973 -------
974 location : `Location`
975 Target location for the newly-ingested dataset.
976 """
977 # Ingesting a file from outside the datastore.
978 # This involves a new name.
979 template = self.templates.getTemplate(ref)
980 location = self.locationFactory.fromPath(template.format(ref))
982 # Get the extension
983 ext = srcUri.getExtension()
985 # Update the destination to include that extension
986 location.updateExtension(ext)
988 # Ask the formatter to validate this extension
989 formatter.validateExtension(location)
991 return location
993 def _write_in_memory_to_artifact(self, inMemoryDataset: Any, ref: DatasetRef) -> StoredFileInfo:
994 """Write out in memory dataset to datastore.
996 Parameters
997 ----------
998 inMemoryDataset : `object`
999 Dataset to write to datastore.
1000 ref : `DatasetRef`
1001 Registry information associated with this dataset.
1003 Returns
1004 -------
1005 info : `StoredFileInfo`
1006 Information describin the artifact written to the datastore.
1007 """
1008 location, formatter = self._prepare_for_put(inMemoryDataset, ref)
1009 uri = location.uri
1011 if not uri.dirname().exists():
1012 log.debug("Folder %s does not exist yet so creating it.", uri.dirname())
1013 uri.dirname().mkdir()
1015 if self._transaction is None: 1015 ↛ 1016line 1015 didn't jump to line 1016, because the condition on line 1015 was never true
1016 raise RuntimeError("Attempting to write artifact without transaction enabled")
1018 def _removeFileExists(uri: ButlerURI) -> None:
1019 """Remove a file and do not complain if it is not there.
1021 This is important since a formatter might fail before the file
1022 is written and we should not confuse people by writing spurious
1023 error messages to the log.
1024 """
1025 try:
1026 uri.remove()
1027 except FileNotFoundError:
1028 pass
1030 # Register a callback to try to delete the uploaded data if
1031 # something fails below
1032 self._transaction.registerUndo("artifactWrite", _removeFileExists, uri)
1034 # For a local file, simply use the formatter directly
1035 if uri.isLocal:
1036 try:
1037 formatter.write(inMemoryDataset)
1038 except Exception as e:
1039 raise RuntimeError(f"Failed to serialize dataset {ref} of type {type(inMemoryDataset)} "
1040 f"to location {uri}") from e
1041 log.debug("Successfully wrote python object to local file at %s", uri)
1042 else:
1043 # This is a remote URI, so first try bytes and write directly else
1044 # fallback to a temporary file
1045 try:
1046 serializedDataset = formatter.toBytes(inMemoryDataset)
1047 except NotImplementedError: 1047 ↛ 1066line 1047 didn't jump to line 1066
1048 with tempfile.NamedTemporaryFile(suffix=uri.getExtension()) as tmpFile:
1049 # Need to configure the formatter to write to a different
1050 # location and that needs us to overwrite internals
1051 tmpLocation = Location(*os.path.split(tmpFile.name))
1052 log.debug("Writing dataset to temporary location at %s", tmpLocation.uri)
1053 with formatter._updateLocation(tmpLocation):
1054 try:
1055 formatter.write(inMemoryDataset)
1056 except Exception as e:
1057 raise RuntimeError(f"Failed to serialize dataset {ref} of type"
1058 f" {type(inMemoryDataset)} to "
1059 f"temporary location {tmpLocation.uri}") from e
1060 uri.transfer_from(tmpLocation.uri, transfer="copy", overwrite=True)
1062 # Cache if required
1063 self.cacheManager.move_to_cache(tmpLocation.uri, ref)
1065 log.debug("Successfully wrote dataset to %s via a temporary file.", uri)
1066 except Exception as e:
1067 raise RuntimeError(f"Failed to serialize dataset {ref} to bytes.") from e
1068 else:
1069 log.debug("Writing bytes directly to %s", uri)
1070 uri.write(serializedDataset, overwrite=True)
1071 log.debug("Successfully wrote bytes directly to %s", uri)
1073 # URI is needed to resolve what ingest case are we dealing with
1074 return self._extractIngestInfo(uri, ref, formatter=formatter)
1076 def _read_artifact_into_memory(self, getInfo: DatastoreFileGetInformation,
1077 ref: DatasetRef, isComponent: bool = False) -> Any:
1078 """Read the artifact from datastore into in memory object.
1080 Parameters
1081 ----------
1082 getInfo : `DatastoreFileGetInformation`
1083 Information about the artifact within the datastore.
1084 ref : `DatasetRef`
1085 The registry information associated with this artifact.
1086 isComponent : `bool`
1087 Flag to indicate if a component is being read from this artifact.
1089 Returns
1090 -------
1091 inMemoryDataset : `object`
1092 The artifact as a python object.
1093 """
1094 location = getInfo.location
1095 uri = location.uri
1096 log.debug("Accessing data from %s", uri)
1098 # Cannot recalculate checksum but can compare size as a quick check
1099 # Do not do this if the size is negative since that indicates
1100 # we do not know.
1101 recorded_size = getInfo.info.file_size
1102 resource_size = uri.size()
1103 if recorded_size >= 0 and resource_size != recorded_size: 1103 ↛ 1104line 1103 didn't jump to line 1104, because the condition on line 1103 was never true
1104 raise RuntimeError("Integrity failure in Datastore. "
1105 f"Size of file {uri} ({resource_size}) "
1106 f"does not match size recorded in registry of {recorded_size}")
1108 # For the general case we have choices for how to proceed.
1109 # 1. Always use a local file (downloading the remote resource to a
1110 # temporary file if needed).
1111 # 2. Use a threshold size and read into memory and use bytes.
1112 # Use both for now with an arbitrary hand off size.
1113 # This allows small datasets to be downloaded from remote object
1114 # stores without requiring a temporary file.
1116 formatter = getInfo.formatter
1117 nbytes_max = 10_000_000 # Arbitrary number that we can tune
1118 if resource_size <= nbytes_max and formatter.can_read_bytes():
1119 with time_this(log, msg="Reading bytes from %s", args=(uri,)):
1120 serializedDataset = uri.read()
1121 log.debug("Deserializing %s from %d bytes from location %s with formatter %s",
1122 f"component {getInfo.component}" if isComponent else "",
1123 len(serializedDataset), uri, formatter.name())
1124 try:
1125 result = formatter.fromBytes(serializedDataset,
1126 component=getInfo.component if isComponent else None)
1127 except Exception as e:
1128 raise ValueError(f"Failure from formatter '{formatter.name()}' for dataset {ref.id}"
1129 f" ({ref.datasetType.name} from {uri}): {e}") from e
1130 else:
1131 # Read from file.
1133 # Have to update the Location associated with the formatter
1134 # because formatter.read does not allow an override.
1135 # This could be improved.
1136 location_updated = False
1137 msg = ""
1139 # First check in cache for local version.
1140 # The cache will only be relevant for remote resources.
1141 if not uri.isLocal:
1142 cached_file = self.cacheManager.find_in_cache(ref, uri.getExtension())
1143 if cached_file is not None: 1143 ↛ 1144line 1143 didn't jump to line 1144, because the condition on line 1143 was never true
1144 msg = f"(via cache read of remote file {uri})"
1145 uri = cached_file
1146 location_updated = True
1148 with uri.as_local() as local_uri:
1150 # URI was remote and file was downloaded
1151 if uri != local_uri:
1152 cache_msg = ""
1153 location_updated = True
1155 # Cache the downloaded file if needed.
1156 cached_uri = self.cacheManager.move_to_cache(local_uri, ref)
1157 if cached_uri is not None: 1157 ↛ 1158line 1157 didn't jump to line 1158, because the condition on line 1157 was never true
1158 local_uri = cached_uri
1159 cache_msg = " and cached"
1161 msg = f"(via download to local file{cache_msg})"
1163 # Calculate the (possibly) new location for the formatter
1164 # to use.
1165 newLocation = Location(*local_uri.split()) if location_updated else None
1167 log.debug("Reading%s from location %s %s with formatter %s",
1168 f" component {getInfo.component}" if isComponent else "",
1169 uri, msg, formatter.name())
1170 try:
1171 with formatter._updateLocation(newLocation):
1172 with time_this(log, msg="Reading%s from location %s %s with formatter %s",
1173 args=(f" component {getInfo.component}" if isComponent else "",
1174 uri, msg, formatter.name())):
1175 result = formatter.read(component=getInfo.component if isComponent else None)
1176 except Exception as e:
1177 raise ValueError(f"Failure from formatter '{formatter.name()}' for dataset {ref.id}"
1178 f" ({ref.datasetType.name} from {uri}): {e}") from e
1180 return self._post_process_get(result, getInfo.readStorageClass, getInfo.assemblerParams,
1181 isComponent=isComponent)
1183 def knows(self, ref: DatasetRef) -> bool:
1184 """Check if the dataset is known to the datastore.
1186 Does not check for existence of any artifact.
1188 Parameters
1189 ----------
1190 ref : `DatasetRef`
1191 Reference to the required dataset.
1193 Returns
1194 -------
1195 exists : `bool`
1196 `True` if the dataset is known to the datastore.
1197 """
1198 fileLocations = self._get_dataset_locations_info(ref)
1199 if fileLocations:
1200 return True
1201 return False
1203 def exists(self, ref: DatasetRef) -> bool:
1204 """Check if the dataset exists in the datastore.
1206 Parameters
1207 ----------
1208 ref : `DatasetRef`
1209 Reference to the required dataset.
1211 Returns
1212 -------
1213 exists : `bool`
1214 `True` if the entity exists in the `Datastore`.
1215 """
1216 fileLocations = self._get_dataset_locations_info(ref)
1218 # if we are being asked to trust that registry might not be correct
1219 # we ask for the expected locations and check them explicitly
1220 if not fileLocations:
1221 if not self.trustGetRequest:
1222 return False
1224 # When we are guessing a dataset location we can not check
1225 # for the existence of every component since we can not
1226 # know if every component was written. Instead we check
1227 # for the existence of any of the expected locations.
1228 for location, _ in self._get_expected_dataset_locations_info(ref): 1228 ↛ 1231line 1228 didn't jump to line 1231, because the loop on line 1228 didn't complete
1229 if self._artifact_exists(location): 1229 ↛ 1228line 1229 didn't jump to line 1228, because the condition on line 1229 was never false
1230 return True
1231 return False
1233 # All listed artifacts must exist.
1234 for location, _ in fileLocations:
1235 if not self._artifact_exists(location):
1236 return False
1238 return True
1240 def getURIs(self, ref: DatasetRef,
1241 predict: bool = False) -> Tuple[Optional[ButlerURI], Dict[str, ButlerURI]]:
1242 """Return URIs associated with dataset.
1244 Parameters
1245 ----------
1246 ref : `DatasetRef`
1247 Reference to the required dataset.
1248 predict : `bool`, optional
1249 If the datastore does not know about the dataset, should it
1250 return a predicted URI or not?
1252 Returns
1253 -------
1254 primary : `ButlerURI`
1255 The URI to the primary artifact associated with this dataset.
1256 If the dataset was disassembled within the datastore this
1257 may be `None`.
1258 components : `dict`
1259 URIs to any components associated with the dataset artifact.
1260 Can be empty if there are no components.
1261 """
1263 primary: Optional[ButlerURI] = None
1264 components: Dict[str, ButlerURI] = {}
1266 # if this has never been written then we have to guess
1267 if not self.exists(ref):
1268 if not predict:
1269 raise FileNotFoundError("Dataset {} not in this datastore".format(ref))
1271 doDisassembly = self.composites.shouldBeDisassembled(ref)
1273 if doDisassembly:
1275 for component, componentStorage in ref.datasetType.storageClass.components.items():
1276 compRef = ref.makeComponentRef(component)
1277 compLocation, _ = self._determine_put_formatter_location(compRef)
1279 # Add a URI fragment to indicate this is a guess
1280 components[component] = ButlerURI(compLocation.uri.geturl() + "#predicted")
1282 else:
1284 location, _ = self._determine_put_formatter_location(ref)
1286 # Add a URI fragment to indicate this is a guess
1287 primary = ButlerURI(location.uri.geturl() + "#predicted")
1289 return primary, components
1291 # If this is a ref that we have written we can get the path.
1292 # Get file metadata and internal metadata
1293 fileLocations = self._get_dataset_locations_info(ref)
1295 guessing = False
1296 if not fileLocations:
1297 if not self.trustGetRequest: 1297 ↛ 1298line 1297 didn't jump to line 1298, because the condition on line 1297 was never true
1298 raise RuntimeError(f"Unexpectedly got no artifacts for dataset {ref}")
1299 fileLocations = self._get_expected_dataset_locations_info(ref)
1300 guessing = True
1302 if len(fileLocations) == 1:
1303 # No disassembly so this is the primary URI
1304 uri = fileLocations[0][0].uri
1305 if guessing and not uri.exists(): 1305 ↛ 1306line 1305 didn't jump to line 1306, because the condition on line 1305 was never true
1306 raise FileNotFoundError(f"Expected URI ({uri}) does not exist")
1307 primary = uri
1309 else:
1310 for location, storedFileInfo in fileLocations:
1311 if storedFileInfo.component is None: 1311 ↛ 1312line 1311 didn't jump to line 1312, because the condition on line 1311 was never true
1312 raise RuntimeError(f"Unexpectedly got no component name for a component at {location}")
1313 uri = location.uri
1314 if guessing and not uri.exists(): 1314 ↛ 1318line 1314 didn't jump to line 1318, because the condition on line 1314 was never true
1315 # If we are trusting then it is entirely possible for
1316 # some components to be missing. In that case we skip
1317 # to the next component.
1318 if self.trustGetRequest:
1319 continue
1320 raise FileNotFoundError(f"Expected URI ({uri}) does not exist")
1321 components[storedFileInfo.component] = uri
1323 return primary, components
1325 def getURI(self, ref: DatasetRef, predict: bool = False) -> ButlerURI:
1326 """URI to the Dataset.
1328 Parameters
1329 ----------
1330 ref : `DatasetRef`
1331 Reference to the required Dataset.
1332 predict : `bool`
1333 If `True`, allow URIs to be returned of datasets that have not
1334 been written.
1336 Returns
1337 -------
1338 uri : `str`
1339 URI pointing to the dataset within the datastore. If the
1340 dataset does not exist in the datastore, and if ``predict`` is
1341 `True`, the URI will be a prediction and will include a URI
1342 fragment "#predicted".
1343 If the datastore does not have entities that relate well
1344 to the concept of a URI the returned URI will be
1345 descriptive. The returned URI is not guaranteed to be obtainable.
1347 Raises
1348 ------
1349 FileNotFoundError
1350 Raised if a URI has been requested for a dataset that does not
1351 exist and guessing is not allowed.
1352 RuntimeError
1353 Raised if a request is made for a single URI but multiple URIs
1354 are associated with this dataset.
1356 Notes
1357 -----
1358 When a predicted URI is requested an attempt will be made to form
1359 a reasonable URI based on file templates and the expected formatter.
1360 """
1361 primary, components = self.getURIs(ref, predict)
1362 if primary is None or components: 1362 ↛ 1363line 1362 didn't jump to line 1363, because the condition on line 1362 was never true
1363 raise RuntimeError(f"Dataset ({ref}) includes distinct URIs for components. "
1364 "Use Dataastore.getURIs() instead.")
1365 return primary
1367 def retrieveArtifacts(self, refs: Iterable[DatasetRef],
1368 destination: ButlerURI, transfer: str = "auto",
1369 preserve_path: bool = True,
1370 overwrite: bool = False) -> List[ButlerURI]:
1371 """Retrieve the file artifacts associated with the supplied refs.
1373 Parameters
1374 ----------
1375 refs : iterable of `DatasetRef`
1376 The datasets for which file artifacts are to be retrieved.
1377 A single ref can result in multiple files. The refs must
1378 be resolved.
1379 destination : `ButlerURI`
1380 Location to write the file artifacts.
1381 transfer : `str`, optional
1382 Method to use to transfer the artifacts. Must be one of the options
1383 supported by `ButlerURI.transfer_from()`. "move" is not allowed.
1384 preserve_path : `bool`, optional
1385 If `True` the full path of the file artifact within the datastore
1386 is preserved. If `False` the final file component of the path
1387 is used.
1388 overwrite : `bool`, optional
1389 If `True` allow transfers to overwrite existing files at the
1390 destination.
1392 Returns
1393 -------
1394 targets : `list` of `ButlerURI`
1395 URIs of file artifacts in destination location. Order is not
1396 preserved.
1397 """
1398 if not destination.isdir(): 1398 ↛ 1399line 1398 didn't jump to line 1399, because the condition on line 1398 was never true
1399 raise ValueError(f"Destination location must refer to a directory. Given {destination}")
1401 if transfer == "move":
1402 raise ValueError("Can not move artifacts out of datastore. Use copy instead.")
1404 # Source -> Destination
1405 # This also helps filter out duplicate DatasetRef in the request
1406 # that will map to the same underlying file transfer.
1407 to_transfer: Dict[ButlerURI, ButlerURI] = {}
1409 for ref in refs:
1410 locations = self._get_dataset_locations_info(ref)
1411 for location, _ in locations:
1412 source_uri = location.uri
1413 target_path: Union[str, ButlerURI]
1414 if preserve_path:
1415 target_path = location.pathInStore
1416 if target_path.isabs(): 1416 ↛ 1419line 1416 didn't jump to line 1419, because the condition on line 1416 was never true
1417 # This is an absolute path to an external file.
1418 # Use the full path.
1419 target_path = target_path.relativeToPathRoot
1420 else:
1421 target_path = source_uri.basename()
1422 target_uri = destination.join(target_path)
1423 to_transfer[source_uri] = target_uri
1425 # In theory can now parallelize the transfer
1426 log.debug("Number of artifacts to transfer to %s: %d",
1427 str(destination), len(to_transfer))
1428 for source_uri, target_uri in to_transfer.items():
1429 target_uri.transfer_from(source_uri, transfer=transfer, overwrite=overwrite)
1431 return list(to_transfer.values())
1433 def get(self, ref: DatasetRef, parameters: Optional[Mapping[str, Any]] = None) -> Any:
1434 """Load an InMemoryDataset from the store.
1436 Parameters
1437 ----------
1438 ref : `DatasetRef`
1439 Reference to the required Dataset.
1440 parameters : `dict`
1441 `StorageClass`-specific parameters that specify, for example,
1442 a slice of the dataset to be loaded.
1444 Returns
1445 -------
1446 inMemoryDataset : `object`
1447 Requested dataset or slice thereof as an InMemoryDataset.
1449 Raises
1450 ------
1451 FileNotFoundError
1452 Requested dataset can not be retrieved.
1453 TypeError
1454 Return value from formatter has unexpected type.
1455 ValueError
1456 Formatter failed to process the dataset.
1457 """
1458 allGetInfo = self._prepare_for_get(ref, parameters)
1459 refComponent = ref.datasetType.component()
1461 # Supplied storage class for the component being read
1462 refStorageClass = ref.datasetType.storageClass
1464 # Create mapping from component name to related info
1465 allComponents = {i.component: i for i in allGetInfo}
1467 # By definition the dataset is disassembled if we have more
1468 # than one record for it.
1469 isDisassembled = len(allGetInfo) > 1
1471 # Look for the special case where we are disassembled but the
1472 # component is a derived component that was not written during
1473 # disassembly. For this scenario we need to check that the
1474 # component requested is listed as a derived component for the
1475 # composite storage class
1476 isDisassembledReadOnlyComponent = False
1477 if isDisassembled and refComponent:
1478 # The composite storage class should be accessible through
1479 # the component dataset type
1480 compositeStorageClass = ref.datasetType.parentStorageClass
1482 # In the unlikely scenario where the composite storage
1483 # class is not known, we can only assume that this is a
1484 # normal component. If that assumption is wrong then the
1485 # branch below that reads a persisted component will fail
1486 # so there is no need to complain here.
1487 if compositeStorageClass is not None: 1487 ↛ 1490line 1487 didn't jump to line 1490, because the condition on line 1487 was never false
1488 isDisassembledReadOnlyComponent = refComponent in compositeStorageClass.derivedComponents
1490 if isDisassembled and not refComponent:
1491 # This was a disassembled dataset spread over multiple files
1492 # and we need to put them all back together again.
1493 # Read into memory and then assemble
1495 # Check that the supplied parameters are suitable for the type read
1496 refStorageClass.validateParameters(parameters)
1498 # We want to keep track of all the parameters that were not used
1499 # by formatters. We assume that if any of the component formatters
1500 # use a parameter that we do not need to apply it again in the
1501 # assembler.
1502 usedParams = set()
1504 components: Dict[str, Any] = {}
1505 for getInfo in allGetInfo:
1506 # assemblerParams are parameters not understood by the
1507 # associated formatter.
1508 usedParams.update(set(getInfo.formatterParams))
1510 component = getInfo.component
1512 if component is None: 1512 ↛ 1513line 1512 didn't jump to line 1513, because the condition on line 1512 was never true
1513 raise RuntimeError(f"Internal error in datastore assembly of {ref}")
1515 # We do not want the formatter to think it's reading
1516 # a component though because it is really reading a
1517 # standalone dataset -- always tell reader it is not a
1518 # component.
1519 components[component] = self._read_artifact_into_memory(getInfo, ref, isComponent=False)
1521 inMemoryDataset = ref.datasetType.storageClass.delegate().assemble(components)
1523 # Any unused parameters will have to be passed to the assembler
1524 if parameters:
1525 unusedParams = {k: v for k, v in parameters.items() if k not in usedParams}
1526 else:
1527 unusedParams = {}
1529 # Process parameters
1530 return ref.datasetType.storageClass.delegate().handleParameters(inMemoryDataset,
1531 parameters=unusedParams)
1533 elif isDisassembledReadOnlyComponent:
1535 compositeStorageClass = ref.datasetType.parentStorageClass
1536 if compositeStorageClass is None: 1536 ↛ 1537line 1536 didn't jump to line 1537, because the condition on line 1536 was never true
1537 raise RuntimeError(f"Unable to retrieve derived component '{refComponent}' since"
1538 "no composite storage class is available.")
1540 if refComponent is None: 1540 ↛ 1542line 1540 didn't jump to line 1542, because the condition on line 1540 was never true
1541 # Mainly for mypy
1542 raise RuntimeError(f"Internal error in datastore {self.name}: component can not be None here")
1544 # Assume that every derived component can be calculated by
1545 # forwarding the request to a single read/write component.
1546 # Rather than guessing which rw component is the right one by
1547 # scanning each for a derived component of the same name,
1548 # we ask the storage class delegate directly which one is best to
1549 # use.
1550 compositeDelegate = compositeStorageClass.delegate()
1551 forwardedComponent = compositeDelegate.selectResponsibleComponent(refComponent,
1552 set(allComponents))
1554 # Select the relevant component
1555 rwInfo = allComponents[forwardedComponent]
1557 # For now assume that read parameters are validated against
1558 # the real component and not the requested component
1559 forwardedStorageClass = rwInfo.formatter.fileDescriptor.readStorageClass
1560 forwardedStorageClass.validateParameters(parameters)
1562 # Unfortunately the FileDescriptor inside the formatter will have
1563 # the wrong write storage class so we need to create a new one
1564 # given the immutability constraint.
1565 writeStorageClass = rwInfo.info.storageClass
1567 # We may need to put some thought into parameters for read
1568 # components but for now forward them on as is
1569 readFormatter = type(rwInfo.formatter)(FileDescriptor(rwInfo.location,
1570 readStorageClass=refStorageClass,
1571 storageClass=writeStorageClass,
1572 parameters=parameters),
1573 ref.dataId)
1575 # The assembler can not receive any parameter requests for a
1576 # derived component at this time since the assembler will
1577 # see the storage class of the derived component and those
1578 # parameters will have to be handled by the formatter on the
1579 # forwarded storage class.
1580 assemblerParams: Dict[str, Any] = {}
1582 # Need to created a new info that specifies the derived
1583 # component and associated storage class
1584 readInfo = DatastoreFileGetInformation(rwInfo.location, readFormatter,
1585 rwInfo.info, assemblerParams, {},
1586 refComponent, refStorageClass)
1588 return self._read_artifact_into_memory(readInfo, ref, isComponent=True)
1590 else:
1591 # Single file request or component from that composite file
1592 for lookup in (refComponent, None): 1592 ↛ 1597line 1592 didn't jump to line 1597, because the loop on line 1592 didn't complete
1593 if lookup in allComponents: 1593 ↛ 1592line 1593 didn't jump to line 1592, because the condition on line 1593 was never false
1594 getInfo = allComponents[lookup]
1595 break
1596 else:
1597 raise FileNotFoundError(f"Component {refComponent} not found "
1598 f"for ref {ref} in datastore {self.name}")
1600 # Do not need the component itself if already disassembled
1601 if isDisassembled:
1602 isComponent = False
1603 else:
1604 isComponent = getInfo.component is not None
1606 # For a disassembled component we can validate parametersagainst
1607 # the component storage class directly
1608 if isDisassembled:
1609 refStorageClass.validateParameters(parameters)
1610 else:
1611 # For an assembled composite this could be a derived
1612 # component derived from a real component. The validity
1613 # of the parameters is not clear. For now validate against
1614 # the composite storage class
1615 getInfo.formatter.fileDescriptor.storageClass.validateParameters(parameters)
1617 return self._read_artifact_into_memory(getInfo, ref, isComponent=isComponent)
1619 @transactional
1620 def put(self, inMemoryDataset: Any, ref: DatasetRef) -> None:
1621 """Write a InMemoryDataset with a given `DatasetRef` to the store.
1623 Parameters
1624 ----------
1625 inMemoryDataset : `object`
1626 The dataset to store.
1627 ref : `DatasetRef`
1628 Reference to the associated Dataset.
1630 Raises
1631 ------
1632 TypeError
1633 Supplied object and storage class are inconsistent.
1634 DatasetTypeNotSupportedError
1635 The associated `DatasetType` is not handled by this datastore.
1637 Notes
1638 -----
1639 If the datastore is configured to reject certain dataset types it
1640 is possible that the put will fail and raise a
1641 `DatasetTypeNotSupportedError`. The main use case for this is to
1642 allow `ChainedDatastore` to put to multiple datastores without
1643 requiring that every datastore accepts the dataset.
1644 """
1646 doDisassembly = self.composites.shouldBeDisassembled(ref)
1647 # doDisassembly = True
1649 artifacts = []
1650 if doDisassembly:
1651 components = ref.datasetType.storageClass.delegate().disassemble(inMemoryDataset)
1652 for component, componentInfo in components.items():
1653 # Don't recurse because we want to take advantage of
1654 # bulk insert -- need a new DatasetRef that refers to the
1655 # same dataset_id but has the component DatasetType
1656 # DatasetType does not refer to the types of components
1657 # So we construct one ourselves.
1658 compRef = ref.makeComponentRef(component)
1659 storedInfo = self._write_in_memory_to_artifact(componentInfo.component, compRef)
1660 artifacts.append((compRef, storedInfo))
1661 else:
1662 # Write the entire thing out
1663 storedInfo = self._write_in_memory_to_artifact(inMemoryDataset, ref)
1664 artifacts.append((ref, storedInfo))
1666 self._register_datasets(artifacts)
1668 @transactional
1669 def trash(self, ref: Union[DatasetRef, Iterable[DatasetRef]], ignore_errors: bool = True) -> None:
1670 # Get file metadata and internal metadata
1671 if not isinstance(ref, DatasetRef):
1672 log.debug("Doing multi-dataset trash in datastore %s", self.name)
1673 # Assumed to be an iterable of refs so bulk mode enabled.
1674 try:
1675 self.bridge.moveToTrash(ref)
1676 except Exception as e:
1677 if ignore_errors:
1678 log.warning("Unexpected issue moving multiple datasets to trash: %s", e)
1679 else:
1680 raise
1681 return
1683 log.debug("Trashing dataset %s in datastore %s", ref, self.name)
1685 fileLocations = self._get_dataset_locations_info(ref)
1687 if not fileLocations:
1688 err_msg = f"Requested dataset to trash ({ref}) is not known to datastore {self.name}"
1689 if ignore_errors: 1689 ↛ 1690line 1689 didn't jump to line 1690, because the condition on line 1689 was never true
1690 log.warning(err_msg)
1691 return
1692 else:
1693 raise FileNotFoundError(err_msg)
1695 for location, storedFileInfo in fileLocations:
1696 if not self._artifact_exists(location): 1696 ↛ 1697line 1696 didn't jump to line 1697, because the condition on line 1696 was never true
1697 err_msg = f"Dataset is known to datastore {self.name} but " \
1698 f"associated artifact ({location.uri}) is missing"
1699 if ignore_errors:
1700 log.warning(err_msg)
1701 return
1702 else:
1703 raise FileNotFoundError(err_msg)
1705 # Mark dataset as trashed
1706 try:
1707 self.bridge.moveToTrash([ref])
1708 except Exception as e:
1709 if ignore_errors:
1710 log.warning("Attempted to mark dataset (%s) to be trashed in datastore %s "
1711 "but encountered an error: %s", ref, self.name, e)
1712 pass
1713 else:
1714 raise
1716 @transactional
1717 def emptyTrash(self, ignore_errors: bool = True) -> None:
1718 """Remove all datasets from the trash.
1720 Parameters
1721 ----------
1722 ignore_errors : `bool`
1723 If `True` return without error even if something went wrong.
1724 Problems could occur if another process is simultaneously trying
1725 to delete.
1726 """
1727 log.debug("Emptying trash in datastore %s", self.name)
1729 # Context manager will empty trash iff we finish it without raising.
1730 # It will also automatically delete the relevant rows from the
1731 # trash table and the records table.
1732 with self.bridge.emptyTrash(self._table, record_class=StoredFileInfo,
1733 record_column="path") as trash_data:
1734 # Removing the artifacts themselves requires that the files are
1735 # not also associated with refs that are not to be trashed.
1736 # Therefore need to do a query with the file paths themselves
1737 # and return all the refs associated with them. Can only delete
1738 # a file if the refs to be trashed are the only refs associated
1739 # with the file.
1740 # This requires multiple copies of the trashed items
1741 trashed, artifacts_to_keep = trash_data
1743 if artifacts_to_keep is None:
1744 # The bridge is not helping us so have to work it out
1745 # ourselves. This is not going to be as efficient.
1746 trashed = list(trashed)
1748 # The instance check is for mypy since up to this point it
1749 # does not know the type of info.
1750 path_map = self._refs_associated_with_artifacts([info.path for _, info in trashed
1751 if isinstance(info, StoredFileInfo)])
1753 for ref, info in trashed:
1755 # Mypy needs to know this is not the base class
1756 assert isinstance(info, StoredFileInfo), f"Unexpectedly got info of class {type(info)}"
1758 # Check for mypy
1759 assert ref.id is not None, f"Internal logic error in emptyTrash with ref {ref}/{info}"
1761 path_map[info.path].remove(ref.id)
1762 if not path_map[info.path]: 1762 ↛ 1753line 1762 didn't jump to line 1753, because the condition on line 1762 was never false
1763 del path_map[info.path]
1765 artifacts_to_keep = set(path_map)
1767 for ref, info in trashed:
1769 # Should not happen for this implementation but need
1770 # to keep mypy happy.
1771 assert info is not None, f"Internal logic error in emptyTrash with ref {ref}."
1773 # Mypy needs to know this is not the base class
1774 assert isinstance(info, StoredFileInfo), f"Unexpectedly got info of class {type(info)}"
1776 # Check for mypy
1777 assert ref.id is not None, f"Internal logic error in emptyTrash with ref {ref}/{info}"
1779 if info.path in artifacts_to_keep:
1780 # This is a multi-dataset artifact and we are not
1781 # removing all associated refs.
1782 continue
1784 # Only trashed refs still known to datastore will be returned.
1785 location = info.file_location(self.locationFactory)
1787 # Point of no return for this artifact
1788 log.debug("Removing artifact %s from datastore %s", location.uri, self.name)
1789 try:
1790 self._delete_artifact(location)
1791 except FileNotFoundError:
1792 # If the file itself has been deleted there is nothing
1793 # we can do about it. It is possible that trash has
1794 # been run in parallel in another process or someone
1795 # decided to delete the file. It is unlikely to come
1796 # back and so we should still continue with the removal
1797 # of the entry from the trash table. It is also possible
1798 # we removed it in a previous iteration if it was
1799 # a multi-dataset artifact. The delete artifact method
1800 # will log a debug message in this scenario.
1801 # Distinguishing file missing before trash started and
1802 # file already removed previously as part of this trash
1803 # is not worth the distinction with regards to potential
1804 # memory cost.
1805 pass
1806 except Exception as e:
1807 if ignore_errors:
1808 # Use a debug message here even though it's not
1809 # a good situation. In some cases this can be
1810 # caused by a race between user A and user B
1811 # and neither of them has permissions for the
1812 # other's files. Butler does not know about users
1813 # and trash has no idea what collections these
1814 # files were in (without guessing from a path).
1815 log.debug("Encountered error removing artifact %s from datastore %s: %s",
1816 location.uri, self.name, e)
1817 else:
1818 raise
1820 @transactional
1821 def transfer_from(self, source_datastore: Datastore, refs: Iterable[DatasetRef],
1822 local_refs: Optional[Iterable[DatasetRef]] = None,
1823 transfer: str = "auto") -> None:
1824 # Docstring inherited
1825 if type(self) is not type(source_datastore): 1825 ↛ 1826line 1825 didn't jump to line 1826, because the condition on line 1825 was never true
1826 raise TypeError(f"Datastore mismatch between this datastore ({type(self)}) and the "
1827 f"source datastore ({type(source_datastore)}).")
1829 # Be explicit for mypy
1830 if not isinstance(source_datastore, FileDatastore): 1830 ↛ 1831line 1830 didn't jump to line 1831, because the condition on line 1830 was never true
1831 raise TypeError("Can only transfer to a FileDatastore from another FileDatastore, not"
1832 f" {type(source_datastore)}")
1834 # Stop early if "direct" transfer mode is requested. That would
1835 # require that the URI inside the source datastore should be stored
1836 # directly in the target datastore, which seems unlikely to be useful
1837 # since at any moment the source datastore could delete the file.
1838 if transfer in ("direct", "split"): 1838 ↛ 1839line 1838 didn't jump to line 1839, because the condition on line 1838 was never true
1839 raise ValueError("Can not transfer from a source datastore using direct mode since"
1840 " those files are controlled by the other datastore.")
1842 # We will go through the list multiple times so must convert
1843 # generators to lists.
1844 refs = list(refs)
1846 if local_refs is None: 1846 ↛ 1847line 1846 didn't jump to line 1847, because the condition on line 1846 was never true
1847 local_refs = refs
1848 else:
1849 local_refs = list(local_refs)
1851 # In order to handle disassembled composites the code works
1852 # at the records level since it can assume that internal APIs
1853 # can be used.
1854 # - If the record already exists in the destination this is assumed
1855 # to be okay.
1856 # - If there is no record but the source and destination URIs are
1857 # identical no transfer is done but the record is added.
1858 # - If the source record refers to an absolute URI currently assume
1859 # that that URI should remain absolute and will be visible to the
1860 # destination butler. May need to have a flag to indicate whether
1861 # the dataset should be transferred. This will only happen if
1862 # the detached Butler has had a local ingest.
1864 # What we really want is all the records in the source datastore
1865 # associated with these refs. Or derived ones if they don't exist
1866 # in the source.
1867 source_records = source_datastore._get_stored_records_associated_with_refs(refs)
1869 # The source dataset_ids are the keys in these records
1870 source_ids = set(source_records)
1871 log.debug("Number of datastore records found in source: %d", len(source_ids))
1873 # The not None check is to appease mypy
1874 requested_ids = set(ref.id for ref in refs if ref.id is not None)
1875 missing_ids = requested_ids - source_ids
1877 # Missing IDs can be okay if that datastore has allowed
1878 # gets based on file existence. Should we transfer what we can
1879 # or complain about it and warn?
1880 if missing_ids and not source_datastore.trustGetRequest: 1880 ↛ 1881line 1880 didn't jump to line 1881, because the condition on line 1880 was never true
1881 raise ValueError(f"Some datasets are missing from source datastore {source_datastore}:"
1882 f" {missing_ids}")
1884 # Need to map these missing IDs to a DatasetRef so we can guess
1885 # the details.
1886 if missing_ids: 1886 ↛ 1887line 1886 didn't jump to line 1887, because the condition on line 1886 was never true
1887 log.info("Number of expected datasets missing from source datastore records: %d out of %d",
1888 len(missing_ids), len(requested_ids))
1889 id_to_ref = {ref.id: ref for ref in refs if ref.id in missing_ids}
1891 for missing in missing_ids:
1892 # Ask the source datastore where the missing artifacts
1893 # should be. An execution butler might not know about the
1894 # artifacts even if they are there.
1895 expected = source_datastore._get_expected_dataset_locations_info(id_to_ref[missing])
1897 # Not all components can be guaranteed to exist so this
1898 # list has to filter those by checking to see if the
1899 # artifact is really there.
1900 records = [info for location, info in expected if location.uri.exists()]
1901 if records:
1902 source_records[missing].extend(records)
1903 else:
1904 log.warning("Asked to transfer dataset %s but no file artifacts exist for it.",
1905 id_to_ref[missing])
1907 # See if we already have these records
1908 target_records = self._get_stored_records_associated_with_refs(local_refs)
1910 # The artifacts to register
1911 artifacts = []
1913 # Refs that already exist
1914 already_present = []
1916 # Now can transfer the artifacts
1917 for source_ref, target_ref in zip(refs, local_refs):
1918 if target_ref.id in target_records: 1918 ↛ 1920line 1918 didn't jump to line 1920, because the condition on line 1918 was never true
1919 # Already have an artifact for this.
1920 already_present.append(target_ref)
1921 continue
1923 # mypy needs to know these are always resolved refs
1924 for info in source_records[source_ref.getCheckedId()]:
1925 source_location = info.file_location(source_datastore.locationFactory)
1926 target_location = info.file_location(self.locationFactory)
1927 if source_location == target_location: 1927 ↛ 1931line 1927 didn't jump to line 1931, because the condition on line 1927 was never true
1928 # Either the dataset is already in the target datastore
1929 # (which is how execution butler currently runs) or
1930 # it is an absolute URI.
1931 if source_location.pathInStore.isabs():
1932 # Just because we can see the artifact when running
1933 # the transfer doesn't mean it will be generally
1934 # accessible to a user of this butler. For now warn
1935 # but assume it will be accessible.
1936 log.warning("Transfer request for an outside-datastore artifact has been found at %s",
1937 source_location)
1938 else:
1939 # Need to transfer it to the new location.
1940 # Assume we should always overwrite. If the artifact
1941 # is there this might indicate that a previous transfer
1942 # was interrupted but was not able to be rolled back
1943 # completely (eg pre-emption) so follow Datastore default
1944 # and overwrite.
1945 target_location.uri.transfer_from(source_location.uri, transfer=transfer,
1946 overwrite=True, transaction=self._transaction)
1948 artifacts.append((target_ref, info))
1950 self._register_datasets(artifacts)
1952 if already_present: 1952 ↛ 1953line 1952 didn't jump to line 1953, because the condition on line 1952 was never true
1953 n_skipped = len(already_present)
1954 log.info("Skipped transfer of %d dataset%s already present in datastore", n_skipped,
1955 "" if n_skipped == 1 else "s")
1957 @transactional
1958 def forget(self, refs: Iterable[DatasetRef]) -> None:
1959 # Docstring inherited.
1960 refs = list(refs)
1961 self.bridge.forget(refs)
1962 self._table.delete(["dataset_id"], *[{"dataset_id": ref.getCheckedId()} for ref in refs])
1964 def validateConfiguration(self, entities: Iterable[Union[DatasetRef, DatasetType, StorageClass]],
1965 logFailures: bool = False) -> None:
1966 """Validate some of the configuration for this datastore.
1968 Parameters
1969 ----------
1970 entities : iterable of `DatasetRef`, `DatasetType`, or `StorageClass`
1971 Entities to test against this configuration. Can be differing
1972 types.
1973 logFailures : `bool`, optional
1974 If `True`, output a log message for every validation error
1975 detected.
1977 Raises
1978 ------
1979 DatastoreValidationError
1980 Raised if there is a validation problem with a configuration.
1981 All the problems are reported in a single exception.
1983 Notes
1984 -----
1985 This method checks that all the supplied entities have valid file
1986 templates and also have formatters defined.
1987 """
1989 templateFailed = None
1990 try:
1991 self.templates.validateTemplates(entities, logFailures=logFailures)
1992 except FileTemplateValidationError as e:
1993 templateFailed = str(e)
1995 formatterFailed = []
1996 for entity in entities:
1997 try:
1998 self.formatterFactory.getFormatterClass(entity)
1999 except KeyError as e:
2000 formatterFailed.append(str(e))
2001 if logFailures: 2001 ↛ 1996line 2001 didn't jump to line 1996, because the condition on line 2001 was never false
2002 log.critical("Formatter failure: %s", e)
2004 if templateFailed or formatterFailed:
2005 messages = []
2006 if templateFailed: 2006 ↛ 2007line 2006 didn't jump to line 2007, because the condition on line 2006 was never true
2007 messages.append(templateFailed)
2008 if formatterFailed: 2008 ↛ 2010line 2008 didn't jump to line 2010, because the condition on line 2008 was never false
2009 messages.append(",".join(formatterFailed))
2010 msg = ";\n".join(messages)
2011 raise DatastoreValidationError(msg)
2013 def getLookupKeys(self) -> Set[LookupKey]:
2014 # Docstring is inherited from base class
2015 return self.templates.getLookupKeys() | self.formatterFactory.getLookupKeys() | \
2016 self.constraints.getLookupKeys()
2018 def validateKey(self, lookupKey: LookupKey,
2019 entity: Union[DatasetRef, DatasetType, StorageClass]) -> None:
2020 # Docstring is inherited from base class
2021 # The key can be valid in either formatters or templates so we can
2022 # only check the template if it exists
2023 if lookupKey in self.templates:
2024 try:
2025 self.templates[lookupKey].validateTemplate(entity)
2026 except FileTemplateValidationError as e:
2027 raise DatastoreValidationError(e) from e
2029 def export(self, refs: Iterable[DatasetRef], *,
2030 directory: Optional[Union[ButlerURI, str]] = None,
2031 transfer: Optional[str] = "auto") -> Iterable[FileDataset]:
2032 # Docstring inherited from Datastore.export.
2033 if transfer is not None and directory is None: 2033 ↛ 2034line 2033 didn't jump to line 2034, because the condition on line 2033 was never true
2034 raise RuntimeError(f"Cannot export using transfer mode {transfer} with no "
2035 "export directory given")
2037 # Force the directory to be a URI object
2038 directoryUri: Optional[ButlerURI] = None
2039 if directory is not None: 2039 ↛ 2042line 2039 didn't jump to line 2042, because the condition on line 2039 was never false
2040 directoryUri = ButlerURI(directory, forceDirectory=True)
2042 if transfer is not None and directoryUri is not None: 2042 ↛ 2047line 2042 didn't jump to line 2047, because the condition on line 2042 was never false
2043 # mypy needs the second test
2044 if not directoryUri.exists(): 2044 ↛ 2045line 2044 didn't jump to line 2045, because the condition on line 2044 was never true
2045 raise FileNotFoundError(f"Export location {directory} does not exist")
2047 progress = Progress("lsst.daf.butler.datastores.FileDatastore.export", level=logging.DEBUG)
2048 for ref in progress.wrap(refs, "Exporting dataset files"):
2049 fileLocations = self._get_dataset_locations_info(ref)
2050 if not fileLocations: 2050 ↛ 2051line 2050 didn't jump to line 2051, because the condition on line 2050 was never true
2051 raise FileNotFoundError(f"Could not retrieve dataset {ref}.")
2052 # For now we can not export disassembled datasets
2053 if len(fileLocations) > 1: 2053 ↛ 2054line 2053 didn't jump to line 2054, because the condition on line 2053 was never true
2054 raise NotImplementedError(f"Can not export disassembled datasets such as {ref}")
2055 location, storedFileInfo = fileLocations[0]
2057 pathInStore = location.pathInStore.path
2058 if transfer is None: 2058 ↛ 2062line 2058 didn't jump to line 2062, because the condition on line 2058 was never true
2059 # TODO: do we also need to return the readStorageClass somehow?
2060 # We will use the path in store directly. If this is an
2061 # absolute URI, preserve it.
2062 if location.pathInStore.isabs():
2063 pathInStore = str(location.uri)
2064 elif transfer == "direct": 2064 ↛ 2066line 2064 didn't jump to line 2066, because the condition on line 2064 was never true
2065 # Use full URIs to the remote store in the export
2066 pathInStore = str(location.uri)
2067 else:
2068 # mypy needs help
2069 assert directoryUri is not None, "directoryUri must be defined to get here"
2070 storeUri = ButlerURI(location.uri)
2072 # if the datastore has an absolute URI to a resource, we
2073 # have two options:
2074 # 1. Keep the absolute URI in the exported YAML
2075 # 2. Allocate a new name in the local datastore and transfer
2076 # it.
2077 # For now go with option 2
2078 if location.pathInStore.isabs(): 2078 ↛ 2079line 2078 didn't jump to line 2079, because the condition on line 2078 was never true
2079 template = self.templates.getTemplate(ref)
2080 newURI = ButlerURI(template.format(ref), forceAbsolute=False)
2081 pathInStore = str(newURI.updatedExtension(location.pathInStore.getExtension()))
2083 exportUri = directoryUri.join(pathInStore)
2084 exportUri.transfer_from(storeUri, transfer=transfer)
2086 yield FileDataset(refs=[ref], path=pathInStore, formatter=storedFileInfo.formatter)
2088 @staticmethod
2089 def computeChecksum(uri: ButlerURI, algorithm: str = "blake2b", block_size: int = 8192) -> Optional[str]:
2090 """Compute the checksum of the supplied file.
2092 Parameters
2093 ----------
2094 uri : `ButlerURI`
2095 Name of resource to calculate checksum from.
2096 algorithm : `str`, optional
2097 Name of algorithm to use. Must be one of the algorithms supported
2098 by :py:class`hashlib`.
2099 block_size : `int`
2100 Number of bytes to read from file at one time.
2102 Returns
2103 -------
2104 hexdigest : `str`
2105 Hex digest of the file.
2107 Notes
2108 -----
2109 Currently returns None if the URI is for a remote resource.
2110 """
2111 if algorithm not in hashlib.algorithms_guaranteed: 2111 ↛ 2112line 2111 didn't jump to line 2112, because the condition on line 2111 was never true
2112 raise NameError("The specified algorithm '{}' is not supported by hashlib".format(algorithm))
2114 if not uri.isLocal: 2114 ↛ 2115line 2114 didn't jump to line 2115, because the condition on line 2114 was never true
2115 return None
2117 hasher = hashlib.new(algorithm)
2119 with uri.as_local() as local_uri:
2120 with open(local_uri.ospath, "rb") as f:
2121 for chunk in iter(lambda: f.read(block_size), b""):
2122 hasher.update(chunk)
2124 return hasher.hexdigest()
2126 def needs_expanded_data_ids(
2127 self,
2128 transfer: Optional[str],
2129 entity: Optional[Union[DatasetRef, DatasetType, StorageClass]] = None,
2130 ) -> bool:
2131 # Docstring inherited.
2132 # This _could_ also use entity to inspect whether the filename template
2133 # involves placeholders other than the required dimensions for its
2134 # dataset type, but that's not necessary for correctness; it just
2135 # enables more optimizations (perhaps only in theory).
2136 return transfer not in ("direct", None)