Coverage for python/lsst/daf/butler/_butler.py: 8%
704 statements
« prev ^ index » next coverage.py v6.5.0, created at 2023-03-01 02:25 -0800
« prev ^ index » next coverage.py v6.5.0, created at 2023-03-01 02:25 -0800
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/>.
22"""
23Butler top level classes.
24"""
25from __future__ import annotations
27__all__ = (
28 "Butler",
29 "ButlerValidationError",
30 "PruneCollectionsArgsError",
31 "PurgeWithoutUnstorePruneCollectionsError",
32 "RunWithoutPurgePruneCollectionsError",
33 "PurgeUnsupportedPruneCollectionsError",
34)
36import collections.abc
37import contextlib
38import logging
39import numbers
40import os
41from collections import defaultdict
42from typing import (
43 Any,
44 ClassVar,
45 Counter,
46 Dict,
47 Iterable,
48 Iterator,
49 List,
50 MutableMapping,
51 Optional,
52 Sequence,
53 Set,
54 TextIO,
55 Tuple,
56 Type,
57 Union,
58)
60from lsst.resources import ResourcePath, ResourcePathExpression
61from lsst.utils import doImportType
62from lsst.utils.introspection import get_class_of
63from lsst.utils.logging import VERBOSE, getLogger
65from ._butlerConfig import ButlerConfig
66from ._butlerRepoIndex import ButlerRepoIndex
67from ._deferredDatasetHandle import DeferredDatasetHandle
68from ._limited_butler import LimitedButler
69from .core import (
70 AmbiguousDatasetError,
71 Config,
72 ConfigSubset,
73 DataCoordinate,
74 DataId,
75 DataIdValue,
76 DatasetRef,
77 DatasetRefURIs,
78 DatasetType,
79 Datastore,
80 Dimension,
81 DimensionConfig,
82 DimensionElement,
83 DimensionRecord,
84 DimensionUniverse,
85 FileDataset,
86 Progress,
87 StorageClass,
88 StorageClassFactory,
89 Timespan,
90 ValidationError,
91)
92from .core.repoRelocation import BUTLER_ROOT_TAG
93from .core.utils import transactional
94from .registry import (
95 CollectionType,
96 ConflictingDefinitionError,
97 DataIdError,
98 DatasetIdGenEnum,
99 MissingDatasetTypeError,
100 Registry,
101 RegistryConfig,
102 RegistryDefaults,
103)
104from .transfers import RepoExportContext
106log = getLogger(__name__)
109class ButlerValidationError(ValidationError):
110 """There is a problem with the Butler configuration."""
112 pass
115class PruneCollectionsArgsError(TypeError):
116 """Base class for errors relating to Butler.pruneCollections input
117 arguments.
118 """
120 pass
123class PurgeWithoutUnstorePruneCollectionsError(PruneCollectionsArgsError):
124 """Raised when purge and unstore are both required to be True, and
125 purge is True but unstore is False.
126 """
128 def __init__(self) -> None:
129 super().__init__("Cannot pass purge=True without unstore=True.")
132class RunWithoutPurgePruneCollectionsError(PruneCollectionsArgsError):
133 """Raised when pruning a RUN collection but purge is False."""
135 def __init__(self, collectionType: CollectionType):
136 self.collectionType = collectionType
137 super().__init__(f"Cannot prune RUN collection {self.collectionType.name} without purge=True.")
140class PurgeUnsupportedPruneCollectionsError(PruneCollectionsArgsError):
141 """Raised when purge is True but is not supported for the given
142 collection."""
144 def __init__(self, collectionType: CollectionType):
145 self.collectionType = collectionType
146 super().__init__(
147 f"Cannot prune {self.collectionType} collection {self.collectionType.name} with purge=True."
148 )
151class Butler(LimitedButler):
152 """Main entry point for the data access system.
154 Parameters
155 ----------
156 config : `ButlerConfig`, `Config` or `str`, optional.
157 Configuration. Anything acceptable to the
158 `ButlerConfig` constructor. If a directory path
159 is given the configuration will be read from a ``butler.yaml`` file in
160 that location. If `None` is given default values will be used.
161 butler : `Butler`, optional.
162 If provided, construct a new Butler that uses the same registry and
163 datastore as the given one, but with the given collection and run.
164 Incompatible with the ``config``, ``searchPaths``, and ``writeable``
165 arguments.
166 collections : `str` or `Iterable` [ `str` ], optional
167 An expression specifying the collections to be searched (in order) when
168 reading datasets.
169 This may be a `str` collection name or an iterable thereof.
170 See :ref:`daf_butler_collection_expressions` for more information.
171 These collections are not registered automatically and must be
172 manually registered before they are used by any method, but they may be
173 manually registered after the `Butler` is initialized.
174 run : `str`, optional
175 Name of the `~CollectionType.RUN` collection new datasets should be
176 inserted into. If ``collections`` is `None` and ``run`` is not `None`,
177 ``collections`` will be set to ``[run]``. If not `None`, this
178 collection will automatically be registered. If this is not set (and
179 ``writeable`` is not set either), a read-only butler will be created.
180 searchPaths : `list` of `str`, optional
181 Directory paths to search when calculating the full Butler
182 configuration. Not used if the supplied config is already a
183 `ButlerConfig`.
184 writeable : `bool`, optional
185 Explicitly sets whether the butler supports write operations. If not
186 provided, a read-write butler is created if any of ``run``, ``tags``,
187 or ``chains`` is non-empty.
188 inferDefaults : `bool`, optional
189 If `True` (default) infer default data ID values from the values
190 present in the datasets in ``collections``: if all collections have the
191 same value (or no value) for a governor dimension, that value will be
192 the default for that dimension. Nonexistent collections are ignored.
193 If a default value is provided explicitly for a governor dimension via
194 ``**kwargs``, no default will be inferred for that dimension.
195 **kwargs : `str`
196 Default data ID key-value pairs. These may only identify "governor"
197 dimensions like ``instrument`` and ``skymap``.
199 Examples
200 --------
201 While there are many ways to control exactly how a `Butler` interacts with
202 the collections in its `Registry`, the most common cases are still simple.
204 For a read-only `Butler` that searches one collection, do::
206 butler = Butler("/path/to/repo", collections=["u/alice/DM-50000"])
208 For a read-write `Butler` that writes to and reads from a
209 `~CollectionType.RUN` collection::
211 butler = Butler("/path/to/repo", run="u/alice/DM-50000/a")
213 The `Butler` passed to a ``PipelineTask`` is often much more complex,
214 because we want to write to one `~CollectionType.RUN` collection but read
215 from several others (as well)::
217 butler = Butler("/path/to/repo", run="u/alice/DM-50000/a",
218 collections=["u/alice/DM-50000/a",
219 "u/bob/DM-49998",
220 "HSC/defaults"])
222 This butler will `put` new datasets to the run ``u/alice/DM-50000/a``.
223 Datasets will be read first from that run (since it appears first in the
224 chain), and then from ``u/bob/DM-49998`` and finally ``HSC/defaults``.
226 Finally, one can always create a `Butler` with no collections::
228 butler = Butler("/path/to/repo", writeable=True)
230 This can be extremely useful when you just want to use ``butler.registry``,
231 e.g. for inserting dimension data or managing collections, or when the
232 collections you want to use with the butler are not consistent.
233 Passing ``writeable`` explicitly here is only necessary if you want to be
234 able to make changes to the repo - usually the value for ``writeable`` can
235 be guessed from the collection arguments provided, but it defaults to
236 `False` when there are not collection arguments.
237 """
239 def __init__(
240 self,
241 config: Union[Config, str, None] = None,
242 *,
243 butler: Optional[Butler] = None,
244 collections: Any = None,
245 run: Optional[str] = None,
246 searchPaths: Optional[List[str]] = None,
247 writeable: Optional[bool] = None,
248 inferDefaults: bool = True,
249 **kwargs: str,
250 ):
251 defaults = RegistryDefaults(collections=collections, run=run, infer=inferDefaults, **kwargs)
252 # Load registry, datastore, etc. from config or existing butler.
253 if butler is not None:
254 if config is not None or searchPaths is not None or writeable is not None:
255 raise TypeError(
256 "Cannot pass 'config', 'searchPaths', or 'writeable' arguments with 'butler' argument."
257 )
258 self.registry = butler.registry.copy(defaults)
259 self.datastore = butler.datastore
260 self.storageClasses = butler.storageClasses
261 self._config: ButlerConfig = butler._config
262 self._allow_put_of_predefined_dataset = butler._allow_put_of_predefined_dataset
263 else:
264 # Can only look for strings in the known repos list.
265 if isinstance(config, str) and config in self.get_known_repos():
266 config = str(self.get_repo_uri(config))
267 try:
268 self._config = ButlerConfig(config, searchPaths=searchPaths)
269 except FileNotFoundError as e:
270 if known := self.get_known_repos():
271 aliases = f"(known aliases: {', '.join(known)})"
272 else:
273 aliases = "(no known aliases)"
274 raise FileNotFoundError(f"{e} {aliases}") from e
275 self._config = ButlerConfig(config, searchPaths=searchPaths)
276 try:
277 if "root" in self._config:
278 butlerRoot = self._config["root"]
279 else:
280 butlerRoot = self._config.configDir
281 if writeable is None:
282 writeable = run is not None
283 self.registry = Registry.fromConfig(
284 self._config, butlerRoot=butlerRoot, writeable=writeable, defaults=defaults
285 )
286 self.datastore = Datastore.fromConfig(
287 self._config, self.registry.getDatastoreBridgeManager(), butlerRoot=butlerRoot
288 )
289 self.storageClasses = StorageClassFactory()
290 self.storageClasses.addFromConfig(self._config)
291 self._allow_put_of_predefined_dataset = self._config.get(
292 "allow_put_of_predefined_dataset", False
293 )
294 except Exception:
295 # Failures here usually mean that configuration is incomplete,
296 # just issue an error message which includes config file URI.
297 log.error(f"Failed to instantiate Butler from config {self._config.configFile}.")
298 raise
300 # For execution butler the datastore needs a special
301 # dependency-inversion trick. This is not used by regular butler,
302 # but we do not have a way to distinguish regular butler from execution
303 # butler.
304 self.datastore.set_retrieve_dataset_type_method(self._retrieve_dataset_type)
306 if "run" in self._config or "collection" in self._config:
307 raise ValueError("Passing a run or collection via configuration is no longer supported.")
309 GENERATION: ClassVar[int] = 3
310 """This is a Generation 3 Butler.
312 This attribute may be removed in the future, once the Generation 2 Butler
313 interface has been fully retired; it should only be used in transitional
314 code.
315 """
317 def _retrieve_dataset_type(self, name: str) -> DatasetType | None:
318 """Return DatasetType defined in registry given dataset type name."""
319 try:
320 return self.registry.getDatasetType(name)
321 except MissingDatasetTypeError:
322 return None
324 @classmethod
325 def get_repo_uri(cls, label: str) -> ResourcePath:
326 """Look up the label in a butler repository index.
328 Parameters
329 ----------
330 label : `str`
331 Label of the Butler repository to look up.
333 Returns
334 -------
335 uri : `lsst.resources.ResourcePath`
336 URI to the Butler repository associated with the given label.
338 Raises
339 ------
340 KeyError
341 Raised if the label is not found in the index, or if an index
342 can not be found at all.
344 Notes
345 -----
346 See `~lsst.daf.butler.ButlerRepoIndex` for details on how the
347 information is discovered.
348 """
349 return ButlerRepoIndex.get_repo_uri(label)
351 @classmethod
352 def get_known_repos(cls) -> Set[str]:
353 """Retrieve the list of known repository labels.
355 Returns
356 -------
357 repos : `set` of `str`
358 All the known labels. Can be empty if no index can be found.
360 Notes
361 -----
362 See `~lsst.daf.butler.ButlerRepoIndex` for details on how the
363 information is discovered.
364 """
365 return ButlerRepoIndex.get_known_repos()
367 @staticmethod
368 def makeRepo(
369 root: ResourcePathExpression,
370 config: Union[Config, str, None] = None,
371 dimensionConfig: Union[Config, str, None] = None,
372 standalone: bool = False,
373 searchPaths: Optional[List[str]] = None,
374 forceConfigRoot: bool = True,
375 outfile: Optional[ResourcePathExpression] = None,
376 overwrite: bool = False,
377 ) -> Config:
378 """Create an empty data repository by adding a butler.yaml config
379 to a repository root directory.
381 Parameters
382 ----------
383 root : `lsst.resources.ResourcePathExpression`
384 Path or URI to the root location of the new repository. Will be
385 created if it does not exist.
386 config : `Config` or `str`, optional
387 Configuration to write to the repository, after setting any
388 root-dependent Registry or Datastore config options. Can not
389 be a `ButlerConfig` or a `ConfigSubset`. If `None`, default
390 configuration will be used. Root-dependent config options
391 specified in this config are overwritten if ``forceConfigRoot``
392 is `True`.
393 dimensionConfig : `Config` or `str`, optional
394 Configuration for dimensions, will be used to initialize registry
395 database.
396 standalone : `bool`
397 If True, write all expanded defaults, not just customized or
398 repository-specific settings.
399 This (mostly) decouples the repository from the default
400 configuration, insulating it from changes to the defaults (which
401 may be good or bad, depending on the nature of the changes).
402 Future *additions* to the defaults will still be picked up when
403 initializing `Butlers` to repos created with ``standalone=True``.
404 searchPaths : `list` of `str`, optional
405 Directory paths to search when calculating the full butler
406 configuration.
407 forceConfigRoot : `bool`, optional
408 If `False`, any values present in the supplied ``config`` that
409 would normally be reset are not overridden and will appear
410 directly in the output config. This allows non-standard overrides
411 of the root directory for a datastore or registry to be given.
412 If this parameter is `True` the values for ``root`` will be
413 forced into the resulting config if appropriate.
414 outfile : `lss.resources.ResourcePathExpression`, optional
415 If not-`None`, the output configuration will be written to this
416 location rather than into the repository itself. Can be a URI
417 string. Can refer to a directory that will be used to write
418 ``butler.yaml``.
419 overwrite : `bool`, optional
420 Create a new configuration file even if one already exists
421 in the specified output location. Default is to raise
422 an exception.
424 Returns
425 -------
426 config : `Config`
427 The updated `Config` instance written to the repo.
429 Raises
430 ------
431 ValueError
432 Raised if a ButlerConfig or ConfigSubset is passed instead of a
433 regular Config (as these subclasses would make it impossible to
434 support ``standalone=False``).
435 FileExistsError
436 Raised if the output config file already exists.
437 os.error
438 Raised if the directory does not exist, exists but is not a
439 directory, or cannot be created.
441 Notes
442 -----
443 Note that when ``standalone=False`` (the default), the configuration
444 search path (see `ConfigSubset.defaultSearchPaths`) that was used to
445 construct the repository should also be used to construct any Butlers
446 to avoid configuration inconsistencies.
447 """
448 if isinstance(config, (ButlerConfig, ConfigSubset)):
449 raise ValueError("makeRepo must be passed a regular Config without defaults applied.")
451 # Ensure that the root of the repository exists or can be made
452 root_uri = ResourcePath(root, forceDirectory=True)
453 root_uri.mkdir()
455 config = Config(config)
457 # If we are creating a new repo from scratch with relative roots,
458 # do not propagate an explicit root from the config file
459 if "root" in config:
460 del config["root"]
462 full = ButlerConfig(config, searchPaths=searchPaths) # this applies defaults
463 imported_class = doImportType(full["datastore", "cls"])
464 if not issubclass(imported_class, Datastore):
465 raise TypeError(f"Imported datastore class {full['datastore', 'cls']} is not a Datastore")
466 datastoreClass: Type[Datastore] = imported_class
467 datastoreClass.setConfigRoot(BUTLER_ROOT_TAG, config, full, overwrite=forceConfigRoot)
469 # if key exists in given config, parse it, otherwise parse the defaults
470 # in the expanded config
471 if config.get(("registry", "db")):
472 registryConfig = RegistryConfig(config)
473 else:
474 registryConfig = RegistryConfig(full)
475 defaultDatabaseUri = registryConfig.makeDefaultDatabaseUri(BUTLER_ROOT_TAG)
476 if defaultDatabaseUri is not None:
477 Config.updateParameters(
478 RegistryConfig, config, full, toUpdate={"db": defaultDatabaseUri}, overwrite=forceConfigRoot
479 )
480 else:
481 Config.updateParameters(RegistryConfig, config, full, toCopy=("db",), overwrite=forceConfigRoot)
483 if standalone:
484 config.merge(full)
485 else:
486 # Always expand the registry.managers section into the per-repo
487 # config, because after the database schema is created, it's not
488 # allowed to change anymore. Note that in the standalone=True
489 # branch, _everything_ in the config is expanded, so there's no
490 # need to special case this.
491 Config.updateParameters(RegistryConfig, config, full, toMerge=("managers",), overwrite=False)
492 configURI: ResourcePathExpression
493 if outfile is not None:
494 # When writing to a separate location we must include
495 # the root of the butler repo in the config else it won't know
496 # where to look.
497 config["root"] = root_uri.geturl()
498 configURI = outfile
499 else:
500 configURI = root_uri
501 # Strip obscore configuration, if it is present, before writing config
502 # to a file, obscore config will be stored in registry.
503 config_to_write = config
504 if (obscore_config_key := ("registry", "managers", "obscore", "config")) in config:
505 config_to_write = config.copy()
506 del config_to_write[obscore_config_key]
507 config_to_write.dumpToUri(configURI, overwrite=overwrite)
509 # Create Registry and populate tables
510 registryConfig = RegistryConfig(config.get("registry"))
511 dimensionConfig = DimensionConfig(dimensionConfig)
512 Registry.createFromConfig(registryConfig, dimensionConfig=dimensionConfig, butlerRoot=root_uri)
514 log.verbose("Wrote new Butler configuration file to %s", configURI)
516 return config
518 @classmethod
519 def _unpickle(
520 cls,
521 config: ButlerConfig,
522 collections: Optional[tuple[str, ...]],
523 run: Optional[str],
524 defaultDataId: Dict[str, str],
525 writeable: bool,
526 ) -> Butler:
527 """Callable used to unpickle a Butler.
529 We prefer not to use ``Butler.__init__`` directly so we can force some
530 of its many arguments to be keyword-only (note that ``__reduce__``
531 can only invoke callables with positional arguments).
533 Parameters
534 ----------
535 config : `ButlerConfig`
536 Butler configuration, already coerced into a true `ButlerConfig`
537 instance (and hence after any search paths for overrides have been
538 utilized).
539 collections : `tuple` [ `str` ]
540 Names of the default collections to read from.
541 run : `str`, optional
542 Name of the default `~CollectionType.RUN` collection to write to.
543 defaultDataId : `dict` [ `str`, `str` ]
544 Default data ID values.
545 writeable : `bool`
546 Whether the Butler should support write operations.
548 Returns
549 -------
550 butler : `Butler`
551 A new `Butler` instance.
552 """
553 # MyPy doesn't recognize that the kwargs below are totally valid; it
554 # seems to think '**defaultDataId* is a _positional_ argument!
555 return cls(
556 config=config,
557 collections=collections,
558 run=run,
559 writeable=writeable,
560 **defaultDataId, # type: ignore
561 )
563 def __reduce__(self) -> tuple:
564 """Support pickling."""
565 return (
566 Butler._unpickle,
567 (
568 self._config,
569 self.collections,
570 self.run,
571 self.registry.defaults.dataId.byName(),
572 self.registry.isWriteable(),
573 ),
574 )
576 def __str__(self) -> str:
577 return "Butler(collections={}, run={}, datastore='{}', registry='{}')".format(
578 self.collections, self.run, self.datastore, self.registry
579 )
581 def isWriteable(self) -> bool:
582 """Return `True` if this `Butler` supports write operations."""
583 return self.registry.isWriteable()
585 @contextlib.contextmanager
586 def transaction(self) -> Iterator[None]:
587 """Context manager supporting `Butler` transactions.
589 Transactions can be nested.
590 """
591 with self.registry.transaction():
592 with self.datastore.transaction():
593 yield
595 def _standardizeArgs(
596 self,
597 datasetRefOrType: Union[DatasetRef, DatasetType, str],
598 dataId: Optional[DataId] = None,
599 for_put: bool = True,
600 **kwargs: Any,
601 ) -> Tuple[DatasetType, Optional[DataId]]:
602 """Standardize the arguments passed to several Butler APIs.
604 Parameters
605 ----------
606 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
607 When `DatasetRef` the `dataId` should be `None`.
608 Otherwise the `DatasetType` or name thereof.
609 dataId : `dict` or `DataCoordinate`
610 A `dict` of `Dimension` link name, value pairs that label the
611 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
612 should be provided as the second argument.
613 for_put : `bool`, optional
614 If `True` this call is invoked as part of a `Butler.put()`.
615 Otherwise it is assumed to be part of a `Butler.get()`. This
616 parameter is only relevant if there is dataset type
617 inconsistency.
618 **kwargs
619 Additional keyword arguments used to augment or construct a
620 `DataCoordinate`. See `DataCoordinate.standardize`
621 parameters.
623 Returns
624 -------
625 datasetType : `DatasetType`
626 A `DatasetType` instance extracted from ``datasetRefOrType``.
627 dataId : `dict` or `DataId`, optional
628 Argument that can be used (along with ``kwargs``) to construct a
629 `DataId`.
631 Notes
632 -----
633 Butler APIs that conceptually need a DatasetRef also allow passing a
634 `DatasetType` (or the name of one) and a `DataId` (or a dict and
635 keyword arguments that can be used to construct one) separately. This
636 method accepts those arguments and always returns a true `DatasetType`
637 and a `DataId` or `dict`.
639 Standardization of `dict` vs `DataId` is best handled by passing the
640 returned ``dataId`` (and ``kwargs``) to `Registry` APIs, which are
641 generally similarly flexible.
642 """
643 externalDatasetType: Optional[DatasetType] = None
644 internalDatasetType: Optional[DatasetType] = None
645 if isinstance(datasetRefOrType, DatasetRef):
646 if dataId is not None or kwargs:
647 raise ValueError("DatasetRef given, cannot use dataId as well")
648 externalDatasetType = datasetRefOrType.datasetType
649 dataId = datasetRefOrType.dataId
650 else:
651 # Don't check whether DataId is provided, because Registry APIs
652 # can usually construct a better error message when it wasn't.
653 if isinstance(datasetRefOrType, DatasetType):
654 externalDatasetType = datasetRefOrType
655 else:
656 internalDatasetType = self.registry.getDatasetType(datasetRefOrType)
658 # Check that they are self-consistent
659 if externalDatasetType is not None:
660 internalDatasetType = self.registry.getDatasetType(externalDatasetType.name)
661 if externalDatasetType != internalDatasetType:
662 # We can allow differences if they are compatible, depending
663 # on whether this is a get or a put. A get requires that
664 # the python type associated with the datastore can be
665 # converted to the user type. A put requires that the user
666 # supplied python type can be converted to the internal
667 # type expected by registry.
668 relevantDatasetType = internalDatasetType
669 if for_put:
670 is_compatible = internalDatasetType.is_compatible_with(externalDatasetType)
671 else:
672 is_compatible = externalDatasetType.is_compatible_with(internalDatasetType)
673 relevantDatasetType = externalDatasetType
674 if not is_compatible:
675 raise ValueError(
676 f"Supplied dataset type ({externalDatasetType}) inconsistent with "
677 f"registry definition ({internalDatasetType})"
678 )
679 # Override the internal definition.
680 internalDatasetType = relevantDatasetType
682 assert internalDatasetType is not None
683 return internalDatasetType, dataId
685 def _rewrite_data_id(
686 self, dataId: Optional[DataId], datasetType: DatasetType, **kwargs: Any
687 ) -> Tuple[Optional[DataId], Dict[str, Any]]:
688 """Rewrite a data ID taking into account dimension records.
690 Take a Data ID and keyword args and rewrite it if necessary to
691 allow the user to specify dimension records rather than dimension
692 primary values.
694 This allows a user to include a dataId dict with keys of
695 ``exposure.day_obs`` and ``exposure.seq_num`` instead of giving
696 the integer exposure ID. It also allows a string to be given
697 for a dimension value rather than the integer ID if that is more
698 convenient. For example, rather than having to specifyin the
699 detector with ``detector.full_name``, a string given for ``detector``
700 will be interpreted as the full name and converted to the integer
701 value.
703 Keyword arguments can also use strings for dimensions like detector
704 and exposure but python does not allow them to include ``.`` and
705 so the ``exposure.day_obs`` syntax can not be used in a keyword
706 argument.
708 Parameters
709 ----------
710 dataId : `dict` or `DataCoordinate`
711 A `dict` of `Dimension` link name, value pairs that will label the
712 `DatasetRef` within a Collection.
713 datasetType : `DatasetType`
714 The dataset type associated with this dataId. Required to
715 determine the relevant dimensions.
716 **kwargs
717 Additional keyword arguments used to augment or construct a
718 `DataId`. See `DataId` parameters.
720 Returns
721 -------
722 dataId : `dict` or `DataCoordinate`
723 The, possibly rewritten, dataId. If given a `DataCoordinate` and
724 no keyword arguments, the original dataId will be returned
725 unchanged.
726 **kwargs : `dict`
727 Any unused keyword arguments (would normally be empty dict).
728 """
729 # Do nothing if we have a standalone DataCoordinate.
730 if isinstance(dataId, DataCoordinate) and not kwargs:
731 return dataId, kwargs
733 # Process dimension records that are using record information
734 # rather than ids
735 newDataId: Dict[str, DataIdValue] = {}
736 byRecord: Dict[str, Dict[str, Any]] = defaultdict(dict)
738 # if all the dataId comes from keyword parameters we do not need
739 # to do anything here because they can't be of the form
740 # exposure.obs_id because a "." is not allowed in a keyword parameter.
741 if dataId:
742 for k, v in dataId.items():
743 # If we have a Dimension we do not need to do anything
744 # because it cannot be a compound key.
745 if isinstance(k, str) and "." in k:
746 # Someone is using a more human-readable dataId
747 dimensionName, record = k.split(".", 1)
748 byRecord[dimensionName][record] = v
749 elif isinstance(k, Dimension):
750 newDataId[k.name] = v
751 else:
752 newDataId[k] = v
754 # Go through the updated dataId and check the type in case someone is
755 # using an alternate key. We have already filtered out the compound
756 # keys dimensions.record format.
757 not_dimensions = {}
759 # Will need to look in the dataId and the keyword arguments
760 # and will remove them if they need to be fixed or are unrecognized.
761 for dataIdDict in (newDataId, kwargs):
762 # Use a list so we can adjust the dict safely in the loop
763 for dimensionName in list(dataIdDict):
764 value = dataIdDict[dimensionName]
765 try:
766 dimension = self.registry.dimensions.getStaticDimensions()[dimensionName]
767 except KeyError:
768 # This is not a real dimension
769 not_dimensions[dimensionName] = value
770 del dataIdDict[dimensionName]
771 continue
773 # Convert an integral type to an explicit int to simplify
774 # comparisons here
775 if isinstance(value, numbers.Integral):
776 value = int(value)
778 if not isinstance(value, dimension.primaryKey.getPythonType()):
779 for alternate in dimension.alternateKeys:
780 if isinstance(value, alternate.getPythonType()):
781 byRecord[dimensionName][alternate.name] = value
782 del dataIdDict[dimensionName]
783 log.debug(
784 "Converting dimension %s to %s.%s=%s",
785 dimensionName,
786 dimensionName,
787 alternate.name,
788 value,
789 )
790 break
791 else:
792 log.warning(
793 "Type mismatch found for value '%r' provided for dimension %s. "
794 "Could not find matching alternative (primary key has type %s) "
795 "so attempting to use as-is.",
796 value,
797 dimensionName,
798 dimension.primaryKey.getPythonType(),
799 )
801 # By this point kwargs and newDataId should only include valid
802 # dimensions. Merge kwargs in to the new dataId and log if there
803 # are dimensions in both (rather than calling update).
804 for k, v in kwargs.items():
805 if k in newDataId and newDataId[k] != v:
806 log.debug(
807 "Keyword arg %s overriding explicit value in dataId of %s with %s", k, newDataId[k], v
808 )
809 newDataId[k] = v
810 # No need to retain any values in kwargs now.
811 kwargs = {}
813 # If we have some unrecognized dimensions we have to try to connect
814 # them to records in other dimensions. This is made more complicated
815 # by some dimensions having records with clashing names. A mitigation
816 # is that we can tell by this point which dimensions are missing
817 # for the DatasetType but this does not work for calibrations
818 # where additional dimensions can be used to constrain the temporal
819 # axis.
820 if not_dimensions:
821 # Search for all dimensions even if we have been given a value
822 # explicitly. In some cases records are given as well as the
823 # actually dimension and this should not be an error if they
824 # match.
825 mandatoryDimensions = datasetType.dimensions.names # - provided
827 candidateDimensions: Set[str] = set()
828 candidateDimensions.update(mandatoryDimensions)
830 # For calibrations we may well be needing temporal dimensions
831 # so rather than always including all dimensions in the scan
832 # restrict things a little. It is still possible for there
833 # to be confusion over day_obs in visit vs exposure for example.
834 # If we are not searching calibration collections things may
835 # fail but they are going to fail anyway because of the
836 # ambiguousness of the dataId...
837 if datasetType.isCalibration():
838 for dim in self.registry.dimensions.getStaticDimensions():
839 if dim.temporal:
840 candidateDimensions.add(str(dim))
842 # Look up table for the first association with a dimension
843 guessedAssociation: Dict[str, Dict[str, Any]] = defaultdict(dict)
845 # Keep track of whether an item is associated with multiple
846 # dimensions.
847 counter: Counter[str] = Counter()
848 assigned: Dict[str, Set[str]] = defaultdict(set)
850 # Go through the missing dimensions and associate the
851 # given names with records within those dimensions
852 matched_dims = set()
853 for dimensionName in candidateDimensions:
854 dimension = self.registry.dimensions.getStaticDimensions()[dimensionName]
855 fields = dimension.metadata.names | dimension.uniqueKeys.names
856 for field in not_dimensions:
857 if field in fields:
858 guessedAssociation[dimensionName][field] = not_dimensions[field]
859 counter[dimensionName] += 1
860 assigned[field].add(dimensionName)
861 matched_dims.add(field)
863 # Calculate the fields that matched nothing.
864 never_found = set(not_dimensions) - matched_dims
866 if never_found:
867 raise ValueError(f"Unrecognized keyword args given: {never_found}")
869 # There is a chance we have allocated a single dataId item
870 # to multiple dimensions. Need to decide which should be retained.
871 # For now assume that the most popular alternative wins.
872 # This means that day_obs with seq_num will result in
873 # exposure.day_obs and not visit.day_obs
874 # Also prefer an explicitly missing dimension over an inferred
875 # temporal dimension.
876 for fieldName, assignedDimensions in assigned.items():
877 if len(assignedDimensions) > 1:
878 # Pick the most popular (preferring mandatory dimensions)
879 requiredButMissing = assignedDimensions.intersection(mandatoryDimensions)
880 if requiredButMissing:
881 candidateDimensions = requiredButMissing
882 else:
883 candidateDimensions = assignedDimensions
885 # If this is a choice between visit and exposure and
886 # neither was a required part of the dataset type,
887 # (hence in this branch) always prefer exposure over
888 # visit since exposures are always defined and visits
889 # are defined from exposures.
890 if candidateDimensions == {"exposure", "visit"}:
891 candidateDimensions = {"exposure"}
893 # Select the relevant items and get a new restricted
894 # counter.
895 theseCounts = {k: v for k, v in counter.items() if k in candidateDimensions}
896 duplicatesCounter: Counter[str] = Counter()
897 duplicatesCounter.update(theseCounts)
899 # Choose the most common. If they are equally common
900 # we will pick the one that was found first.
901 # Returns a list of tuples
902 selected = duplicatesCounter.most_common(1)[0][0]
904 log.debug(
905 "Ambiguous dataId entry '%s' associated with multiple dimensions: %s."
906 " Removed ambiguity by choosing dimension %s.",
907 fieldName,
908 ", ".join(assignedDimensions),
909 selected,
910 )
912 for candidateDimension in assignedDimensions:
913 if candidateDimension != selected:
914 del guessedAssociation[candidateDimension][fieldName]
916 # Update the record look up dict with the new associations
917 for dimensionName, values in guessedAssociation.items():
918 if values: # A dict might now be empty
919 log.debug("Assigned non-dimension dataId keys to dimension %s: %s", dimensionName, values)
920 byRecord[dimensionName].update(values)
922 if byRecord:
923 # Some record specifiers were found so we need to convert
924 # them to the Id form
925 for dimensionName, values in byRecord.items():
926 if dimensionName in newDataId:
927 log.debug(
928 "DataId specified explicit %s dimension value of %s in addition to"
929 " general record specifiers for it of %s. Ignoring record information.",
930 dimensionName,
931 newDataId[dimensionName],
932 str(values),
933 )
934 # Get the actual record and compare with these values.
935 try:
936 recs = list(self.registry.queryDimensionRecords(dimensionName, dataId=newDataId))
937 except DataIdError:
938 raise ValueError(
939 f"Could not find dimension '{dimensionName}'"
940 f" with dataId {newDataId} as part of comparing with"
941 f" record values {byRecord[dimensionName]}"
942 ) from None
943 if len(recs) == 1:
944 errmsg: List[str] = []
945 for k, v in values.items():
946 if (recval := getattr(recs[0], k)) != v:
947 errmsg.append(f"{k}({recval} != {v})")
948 if errmsg:
949 raise ValueError(
950 f"Dimension {dimensionName} in dataId has explicit value"
951 " inconsistent with records: " + ", ".join(errmsg)
952 )
953 else:
954 # Multiple matches for an explicit dimension
955 # should never happen but let downstream complain.
956 pass
957 continue
959 # Build up a WHERE expression
960 bind = {k: v for k, v in values.items()}
961 where = " AND ".join(f"{dimensionName}.{k} = {k}" for k in bind)
963 # Hopefully we get a single record that matches
964 records = set(
965 self.registry.queryDimensionRecords(
966 dimensionName, dataId=newDataId, where=where, bind=bind, **kwargs
967 )
968 )
970 if len(records) != 1:
971 if len(records) > 1:
972 # visit can have an ambiguous answer without involving
973 # visit_system. The default visit_system is defined
974 # by the instrument.
975 if (
976 dimensionName == "visit"
977 and "visit_system_membership" in self.registry.dimensions
978 and "visit_system" in self.registry.dimensions["instrument"].metadata
979 ):
980 instrument_records = list(
981 self.registry.queryDimensionRecords(
982 "instrument",
983 dataId=newDataId,
984 **kwargs,
985 )
986 )
987 if len(instrument_records) == 1:
988 visit_system = instrument_records[0].visit_system
989 if visit_system is None:
990 # Set to a value that will never match.
991 visit_system = -1
993 # Look up each visit in the
994 # visit_system_membership records.
995 for rec in records:
996 membership = list(
997 self.registry.queryDimensionRecords(
998 # Use bind to allow zero results.
999 # This is a fully-specified query.
1000 "visit_system_membership",
1001 where="instrument = inst AND visit_system = system AND visit = v",
1002 bind=dict(
1003 inst=instrument_records[0].name, system=visit_system, v=rec.id
1004 ),
1005 )
1006 )
1007 if membership:
1008 # This record is the right answer.
1009 records = set([rec])
1010 break
1012 # The ambiguity may have been resolved so check again.
1013 if len(records) > 1:
1014 log.debug("Received %d records from constraints of %s", len(records), str(values))
1015 for r in records:
1016 log.debug("- %s", str(r))
1017 raise ValueError(
1018 f"DataId specification for dimension {dimensionName} is not"
1019 f" uniquely constrained to a single dataset by {values}."
1020 f" Got {len(records)} results."
1021 )
1022 else:
1023 raise ValueError(
1024 f"DataId specification for dimension {dimensionName} matched no"
1025 f" records when constrained by {values}"
1026 )
1028 # Get the primary key from the real dimension object
1029 dimension = self.registry.dimensions.getStaticDimensions()[dimensionName]
1030 if not isinstance(dimension, Dimension):
1031 raise RuntimeError(
1032 f"{dimension.name} is not a true dimension, and cannot be used in data IDs."
1033 )
1034 newDataId[dimensionName] = getattr(records.pop(), dimension.primaryKey.name)
1036 return newDataId, kwargs
1038 def _findDatasetRef(
1039 self,
1040 datasetRefOrType: Union[DatasetRef, DatasetType, str],
1041 dataId: Optional[DataId] = None,
1042 *,
1043 collections: Any = None,
1044 allowUnresolved: bool = False,
1045 **kwargs: Any,
1046 ) -> DatasetRef:
1047 """Shared logic for methods that start with a search for a dataset in
1048 the registry.
1050 Parameters
1051 ----------
1052 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
1053 When `DatasetRef` the `dataId` should be `None`.
1054 Otherwise the `DatasetType` or name thereof.
1055 dataId : `dict` or `DataCoordinate`, optional
1056 A `dict` of `Dimension` link name, value pairs that label the
1057 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
1058 should be provided as the first argument.
1059 collections : Any, optional
1060 Collections to be searched, overriding ``self.collections``.
1061 Can be any of the types supported by the ``collections`` argument
1062 to butler construction.
1063 allowUnresolved : `bool`, optional
1064 If `True`, return an unresolved `DatasetRef` if finding a resolved
1065 one in the `Registry` fails. Defaults to `False`.
1066 **kwargs
1067 Additional keyword arguments used to augment or construct a
1068 `DataId`. See `DataId` parameters.
1070 Returns
1071 -------
1072 ref : `DatasetRef`
1073 A reference to the dataset identified by the given arguments.
1075 Raises
1076 ------
1077 LookupError
1078 Raised if no matching dataset exists in the `Registry` (and
1079 ``allowUnresolved is False``).
1080 ValueError
1081 Raised if a resolved `DatasetRef` was passed as an input, but it
1082 differs from the one found in the registry.
1083 TypeError
1084 Raised if no collections were provided.
1085 """
1086 datasetType, dataId = self._standardizeArgs(datasetRefOrType, dataId, for_put=False, **kwargs)
1087 if isinstance(datasetRefOrType, DatasetRef):
1088 idNumber = datasetRefOrType.id
1089 else:
1090 idNumber = None
1091 timespan: Optional[Timespan] = None
1093 dataId, kwargs = self._rewrite_data_id(dataId, datasetType, **kwargs)
1095 if datasetType.isCalibration():
1096 # Because this is a calibration dataset, first try to make a
1097 # standardize the data ID without restricting the dimensions to
1098 # those of the dataset type requested, because there may be extra
1099 # dimensions that provide temporal information for a validity-range
1100 # lookup.
1101 dataId = DataCoordinate.standardize(
1102 dataId, universe=self.registry.dimensions, defaults=self.registry.defaults.dataId, **kwargs
1103 )
1104 if dataId.graph.temporal:
1105 dataId = self.registry.expandDataId(dataId)
1106 timespan = dataId.timespan
1107 else:
1108 # Standardize the data ID to just the dimensions of the dataset
1109 # type instead of letting registry.findDataset do it, so we get the
1110 # result even if no dataset is found.
1111 dataId = DataCoordinate.standardize(
1112 dataId, graph=datasetType.dimensions, defaults=self.registry.defaults.dataId, **kwargs
1113 )
1114 # Always lookup the DatasetRef, even if one is given, to ensure it is
1115 # present in the current collection.
1116 ref = self.registry.findDataset(datasetType, dataId, collections=collections, timespan=timespan)
1117 if ref is None:
1118 if allowUnresolved:
1119 return DatasetRef(datasetType, dataId)
1120 else:
1121 if collections is None:
1122 collections = self.registry.defaults.collections
1123 raise LookupError(
1124 f"Dataset {datasetType.name} with data ID {dataId} "
1125 f"could not be found in collections {collections}."
1126 )
1127 if idNumber is not None and idNumber != ref.id:
1128 if collections is None:
1129 collections = self.registry.defaults.collections
1130 raise ValueError(
1131 f"DatasetRef.id provided ({idNumber}) does not match "
1132 f"id ({ref.id}) in registry in collections {collections}."
1133 )
1134 if datasetType != ref.datasetType:
1135 # If they differ it is because the user explicitly specified
1136 # a compatible dataset type to this call rather than using the
1137 # registry definition. The DatasetRef must therefore be recreated
1138 # using the user definition such that the expected type is
1139 # returned.
1140 ref = DatasetRef(datasetType, ref.dataId, run=ref.run, id=ref.id)
1142 return ref
1144 @transactional
1145 def putDirect(self, obj: Any, ref: DatasetRef) -> DatasetRef:
1146 # Docstring inherited.
1147 (imported_ref,) = self.registry._importDatasets(
1148 [ref],
1149 expand=True,
1150 )
1151 if imported_ref.id != ref.getCheckedId():
1152 raise RuntimeError("This registry configuration does not support putDirect.")
1153 self.datastore.put(obj, ref)
1154 return ref
1156 @transactional
1157 def put(
1158 self,
1159 obj: Any,
1160 datasetRefOrType: Union[DatasetRef, DatasetType, str],
1161 dataId: Optional[DataId] = None,
1162 *,
1163 run: Optional[str] = None,
1164 **kwargs: Any,
1165 ) -> DatasetRef:
1166 """Store and register a dataset.
1168 Parameters
1169 ----------
1170 obj : `object`
1171 The dataset.
1172 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
1173 When `DatasetRef` is provided, ``dataId`` should be `None`.
1174 Otherwise the `DatasetType` or name thereof.
1175 dataId : `dict` or `DataCoordinate`
1176 A `dict` of `Dimension` link name, value pairs that label the
1177 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
1178 should be provided as the second argument.
1179 run : `str`, optional
1180 The name of the run the dataset should be added to, overriding
1181 ``self.run``.
1182 **kwargs
1183 Additional keyword arguments used to augment or construct a
1184 `DataCoordinate`. See `DataCoordinate.standardize`
1185 parameters.
1187 Returns
1188 -------
1189 ref : `DatasetRef`
1190 A reference to the stored dataset, updated with the correct id if
1191 given.
1193 Raises
1194 ------
1195 TypeError
1196 Raised if the butler is read-only or if no run has been provided.
1197 """
1198 log.debug("Butler put: %s, dataId=%s, run=%s", datasetRefOrType, dataId, run)
1199 if not self.isWriteable():
1200 raise TypeError("Butler is read-only.")
1201 datasetType, dataId = self._standardizeArgs(datasetRefOrType, dataId, **kwargs)
1202 if isinstance(datasetRefOrType, DatasetRef) and datasetRefOrType.id is not None:
1203 raise ValueError("DatasetRef must not be in registry, must have None id")
1205 # Handle dimension records in dataId
1206 dataId, kwargs = self._rewrite_data_id(dataId, datasetType, **kwargs)
1208 # Add Registry Dataset entry.
1209 dataId = self.registry.expandDataId(dataId, graph=datasetType.dimensions, **kwargs)
1211 # For an execution butler the datasets will be pre-defined.
1212 # If the butler is configured that way datasets should only be inserted
1213 # if they do not already exist in registry. Trying and catching
1214 # ConflictingDefinitionError will not work because the transaction
1215 # will be corrupted. Instead, in this mode always check first.
1216 ref = None
1217 ref_is_predefined = False
1218 if self._allow_put_of_predefined_dataset:
1219 # Get the matching ref for this run.
1220 ref = self.registry.findDataset(datasetType, collections=run, dataId=dataId)
1222 if ref:
1223 # Must be expanded form for datastore templating
1224 dataId = self.registry.expandDataId(dataId, graph=datasetType.dimensions)
1225 ref = ref.expanded(dataId)
1226 ref_is_predefined = True
1228 if not ref:
1229 (ref,) = self.registry.insertDatasets(datasetType, run=run, dataIds=[dataId])
1231 # If the ref is predefined it is possible that the datastore also
1232 # has the record. Asking datastore to put it again will result in
1233 # the artifact being recreated, overwriting previous, then will cause
1234 # a failure in writing the record which will cause the artifact
1235 # to be removed. Much safer to ask first before attempting to
1236 # overwrite. Race conditions should not be an issue for the
1237 # execution butler environment.
1238 if ref_is_predefined:
1239 if self.datastore.knows(ref):
1240 raise ConflictingDefinitionError(f"Dataset associated {ref} already exists.")
1242 self.datastore.put(obj, ref)
1244 return ref
1246 def getDirect(
1247 self,
1248 ref: DatasetRef,
1249 *,
1250 parameters: Optional[Dict[str, Any]] = None,
1251 storageClass: Optional[Union[StorageClass, str]] = None,
1252 ) -> Any:
1253 """Retrieve a stored dataset.
1255 Unlike `Butler.get`, this method allows datasets outside the Butler's
1256 collection to be read as long as the `DatasetRef` that identifies them
1257 can be obtained separately.
1259 Parameters
1260 ----------
1261 ref : `DatasetRef`
1262 Resolved reference to an already stored dataset.
1263 parameters : `dict`
1264 Additional StorageClass-defined options to control reading,
1265 typically used to efficiently read only a subset of the dataset.
1266 storageClass : `StorageClass` or `str`, optional
1267 The storage class to be used to override the Python type
1268 returned by this method. By default the returned type matches
1269 the dataset type definition for this dataset. Specifying a
1270 read `StorageClass` can force a different type to be returned.
1271 This type must be compatible with the original type.
1273 Returns
1274 -------
1275 obj : `object`
1276 The dataset.
1277 """
1278 return self.datastore.get(ref, parameters=parameters, storageClass=storageClass)
1280 def getDirectDeferred(
1281 self,
1282 ref: DatasetRef,
1283 *,
1284 parameters: Union[dict, None] = None,
1285 storageClass: str | StorageClass | None = None,
1286 ) -> DeferredDatasetHandle:
1287 """Create a `DeferredDatasetHandle` which can later retrieve a dataset,
1288 from a resolved `DatasetRef`.
1290 Parameters
1291 ----------
1292 ref : `DatasetRef`
1293 Resolved reference to an already stored dataset.
1294 parameters : `dict`
1295 Additional StorageClass-defined options to control reading,
1296 typically used to efficiently read only a subset of the dataset.
1297 storageClass : `StorageClass` or `str`, optional
1298 The storage class to be used to override the Python type
1299 returned by this method. By default the returned type matches
1300 the dataset type definition for this dataset. Specifying a
1301 read `StorageClass` can force a different type to be returned.
1302 This type must be compatible with the original type.
1304 Returns
1305 -------
1306 obj : `DeferredDatasetHandle`
1307 A handle which can be used to retrieve a dataset at a later time.
1309 Raises
1310 ------
1311 AmbiguousDatasetError
1312 Raised if ``ref.id is None``, i.e. the reference is unresolved.
1313 """
1314 if ref.id is None:
1315 raise AmbiguousDatasetError(
1316 f"Dataset of type {ref.datasetType.name} with data ID {ref.dataId} is not resolved."
1317 )
1318 return DeferredDatasetHandle(butler=self, ref=ref, parameters=parameters, storageClass=storageClass)
1320 def getDeferred(
1321 self,
1322 datasetRefOrType: Union[DatasetRef, DatasetType, str],
1323 dataId: Optional[DataId] = None,
1324 *,
1325 parameters: Union[dict, None] = None,
1326 collections: Any = None,
1327 storageClass: str | StorageClass | None = None,
1328 **kwargs: Any,
1329 ) -> DeferredDatasetHandle:
1330 """Create a `DeferredDatasetHandle` which can later retrieve a dataset,
1331 after an immediate registry lookup.
1333 Parameters
1334 ----------
1335 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
1336 When `DatasetRef` the `dataId` should be `None`.
1337 Otherwise the `DatasetType` or name thereof.
1338 dataId : `dict` or `DataCoordinate`, optional
1339 A `dict` of `Dimension` link name, value pairs that label the
1340 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
1341 should be provided as the first argument.
1342 parameters : `dict`
1343 Additional StorageClass-defined options to control reading,
1344 typically used to efficiently read only a subset of the dataset.
1345 collections : Any, optional
1346 Collections to be searched, overriding ``self.collections``.
1347 Can be any of the types supported by the ``collections`` argument
1348 to butler construction.
1349 storageClass : `StorageClass` or `str`, optional
1350 The storage class to be used to override the Python type
1351 returned by this method. By default the returned type matches
1352 the dataset type definition for this dataset. Specifying a
1353 read `StorageClass` can force a different type to be returned.
1354 This type must be compatible with the original type.
1355 **kwargs
1356 Additional keyword arguments used to augment or construct a
1357 `DataId`. See `DataId` parameters.
1359 Returns
1360 -------
1361 obj : `DeferredDatasetHandle`
1362 A handle which can be used to retrieve a dataset at a later time.
1364 Raises
1365 ------
1366 LookupError
1367 Raised if no matching dataset exists in the `Registry` (and
1368 ``allowUnresolved is False``).
1369 ValueError
1370 Raised if a resolved `DatasetRef` was passed as an input, but it
1371 differs from the one found in the registry.
1372 TypeError
1373 Raised if no collections were provided.
1374 """
1375 ref = self._findDatasetRef(datasetRefOrType, dataId, collections=collections, **kwargs)
1376 return DeferredDatasetHandle(butler=self, ref=ref, parameters=parameters, storageClass=storageClass)
1378 def get(
1379 self,
1380 datasetRefOrType: Union[DatasetRef, DatasetType, str],
1381 dataId: Optional[DataId] = None,
1382 *,
1383 parameters: Optional[Dict[str, Any]] = None,
1384 collections: Any = None,
1385 storageClass: Optional[Union[StorageClass, str]] = None,
1386 **kwargs: Any,
1387 ) -> Any:
1388 """Retrieve a stored dataset.
1390 Parameters
1391 ----------
1392 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
1393 When `DatasetRef` the `dataId` should be `None`.
1394 Otherwise the `DatasetType` or name thereof.
1395 dataId : `dict` or `DataCoordinate`
1396 A `dict` of `Dimension` link name, value pairs that label the
1397 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
1398 should be provided as the first argument.
1399 parameters : `dict`
1400 Additional StorageClass-defined options to control reading,
1401 typically used to efficiently read only a subset of the dataset.
1402 collections : Any, optional
1403 Collections to be searched, overriding ``self.collections``.
1404 Can be any of the types supported by the ``collections`` argument
1405 to butler construction.
1406 storageClass : `StorageClass` or `str`, optional
1407 The storage class to be used to override the Python type
1408 returned by this method. By default the returned type matches
1409 the dataset type definition for this dataset. Specifying a
1410 read `StorageClass` can force a different type to be returned.
1411 This type must be compatible with the original type.
1412 **kwargs
1413 Additional keyword arguments used to augment or construct a
1414 `DataCoordinate`. See `DataCoordinate.standardize`
1415 parameters.
1417 Returns
1418 -------
1419 obj : `object`
1420 The dataset.
1422 Raises
1423 ------
1424 ValueError
1425 Raised if a resolved `DatasetRef` was passed as an input, but it
1426 differs from the one found in the registry.
1427 LookupError
1428 Raised if no matching dataset exists in the `Registry`.
1429 TypeError
1430 Raised if no collections were provided.
1432 Notes
1433 -----
1434 When looking up datasets in a `~CollectionType.CALIBRATION` collection,
1435 this method requires that the given data ID include temporal dimensions
1436 beyond the dimensions of the dataset type itself, in order to find the
1437 dataset with the appropriate validity range. For example, a "bias"
1438 dataset with native dimensions ``{instrument, detector}`` could be
1439 fetched with a ``{instrument, detector, exposure}`` data ID, because
1440 ``exposure`` is a temporal dimension.
1441 """
1442 log.debug("Butler get: %s, dataId=%s, parameters=%s", datasetRefOrType, dataId, parameters)
1443 ref = self._findDatasetRef(datasetRefOrType, dataId, collections=collections, **kwargs)
1444 return self.getDirect(ref, parameters=parameters, storageClass=storageClass)
1446 def getURIs(
1447 self,
1448 datasetRefOrType: Union[DatasetRef, DatasetType, str],
1449 dataId: Optional[DataId] = None,
1450 *,
1451 predict: bool = False,
1452 collections: Any = None,
1453 run: Optional[str] = None,
1454 **kwargs: Any,
1455 ) -> DatasetRefURIs:
1456 """Returns the URIs associated with the dataset.
1458 Parameters
1459 ----------
1460 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
1461 When `DatasetRef` the `dataId` should be `None`.
1462 Otherwise the `DatasetType` or name thereof.
1463 dataId : `dict` or `DataCoordinate`
1464 A `dict` of `Dimension` link name, value pairs that label the
1465 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
1466 should be provided as the first argument.
1467 predict : `bool`
1468 If `True`, allow URIs to be returned of datasets that have not
1469 been written.
1470 collections : Any, optional
1471 Collections to be searched, overriding ``self.collections``.
1472 Can be any of the types supported by the ``collections`` argument
1473 to butler construction.
1474 run : `str`, optional
1475 Run to use for predictions, overriding ``self.run``.
1476 **kwargs
1477 Additional keyword arguments used to augment or construct a
1478 `DataCoordinate`. See `DataCoordinate.standardize`
1479 parameters.
1481 Returns
1482 -------
1483 uris : `DatasetRefURIs`
1484 The URI to the primary artifact associated with this dataset (if
1485 the dataset was disassembled within the datastore this may be
1486 `None`), and the URIs to any components associated with the dataset
1487 artifact. (can be empty if there are no components).
1488 """
1489 ref = self._findDatasetRef(
1490 datasetRefOrType, dataId, allowUnresolved=predict, collections=collections, **kwargs
1491 )
1492 if ref.id is None: # only possible if predict is True
1493 if run is None:
1494 run = self.run
1495 if run is None:
1496 raise TypeError("Cannot predict location with run=None.")
1497 # Lie about ID, because we can't guess it, and only
1498 # Datastore.getURIs() will ever see it (and it doesn't use it).
1499 ref = ref.resolved(id=0, run=run)
1500 return self.datastore.getURIs(ref, predict)
1502 def getURI(
1503 self,
1504 datasetRefOrType: Union[DatasetRef, DatasetType, str],
1505 dataId: Optional[DataId] = None,
1506 *,
1507 predict: bool = False,
1508 collections: Any = None,
1509 run: Optional[str] = None,
1510 **kwargs: Any,
1511 ) -> ResourcePath:
1512 """Return the URI to the Dataset.
1514 Parameters
1515 ----------
1516 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
1517 When `DatasetRef` the `dataId` should be `None`.
1518 Otherwise the `DatasetType` or name thereof.
1519 dataId : `dict` or `DataCoordinate`
1520 A `dict` of `Dimension` link name, value pairs that label the
1521 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
1522 should be provided as the first argument.
1523 predict : `bool`
1524 If `True`, allow URIs to be returned of datasets that have not
1525 been written.
1526 collections : Any, optional
1527 Collections to be searched, overriding ``self.collections``.
1528 Can be any of the types supported by the ``collections`` argument
1529 to butler construction.
1530 run : `str`, optional
1531 Run to use for predictions, overriding ``self.run``.
1532 **kwargs
1533 Additional keyword arguments used to augment or construct a
1534 `DataCoordinate`. See `DataCoordinate.standardize`
1535 parameters.
1537 Returns
1538 -------
1539 uri : `lsst.resources.ResourcePath`
1540 URI pointing to the Dataset within the datastore. If the
1541 Dataset does not exist in the datastore, and if ``predict`` is
1542 `True`, the URI will be a prediction and will include a URI
1543 fragment "#predicted".
1544 If the datastore does not have entities that relate well
1545 to the concept of a URI the returned URI string will be
1546 descriptive. The returned URI is not guaranteed to be obtainable.
1548 Raises
1549 ------
1550 LookupError
1551 A URI has been requested for a dataset that does not exist and
1552 guessing is not allowed.
1553 ValueError
1554 Raised if a resolved `DatasetRef` was passed as an input, but it
1555 differs from the one found in the registry.
1556 TypeError
1557 Raised if no collections were provided.
1558 RuntimeError
1559 Raised if a URI is requested for a dataset that consists of
1560 multiple artifacts.
1561 """
1562 primary, components = self.getURIs(
1563 datasetRefOrType, dataId=dataId, predict=predict, collections=collections, run=run, **kwargs
1564 )
1566 if primary is None or components:
1567 raise RuntimeError(
1568 f"Dataset ({datasetRefOrType}) includes distinct URIs for components. "
1569 "Use Butler.getURIs() instead."
1570 )
1571 return primary
1573 def retrieveArtifacts(
1574 self,
1575 refs: Iterable[DatasetRef],
1576 destination: ResourcePathExpression,
1577 transfer: str = "auto",
1578 preserve_path: bool = True,
1579 overwrite: bool = False,
1580 ) -> List[ResourcePath]:
1581 """Retrieve the artifacts associated with the supplied refs.
1583 Parameters
1584 ----------
1585 refs : iterable of `DatasetRef`
1586 The datasets for which artifacts are to be retrieved.
1587 A single ref can result in multiple artifacts. The refs must
1588 be resolved.
1589 destination : `lsst.resources.ResourcePath` or `str`
1590 Location to write the artifacts.
1591 transfer : `str`, optional
1592 Method to use to transfer the artifacts. Must be one of the options
1593 supported by `~lsst.resources.ResourcePath.transfer_from()`.
1594 "move" is not allowed.
1595 preserve_path : `bool`, optional
1596 If `True` the full path of the artifact within the datastore
1597 is preserved. If `False` the final file component of the path
1598 is used.
1599 overwrite : `bool`, optional
1600 If `True` allow transfers to overwrite existing files at the
1601 destination.
1603 Returns
1604 -------
1605 targets : `list` of `lsst.resources.ResourcePath`
1606 URIs of file artifacts in destination location. Order is not
1607 preserved.
1609 Notes
1610 -----
1611 For non-file datastores the artifacts written to the destination
1612 may not match the representation inside the datastore. For example
1613 a hierarchical data structure in a NoSQL database may well be stored
1614 as a JSON file.
1615 """
1616 return self.datastore.retrieveArtifacts(
1617 refs,
1618 ResourcePath(destination),
1619 transfer=transfer,
1620 preserve_path=preserve_path,
1621 overwrite=overwrite,
1622 )
1624 def datasetExists(
1625 self,
1626 datasetRefOrType: Union[DatasetRef, DatasetType, str],
1627 dataId: Optional[DataId] = None,
1628 *,
1629 collections: Any = None,
1630 **kwargs: Any,
1631 ) -> bool:
1632 """Return True if the Dataset is actually present in the Datastore.
1634 Parameters
1635 ----------
1636 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
1637 When `DatasetRef` the `dataId` should be `None`.
1638 Otherwise the `DatasetType` or name thereof.
1639 dataId : `dict` or `DataCoordinate`
1640 A `dict` of `Dimension` link name, value pairs that label the
1641 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
1642 should be provided as the first argument.
1643 collections : Any, optional
1644 Collections to be searched, overriding ``self.collections``.
1645 Can be any of the types supported by the ``collections`` argument
1646 to butler construction.
1647 **kwargs
1648 Additional keyword arguments used to augment or construct a
1649 `DataCoordinate`. See `DataCoordinate.standardize`
1650 parameters.
1652 Raises
1653 ------
1654 LookupError
1655 Raised if the dataset is not even present in the Registry.
1656 ValueError
1657 Raised if a resolved `DatasetRef` was passed as an input, but it
1658 differs from the one found in the registry.
1659 TypeError
1660 Raised if no collections were provided.
1661 """
1662 ref = self._findDatasetRef(datasetRefOrType, dataId, collections=collections, **kwargs)
1663 return self.datastore.exists(ref)
1665 def removeRuns(self, names: Iterable[str], unstore: bool = True) -> None:
1666 """Remove one or more `~CollectionType.RUN` collections and the
1667 datasets within them.
1669 Parameters
1670 ----------
1671 names : `Iterable` [ `str` ]
1672 The names of the collections to remove.
1673 unstore : `bool`, optional
1674 If `True` (default), delete datasets from all datastores in which
1675 they are present, and attempt to rollback the registry deletions if
1676 datastore deletions fail (which may not always be possible). If
1677 `False`, datastore records for these datasets are still removed,
1678 but any artifacts (e.g. files) will not be.
1680 Raises
1681 ------
1682 TypeError
1683 Raised if one or more collections are not of type
1684 `~CollectionType.RUN`.
1685 """
1686 if not self.isWriteable():
1687 raise TypeError("Butler is read-only.")
1688 names = list(names)
1689 refs: List[DatasetRef] = []
1690 for name in names:
1691 collectionType = self.registry.getCollectionType(name)
1692 if collectionType is not CollectionType.RUN:
1693 raise TypeError(f"The collection type of '{name}' is {collectionType.name}, not RUN.")
1694 refs.extend(self.registry.queryDatasets(..., collections=name, findFirst=True))
1695 with self.datastore.transaction():
1696 with self.registry.transaction():
1697 if unstore:
1698 self.datastore.trash(refs)
1699 else:
1700 self.datastore.forget(refs)
1701 for name in names:
1702 self.registry.removeCollection(name)
1703 if unstore:
1704 # Point of no return for removing artifacts
1705 self.datastore.emptyTrash()
1707 def pruneCollection(
1708 self, name: str, purge: bool = False, unstore: bool = False, unlink: Optional[List[str]] = None
1709 ) -> None:
1710 """Remove a collection and possibly prune datasets within it.
1712 Parameters
1713 ----------
1714 name : `str`
1715 Name of the collection to remove. If this is a
1716 `~CollectionType.TAGGED` or `~CollectionType.CHAINED` collection,
1717 datasets within the collection are not modified unless ``unstore``
1718 is `True`. If this is a `~CollectionType.RUN` collection,
1719 ``purge`` and ``unstore`` must be `True`, and all datasets in it
1720 are fully removed from the data repository.
1721 purge : `bool`, optional
1722 If `True`, permit `~CollectionType.RUN` collections to be removed,
1723 fully removing datasets within them. Requires ``unstore=True`` as
1724 well as an added precaution against accidental deletion. Must be
1725 `False` (default) if the collection is not a ``RUN``.
1726 unstore: `bool`, optional
1727 If `True`, remove all datasets in the collection from all
1728 datastores in which they appear.
1729 unlink: `list` [`str`], optional
1730 Before removing the given `collection` unlink it from from these
1731 parent collections.
1733 Raises
1734 ------
1735 TypeError
1736 Raised if the butler is read-only or arguments are mutually
1737 inconsistent.
1738 """
1739 # See pruneDatasets comments for more information about the logic here;
1740 # the cases are almost the same, but here we can rely on Registry to
1741 # take care everything but Datastore deletion when we remove the
1742 # collection.
1743 if not self.isWriteable():
1744 raise TypeError("Butler is read-only.")
1745 collectionType = self.registry.getCollectionType(name)
1746 if purge and not unstore:
1747 raise PurgeWithoutUnstorePruneCollectionsError()
1748 if collectionType is CollectionType.RUN and not purge:
1749 raise RunWithoutPurgePruneCollectionsError(collectionType)
1750 if collectionType is not CollectionType.RUN and purge:
1751 raise PurgeUnsupportedPruneCollectionsError(collectionType)
1753 def remove(child: str, parent: str) -> None:
1754 """Remove a child collection from a parent collection."""
1755 # Remove child from parent.
1756 chain = list(self.registry.getCollectionChain(parent))
1757 try:
1758 chain.remove(name)
1759 except ValueError as e:
1760 raise RuntimeError(f"{name} is not a child of {parent}") from e
1761 self.registry.setCollectionChain(parent, chain)
1763 with self.datastore.transaction():
1764 with self.registry.transaction():
1765 if unlink:
1766 for parent in unlink:
1767 remove(name, parent)
1768 if unstore:
1769 refs = self.registry.queryDatasets(..., collections=name, findFirst=True)
1770 self.datastore.trash(refs)
1771 self.registry.removeCollection(name)
1773 if unstore:
1774 # Point of no return for removing artifacts
1775 self.datastore.emptyTrash()
1777 def pruneDatasets(
1778 self,
1779 refs: Iterable[DatasetRef],
1780 *,
1781 disassociate: bool = True,
1782 unstore: bool = False,
1783 tags: Iterable[str] = (),
1784 purge: bool = False,
1785 ) -> None:
1786 # docstring inherited from LimitedButler
1788 if not self.isWriteable():
1789 raise TypeError("Butler is read-only.")
1790 if purge:
1791 if not disassociate:
1792 raise TypeError("Cannot pass purge=True without disassociate=True.")
1793 if not unstore:
1794 raise TypeError("Cannot pass purge=True without unstore=True.")
1795 elif disassociate:
1796 tags = tuple(tags)
1797 if not tags:
1798 raise TypeError("No tags provided but disassociate=True.")
1799 for tag in tags:
1800 collectionType = self.registry.getCollectionType(tag)
1801 if collectionType is not CollectionType.TAGGED:
1802 raise TypeError(
1803 f"Cannot disassociate from collection '{tag}' "
1804 f"of non-TAGGED type {collectionType.name}."
1805 )
1806 # For an execution butler we want to keep existing UUIDs for the
1807 # datasets, for that we need to keep them in the collections but
1808 # remove from datastore.
1809 if self._allow_put_of_predefined_dataset and purge:
1810 purge = False
1811 disassociate = False
1812 # Transform possibly-single-pass iterable into something we can iterate
1813 # over multiple times.
1814 refs = list(refs)
1815 # Pruning a component of a DatasetRef makes no sense since registry
1816 # doesn't know about components and datastore might not store
1817 # components in a separate file
1818 for ref in refs:
1819 if ref.datasetType.component():
1820 raise ValueError(f"Can not prune a component of a dataset (ref={ref})")
1821 # We don't need an unreliable Datastore transaction for this, because
1822 # we've been extra careful to ensure that Datastore.trash only involves
1823 # mutating the Registry (it can _look_ at Datastore-specific things,
1824 # but shouldn't change them), and hence all operations here are
1825 # Registry operations.
1826 with self.datastore.transaction():
1827 with self.registry.transaction():
1828 if unstore:
1829 self.datastore.trash(refs)
1830 if purge:
1831 self.registry.removeDatasets(refs)
1832 elif disassociate:
1833 assert tags, "Guaranteed by earlier logic in this function."
1834 for tag in tags:
1835 self.registry.disassociate(tag, refs)
1836 # We've exited the Registry transaction, and apparently committed.
1837 # (if there was an exception, everything rolled back, and it's as if
1838 # nothing happened - and we never get here).
1839 # Datastore artifacts are not yet gone, but they're clearly marked
1840 # as trash, so if we fail to delete now because of (e.g.) filesystem
1841 # problems we can try again later, and if manual administrative
1842 # intervention is required, it's pretty clear what that should entail:
1843 # deleting everything on disk and in private Datastore tables that is
1844 # in the dataset_location_trash table.
1845 if unstore:
1846 # Point of no return for removing artifacts
1847 self.datastore.emptyTrash()
1849 @transactional
1850 def ingest(
1851 self,
1852 *datasets: FileDataset,
1853 transfer: Optional[str] = "auto",
1854 run: Optional[str] = None,
1855 idGenerationMode: DatasetIdGenEnum = DatasetIdGenEnum.UNIQUE,
1856 record_validation_info: bool = True,
1857 ) -> None:
1858 """Store and register one or more datasets that already exist on disk.
1860 Parameters
1861 ----------
1862 datasets : `FileDataset`
1863 Each positional argument is a struct containing information about
1864 a file to be ingested, including its URI (either absolute or
1865 relative to the datastore root, if applicable), a `DatasetRef`,
1866 and optionally a formatter class or its fully-qualified string
1867 name. If a formatter is not provided, the formatter that would be
1868 used for `put` is assumed. On successful return, all
1869 `FileDataset.ref` attributes will have their `DatasetRef.id`
1870 attribute populated and all `FileDataset.formatter` attributes will
1871 be set to the formatter class used. `FileDataset.path` attributes
1872 may be modified to put paths in whatever the datastore considers a
1873 standardized form.
1874 transfer : `str`, optional
1875 If not `None`, must be one of 'auto', 'move', 'copy', 'direct',
1876 'split', 'hardlink', 'relsymlink' or 'symlink', indicating how to
1877 transfer the file.
1878 run : `str`, optional
1879 The name of the run ingested datasets should be added to,
1880 overriding ``self.run``.
1881 idGenerationMode : `DatasetIdGenEnum`, optional
1882 Specifies option for generating dataset IDs. By default unique IDs
1883 are generated for each inserted dataset.
1884 record_validation_info : `bool`, optional
1885 If `True`, the default, the datastore can record validation
1886 information associated with the file. If `False` the datastore
1887 will not attempt to track any information such as checksums
1888 or file sizes. This can be useful if such information is tracked
1889 in an external system or if the file is to be compressed in place.
1890 It is up to the datastore whether this parameter is relevant.
1892 Raises
1893 ------
1894 TypeError
1895 Raised if the butler is read-only or if no run was provided.
1896 NotImplementedError
1897 Raised if the `Datastore` does not support the given transfer mode.
1898 DatasetTypeNotSupportedError
1899 Raised if one or more files to be ingested have a dataset type that
1900 is not supported by the `Datastore`..
1901 FileNotFoundError
1902 Raised if one of the given files does not exist.
1903 FileExistsError
1904 Raised if transfer is not `None` but the (internal) location the
1905 file would be moved to is already occupied.
1907 Notes
1908 -----
1909 This operation is not fully exception safe: if a database operation
1910 fails, the given `FileDataset` instances may be only partially updated.
1912 It is atomic in terms of database operations (they will either all
1913 succeed or all fail) providing the database engine implements
1914 transactions correctly. It will attempt to be atomic in terms of
1915 filesystem operations as well, but this cannot be implemented
1916 rigorously for most datastores.
1917 """
1918 if not self.isWriteable():
1919 raise TypeError("Butler is read-only.")
1920 progress = Progress("lsst.daf.butler.Butler.ingest", level=logging.DEBUG)
1921 # Reorganize the inputs so they're grouped by DatasetType and then
1922 # data ID. We also include a list of DatasetRefs for each FileDataset
1923 # to hold the resolved DatasetRefs returned by the Registry, before
1924 # it's safe to swap them into FileDataset.refs.
1925 # Some type annotation aliases to make that clearer:
1926 GroupForType = Dict[DataCoordinate, Tuple[FileDataset, List[DatasetRef]]]
1927 GroupedData = MutableMapping[DatasetType, GroupForType]
1928 # The actual data structure:
1929 groupedData: GroupedData = defaultdict(dict)
1930 # And the nested loop that populates it:
1931 for dataset in progress.wrap(datasets, desc="Grouping by dataset type"):
1932 # This list intentionally shared across the inner loop, since it's
1933 # associated with `dataset`.
1934 resolvedRefs: List[DatasetRef] = []
1936 # Somewhere to store pre-existing refs if we have an
1937 # execution butler.
1938 existingRefs: List[DatasetRef] = []
1940 for ref in dataset.refs:
1941 if ref.dataId in groupedData[ref.datasetType]:
1942 raise ConflictingDefinitionError(
1943 f"Ingest conflict. Dataset {dataset.path} has same"
1944 " DataId as other ingest dataset"
1945 f" {groupedData[ref.datasetType][ref.dataId][0].path} "
1946 f" ({ref.dataId})"
1947 )
1948 if self._allow_put_of_predefined_dataset:
1949 existing_ref = self.registry.findDataset(
1950 ref.datasetType, dataId=ref.dataId, collections=run
1951 )
1952 if existing_ref:
1953 if self.datastore.knows(existing_ref):
1954 raise ConflictingDefinitionError(
1955 f"Dataset associated with path {dataset.path}"
1956 f" already exists as {existing_ref}."
1957 )
1958 # Store this ref elsewhere since it already exists
1959 # and we do not want to remake it but we do want
1960 # to store it in the datastore.
1961 existingRefs.append(existing_ref)
1963 # Nothing else to do until we have finished
1964 # iterating.
1965 continue
1967 groupedData[ref.datasetType][ref.dataId] = (dataset, resolvedRefs)
1969 if existingRefs:
1970 if len(dataset.refs) != len(existingRefs):
1971 # Keeping track of partially pre-existing datasets is hard
1972 # and should generally never happen. For now don't allow
1973 # it.
1974 raise ConflictingDefinitionError(
1975 f"For dataset {dataset.path} some dataIds already exist"
1976 " in registry but others do not. This is not supported."
1977 )
1979 # Attach the resolved refs if we found them.
1980 dataset.refs = existingRefs
1982 # Now we can bulk-insert into Registry for each DatasetType.
1983 for datasetType, groupForType in progress.iter_item_chunks(
1984 groupedData.items(), desc="Bulk-inserting datasets by type"
1985 ):
1986 refs = self.registry.insertDatasets(
1987 datasetType,
1988 dataIds=groupForType.keys(),
1989 run=run,
1990 expand=self.datastore.needs_expanded_data_ids(transfer, datasetType),
1991 idGenerationMode=idGenerationMode,
1992 )
1993 # Append those resolved DatasetRefs to the new lists we set up for
1994 # them.
1995 for ref, (_, resolvedRefs) in zip(refs, groupForType.values()):
1996 resolvedRefs.append(ref)
1998 # Go back to the original FileDatasets to replace their refs with the
1999 # new resolved ones.
2000 for groupForType in progress.iter_chunks(
2001 groupedData.values(), desc="Reassociating resolved dataset refs with files"
2002 ):
2003 for dataset, resolvedRefs in groupForType.values():
2004 dataset.refs = resolvedRefs
2006 # Bulk-insert everything into Datastore.
2007 self.datastore.ingest(*datasets, transfer=transfer, record_validation_info=record_validation_info)
2009 @contextlib.contextmanager
2010 def export(
2011 self,
2012 *,
2013 directory: Optional[str] = None,
2014 filename: Optional[str] = None,
2015 format: Optional[str] = None,
2016 transfer: Optional[str] = None,
2017 ) -> Iterator[RepoExportContext]:
2018 """Export datasets from the repository represented by this `Butler`.
2020 This method is a context manager that returns a helper object
2021 (`RepoExportContext`) that is used to indicate what information from
2022 the repository should be exported.
2024 Parameters
2025 ----------
2026 directory : `str`, optional
2027 Directory dataset files should be written to if ``transfer`` is not
2028 `None`.
2029 filename : `str`, optional
2030 Name for the file that will include database information associated
2031 with the exported datasets. If this is not an absolute path and
2032 ``directory`` is not `None`, it will be written to ``directory``
2033 instead of the current working directory. Defaults to
2034 "export.{format}".
2035 format : `str`, optional
2036 File format for the database information file. If `None`, the
2037 extension of ``filename`` will be used.
2038 transfer : `str`, optional
2039 Transfer mode passed to `Datastore.export`.
2041 Raises
2042 ------
2043 TypeError
2044 Raised if the set of arguments passed is inconsistent.
2046 Examples
2047 --------
2048 Typically the `Registry.queryDataIds` and `Registry.queryDatasets`
2049 methods are used to provide the iterables over data IDs and/or datasets
2050 to be exported::
2052 with butler.export("exports.yaml") as export:
2053 # Export all flats, but none of the dimension element rows
2054 # (i.e. data ID information) associated with them.
2055 export.saveDatasets(butler.registry.queryDatasets("flat"),
2056 elements=())
2057 # Export all datasets that start with "deepCoadd_" and all of
2058 # their associated data ID information.
2059 export.saveDatasets(butler.registry.queryDatasets("deepCoadd_*"))
2060 """
2061 if directory is None and transfer is not None:
2062 raise TypeError("Cannot transfer without providing a directory.")
2063 if transfer == "move":
2064 raise TypeError("Transfer may not be 'move': export is read-only")
2065 if format is None:
2066 if filename is None:
2067 raise TypeError("At least one of 'filename' or 'format' must be provided.")
2068 else:
2069 _, format = os.path.splitext(filename)
2070 if not format:
2071 raise ValueError("Please specify a file extension to determine export format.")
2072 format = format[1:] # Strip leading ".""
2073 elif filename is None:
2074 filename = f"export.{format}"
2075 if directory is not None:
2076 filename = os.path.join(directory, filename)
2077 formats = self._config["repo_transfer_formats"]
2078 if format not in formats:
2079 raise ValueError(f"Unknown export format {format!r}, allowed: {','.join(formats.keys())}")
2080 BackendClass = get_class_of(formats[format, "export"])
2081 with open(filename, "w") as stream:
2082 backend = BackendClass(stream, universe=self.registry.dimensions)
2083 try:
2084 helper = RepoExportContext(
2085 self.registry, self.datastore, backend=backend, directory=directory, transfer=transfer
2086 )
2087 yield helper
2088 except BaseException:
2089 raise
2090 else:
2091 helper._finish()
2093 def import_(
2094 self,
2095 *,
2096 directory: Optional[str] = None,
2097 filename: Union[str, TextIO, None] = None,
2098 format: Optional[str] = None,
2099 transfer: Optional[str] = None,
2100 skip_dimensions: Optional[Set] = None,
2101 idGenerationMode: DatasetIdGenEnum = DatasetIdGenEnum.UNIQUE,
2102 reuseIds: bool = False,
2103 ) -> None:
2104 """Import datasets into this repository that were exported from a
2105 different butler repository via `~lsst.daf.butler.Butler.export`.
2107 Parameters
2108 ----------
2109 directory : `str`, optional
2110 Directory containing dataset files to import from. If `None`,
2111 ``filename`` and all dataset file paths specified therein must
2112 be absolute.
2113 filename : `str` or `TextIO`, optional
2114 A stream or name of file that contains database information
2115 associated with the exported datasets, typically generated by
2116 `~lsst.daf.butler.Butler.export`. If this a string (name) and
2117 is not an absolute path, does not exist in the current working
2118 directory, and ``directory`` is not `None`, it is assumed to be in
2119 ``directory``. Defaults to "export.{format}".
2120 format : `str`, optional
2121 File format for ``filename``. If `None`, the extension of
2122 ``filename`` will be used.
2123 transfer : `str`, optional
2124 Transfer mode passed to `~lsst.daf.butler.Datastore.ingest`.
2125 skip_dimensions : `set`, optional
2126 Names of dimensions that should be skipped and not imported.
2127 idGenerationMode : `DatasetIdGenEnum`, optional
2128 Specifies option for generating dataset IDs when IDs are not
2129 provided or their type does not match backend type. By default
2130 unique IDs are generated for each inserted dataset.
2131 reuseIds : `bool`, optional
2132 If `True` then forces re-use of imported dataset IDs for integer
2133 IDs which are normally generated as auto-incremented; exception
2134 will be raised if imported IDs clash with existing ones. This
2135 option has no effect on the use of globally-unique IDs which are
2136 always re-used (or generated if integer IDs are being imported).
2138 Raises
2139 ------
2140 TypeError
2141 Raised if the set of arguments passed is inconsistent, or if the
2142 butler is read-only.
2143 """
2144 if not self.isWriteable():
2145 raise TypeError("Butler is read-only.")
2146 if format is None:
2147 if filename is None:
2148 raise TypeError("At least one of 'filename' or 'format' must be provided.")
2149 else:
2150 _, format = os.path.splitext(filename) # type: ignore
2151 elif filename is None:
2152 filename = f"export.{format}"
2153 if isinstance(filename, str) and directory is not None and not os.path.exists(filename):
2154 filename = os.path.join(directory, filename)
2155 BackendClass = get_class_of(self._config["repo_transfer_formats"][format]["import"])
2157 def doImport(importStream: TextIO) -> None:
2158 backend = BackendClass(importStream, self.registry)
2159 backend.register()
2160 with self.transaction():
2161 backend.load(
2162 self.datastore,
2163 directory=directory,
2164 transfer=transfer,
2165 skip_dimensions=skip_dimensions,
2166 idGenerationMode=idGenerationMode,
2167 reuseIds=reuseIds,
2168 )
2170 if isinstance(filename, str):
2171 with open(filename, "r") as stream:
2172 doImport(stream)
2173 else:
2174 doImport(filename)
2176 def transfer_from(
2177 self,
2178 source_butler: LimitedButler,
2179 source_refs: Iterable[DatasetRef],
2180 transfer: str = "auto",
2181 id_gen_map: Dict[str, DatasetIdGenEnum] | None = None,
2182 skip_missing: bool = True,
2183 register_dataset_types: bool = False,
2184 transfer_dimensions: bool = False,
2185 ) -> List[DatasetRef]:
2186 """Transfer datasets to this Butler from a run in another Butler.
2188 Parameters
2189 ----------
2190 source_butler : `LimitedButler`
2191 Butler from which the datasets are to be transferred. If data IDs
2192 in ``source_refs`` are not expanded then this has to be a full
2193 `Butler` whose registry will be used to expand data IDs.
2194 source_refs : iterable of `DatasetRef`
2195 Datasets defined in the source butler that should be transferred to
2196 this butler.
2197 transfer : `str`, optional
2198 Transfer mode passed to `~lsst.daf.butler.Datastore.transfer_from`.
2199 id_gen_map : `dict` of [`str`, `DatasetIdGenEnum`], optional
2200 A mapping of dataset type to ID generation mode. Only used if
2201 the source butler is using integer IDs. Should not be used
2202 if this receiving butler uses integer IDs. Without this dataset
2203 import always uses unique.
2204 skip_missing : `bool`
2205 If `True`, datasets with no datastore artifact associated with
2206 them are not transferred. If `False` a registry entry will be
2207 created even if no datastore record is created (and so will
2208 look equivalent to the dataset being unstored).
2209 register_dataset_types : `bool`
2210 If `True` any missing dataset types are registered. Otherwise
2211 an exception is raised.
2212 transfer_dimensions : `bool`, optional
2213 If `True`, dimension record data associated with the new datasets
2214 will be transferred.
2216 Returns
2217 -------
2218 refs : `list` of `DatasetRef`
2219 The refs added to this Butler.
2221 Notes
2222 -----
2223 Requires that any dimension definitions are already present in the
2224 receiving Butler. The datastore artifact has to exist for a transfer
2225 to be made but non-existence is not an error.
2227 Datasets that already exist in this run will be skipped.
2229 The datasets are imported as part of a transaction, although
2230 dataset types are registered before the transaction is started.
2231 This means that it is possible for a dataset type to be registered
2232 even though transfer has failed.
2233 """
2234 if not self.isWriteable():
2235 raise TypeError("Butler is read-only.")
2236 progress = Progress("lsst.daf.butler.Butler.transfer_from", level=VERBOSE)
2238 # Will iterate through the refs multiple times so need to convert
2239 # to a list if this isn't a collection.
2240 if not isinstance(source_refs, collections.abc.Collection):
2241 source_refs = list(source_refs)
2243 original_count = len(source_refs)
2244 log.info("Transferring %d datasets into %s", original_count, str(self))
2246 if id_gen_map is None:
2247 id_gen_map = {}
2249 # In some situations the datastore artifact may be missing
2250 # and we do not want that registry entry to be imported.
2251 # Asking datastore is not sufficient, the records may have been
2252 # purged, we have to ask for the (predicted) URI and check
2253 # existence explicitly. Execution butler is set up exactly like
2254 # this with no datastore records.
2255 artifact_existence: Dict[ResourcePath, bool] = {}
2256 if skip_missing:
2257 dataset_existence = source_butler.datastore.mexists(
2258 source_refs, artifact_existence=artifact_existence
2259 )
2260 source_refs = [ref for ref, exists in dataset_existence.items() if exists]
2261 filtered_count = len(source_refs)
2262 log.verbose(
2263 "%d datasets removed because the artifact does not exist. Now have %d.",
2264 original_count - filtered_count,
2265 filtered_count,
2266 )
2268 # Importing requires that we group the refs by dataset type and run
2269 # before doing the import.
2270 source_dataset_types = set()
2271 grouped_refs = defaultdict(list)
2272 grouped_indices = defaultdict(list)
2273 for i, ref in enumerate(source_refs):
2274 grouped_refs[ref.datasetType, ref.run].append(ref)
2275 grouped_indices[ref.datasetType, ref.run].append(i)
2276 source_dataset_types.add(ref.datasetType)
2278 # Check to see if the dataset type in the source butler has
2279 # the same definition in the target butler and register missing
2280 # ones if requested. Registration must happen outside a transaction.
2281 newly_registered_dataset_types = set()
2282 for datasetType in source_dataset_types:
2283 if register_dataset_types:
2284 # Let this raise immediately if inconsistent. Continuing
2285 # on to find additional inconsistent dataset types
2286 # might result in additional unwanted dataset types being
2287 # registered.
2288 if self.registry.registerDatasetType(datasetType):
2289 newly_registered_dataset_types.add(datasetType)
2290 else:
2291 # If the dataset type is missing, let it fail immediately.
2292 target_dataset_type = self.registry.getDatasetType(datasetType.name)
2293 if target_dataset_type != datasetType:
2294 raise ConflictingDefinitionError(
2295 "Source butler dataset type differs from definition"
2296 f" in target butler: {datasetType} !="
2297 f" {target_dataset_type}"
2298 )
2299 if newly_registered_dataset_types:
2300 # We may have registered some even if there were inconsistencies
2301 # but should let people know (or else remove them again).
2302 log.log(
2303 VERBOSE,
2304 "Registered the following dataset types in the target Butler: %s",
2305 ", ".join(d.name for d in newly_registered_dataset_types),
2306 )
2307 else:
2308 log.log(VERBOSE, "All required dataset types are known to the target Butler")
2310 dimension_records: Dict[DimensionElement, Dict[DataCoordinate, DimensionRecord]] = defaultdict(dict)
2311 if transfer_dimensions:
2312 # Collect all the dimension records for these refs.
2313 # All dimensions are to be copied but the list of valid dimensions
2314 # come from this butler's universe.
2315 elements = frozenset(
2316 element
2317 for element in self.registry.dimensions.getStaticElements()
2318 if element.hasTable() and element.viewOf is None
2319 )
2320 dataIds = set(ref.dataId for ref in source_refs)
2321 # This logic comes from saveDataIds.
2322 for dataId in dataIds:
2323 # Need an expanded record, if not expanded that we need a full
2324 # butler with registry (allow mocks with registry too).
2325 if not dataId.hasRecords():
2326 if registry := getattr(source_butler, "registry", None):
2327 dataId = registry.expandDataId(dataId)
2328 else:
2329 raise TypeError("Input butler needs to be a full butler to expand DataId.")
2330 # If this butler doesn't know about a dimension in the source
2331 # butler things will break later.
2332 for record in dataId.records.values():
2333 if record is not None and record.definition in elements:
2334 dimension_records[record.definition].setdefault(record.dataId, record)
2336 # The returned refs should be identical for UUIDs.
2337 # For now must also support integers and so need to retain the
2338 # newly-created refs from this registry.
2339 # Pre-size it so we can assign refs into the correct slots
2340 transferred_refs_tmp: List[Optional[DatasetRef]] = [None] * len(source_refs)
2341 default_id_gen = DatasetIdGenEnum.UNIQUE
2343 handled_collections: Set[str] = set()
2345 # Do all the importing in a single transaction.
2346 with self.transaction():
2347 if dimension_records:
2348 log.verbose("Ensuring that dimension records exist for transferred datasets.")
2349 for element, r in dimension_records.items():
2350 records = [r[dataId] for dataId in r]
2351 # Assume that if the record is already present that we can
2352 # use it without having to check that the record metadata
2353 # is consistent.
2354 self.registry.insertDimensionData(element, *records, skip_existing=True)
2356 for (datasetType, run), refs_to_import in progress.iter_item_chunks(
2357 grouped_refs.items(), desc="Importing to registry by run and dataset type"
2358 ):
2359 if run not in handled_collections:
2360 # May need to create output collection. If source butler
2361 # has a registry, ask for documentation string.
2362 run_doc = None
2363 if registry := getattr(source_butler, "registry", None):
2364 run_doc = registry.getCollectionDocumentation(run)
2365 registered = self.registry.registerRun(run, doc=run_doc)
2366 handled_collections.add(run)
2367 if registered:
2368 log.log(VERBOSE, "Creating output run %s", run)
2370 id_generation_mode = default_id_gen
2371 if isinstance(refs_to_import[0].id, int):
2372 # ID generation mode might need to be overridden when
2373 # targetting UUID
2374 id_generation_mode = id_gen_map.get(datasetType.name, default_id_gen)
2376 n_refs = len(refs_to_import)
2377 log.verbose(
2378 "Importing %d ref%s of dataset type %s into run %s",
2379 n_refs,
2380 "" if n_refs == 1 else "s",
2381 datasetType.name,
2382 run,
2383 )
2385 # No way to know if this butler's registry uses UUID.
2386 # We have to trust the caller on this. If it fails they will
2387 # have to change their approach. We can't catch the exception
2388 # and retry with unique because that will mess up the
2389 # transaction handling. We aren't allowed to ask the registry
2390 # manager what type of ID it is using.
2391 imported_refs = self.registry._importDatasets(
2392 refs_to_import, idGenerationMode=id_generation_mode, expand=False
2393 )
2395 # Map them into the correct slots to match the initial order
2396 for i, ref in zip(grouped_indices[datasetType, run], imported_refs):
2397 transferred_refs_tmp[i] = ref
2399 # Mypy insists that we might have None in here so we have to make
2400 # that explicit by assigning to a new variable and filtering out
2401 # something that won't be there.
2402 transferred_refs = [ref for ref in transferred_refs_tmp if ref is not None]
2404 # Check consistency
2405 assert len(source_refs) == len(transferred_refs), "Different number of refs imported than given"
2407 log.verbose("Imported %d datasets into destination butler", len(transferred_refs))
2409 # The transferred refs need to be reordered to match the original
2410 # ordering given by the caller. Without this the datastore transfer
2411 # will be broken.
2413 # Ask the datastore to transfer. The datastore has to check that
2414 # the source datastore is compatible with the target datastore.
2415 self.datastore.transfer_from(
2416 source_butler.datastore,
2417 source_refs,
2418 local_refs=transferred_refs,
2419 transfer=transfer,
2420 artifact_existence=artifact_existence,
2421 )
2423 return transferred_refs
2425 def validateConfiguration(
2426 self,
2427 logFailures: bool = False,
2428 datasetTypeNames: Optional[Iterable[str]] = None,
2429 ignore: Iterable[str] | None = None,
2430 ) -> None:
2431 """Validate butler configuration.
2433 Checks that each `DatasetType` can be stored in the `Datastore`.
2435 Parameters
2436 ----------
2437 logFailures : `bool`, optional
2438 If `True`, output a log message for every validation error
2439 detected.
2440 datasetTypeNames : iterable of `str`, optional
2441 The `DatasetType` names that should be checked. This allows
2442 only a subset to be selected.
2443 ignore : iterable of `str`, optional
2444 Names of DatasetTypes to skip over. This can be used to skip
2445 known problems. If a named `DatasetType` corresponds to a
2446 composite, all components of that `DatasetType` will also be
2447 ignored.
2449 Raises
2450 ------
2451 ButlerValidationError
2452 Raised if there is some inconsistency with how this Butler
2453 is configured.
2454 """
2455 if datasetTypeNames:
2456 datasetTypes = [self.registry.getDatasetType(name) for name in datasetTypeNames]
2457 else:
2458 datasetTypes = list(self.registry.queryDatasetTypes())
2460 # filter out anything from the ignore list
2461 if ignore:
2462 ignore = set(ignore)
2463 datasetTypes = [
2464 e for e in datasetTypes if e.name not in ignore and e.nameAndComponent()[0] not in ignore
2465 ]
2466 else:
2467 ignore = set()
2469 # Find all the registered instruments
2470 instruments = set(record.name for record in self.registry.queryDimensionRecords("instrument"))
2472 # For each datasetType that has an instrument dimension, create
2473 # a DatasetRef for each defined instrument
2474 datasetRefs = []
2476 for datasetType in datasetTypes:
2477 if "instrument" in datasetType.dimensions:
2478 for instrument in instruments:
2479 datasetRef = DatasetRef(
2480 datasetType, {"instrument": instrument}, conform=False # type: ignore
2481 )
2482 datasetRefs.append(datasetRef)
2484 entities: List[Union[DatasetType, DatasetRef]] = []
2485 entities.extend(datasetTypes)
2486 entities.extend(datasetRefs)
2488 datastoreErrorStr = None
2489 try:
2490 self.datastore.validateConfiguration(entities, logFailures=logFailures)
2491 except ValidationError as e:
2492 datastoreErrorStr = str(e)
2494 # Also check that the LookupKeys used by the datastores match
2495 # registry and storage class definitions
2496 keys = self.datastore.getLookupKeys()
2498 failedNames = set()
2499 failedDataId = set()
2500 for key in keys:
2501 if key.name is not None:
2502 if key.name in ignore:
2503 continue
2505 # skip if specific datasetType names were requested and this
2506 # name does not match
2507 if datasetTypeNames and key.name not in datasetTypeNames:
2508 continue
2510 # See if it is a StorageClass or a DatasetType
2511 if key.name in self.storageClasses:
2512 pass
2513 else:
2514 try:
2515 self.registry.getDatasetType(key.name)
2516 except KeyError:
2517 if logFailures:
2518 log.critical("Key '%s' does not correspond to a DatasetType or StorageClass", key)
2519 failedNames.add(key)
2520 else:
2521 # Dimensions are checked for consistency when the Butler
2522 # is created and rendezvoused with a universe.
2523 pass
2525 # Check that the instrument is a valid instrument
2526 # Currently only support instrument so check for that
2527 if key.dataId:
2528 dataIdKeys = set(key.dataId)
2529 if set(["instrument"]) != dataIdKeys:
2530 if logFailures:
2531 log.critical("Key '%s' has unsupported DataId override", key)
2532 failedDataId.add(key)
2533 elif key.dataId["instrument"] not in instruments:
2534 if logFailures:
2535 log.critical("Key '%s' has unknown instrument", key)
2536 failedDataId.add(key)
2538 messages = []
2540 if datastoreErrorStr:
2541 messages.append(datastoreErrorStr)
2543 for failed, msg in (
2544 (failedNames, "Keys without corresponding DatasetType or StorageClass entry: "),
2545 (failedDataId, "Keys with bad DataId entries: "),
2546 ):
2547 if failed:
2548 msg += ", ".join(str(k) for k in failed)
2549 messages.append(msg)
2551 if messages:
2552 raise ValidationError(";\n".join(messages))
2554 @property
2555 def collections(self) -> Sequence[str]:
2556 """The collections to search by default, in order
2557 (`Sequence` [ `str` ]).
2559 This is an alias for ``self.registry.defaults.collections``. It cannot
2560 be set directly in isolation, but all defaults may be changed together
2561 by assigning a new `RegistryDefaults` instance to
2562 ``self.registry.defaults``.
2563 """
2564 return self.registry.defaults.collections
2566 @property
2567 def run(self) -> Optional[str]:
2568 """Name of the run this butler writes outputs to by default (`str` or
2569 `None`).
2571 This is an alias for ``self.registry.defaults.run``. It cannot be set
2572 directly in isolation, but all defaults may be changed together by
2573 assigning a new `RegistryDefaults` instance to
2574 ``self.registry.defaults``.
2575 """
2576 return self.registry.defaults.run
2578 @property
2579 def dimensions(self) -> DimensionUniverse:
2580 # Docstring inherited.
2581 return self.registry.dimensions
2583 registry: Registry
2584 """The object that manages dataset metadata and relationships (`Registry`).
2586 Most operations that don't involve reading or writing butler datasets are
2587 accessible only via `Registry` methods.
2588 """
2590 datastore: Datastore
2591 """The object that manages actual dataset storage (`Datastore`).
2593 Direct user access to the datastore should rarely be necessary; the primary
2594 exception is the case where a `Datastore` implementation provides extra
2595 functionality beyond what the base class defines.
2596 """
2598 storageClasses: StorageClassFactory
2599 """An object that maps known storage class names to objects that fully
2600 describe them (`StorageClassFactory`).
2601 """
2603 _allow_put_of_predefined_dataset: bool
2604 """Allow a put to succeed even if there is already a registry entry for it
2605 but not a datastore record. (`bool`)."""