Coverage for python/lsst/daf/butler/_butler.py : 9%

Hot-keys on this page
r m x p toggle line displays
j k next/prev highlighted chunk
0 (zero) top of page
1 (one) first highlighted chunk
1# This file is part of daf_butler.
2#
3# Developed for the LSST Data Management System.
4# This product includes software developed by the LSST Project
5# (http://www.lsst.org).
6# See the COPYRIGHT file at the top-level directory of this distribution
7# for details of code ownership.
8#
9# This program is free software: you can redistribute it and/or modify
10# it under the terms of the GNU General Public License as published by
11# the Free Software Foundation, either version 3 of the License, or
12# (at your option) any later version.
13#
14# This program is distributed in the hope that it will be useful,
15# but WITHOUT ANY WARRANTY; without even the implied warranty of
16# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17# GNU General Public License for more details.
18#
19# You should have received a copy of the GNU General Public License
20# along with this program. If not, see <http://www.gnu.org/licenses/>.
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)
37from collections import defaultdict
38import contextlib
39import logging
40import numbers
41import os
42from typing import (
43 Any,
44 ClassVar,
45 Counter,
46 Dict,
47 Iterable,
48 Iterator,
49 List,
50 MutableMapping,
51 Optional,
52 Set,
53 TextIO,
54 Tuple,
55 Type,
56 Union,
57)
59try:
60 import boto3
61except ImportError:
62 boto3 = None
64from lsst.utils import doImport
65from .core import (
66 AmbiguousDatasetError,
67 ButlerURI,
68 Config,
69 ConfigSubset,
70 DataCoordinate,
71 DataId,
72 DataIdValue,
73 DatasetRef,
74 DatasetType,
75 Datastore,
76 Dimension,
77 DimensionConfig,
78 FileDataset,
79 Progress,
80 StorageClassFactory,
81 Timespan,
82 ValidationError,
83)
84from .core.repoRelocation import BUTLER_ROOT_TAG
85from .core.utils import transactional, getClassOf
86from ._deferredDatasetHandle import DeferredDatasetHandle
87from ._butlerConfig import ButlerConfig
88from .registry import (
89 Registry,
90 RegistryConfig,
91 RegistryDefaults,
92 CollectionSearch,
93 CollectionType,
94 ConflictingDefinitionError,
95 DatasetIdGenEnum,
96)
97from .transfers import RepoExportContext
99log = logging.getLogger(__name__)
102class ButlerValidationError(ValidationError):
103 """There is a problem with the Butler configuration."""
104 pass
107class PruneCollectionsArgsError(TypeError):
108 """Base class for errors relating to Butler.pruneCollections input
109 arguments.
110 """
111 pass
114class PurgeWithoutUnstorePruneCollectionsError(PruneCollectionsArgsError):
115 """Raised when purge and unstore are both required to be True, and
116 purge is True but unstore is False.
117 """
119 def __init__(self) -> None:
120 super().__init__("Cannot pass purge=True without unstore=True.")
123class RunWithoutPurgePruneCollectionsError(PruneCollectionsArgsError):
124 """Raised when pruning a RUN collection but purge is False."""
126 def __init__(self, collectionType: CollectionType):
127 self.collectionType = collectionType
128 super().__init__(f"Cannot prune RUN collection {self.collectionType.name} without purge=True.")
131class PurgeUnsupportedPruneCollectionsError(PruneCollectionsArgsError):
132 """Raised when purge is True but is not supported for the given
133 collection."""
135 def __init__(self, collectionType: CollectionType):
136 self.collectionType = collectionType
137 super().__init__(
138 f"Cannot prune {self.collectionType} collection {self.collectionType.name} with purge=True.")
141class Butler:
142 """Main entry point for the data access system.
144 Parameters
145 ----------
146 config : `ButlerConfig`, `Config` or `str`, optional.
147 Configuration. Anything acceptable to the
148 `ButlerConfig` constructor. If a directory path
149 is given the configuration will be read from a ``butler.yaml`` file in
150 that location. If `None` is given default values will be used.
151 butler : `Butler`, optional.
152 If provided, construct a new Butler that uses the same registry and
153 datastore as the given one, but with the given collection and run.
154 Incompatible with the ``config``, ``searchPaths``, and ``writeable``
155 arguments.
156 collections : `str` or `Iterable` [ `str` ], optional
157 An expression specifying the collections to be searched (in order) when
158 reading datasets.
159 This may be a `str` collection name or an iterable thereof.
160 See :ref:`daf_butler_collection_expressions` for more information.
161 These collections are not registered automatically and must be
162 manually registered before they are used by any method, but they may be
163 manually registered after the `Butler` is initialized.
164 run : `str`, optional
165 Name of the `~CollectionType.RUN` collection new datasets should be
166 inserted into. If ``collections`` is `None` and ``run`` is not `None`,
167 ``collections`` will be set to ``[run]``. If not `None`, this
168 collection will automatically be registered. If this is not set (and
169 ``writeable`` is not set either), a read-only butler will be created.
170 searchPaths : `list` of `str`, optional
171 Directory paths to search when calculating the full Butler
172 configuration. Not used if the supplied config is already a
173 `ButlerConfig`.
174 writeable : `bool`, optional
175 Explicitly sets whether the butler supports write operations. If not
176 provided, a read-write butler is created if any of ``run``, ``tags``,
177 or ``chains`` is non-empty.
178 inferDefaults : `bool`, optional
179 If `True` (default) infer default data ID values from the values
180 present in the datasets in ``collections``: if all collections have the
181 same value (or no value) for a governor dimension, that value will be
182 the default for that dimension. Nonexistent collections are ignored.
183 If a default value is provided explicitly for a governor dimension via
184 ``**kwargs``, no default will be inferred for that dimension.
185 **kwargs : `str`
186 Default data ID key-value pairs. These may only identify "governor"
187 dimensions like ``instrument`` and ``skymap``.
189 Examples
190 --------
191 While there are many ways to control exactly how a `Butler` interacts with
192 the collections in its `Registry`, the most common cases are still simple.
194 For a read-only `Butler` that searches one collection, do::
196 butler = Butler("/path/to/repo", collections=["u/alice/DM-50000"])
198 For a read-write `Butler` that writes to and reads from a
199 `~CollectionType.RUN` collection::
201 butler = Butler("/path/to/repo", run="u/alice/DM-50000/a")
203 The `Butler` passed to a ``PipelineTask`` is often much more complex,
204 because we want to write to one `~CollectionType.RUN` collection but read
205 from several others (as well)::
207 butler = Butler("/path/to/repo", run="u/alice/DM-50000/a",
208 collections=["u/alice/DM-50000/a",
209 "u/bob/DM-49998",
210 "HSC/defaults"])
212 This butler will `put` new datasets to the run ``u/alice/DM-50000/a``.
213 Datasets will be read first from that run (since it appears first in the
214 chain), and then from ``u/bob/DM-49998`` and finally ``HSC/defaults``.
216 Finally, one can always create a `Butler` with no collections::
218 butler = Butler("/path/to/repo", writeable=True)
220 This can be extremely useful when you just want to use ``butler.registry``,
221 e.g. for inserting dimension data or managing collections, or when the
222 collections you want to use with the butler are not consistent.
223 Passing ``writeable`` explicitly here is only necessary if you want to be
224 able to make changes to the repo - usually the value for ``writeable`` can
225 be guessed from the collection arguments provided, but it defaults to
226 `False` when there are not collection arguments.
227 """
228 def __init__(self, config: Union[Config, str, None] = None, *,
229 butler: Optional[Butler] = None,
230 collections: Any = None,
231 run: Optional[str] = None,
232 searchPaths: Optional[List[str]] = None,
233 writeable: Optional[bool] = None,
234 inferDefaults: bool = True,
235 **kwargs: str,
236 ):
237 defaults = RegistryDefaults(collections=collections, run=run, infer=inferDefaults, **kwargs)
238 # Load registry, datastore, etc. from config or existing butler.
239 if butler is not None:
240 if config is not None or searchPaths is not None or writeable is not None:
241 raise TypeError("Cannot pass 'config', 'searchPaths', or 'writeable' "
242 "arguments with 'butler' argument.")
243 self.registry = butler.registry.copy(defaults)
244 self.datastore = butler.datastore
245 self.storageClasses = butler.storageClasses
246 self._config: ButlerConfig = butler._config
247 else:
248 self._config = ButlerConfig(config, searchPaths=searchPaths)
249 if "root" in self._config:
250 butlerRoot = self._config["root"]
251 else:
252 butlerRoot = self._config.configDir
253 if writeable is None:
254 writeable = run is not None
255 self.registry = Registry.fromConfig(self._config, butlerRoot=butlerRoot, writeable=writeable,
256 defaults=defaults)
257 self.datastore = Datastore.fromConfig(self._config, self.registry.getDatastoreBridgeManager(),
258 butlerRoot=butlerRoot)
259 self.storageClasses = StorageClassFactory()
260 self.storageClasses.addFromConfig(self._config)
261 if "run" in self._config or "collection" in self._config:
262 raise ValueError("Passing a run or collection via configuration is no longer supported.")
264 GENERATION: ClassVar[int] = 3
265 """This is a Generation 3 Butler.
267 This attribute may be removed in the future, once the Generation 2 Butler
268 interface has been fully retired; it should only be used in transitional
269 code.
270 """
272 @staticmethod
273 def makeRepo(root: str, config: Union[Config, str, None] = None,
274 dimensionConfig: Union[Config, str, None] = None, standalone: bool = False,
275 searchPaths: Optional[List[str]] = None, forceConfigRoot: bool = True,
276 outfile: Optional[str] = None, overwrite: bool = False) -> Config:
277 """Create an empty data repository by adding a butler.yaml config
278 to a repository root directory.
280 Parameters
281 ----------
282 root : `str` or `ButlerURI`
283 Path or URI to the root location of the new repository. Will be
284 created if it does not exist.
285 config : `Config` or `str`, optional
286 Configuration to write to the repository, after setting any
287 root-dependent Registry or Datastore config options. Can not
288 be a `ButlerConfig` or a `ConfigSubset`. If `None`, default
289 configuration will be used. Root-dependent config options
290 specified in this config are overwritten if ``forceConfigRoot``
291 is `True`.
292 dimensionConfig : `Config` or `str`, optional
293 Configuration for dimensions, will be used to initialize registry
294 database.
295 standalone : `bool`
296 If True, write all expanded defaults, not just customized or
297 repository-specific settings.
298 This (mostly) decouples the repository from the default
299 configuration, insulating it from changes to the defaults (which
300 may be good or bad, depending on the nature of the changes).
301 Future *additions* to the defaults will still be picked up when
302 initializing `Butlers` to repos created with ``standalone=True``.
303 searchPaths : `list` of `str`, optional
304 Directory paths to search when calculating the full butler
305 configuration.
306 forceConfigRoot : `bool`, optional
307 If `False`, any values present in the supplied ``config`` that
308 would normally be reset are not overridden and will appear
309 directly in the output config. This allows non-standard overrides
310 of the root directory for a datastore or registry to be given.
311 If this parameter is `True` the values for ``root`` will be
312 forced into the resulting config if appropriate.
313 outfile : `str`, optional
314 If not-`None`, the output configuration will be written to this
315 location rather than into the repository itself. Can be a URI
316 string. Can refer to a directory that will be used to write
317 ``butler.yaml``.
318 overwrite : `bool`, optional
319 Create a new configuration file even if one already exists
320 in the specified output location. Default is to raise
321 an exception.
323 Returns
324 -------
325 config : `Config`
326 The updated `Config` instance written to the repo.
328 Raises
329 ------
330 ValueError
331 Raised if a ButlerConfig or ConfigSubset is passed instead of a
332 regular Config (as these subclasses would make it impossible to
333 support ``standalone=False``).
334 FileExistsError
335 Raised if the output config file already exists.
336 os.error
337 Raised if the directory does not exist, exists but is not a
338 directory, or cannot be created.
340 Notes
341 -----
342 Note that when ``standalone=False`` (the default), the configuration
343 search path (see `ConfigSubset.defaultSearchPaths`) that was used to
344 construct the repository should also be used to construct any Butlers
345 to avoid configuration inconsistencies.
346 """
347 if isinstance(config, (ButlerConfig, ConfigSubset)):
348 raise ValueError("makeRepo must be passed a regular Config without defaults applied.")
350 # Ensure that the root of the repository exists or can be made
351 uri = ButlerURI(root, forceDirectory=True)
352 uri.mkdir()
354 config = Config(config)
356 # If we are creating a new repo from scratch with relative roots,
357 # do not propagate an explicit root from the config file
358 if "root" in config:
359 del config["root"]
361 full = ButlerConfig(config, searchPaths=searchPaths) # this applies defaults
362 datastoreClass: Type[Datastore] = doImport(full["datastore", "cls"])
363 datastoreClass.setConfigRoot(BUTLER_ROOT_TAG, config, full, overwrite=forceConfigRoot)
365 # if key exists in given config, parse it, otherwise parse the defaults
366 # in the expanded config
367 if config.get(("registry", "db")):
368 registryConfig = RegistryConfig(config)
369 else:
370 registryConfig = RegistryConfig(full)
371 defaultDatabaseUri = registryConfig.makeDefaultDatabaseUri(BUTLER_ROOT_TAG)
372 if defaultDatabaseUri is not None:
373 Config.updateParameters(RegistryConfig, config, full,
374 toUpdate={"db": defaultDatabaseUri},
375 overwrite=forceConfigRoot)
376 else:
377 Config.updateParameters(RegistryConfig, config, full, toCopy=("db",),
378 overwrite=forceConfigRoot)
380 if standalone:
381 config.merge(full)
382 else:
383 # Always expand the registry.managers section into the per-repo
384 # config, because after the database schema is created, it's not
385 # allowed to change anymore. Note that in the standalone=True
386 # branch, _everything_ in the config is expanded, so there's no
387 # need to special case this.
388 Config.updateParameters(RegistryConfig, config, full, toCopy=("managers",), overwrite=False)
389 configURI: Union[str, ButlerURI]
390 if outfile is not None:
391 # When writing to a separate location we must include
392 # the root of the butler repo in the config else it won't know
393 # where to look.
394 config["root"] = uri.geturl()
395 configURI = outfile
396 else:
397 configURI = uri
398 config.dumpToUri(configURI, overwrite=overwrite)
400 # Create Registry and populate tables
401 registryConfig = RegistryConfig(config.get("registry"))
402 dimensionConfig = DimensionConfig(dimensionConfig)
403 Registry.createFromConfig(registryConfig, dimensionConfig=dimensionConfig, butlerRoot=root)
405 return config
407 @classmethod
408 def _unpickle(cls, config: ButlerConfig, collections: Optional[CollectionSearch], run: Optional[str],
409 defaultDataId: Dict[str, str], writeable: bool) -> Butler:
410 """Callable used to unpickle a Butler.
412 We prefer not to use ``Butler.__init__`` directly so we can force some
413 of its many arguments to be keyword-only (note that ``__reduce__``
414 can only invoke callables with positional arguments).
416 Parameters
417 ----------
418 config : `ButlerConfig`
419 Butler configuration, already coerced into a true `ButlerConfig`
420 instance (and hence after any search paths for overrides have been
421 utilized).
422 collections : `CollectionSearch`
423 Names of the default collections to read from.
424 run : `str`, optional
425 Name of the default `~CollectionType.RUN` collection to write to.
426 defaultDataId : `dict` [ `str`, `str` ]
427 Default data ID values.
428 writeable : `bool`
429 Whether the Butler should support write operations.
431 Returns
432 -------
433 butler : `Butler`
434 A new `Butler` instance.
435 """
436 # MyPy doesn't recognize that the kwargs below are totally valid; it
437 # seems to think '**defaultDataId* is a _positional_ argument!
438 return cls(config=config, collections=collections, run=run, writeable=writeable,
439 **defaultDataId) # type: ignore
441 def __reduce__(self) -> tuple:
442 """Support pickling.
443 """
444 return (Butler._unpickle, (self._config, self.collections, self.run,
445 self.registry.defaults.dataId.byName(),
446 self.registry.isWriteable()))
448 def __str__(self) -> str:
449 return "Butler(collections={}, run={}, datastore='{}', registry='{}')".format(
450 self.collections, self.run, self.datastore, self.registry)
452 def isWriteable(self) -> bool:
453 """Return `True` if this `Butler` supports write operations.
454 """
455 return self.registry.isWriteable()
457 @contextlib.contextmanager
458 def transaction(self) -> Iterator[None]:
459 """Context manager supporting `Butler` transactions.
461 Transactions can be nested.
462 """
463 with self.registry.transaction():
464 with self.datastore.transaction():
465 yield
467 def _standardizeArgs(self, datasetRefOrType: Union[DatasetRef, DatasetType, str],
468 dataId: Optional[DataId] = None, **kwds: Any
469 ) -> Tuple[DatasetType, Optional[DataId]]:
470 """Standardize the arguments passed to several Butler APIs.
472 Parameters
473 ----------
474 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
475 When `DatasetRef` the `dataId` should be `None`.
476 Otherwise the `DatasetType` or name thereof.
477 dataId : `dict` or `DataCoordinate`
478 A `dict` of `Dimension` link name, value pairs that label the
479 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
480 should be provided as the second argument.
481 kwds
482 Additional keyword arguments used to augment or construct a
483 `DataCoordinate`. See `DataCoordinate.standardize`
484 parameters.
486 Returns
487 -------
488 datasetType : `DatasetType`
489 A `DatasetType` instance extracted from ``datasetRefOrType``.
490 dataId : `dict` or `DataId`, optional
491 Argument that can be used (along with ``kwds``) to construct a
492 `DataId`.
494 Notes
495 -----
496 Butler APIs that conceptually need a DatasetRef also allow passing a
497 `DatasetType` (or the name of one) and a `DataId` (or a dict and
498 keyword arguments that can be used to construct one) separately. This
499 method accepts those arguments and always returns a true `DatasetType`
500 and a `DataId` or `dict`.
502 Standardization of `dict` vs `DataId` is best handled by passing the
503 returned ``dataId`` (and ``kwds``) to `Registry` APIs, which are
504 generally similarly flexible.
505 """
506 externalDatasetType: Optional[DatasetType] = None
507 internalDatasetType: Optional[DatasetType] = None
508 if isinstance(datasetRefOrType, DatasetRef):
509 if dataId is not None or kwds:
510 raise ValueError("DatasetRef given, cannot use dataId as well")
511 externalDatasetType = datasetRefOrType.datasetType
512 dataId = datasetRefOrType.dataId
513 else:
514 # Don't check whether DataId is provided, because Registry APIs
515 # can usually construct a better error message when it wasn't.
516 if isinstance(datasetRefOrType, DatasetType):
517 externalDatasetType = datasetRefOrType
518 else:
519 internalDatasetType = self.registry.getDatasetType(datasetRefOrType)
521 # Check that they are self-consistent
522 if externalDatasetType is not None:
523 internalDatasetType = self.registry.getDatasetType(externalDatasetType.name)
524 if externalDatasetType != internalDatasetType:
525 raise ValueError(f"Supplied dataset type ({externalDatasetType}) inconsistent with "
526 f"registry definition ({internalDatasetType})")
528 assert internalDatasetType is not None
529 return internalDatasetType, dataId
531 def _findDatasetRef(self, datasetRefOrType: Union[DatasetRef, DatasetType, str],
532 dataId: Optional[DataId] = None, *,
533 collections: Any = None,
534 allowUnresolved: bool = False,
535 **kwds: Any) -> DatasetRef:
536 """Shared logic for methods that start with a search for a dataset in
537 the registry.
539 Parameters
540 ----------
541 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
542 When `DatasetRef` the `dataId` should be `None`.
543 Otherwise the `DatasetType` or name thereof.
544 dataId : `dict` or `DataCoordinate`, optional
545 A `dict` of `Dimension` link name, value pairs that label the
546 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
547 should be provided as the first argument.
548 collections : Any, optional
549 Collections to be searched, overriding ``self.collections``.
550 Can be any of the types supported by the ``collections`` argument
551 to butler construction.
552 allowUnresolved : `bool`, optional
553 If `True`, return an unresolved `DatasetRef` if finding a resolved
554 one in the `Registry` fails. Defaults to `False`.
555 kwds
556 Additional keyword arguments used to augment or construct a
557 `DataId`. See `DataId` parameters.
559 Returns
560 -------
561 ref : `DatasetRef`
562 A reference to the dataset identified by the given arguments.
564 Raises
565 ------
566 LookupError
567 Raised if no matching dataset exists in the `Registry` (and
568 ``allowUnresolved is False``).
569 ValueError
570 Raised if a resolved `DatasetRef` was passed as an input, but it
571 differs from the one found in the registry.
572 TypeError
573 Raised if no collections were provided.
574 """
575 datasetType, dataId = self._standardizeArgs(datasetRefOrType, dataId, **kwds)
576 if isinstance(datasetRefOrType, DatasetRef):
577 idNumber = datasetRefOrType.id
578 else:
579 idNumber = None
580 timespan: Optional[Timespan] = None
582 # Process dimension records that are using record information
583 # rather than ids
584 newDataId: Dict[str, DataIdValue] = {}
585 byRecord: Dict[str, Dict[str, Any]] = defaultdict(dict)
587 # if all the dataId comes from keyword parameters we do not need
588 # to do anything here because they can't be of the form
589 # exposure.obs_id because a "." is not allowed in a keyword parameter.
590 if dataId:
591 for k, v in dataId.items():
592 # If we have a Dimension we do not need to do anything
593 # because it cannot be a compound key.
594 if isinstance(k, str) and "." in k:
595 # Someone is using a more human-readable dataId
596 dimensionName, record = k.split(".", 1)
597 byRecord[dimensionName][record] = v
598 elif isinstance(k, Dimension):
599 newDataId[k.name] = v
600 else:
601 newDataId[k] = v
603 # Go through the updated dataId and check the type in case someone is
604 # using an alternate key. We have already filtered out the compound
605 # keys dimensions.record format.
606 not_dimensions = {}
608 # Will need to look in the dataId and the keyword arguments
609 # and will remove them if they need to be fixed or are unrecognized.
610 for dataIdDict in (newDataId, kwds):
611 # Use a list so we can adjust the dict safely in the loop
612 for dimensionName in list(dataIdDict):
613 value = dataIdDict[dimensionName]
614 try:
615 dimension = self.registry.dimensions.getStaticDimensions()[dimensionName]
616 except KeyError:
617 # This is not a real dimension
618 not_dimensions[dimensionName] = value
619 del dataIdDict[dimensionName]
620 continue
622 # Convert an integral type to an explicit int to simplify
623 # comparisons here
624 if isinstance(value, numbers.Integral):
625 value = int(value)
627 if not isinstance(value, dimension.primaryKey.getPythonType()):
628 for alternate in dimension.alternateKeys:
629 if isinstance(value, alternate.getPythonType()):
630 byRecord[dimensionName][alternate.name] = value
631 del dataIdDict[dimensionName]
632 log.debug("Converting dimension %s to %s.%s=%s",
633 dimensionName, dimensionName, alternate.name, value)
634 break
635 else:
636 log.warning("Type mismatch found for value '%r' provided for dimension %s. "
637 "Could not find matching alternative (primary key has type %s) "
638 "so attempting to use as-is.",
639 value, dimensionName, dimension.primaryKey.getPythonType())
641 # If we have some unrecognized dimensions we have to try to connect
642 # them to records in other dimensions. This is made more complicated
643 # by some dimensions having records with clashing names. A mitigation
644 # is that we can tell by this point which dimensions are missing
645 # for the DatasetType but this does not work for calibrations
646 # where additional dimensions can be used to constrain the temporal
647 # axis.
648 if not_dimensions:
649 # Calculate missing dimensions
650 provided = set(newDataId) | set(kwds) | set(byRecord)
651 missingDimensions = datasetType.dimensions.names - provided
653 # For calibrations we may well be needing temporal dimensions
654 # so rather than always including all dimensions in the scan
655 # restrict things a little. It is still possible for there
656 # to be confusion over day_obs in visit vs exposure for example.
657 # If we are not searching calibration collections things may
658 # fail but they are going to fail anyway because of the
659 # ambiguousness of the dataId...
660 candidateDimensions: Set[str] = set()
661 candidateDimensions.update(missingDimensions)
662 if datasetType.isCalibration():
663 for dim in self.registry.dimensions.getStaticDimensions():
664 if dim.temporal:
665 candidateDimensions.add(str(dim))
667 # Look up table for the first association with a dimension
668 guessedAssociation: Dict[str, Dict[str, Any]] = defaultdict(dict)
670 # Keep track of whether an item is associated with multiple
671 # dimensions.
672 counter: Counter[str] = Counter()
673 assigned: Dict[str, Set[str]] = defaultdict(set)
675 # Go through the missing dimensions and associate the
676 # given names with records within those dimensions
677 for dimensionName in candidateDimensions:
678 dimension = self.registry.dimensions.getStaticDimensions()[dimensionName]
679 fields = dimension.metadata.names | dimension.uniqueKeys.names
680 for field in not_dimensions:
681 if field in fields:
682 guessedAssociation[dimensionName][field] = not_dimensions[field]
683 counter[dimensionName] += 1
684 assigned[field].add(dimensionName)
686 # There is a chance we have allocated a single dataId item
687 # to multiple dimensions. Need to decide which should be retained.
688 # For now assume that the most popular alternative wins.
689 # This means that day_obs with seq_num will result in
690 # exposure.day_obs and not visit.day_obs
691 # Also prefer an explicitly missing dimension over an inferred
692 # temporal dimension.
693 for fieldName, assignedDimensions in assigned.items():
694 if len(assignedDimensions) > 1:
695 # Pick the most popular (preferring mandatory dimensions)
696 requiredButMissing = assignedDimensions.intersection(missingDimensions)
697 if requiredButMissing:
698 candidateDimensions = requiredButMissing
699 else:
700 candidateDimensions = assignedDimensions
702 # Select the relevant items and get a new restricted
703 # counter.
704 theseCounts = {k: v for k, v in counter.items() if k in candidateDimensions}
705 duplicatesCounter: Counter[str] = Counter()
706 duplicatesCounter.update(theseCounts)
708 # Choose the most common. If they are equally common
709 # we will pick the one that was found first.
710 # Returns a list of tuples
711 selected = duplicatesCounter.most_common(1)[0][0]
713 log.debug("Ambiguous dataId entry '%s' associated with multiple dimensions: %s."
714 " Removed ambiguity by choosing dimension %s.",
715 fieldName, ", ".join(assignedDimensions), selected)
717 for candidateDimension in assignedDimensions:
718 if candidateDimension != selected:
719 del guessedAssociation[candidateDimension][fieldName]
721 # Update the record look up dict with the new associations
722 for dimensionName, values in guessedAssociation.items():
723 if values: # A dict might now be empty
724 log.debug("Assigned non-dimension dataId keys to dimension %s: %s",
725 dimensionName, values)
726 byRecord[dimensionName].update(values)
728 if byRecord:
729 # Some record specifiers were found so we need to convert
730 # them to the Id form
731 for dimensionName, values in byRecord.items():
732 if dimensionName in newDataId:
733 log.warning("DataId specified explicit %s dimension value of %s in addition to"
734 " general record specifiers for it of %s. Ignoring record information.",
735 dimensionName, newDataId[dimensionName], str(values))
736 continue
738 # Build up a WHERE expression -- use single quotes
739 def quote(s: Any) -> str:
740 if isinstance(s, str):
741 return f"'{s}'"
742 else:
743 return s
745 where = " AND ".join(f"{dimensionName}.{k} = {quote(v)}"
746 for k, v in values.items())
748 # Hopefully we get a single record that matches
749 records = set(self.registry.queryDimensionRecords(dimensionName, dataId=newDataId,
750 where=where, **kwds))
752 if len(records) != 1:
753 if len(records) > 1:
754 log.debug("Received %d records from constraints of %s", len(records), str(values))
755 for r in records:
756 log.debug("- %s", str(r))
757 raise RuntimeError(f"DataId specification for dimension {dimensionName} is not"
758 f" uniquely constrained to a single dataset by {values}."
759 f" Got {len(records)} results.")
760 raise RuntimeError(f"DataId specification for dimension {dimensionName} matched no"
761 f" records when constrained by {values}")
763 # Get the primary key from the real dimension object
764 dimension = self.registry.dimensions.getStaticDimensions()[dimensionName]
765 if not isinstance(dimension, Dimension):
766 raise RuntimeError(
767 f"{dimension.name} is not a true dimension, and cannot be used in data IDs."
768 )
769 newDataId[dimensionName] = getattr(records.pop(), dimension.primaryKey.name)
771 # We have modified the dataId so need to switch to it
772 dataId = newDataId
774 if datasetType.isCalibration():
775 # Because this is a calibration dataset, first try to make a
776 # standardize the data ID without restricting the dimensions to
777 # those of the dataset type requested, because there may be extra
778 # dimensions that provide temporal information for a validity-range
779 # lookup.
780 dataId = DataCoordinate.standardize(dataId, universe=self.registry.dimensions,
781 defaults=self.registry.defaults.dataId, **kwds)
782 if dataId.graph.temporal:
783 dataId = self.registry.expandDataId(dataId)
784 timespan = dataId.timespan
785 else:
786 # Standardize the data ID to just the dimensions of the dataset
787 # type instead of letting registry.findDataset do it, so we get the
788 # result even if no dataset is found.
789 dataId = DataCoordinate.standardize(dataId, graph=datasetType.dimensions,
790 defaults=self.registry.defaults.dataId, **kwds)
791 # Always lookup the DatasetRef, even if one is given, to ensure it is
792 # present in the current collection.
793 ref = self.registry.findDataset(datasetType, dataId, collections=collections, timespan=timespan)
794 if ref is None:
795 if allowUnresolved:
796 return DatasetRef(datasetType, dataId)
797 else:
798 if collections is None:
799 collections = self.registry.defaults.collections
800 raise LookupError(f"Dataset {datasetType.name} with data ID {dataId} "
801 f"could not be found in collections {collections}.")
802 if idNumber is not None and idNumber != ref.id:
803 if collections is None:
804 collections = self.registry.defaults.collections
805 raise ValueError(f"DatasetRef.id provided ({idNumber}) does not match "
806 f"id ({ref.id}) in registry in collections {collections}.")
807 return ref
809 @transactional
810 def put(self, obj: Any, datasetRefOrType: Union[DatasetRef, DatasetType, str],
811 dataId: Optional[DataId] = None, *,
812 run: Optional[str] = None,
813 **kwds: Any) -> DatasetRef:
814 """Store and register a dataset.
816 Parameters
817 ----------
818 obj : `object`
819 The dataset.
820 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
821 When `DatasetRef` is provided, ``dataId`` should be `None`.
822 Otherwise the `DatasetType` or name thereof.
823 dataId : `dict` or `DataCoordinate`
824 A `dict` of `Dimension` link name, value pairs that label the
825 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
826 should be provided as the second argument.
827 run : `str`, optional
828 The name of the run the dataset should be added to, overriding
829 ``self.run``.
830 kwds
831 Additional keyword arguments used to augment or construct a
832 `DataCoordinate`. See `DataCoordinate.standardize`
833 parameters.
835 Returns
836 -------
837 ref : `DatasetRef`
838 A reference to the stored dataset, updated with the correct id if
839 given.
841 Raises
842 ------
843 TypeError
844 Raised if the butler is read-only or if no run has been provided.
845 """
846 log.debug("Butler put: %s, dataId=%s, run=%s", datasetRefOrType, dataId, run)
847 if not self.isWriteable():
848 raise TypeError("Butler is read-only.")
849 datasetType, dataId = self._standardizeArgs(datasetRefOrType, dataId, **kwds)
850 if isinstance(datasetRefOrType, DatasetRef) and datasetRefOrType.id is not None:
851 raise ValueError("DatasetRef must not be in registry, must have None id")
853 # Add Registry Dataset entry.
854 dataId = self.registry.expandDataId(dataId, graph=datasetType.dimensions, **kwds)
855 ref, = self.registry.insertDatasets(datasetType, run=run, dataIds=[dataId])
857 # Add Datastore entry.
858 self.datastore.put(obj, ref)
860 return ref
862 def getDirect(self, ref: DatasetRef, *, parameters: Optional[Dict[str, Any]] = None) -> Any:
863 """Retrieve a stored dataset.
865 Unlike `Butler.get`, this method allows datasets outside the Butler's
866 collection to be read as long as the `DatasetRef` that identifies them
867 can be obtained separately.
869 Parameters
870 ----------
871 ref : `DatasetRef`
872 Resolved reference to an already stored dataset.
873 parameters : `dict`
874 Additional StorageClass-defined options to control reading,
875 typically used to efficiently read only a subset of the dataset.
877 Returns
878 -------
879 obj : `object`
880 The dataset.
881 """
882 return self.datastore.get(ref, parameters=parameters)
884 def getDirectDeferred(self, ref: DatasetRef, *,
885 parameters: Union[dict, None] = None) -> DeferredDatasetHandle:
886 """Create a `DeferredDatasetHandle` which can later retrieve a dataset,
887 from a resolved `DatasetRef`.
889 Parameters
890 ----------
891 ref : `DatasetRef`
892 Resolved reference to an already stored dataset.
893 parameters : `dict`
894 Additional StorageClass-defined options to control reading,
895 typically used to efficiently read only a subset of the dataset.
897 Returns
898 -------
899 obj : `DeferredDatasetHandle`
900 A handle which can be used to retrieve a dataset at a later time.
902 Raises
903 ------
904 AmbiguousDatasetError
905 Raised if ``ref.id is None``, i.e. the reference is unresolved.
906 """
907 if ref.id is None:
908 raise AmbiguousDatasetError(
909 f"Dataset of type {ref.datasetType.name} with data ID {ref.dataId} is not resolved."
910 )
911 return DeferredDatasetHandle(butler=self, ref=ref, parameters=parameters)
913 def getDeferred(self, datasetRefOrType: Union[DatasetRef, DatasetType, str],
914 dataId: Optional[DataId] = None, *,
915 parameters: Union[dict, None] = None,
916 collections: Any = None,
917 **kwds: Any) -> DeferredDatasetHandle:
918 """Create a `DeferredDatasetHandle` which can later retrieve a dataset,
919 after an immediate registry lookup.
921 Parameters
922 ----------
923 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
924 When `DatasetRef` the `dataId` should be `None`.
925 Otherwise the `DatasetType` or name thereof.
926 dataId : `dict` or `DataCoordinate`, optional
927 A `dict` of `Dimension` link name, value pairs that label the
928 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
929 should be provided as the first argument.
930 parameters : `dict`
931 Additional StorageClass-defined options to control reading,
932 typically used to efficiently read only a subset of the dataset.
933 collections : Any, optional
934 Collections to be searched, overriding ``self.collections``.
935 Can be any of the types supported by the ``collections`` argument
936 to butler construction.
937 kwds
938 Additional keyword arguments used to augment or construct a
939 `DataId`. See `DataId` parameters.
941 Returns
942 -------
943 obj : `DeferredDatasetHandle`
944 A handle which can be used to retrieve a dataset at a later time.
946 Raises
947 ------
948 LookupError
949 Raised if no matching dataset exists in the `Registry` (and
950 ``allowUnresolved is False``).
951 ValueError
952 Raised if a resolved `DatasetRef` was passed as an input, but it
953 differs from the one found in the registry.
954 TypeError
955 Raised if no collections were provided.
956 """
957 ref = self._findDatasetRef(datasetRefOrType, dataId, collections=collections, **kwds)
958 return DeferredDatasetHandle(butler=self, ref=ref, parameters=parameters)
960 def get(self, datasetRefOrType: Union[DatasetRef, DatasetType, str],
961 dataId: Optional[DataId] = None, *,
962 parameters: Optional[Dict[str, Any]] = None,
963 collections: Any = None,
964 **kwds: Any) -> Any:
965 """Retrieve a stored dataset.
967 Parameters
968 ----------
969 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
970 When `DatasetRef` the `dataId` should be `None`.
971 Otherwise the `DatasetType` or name thereof.
972 dataId : `dict` or `DataCoordinate`
973 A `dict` of `Dimension` link name, value pairs that label the
974 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
975 should be provided as the first argument.
976 parameters : `dict`
977 Additional StorageClass-defined options to control reading,
978 typically used to efficiently read only a subset of the dataset.
979 collections : Any, optional
980 Collections to be searched, overriding ``self.collections``.
981 Can be any of the types supported by the ``collections`` argument
982 to butler construction.
983 kwds
984 Additional keyword arguments used to augment or construct a
985 `DataCoordinate`. See `DataCoordinate.standardize`
986 parameters.
988 Returns
989 -------
990 obj : `object`
991 The dataset.
993 Raises
994 ------
995 ValueError
996 Raised if a resolved `DatasetRef` was passed as an input, but it
997 differs from the one found in the registry.
998 LookupError
999 Raised if no matching dataset exists in the `Registry`.
1000 TypeError
1001 Raised if no collections were provided.
1003 Notes
1004 -----
1005 When looking up datasets in a `~CollectionType.CALIBRATION` collection,
1006 this method requires that the given data ID include temporal dimensions
1007 beyond the dimensions of the dataset type itself, in order to find the
1008 dataset with the appropriate validity range. For example, a "bias"
1009 dataset with native dimensions ``{instrument, detector}`` could be
1010 fetched with a ``{instrument, detector, exposure}`` data ID, because
1011 ``exposure`` is a temporal dimension.
1012 """
1013 log.debug("Butler get: %s, dataId=%s, parameters=%s", datasetRefOrType, dataId, parameters)
1014 ref = self._findDatasetRef(datasetRefOrType, dataId, collections=collections, **kwds)
1015 return self.getDirect(ref, parameters=parameters)
1017 def getURIs(self, datasetRefOrType: Union[DatasetRef, DatasetType, str],
1018 dataId: Optional[DataId] = None, *,
1019 predict: bool = False,
1020 collections: Any = None,
1021 run: Optional[str] = None,
1022 **kwds: Any) -> Tuple[Optional[ButlerURI], Dict[str, ButlerURI]]:
1023 """Returns the URIs associated with the dataset.
1025 Parameters
1026 ----------
1027 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
1028 When `DatasetRef` the `dataId` should be `None`.
1029 Otherwise the `DatasetType` or name thereof.
1030 dataId : `dict` or `DataCoordinate`
1031 A `dict` of `Dimension` link name, value pairs that label the
1032 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
1033 should be provided as the first argument.
1034 predict : `bool`
1035 If `True`, allow URIs to be returned of datasets that have not
1036 been written.
1037 collections : Any, optional
1038 Collections to be searched, overriding ``self.collections``.
1039 Can be any of the types supported by the ``collections`` argument
1040 to butler construction.
1041 run : `str`, optional
1042 Run to use for predictions, overriding ``self.run``.
1043 kwds
1044 Additional keyword arguments used to augment or construct a
1045 `DataCoordinate`. See `DataCoordinate.standardize`
1046 parameters.
1048 Returns
1049 -------
1050 primary : `ButlerURI`
1051 The URI to the primary artifact associated with this dataset.
1052 If the dataset was disassembled within the datastore this
1053 may be `None`.
1054 components : `dict`
1055 URIs to any components associated with the dataset artifact.
1056 Can be empty if there are no components.
1057 """
1058 ref = self._findDatasetRef(datasetRefOrType, dataId, allowUnresolved=predict,
1059 collections=collections, **kwds)
1060 if ref.id is None: # only possible if predict is True
1061 if run is None:
1062 run = self.run
1063 if run is None:
1064 raise TypeError("Cannot predict location with run=None.")
1065 # Lie about ID, because we can't guess it, and only
1066 # Datastore.getURIs() will ever see it (and it doesn't use it).
1067 ref = ref.resolved(id=0, run=run)
1068 return self.datastore.getURIs(ref, predict)
1070 def getURI(self, datasetRefOrType: Union[DatasetRef, DatasetType, str],
1071 dataId: Optional[DataId] = None, *,
1072 predict: bool = False,
1073 collections: Any = None,
1074 run: Optional[str] = None,
1075 **kwds: Any) -> ButlerURI:
1076 """Return the URI to the Dataset.
1078 Parameters
1079 ----------
1080 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
1081 When `DatasetRef` the `dataId` should be `None`.
1082 Otherwise the `DatasetType` or name thereof.
1083 dataId : `dict` or `DataCoordinate`
1084 A `dict` of `Dimension` link name, value pairs that label the
1085 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
1086 should be provided as the first argument.
1087 predict : `bool`
1088 If `True`, allow URIs to be returned of datasets that have not
1089 been written.
1090 collections : Any, optional
1091 Collections to be searched, overriding ``self.collections``.
1092 Can be any of the types supported by the ``collections`` argument
1093 to butler construction.
1094 run : `str`, optional
1095 Run to use for predictions, overriding ``self.run``.
1096 kwds
1097 Additional keyword arguments used to augment or construct a
1098 `DataCoordinate`. See `DataCoordinate.standardize`
1099 parameters.
1101 Returns
1102 -------
1103 uri : `ButlerURI`
1104 URI pointing to the Dataset within the datastore. If the
1105 Dataset does not exist in the datastore, and if ``predict`` is
1106 `True`, the URI will be a prediction and will include a URI
1107 fragment "#predicted".
1108 If the datastore does not have entities that relate well
1109 to the concept of a URI the returned URI string will be
1110 descriptive. The returned URI is not guaranteed to be obtainable.
1112 Raises
1113 ------
1114 LookupError
1115 A URI has been requested for a dataset that does not exist and
1116 guessing is not allowed.
1117 ValueError
1118 Raised if a resolved `DatasetRef` was passed as an input, but it
1119 differs from the one found in the registry.
1120 TypeError
1121 Raised if no collections were provided.
1122 RuntimeError
1123 Raised if a URI is requested for a dataset that consists of
1124 multiple artifacts.
1125 """
1126 primary, components = self.getURIs(datasetRefOrType, dataId=dataId, predict=predict,
1127 collections=collections, run=run, **kwds)
1129 if primary is None or components:
1130 raise RuntimeError(f"Dataset ({datasetRefOrType}) includes distinct URIs for components. "
1131 "Use Butler.getURIs() instead.")
1132 return primary
1134 def datasetExists(self, datasetRefOrType: Union[DatasetRef, DatasetType, str],
1135 dataId: Optional[DataId] = None, *,
1136 collections: Any = None,
1137 **kwds: Any) -> bool:
1138 """Return True if the Dataset is actually present in the Datastore.
1140 Parameters
1141 ----------
1142 datasetRefOrType : `DatasetRef`, `DatasetType`, or `str`
1143 When `DatasetRef` the `dataId` should be `None`.
1144 Otherwise the `DatasetType` or name thereof.
1145 dataId : `dict` or `DataCoordinate`
1146 A `dict` of `Dimension` link name, value pairs that label the
1147 `DatasetRef` within a Collection. When `None`, a `DatasetRef`
1148 should be provided as the first argument.
1149 collections : Any, optional
1150 Collections to be searched, overriding ``self.collections``.
1151 Can be any of the types supported by the ``collections`` argument
1152 to butler construction.
1153 kwds
1154 Additional keyword arguments used to augment or construct a
1155 `DataCoordinate`. See `DataCoordinate.standardize`
1156 parameters.
1158 Raises
1159 ------
1160 LookupError
1161 Raised if the dataset is not even present in the Registry.
1162 ValueError
1163 Raised if a resolved `DatasetRef` was passed as an input, but it
1164 differs from the one found in the registry.
1165 TypeError
1166 Raised if no collections were provided.
1167 """
1168 ref = self._findDatasetRef(datasetRefOrType, dataId, collections=collections, **kwds)
1169 return self.datastore.exists(ref)
1171 def removeRuns(self, names: Iterable[str], unstore: bool = True) -> None:
1172 """Remove one or more `~CollectionType.RUN` collections and the
1173 datasets within them.
1175 Parameters
1176 ----------
1177 names : `Iterable` [ `str` ]
1178 The names of the collections to remove.
1179 unstore : `bool`, optional
1180 If `True` (default), delete datasets from all datastores in which
1181 they are present, and attempt to rollback the registry deletions if
1182 datastore deletions fail (which may not always be possible). If
1183 `False`, datastore records for these datasets are still removed,
1184 but any artifacts (e.g. files) will not be.
1186 Raises
1187 ------
1188 TypeError
1189 Raised if one or more collections are not of type
1190 `~CollectionType.RUN`.
1191 """
1192 if not self.isWriteable():
1193 raise TypeError("Butler is read-only.")
1194 names = list(names)
1195 refs: List[DatasetRef] = []
1196 for name in names:
1197 collectionType = self.registry.getCollectionType(name)
1198 if collectionType is not CollectionType.RUN:
1199 raise TypeError(f"The collection type of '{name}' is {collectionType.name}, not RUN.")
1200 refs.extend(self.registry.queryDatasets(..., collections=name, findFirst=True))
1201 with self.registry.transaction():
1202 if unstore:
1203 for ref in refs:
1204 if self.datastore.exists(ref):
1205 self.datastore.trash(ref)
1206 else:
1207 self.datastore.forget(refs)
1208 for name in names:
1209 self.registry.removeCollection(name)
1210 if unstore:
1211 # Point of no return for removing artifacts
1212 self.datastore.emptyTrash()
1214 def pruneCollection(self, name: str, purge: bool = False, unstore: bool = False,
1215 unlink: Optional[List[str]] = None) -> None:
1216 """Remove a collection and possibly prune datasets within it.
1218 Parameters
1219 ----------
1220 name : `str`
1221 Name of the collection to remove. If this is a
1222 `~CollectionType.TAGGED` or `~CollectionType.CHAINED` collection,
1223 datasets within the collection are not modified unless ``unstore``
1224 is `True`. If this is a `~CollectionType.RUN` collection,
1225 ``purge`` and ``unstore`` must be `True`, and all datasets in it
1226 are fully removed from the data repository.
1227 purge : `bool`, optional
1228 If `True`, permit `~CollectionType.RUN` collections to be removed,
1229 fully removing datasets within them. Requires ``unstore=True`` as
1230 well as an added precaution against accidental deletion. Must be
1231 `False` (default) if the collection is not a ``RUN``.
1232 unstore: `bool`, optional
1233 If `True`, remove all datasets in the collection from all
1234 datastores in which they appear.
1235 unlink: `list` [`str`], optional
1236 Before removing the given `collection` unlink it from from these
1237 parent collections.
1239 Raises
1240 ------
1241 TypeError
1242 Raised if the butler is read-only or arguments are mutually
1243 inconsistent.
1244 """
1246 # See pruneDatasets comments for more information about the logic here;
1247 # the cases are almost the same, but here we can rely on Registry to
1248 # take care everything but Datastore deletion when we remove the
1249 # collection.
1250 if not self.isWriteable():
1251 raise TypeError("Butler is read-only.")
1252 collectionType = self.registry.getCollectionType(name)
1253 if purge and not unstore:
1254 raise PurgeWithoutUnstorePruneCollectionsError()
1255 if collectionType is CollectionType.RUN and not purge:
1256 raise RunWithoutPurgePruneCollectionsError(collectionType)
1257 if collectionType is not CollectionType.RUN and purge:
1258 raise PurgeUnsupportedPruneCollectionsError(collectionType)
1260 def remove(child: str, parent: str) -> None:
1261 """Remove a child collection from a parent collection."""
1262 # Remove child from parent.
1263 chain = list(self.registry.getCollectionChain(parent))
1264 try:
1265 chain.remove(name)
1266 except ValueError as e:
1267 raise RuntimeError(f"{name} is not a child of {parent}") from e
1268 self.registry.setCollectionChain(parent, chain)
1270 with self.registry.transaction():
1271 if (unlink):
1272 for parent in unlink:
1273 remove(name, parent)
1274 if unstore:
1275 for ref in self.registry.queryDatasets(..., collections=name, findFirst=True):
1276 if self.datastore.exists(ref):
1277 self.datastore.trash(ref)
1278 self.registry.removeCollection(name)
1279 if unstore:
1280 # Point of no return for removing artifacts
1281 self.datastore.emptyTrash()
1283 def pruneDatasets(self, refs: Iterable[DatasetRef], *,
1284 disassociate: bool = True,
1285 unstore: bool = False,
1286 tags: Iterable[str] = (),
1287 purge: bool = False,
1288 run: Optional[str] = None) -> None:
1289 """Remove one or more datasets from a collection and/or storage.
1291 Parameters
1292 ----------
1293 refs : `~collections.abc.Iterable` of `DatasetRef`
1294 Datasets to prune. These must be "resolved" references (not just
1295 a `DatasetType` and data ID).
1296 disassociate : `bool`, optional
1297 Disassociate pruned datasets from ``tags``, or from all collections
1298 if ``purge=True``.
1299 unstore : `bool`, optional
1300 If `True` (`False` is default) remove these datasets from all
1301 datastores known to this butler. Note that this will make it
1302 impossible to retrieve these datasets even via other collections.
1303 Datasets that are already not stored are ignored by this option.
1304 tags : `Iterable` [ `str` ], optional
1305 `~CollectionType.TAGGED` collections to disassociate the datasets
1306 from. Ignored if ``disassociate`` is `False` or ``purge`` is
1307 `True`.
1308 purge : `bool`, optional
1309 If `True` (`False` is default), completely remove the dataset from
1310 the `Registry`. To prevent accidental deletions, ``purge`` may
1311 only be `True` if all of the following conditions are met:
1313 - All given datasets are in the given run.
1314 - ``disassociate`` is `True`;
1315 - ``unstore`` is `True`.
1317 This mode may remove provenance information from datasets other
1318 than those provided, and should be used with extreme care.
1320 Raises
1321 ------
1322 TypeError
1323 Raised if the butler is read-only, if no collection was provided,
1324 or the conditions for ``purge=True`` were not met.
1325 """
1326 if not self.isWriteable():
1327 raise TypeError("Butler is read-only.")
1328 if purge:
1329 if not disassociate:
1330 raise TypeError("Cannot pass purge=True without disassociate=True.")
1331 if not unstore:
1332 raise TypeError("Cannot pass purge=True without unstore=True.")
1333 elif disassociate:
1334 tags = tuple(tags)
1335 if not tags:
1336 raise TypeError("No tags provided but disassociate=True.")
1337 for tag in tags:
1338 collectionType = self.registry.getCollectionType(tag)
1339 if collectionType is not CollectionType.TAGGED:
1340 raise TypeError(f"Cannot disassociate from collection '{tag}' "
1341 f"of non-TAGGED type {collectionType.name}.")
1342 # Transform possibly-single-pass iterable into something we can iterate
1343 # over multiple times.
1344 refs = list(refs)
1345 # Pruning a component of a DatasetRef makes no sense since registry
1346 # doesn't know about components and datastore might not store
1347 # components in a separate file
1348 for ref in refs:
1349 if ref.datasetType.component():
1350 raise ValueError(f"Can not prune a component of a dataset (ref={ref})")
1351 # We don't need an unreliable Datastore transaction for this, because
1352 # we've been extra careful to ensure that Datastore.trash only involves
1353 # mutating the Registry (it can _look_ at Datastore-specific things,
1354 # but shouldn't change them), and hence all operations here are
1355 # Registry operations.
1356 with self.registry.transaction():
1357 if unstore:
1358 for ref in refs:
1359 # There is a difference between a concrete composite
1360 # and virtual composite. In a virtual composite the
1361 # datastore is never given the top level DatasetRef. In
1362 # the concrete composite the datastore knows all the
1363 # refs and will clean up itself if asked to remove the
1364 # parent ref. We can not check configuration for this
1365 # since we can not trust that the configuration is the
1366 # same. We therefore have to ask if the ref exists or
1367 # not. This is consistent with the fact that we want
1368 # to ignore already-removed-from-datastore datasets
1369 # anyway.
1370 if self.datastore.exists(ref):
1371 self.datastore.trash(ref)
1372 if purge:
1373 self.registry.removeDatasets(refs)
1374 elif disassociate:
1375 assert tags, "Guaranteed by earlier logic in this function."
1376 for tag in tags:
1377 self.registry.disassociate(tag, refs)
1378 # We've exited the Registry transaction, and apparently committed.
1379 # (if there was an exception, everything rolled back, and it's as if
1380 # nothing happened - and we never get here).
1381 # Datastore artifacts are not yet gone, but they're clearly marked
1382 # as trash, so if we fail to delete now because of (e.g.) filesystem
1383 # problems we can try again later, and if manual administrative
1384 # intervention is required, it's pretty clear what that should entail:
1385 # deleting everything on disk and in private Datastore tables that is
1386 # in the dataset_location_trash table.
1387 if unstore:
1388 # Point of no return for removing artifacts
1389 self.datastore.emptyTrash()
1391 @transactional
1392 def ingest(self, *datasets: FileDataset, transfer: Optional[str] = "auto", run: Optional[str] = None,
1393 idGenerationMode: DatasetIdGenEnum = DatasetIdGenEnum.UNIQUE,
1394 ) -> None:
1395 """Store and register one or more datasets that already exist on disk.
1397 Parameters
1398 ----------
1399 datasets : `FileDataset`
1400 Each positional argument is a struct containing information about
1401 a file to be ingested, including its path (either absolute or
1402 relative to the datastore root, if applicable), a `DatasetRef`,
1403 and optionally a formatter class or its fully-qualified string
1404 name. If a formatter is not provided, the formatter that would be
1405 used for `put` is assumed. On successful return, all
1406 `FileDataset.ref` attributes will have their `DatasetRef.id`
1407 attribute populated and all `FileDataset.formatter` attributes will
1408 be set to the formatter class used. `FileDataset.path` attributes
1409 may be modified to put paths in whatever the datastore considers a
1410 standardized form.
1411 transfer : `str`, optional
1412 If not `None`, must be one of 'auto', 'move', 'copy', 'direct',
1413 'hardlink', 'relsymlink' or 'symlink', indicating how to transfer
1414 the file.
1415 run : `str`, optional
1416 The name of the run ingested datasets should be added to,
1417 overriding ``self.run``.
1418 idGenerationMode : `DatasetIdGenEnum`, optional
1419 Specifies option for generating dataset IDs. By default unique IDs
1420 are generated for each inserted dataset.
1422 Raises
1423 ------
1424 TypeError
1425 Raised if the butler is read-only or if no run was provided.
1426 NotImplementedError
1427 Raised if the `Datastore` does not support the given transfer mode.
1428 DatasetTypeNotSupportedError
1429 Raised if one or more files to be ingested have a dataset type that
1430 is not supported by the `Datastore`..
1431 FileNotFoundError
1432 Raised if one of the given files does not exist.
1433 FileExistsError
1434 Raised if transfer is not `None` but the (internal) location the
1435 file would be moved to is already occupied.
1437 Notes
1438 -----
1439 This operation is not fully exception safe: if a database operation
1440 fails, the given `FileDataset` instances may be only partially updated.
1442 It is atomic in terms of database operations (they will either all
1443 succeed or all fail) providing the database engine implements
1444 transactions correctly. It will attempt to be atomic in terms of
1445 filesystem operations as well, but this cannot be implemented
1446 rigorously for most datastores.
1447 """
1448 if not self.isWriteable():
1449 raise TypeError("Butler is read-only.")
1450 progress = Progress("lsst.daf.butler.Butler.ingest", level=logging.DEBUG)
1451 # Reorganize the inputs so they're grouped by DatasetType and then
1452 # data ID. We also include a list of DatasetRefs for each FileDataset
1453 # to hold the resolved DatasetRefs returned by the Registry, before
1454 # it's safe to swap them into FileDataset.refs.
1455 # Some type annotation aliases to make that clearer:
1456 GroupForType = Dict[DataCoordinate, Tuple[FileDataset, List[DatasetRef]]]
1457 GroupedData = MutableMapping[DatasetType, GroupForType]
1458 # The actual data structure:
1459 groupedData: GroupedData = defaultdict(dict)
1460 # And the nested loop that populates it:
1461 for dataset in progress.wrap(datasets, desc="Grouping by dataset type"):
1462 # This list intentionally shared across the inner loop, since it's
1463 # associated with `dataset`.
1464 resolvedRefs: List[DatasetRef] = []
1465 for ref in dataset.refs:
1466 if ref.dataId in groupedData[ref.datasetType]:
1467 raise ConflictingDefinitionError(f"Ingest conflict. Dataset {dataset.path} has same"
1468 " DataId as other ingest dataset"
1469 f" {groupedData[ref.datasetType][ref.dataId][0].path} "
1470 f" ({ref.dataId})")
1471 groupedData[ref.datasetType][ref.dataId] = (dataset, resolvedRefs)
1473 # Now we can bulk-insert into Registry for each DatasetType.
1474 allResolvedRefs: List[DatasetRef] = []
1475 for datasetType, groupForType in progress.iter_item_chunks(groupedData.items(),
1476 desc="Bulk-inserting datasets by type"):
1477 refs = self.registry.insertDatasets(
1478 datasetType,
1479 dataIds=groupForType.keys(),
1480 run=run,
1481 expand=self.datastore.needs_expanded_data_ids(transfer, datasetType),
1482 idGenerationMode=idGenerationMode,
1483 )
1484 # Append those resolved DatasetRefs to the new lists we set up for
1485 # them.
1486 for ref, (_, resolvedRefs) in zip(refs, groupForType.values()):
1487 resolvedRefs.append(ref)
1489 # Go back to the original FileDatasets to replace their refs with the
1490 # new resolved ones, and also build a big list of all refs.
1491 allResolvedRefs = []
1492 for groupForType in progress.iter_chunks(groupedData.values(),
1493 desc="Reassociating resolved dataset refs with files"):
1494 for dataset, resolvedRefs in groupForType.values():
1495 dataset.refs = resolvedRefs
1496 allResolvedRefs.extend(resolvedRefs)
1498 # Bulk-insert everything into Datastore.
1499 self.datastore.ingest(*datasets, transfer=transfer)
1501 @contextlib.contextmanager
1502 def export(self, *, directory: Optional[str] = None,
1503 filename: Optional[str] = None,
1504 format: Optional[str] = None,
1505 transfer: Optional[str] = None) -> Iterator[RepoExportContext]:
1506 """Export datasets from the repository represented by this `Butler`.
1508 This method is a context manager that returns a helper object
1509 (`RepoExportContext`) that is used to indicate what information from
1510 the repository should be exported.
1512 Parameters
1513 ----------
1514 directory : `str`, optional
1515 Directory dataset files should be written to if ``transfer`` is not
1516 `None`.
1517 filename : `str`, optional
1518 Name for the file that will include database information associated
1519 with the exported datasets. If this is not an absolute path and
1520 ``directory`` is not `None`, it will be written to ``directory``
1521 instead of the current working directory. Defaults to
1522 "export.{format}".
1523 format : `str`, optional
1524 File format for the database information file. If `None`, the
1525 extension of ``filename`` will be used.
1526 transfer : `str`, optional
1527 Transfer mode passed to `Datastore.export`.
1529 Raises
1530 ------
1531 TypeError
1532 Raised if the set of arguments passed is inconsistent.
1534 Examples
1535 --------
1536 Typically the `Registry.queryDataIds` and `Registry.queryDatasets`
1537 methods are used to provide the iterables over data IDs and/or datasets
1538 to be exported::
1540 with butler.export("exports.yaml") as export:
1541 # Export all flats, but none of the dimension element rows
1542 # (i.e. data ID information) associated with them.
1543 export.saveDatasets(butler.registry.queryDatasets("flat"),
1544 elements=())
1545 # Export all datasets that start with "deepCoadd_" and all of
1546 # their associated data ID information.
1547 export.saveDatasets(butler.registry.queryDatasets("deepCoadd_*"))
1548 """
1549 if directory is None and transfer is not None:
1550 raise TypeError("Cannot transfer without providing a directory.")
1551 if transfer == "move":
1552 raise TypeError("Transfer may not be 'move': export is read-only")
1553 if format is None:
1554 if filename is None:
1555 raise TypeError("At least one of 'filename' or 'format' must be provided.")
1556 else:
1557 _, format = os.path.splitext(filename)
1558 elif filename is None:
1559 filename = f"export.{format}"
1560 if directory is not None:
1561 filename = os.path.join(directory, filename)
1562 BackendClass = getClassOf(self._config["repo_transfer_formats"][format]["export"])
1563 with open(filename, 'w') as stream:
1564 backend = BackendClass(stream)
1565 try:
1566 helper = RepoExportContext(self.registry, self.datastore, backend=backend,
1567 directory=directory, transfer=transfer)
1568 yield helper
1569 except BaseException:
1570 raise
1571 else:
1572 helper._finish()
1574 def import_(self, *, directory: Optional[str] = None,
1575 filename: Union[str, TextIO, None] = None,
1576 format: Optional[str] = None,
1577 transfer: Optional[str] = None,
1578 skip_dimensions: Optional[Set] = None,
1579 idGenerationMode: DatasetIdGenEnum = DatasetIdGenEnum.UNIQUE,
1580 reuseIds: bool = False) -> None:
1581 """Import datasets into this repository that were exported from a
1582 different butler repository via `~lsst.daf.butler.Butler.export`.
1584 Parameters
1585 ----------
1586 directory : `str`, optional
1587 Directory containing dataset files to import from. If `None`,
1588 ``filename`` and all dataset file paths specified therein must
1589 be absolute.
1590 filename : `str` or `TextIO`, optional
1591 A stream or name of file that contains database information
1592 associated with the exported datasets, typically generated by
1593 `~lsst.daf.butler.Butler.export`. If this a string (name) and
1594 is not an absolute path, does not exist in the current working
1595 directory, and ``directory`` is not `None`, it is assumed to be in
1596 ``directory``. Defaults to "export.{format}".
1597 format : `str`, optional
1598 File format for ``filename``. If `None`, the extension of
1599 ``filename`` will be used.
1600 transfer : `str`, optional
1601 Transfer mode passed to `~lsst.daf.butler.Datastore.ingest`.
1602 skip_dimensions : `set`, optional
1603 Names of dimensions that should be skipped and not imported.
1604 idGenerationMode : `DatasetIdGenEnum`, optional
1605 Specifies option for generating dataset IDs when IDs are not
1606 provided or their type does not match backend type. By default
1607 unique IDs are generated for each inserted dataset.
1608 reuseIds : `bool`, optional
1609 If `True` then forces re-use of imported dataset IDs for integer
1610 IDs which are normally generated as auto-incremented; exception
1611 will be raised if imported IDs clash with existing ones. This
1612 option has no effect on the use of globally-unique IDs which are
1613 always re-used (or generated if integer IDs are being imported).
1615 Raises
1616 ------
1617 TypeError
1618 Raised if the set of arguments passed is inconsistent, or if the
1619 butler is read-only.
1620 """
1621 if not self.isWriteable():
1622 raise TypeError("Butler is read-only.")
1623 if format is None:
1624 if filename is None:
1625 raise TypeError("At least one of 'filename' or 'format' must be provided.")
1626 else:
1627 _, format = os.path.splitext(filename) # type: ignore
1628 elif filename is None:
1629 filename = f"export.{format}"
1630 if isinstance(filename, str) and directory is not None and not os.path.exists(filename):
1631 filename = os.path.join(directory, filename)
1632 BackendClass = getClassOf(self._config["repo_transfer_formats"][format]["import"])
1634 def doImport(importStream: TextIO) -> None:
1635 backend = BackendClass(importStream, self.registry)
1636 backend.register()
1637 with self.transaction():
1638 backend.load(self.datastore, directory=directory, transfer=transfer,
1639 skip_dimensions=skip_dimensions, idGenerationMode=idGenerationMode,
1640 reuseIds=reuseIds)
1642 if isinstance(filename, str):
1643 with open(filename, "r") as stream:
1644 doImport(stream)
1645 else:
1646 doImport(filename)
1648 def validateConfiguration(self, logFailures: bool = False,
1649 datasetTypeNames: Optional[Iterable[str]] = None,
1650 ignore: Iterable[str] = None) -> None:
1651 """Validate butler configuration.
1653 Checks that each `DatasetType` can be stored in the `Datastore`.
1655 Parameters
1656 ----------
1657 logFailures : `bool`, optional
1658 If `True`, output a log message for every validation error
1659 detected.
1660 datasetTypeNames : iterable of `str`, optional
1661 The `DatasetType` names that should be checked. This allows
1662 only a subset to be selected.
1663 ignore : iterable of `str`, optional
1664 Names of DatasetTypes to skip over. This can be used to skip
1665 known problems. If a named `DatasetType` corresponds to a
1666 composite, all components of that `DatasetType` will also be
1667 ignored.
1669 Raises
1670 ------
1671 ButlerValidationError
1672 Raised if there is some inconsistency with how this Butler
1673 is configured.
1674 """
1675 if datasetTypeNames:
1676 datasetTypes = [self.registry.getDatasetType(name) for name in datasetTypeNames]
1677 else:
1678 datasetTypes = list(self.registry.queryDatasetTypes())
1680 # filter out anything from the ignore list
1681 if ignore:
1682 ignore = set(ignore)
1683 datasetTypes = [e for e in datasetTypes
1684 if e.name not in ignore and e.nameAndComponent()[0] not in ignore]
1685 else:
1686 ignore = set()
1688 # Find all the registered instruments
1689 instruments = set(
1690 record.name for record in self.registry.queryDimensionRecords("instrument")
1691 )
1693 # For each datasetType that has an instrument dimension, create
1694 # a DatasetRef for each defined instrument
1695 datasetRefs = []
1697 for datasetType in datasetTypes:
1698 if "instrument" in datasetType.dimensions:
1699 for instrument in instruments:
1700 datasetRef = DatasetRef(datasetType, {"instrument": instrument}, # type: ignore
1701 conform=False)
1702 datasetRefs.append(datasetRef)
1704 entities: List[Union[DatasetType, DatasetRef]] = []
1705 entities.extend(datasetTypes)
1706 entities.extend(datasetRefs)
1708 datastoreErrorStr = None
1709 try:
1710 self.datastore.validateConfiguration(entities, logFailures=logFailures)
1711 except ValidationError as e:
1712 datastoreErrorStr = str(e)
1714 # Also check that the LookupKeys used by the datastores match
1715 # registry and storage class definitions
1716 keys = self.datastore.getLookupKeys()
1718 failedNames = set()
1719 failedDataId = set()
1720 for key in keys:
1721 if key.name is not None:
1722 if key.name in ignore:
1723 continue
1725 # skip if specific datasetType names were requested and this
1726 # name does not match
1727 if datasetTypeNames and key.name not in datasetTypeNames:
1728 continue
1730 # See if it is a StorageClass or a DatasetType
1731 if key.name in self.storageClasses:
1732 pass
1733 else:
1734 try:
1735 self.registry.getDatasetType(key.name)
1736 except KeyError:
1737 if logFailures:
1738 log.critical("Key '%s' does not correspond to a DatasetType or StorageClass", key)
1739 failedNames.add(key)
1740 else:
1741 # Dimensions are checked for consistency when the Butler
1742 # is created and rendezvoused with a universe.
1743 pass
1745 # Check that the instrument is a valid instrument
1746 # Currently only support instrument so check for that
1747 if key.dataId:
1748 dataIdKeys = set(key.dataId)
1749 if set(["instrument"]) != dataIdKeys:
1750 if logFailures:
1751 log.critical("Key '%s' has unsupported DataId override", key)
1752 failedDataId.add(key)
1753 elif key.dataId["instrument"] not in instruments:
1754 if logFailures:
1755 log.critical("Key '%s' has unknown instrument", key)
1756 failedDataId.add(key)
1758 messages = []
1760 if datastoreErrorStr:
1761 messages.append(datastoreErrorStr)
1763 for failed, msg in ((failedNames, "Keys without corresponding DatasetType or StorageClass entry: "),
1764 (failedDataId, "Keys with bad DataId entries: ")):
1765 if failed:
1766 msg += ", ".join(str(k) for k in failed)
1767 messages.append(msg)
1769 if messages:
1770 raise ValidationError(";\n".join(messages))
1772 @property
1773 def collections(self) -> CollectionSearch:
1774 """The collections to search by default, in order (`CollectionSearch`).
1776 This is an alias for ``self.registry.defaults.collections``. It cannot
1777 be set directly in isolation, but all defaults may be changed together
1778 by assigning a new `RegistryDefaults` instance to
1779 ``self.registry.defaults``.
1780 """
1781 return self.registry.defaults.collections
1783 @property
1784 def run(self) -> Optional[str]:
1785 """Name of the run this butler writes outputs to by default (`str` or
1786 `None`).
1788 This is an alias for ``self.registry.defaults.run``. It cannot be set
1789 directly in isolation, but all defaults may be changed together by
1790 assigning a new `RegistryDefaults` instance to
1791 ``self.registry.defaults``.
1792 """
1793 return self.registry.defaults.run
1795 registry: Registry
1796 """The object that manages dataset metadata and relationships (`Registry`).
1798 Most operations that don't involve reading or writing butler datasets are
1799 accessible only via `Registry` methods.
1800 """
1802 datastore: Datastore
1803 """The object that manages actual dataset storage (`Datastore`).
1805 Direct user access to the datastore should rarely be necessary; the primary
1806 exception is the case where a `Datastore` implementation provides extra
1807 functionality beyond what the base class defines.
1808 """
1810 storageClasses: StorageClassFactory
1811 """An object that maps known storage class names to objects that fully
1812 describe them (`StorageClassFactory`).
1813 """