Coverage for python/lsst/daf/butler/registry/tests/_registry.py: 4%
1343 statements
« prev ^ index » next coverage.py v6.5.0, created at 2023-01-05 10:36 +0000
« prev ^ index » next coverage.py v6.5.0, created at 2023-01-05 10:36 +0000
1# This file is part of daf_butler.
2#
3# Developed for the LSST Data Management System.
4# This product includes software developed by the LSST Project
5# (http://www.lsst.org).
6# See the COPYRIGHT file at the top-level directory of this distribution
7# for details of code ownership.
8#
9# This program is free software: you can redistribute it and/or modify
10# it under the terms of the GNU General Public License as published by
11# the Free Software Foundation, either version 3 of the License, or
12# (at your option) any later version.
13#
14# This program is distributed in the hope that it will be useful,
15# but WITHOUT ANY WARRANTY; without even the implied warranty of
16# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17# GNU General Public License for more details.
18#
19# You should have received a copy of the GNU General Public License
20# along with this program. If not, see <http://www.gnu.org/licenses/>.
21from __future__ import annotations
23__all__ = ["RegistryTests"]
25import itertools
26import logging
27import os
28import re
29import unittest
30import uuid
31from abc import ABC, abstractmethod
32from collections import defaultdict, namedtuple
33from datetime import datetime, timedelta
34from typing import TYPE_CHECKING, Iterator, Optional, Type, Union
36import astropy.time
37import sqlalchemy
39try:
40 import numpy as np
41except ImportError:
42 np = None
44import lsst.sphgeom
46from ...core import (
47 DataCoordinate,
48 DataCoordinateSet,
49 DatasetAssociation,
50 DatasetRef,
51 DatasetType,
52 DimensionGraph,
53 NamedValueSet,
54 StorageClass,
55 Timespan,
56 ddl,
57)
58from .._collection_summary import CollectionSummary
59from .._collectionType import CollectionType
60from .._config import RegistryConfig
61from .._exceptions import (
62 ArgumentError,
63 CollectionError,
64 CollectionTypeError,
65 ConflictingDefinitionError,
66 DataIdValueError,
67 DatasetTypeError,
68 InconsistentDataIdError,
69 MissingCollectionError,
70 MissingDatasetTypeError,
71 OrphanedRecordError,
72)
73from ..interfaces import ButlerAttributeExistsError, DatasetIdGenEnum
75if TYPE_CHECKING: 75 ↛ 76line 75 didn't jump to line 76, because the condition on line 75 was never true
76 from .._registry import Registry
79class RegistryTests(ABC):
80 """Generic tests for the `Registry` class that can be subclassed to
81 generate tests for different configurations.
82 """
84 collectionsManager: Optional[str] = None
85 """Name of the collections manager class, if subclass provides value for
86 this member then it overrides name specified in default configuration
87 (`str`).
88 """
90 datasetsManager: Optional[str] = None
91 """Name of the datasets manager class, if subclass provides value for
92 this member then it overrides name specified in default configuration
93 (`str`).
94 """
96 @classmethod
97 @abstractmethod
98 def getDataDir(cls) -> str:
99 """Return the root directory containing test data YAML files."""
100 raise NotImplementedError()
102 def makeRegistryConfig(self) -> RegistryConfig:
103 """Create RegistryConfig used to create a registry.
105 This method should be called by a subclass from `makeRegistry`.
106 Returned instance will be pre-configured based on the values of class
107 members, and default-configured for all other parameters. Subclasses
108 that need default configuration should just instantiate
109 `RegistryConfig` directly.
110 """
111 config = RegistryConfig()
112 if self.collectionsManager:
113 config["managers", "collections"] = self.collectionsManager
114 if self.datasetsManager:
115 config["managers", "datasets"] = self.datasetsManager
116 return config
118 @abstractmethod
119 def makeRegistry(self, share_repo_with: Optional[Registry] = None) -> Optional[Registry]:
120 """Return the Registry instance to be tested.
122 Parameters
123 ----------
124 share_repo_with : `Registry`, optional
125 If provided, the new registry should point to the same data
126 repository as this existing registry.
128 Returns
129 -------
130 registry : `Registry`
131 New `Registry` instance, or `None` *only* if `share_repo_with` is
132 not `None` and this test case does not support that argument
133 (e.g. it is impossible with in-memory SQLite DBs).
134 """
135 raise NotImplementedError()
137 def loadData(self, registry: Registry, filename: str):
138 """Load registry test data from ``getDataDir/<filename>``,
139 which should be a YAML import/export file.
140 """
141 from ...transfers import YamlRepoImportBackend
143 with open(os.path.join(self.getDataDir(), filename), "r") as stream:
144 backend = YamlRepoImportBackend(stream, registry)
145 backend.register()
146 backend.load(datastore=None)
148 def checkQueryResults(self, results, expected):
149 """Check that a query results object contains expected values.
151 Parameters
152 ----------
153 results : `DataCoordinateQueryResults` or `DatasetQueryResults`
154 A lazy-evaluation query results object.
155 expected : `list`
156 A list of `DataCoordinate` o `DatasetRef` objects that should be
157 equal to results of the query, aside from ordering.
158 """
159 self.assertCountEqual(list(results), expected)
160 self.assertEqual(results.count(), len(expected))
161 if expected:
162 self.assertTrue(results.any())
163 else:
164 self.assertFalse(results.any())
166 def testOpaque(self):
167 """Tests for `Registry.registerOpaqueTable`,
168 `Registry.insertOpaqueData`, `Registry.fetchOpaqueData`, and
169 `Registry.deleteOpaqueData`.
170 """
171 registry = self.makeRegistry()
172 table = "opaque_table_for_testing"
173 registry.registerOpaqueTable(
174 table,
175 spec=ddl.TableSpec(
176 fields=[
177 ddl.FieldSpec("id", dtype=sqlalchemy.BigInteger, primaryKey=True),
178 ddl.FieldSpec("name", dtype=sqlalchemy.String, length=16, nullable=False),
179 ddl.FieldSpec("count", dtype=sqlalchemy.SmallInteger, nullable=True),
180 ],
181 ),
182 )
183 rows = [
184 {"id": 1, "name": "one", "count": None},
185 {"id": 2, "name": "two", "count": 5},
186 {"id": 3, "name": "three", "count": 6},
187 ]
188 registry.insertOpaqueData(table, *rows)
189 self.assertCountEqual(rows, list(registry.fetchOpaqueData(table)))
190 self.assertEqual(rows[0:1], list(registry.fetchOpaqueData(table, id=1)))
191 self.assertEqual(rows[1:2], list(registry.fetchOpaqueData(table, name="two")))
192 self.assertEqual(rows[0:1], list(registry.fetchOpaqueData(table, id=(1, 3), name=("one", "two"))))
193 self.assertEqual(rows, list(registry.fetchOpaqueData(table, id=(1, 2, 3))))
194 # Test very long IN clause which exceeds sqlite limit on number of
195 # parameters. SQLite says the limit is 32k but it looks like it is
196 # much higher.
197 self.assertEqual(rows, list(registry.fetchOpaqueData(table, id=list(range(300_000)))))
198 # Two IN clauses, each longer than 1k batch size, first with
199 # duplicates, second has matching elements in different batches (after
200 # sorting).
201 self.assertEqual(
202 rows[0:2],
203 list(
204 registry.fetchOpaqueData(
205 table,
206 id=list(range(1000)) + list(range(100, 0, -1)),
207 name=["one"] + [f"q{i}" for i in range(2200)] + ["two"],
208 )
209 ),
210 )
211 self.assertEqual([], list(registry.fetchOpaqueData(table, id=1, name="two")))
212 registry.deleteOpaqueData(table, id=3)
213 self.assertCountEqual(rows[:2], list(registry.fetchOpaqueData(table)))
214 registry.deleteOpaqueData(table)
215 self.assertEqual([], list(registry.fetchOpaqueData(table)))
217 def testDatasetType(self):
218 """Tests for `Registry.registerDatasetType` and
219 `Registry.getDatasetType`.
220 """
221 registry = self.makeRegistry()
222 # Check valid insert
223 datasetTypeName = "test"
224 storageClass = StorageClass("testDatasetType")
225 registry.storageClasses.registerStorageClass(storageClass)
226 dimensions = registry.dimensions.extract(("instrument", "visit"))
227 differentDimensions = registry.dimensions.extract(("instrument", "patch"))
228 inDatasetType = DatasetType(datasetTypeName, dimensions, storageClass)
229 # Inserting for the first time should return True
230 self.assertTrue(registry.registerDatasetType(inDatasetType))
231 outDatasetType1 = registry.getDatasetType(datasetTypeName)
232 self.assertEqual(outDatasetType1, inDatasetType)
234 # Re-inserting should work
235 self.assertFalse(registry.registerDatasetType(inDatasetType))
236 # Except when they are not identical
237 with self.assertRaises(ConflictingDefinitionError):
238 nonIdenticalDatasetType = DatasetType(datasetTypeName, differentDimensions, storageClass)
239 registry.registerDatasetType(nonIdenticalDatasetType)
241 # Template can be None
242 datasetTypeName = "testNoneTemplate"
243 storageClass = StorageClass("testDatasetType2")
244 registry.storageClasses.registerStorageClass(storageClass)
245 dimensions = registry.dimensions.extract(("instrument", "visit"))
246 inDatasetType = DatasetType(datasetTypeName, dimensions, storageClass)
247 registry.registerDatasetType(inDatasetType)
248 outDatasetType2 = registry.getDatasetType(datasetTypeName)
249 self.assertEqual(outDatasetType2, inDatasetType)
251 allTypes = set(registry.queryDatasetTypes())
252 self.assertEqual(allTypes, {outDatasetType1, outDatasetType2})
254 def testDimensions(self):
255 """Tests for `Registry.insertDimensionData`,
256 `Registry.syncDimensionData`, and `Registry.expandDataId`.
257 """
258 registry = self.makeRegistry()
259 dimensionName = "instrument"
260 dimension = registry.dimensions[dimensionName]
261 dimensionValue = {
262 "name": "DummyCam",
263 "visit_max": 10,
264 "visit_system": 0,
265 "exposure_max": 10,
266 "detector_max": 2,
267 "class_name": "lsst.pipe.base.Instrument",
268 }
269 registry.insertDimensionData(dimensionName, dimensionValue)
270 # Inserting the same value twice should fail
271 with self.assertRaises(sqlalchemy.exc.IntegrityError):
272 registry.insertDimensionData(dimensionName, dimensionValue)
273 # expandDataId should retrieve the record we just inserted
274 self.assertEqual(
275 registry.expandDataId(instrument="DummyCam", graph=dimension.graph)
276 .records[dimensionName]
277 .toDict(),
278 dimensionValue,
279 )
280 # expandDataId should raise if there is no record with the given ID.
281 with self.assertRaises(DataIdValueError):
282 registry.expandDataId({"instrument": "Unknown"}, graph=dimension.graph)
283 # band doesn't have a table; insert should fail.
284 with self.assertRaises(TypeError):
285 registry.insertDimensionData("band", {"band": "i"})
286 dimensionName2 = "physical_filter"
287 dimension2 = registry.dimensions[dimensionName2]
288 dimensionValue2 = {"name": "DummyCam_i", "band": "i"}
289 # Missing required dependency ("instrument") should fail
290 with self.assertRaises(KeyError):
291 registry.insertDimensionData(dimensionName2, dimensionValue2)
292 # Adding required dependency should fix the failure
293 dimensionValue2["instrument"] = "DummyCam"
294 registry.insertDimensionData(dimensionName2, dimensionValue2)
295 # expandDataId should retrieve the record we just inserted.
296 self.assertEqual(
297 registry.expandDataId(instrument="DummyCam", physical_filter="DummyCam_i", graph=dimension2.graph)
298 .records[dimensionName2]
299 .toDict(),
300 dimensionValue2,
301 )
302 # Use syncDimensionData to insert a new record successfully.
303 dimensionName3 = "detector"
304 dimensionValue3 = {
305 "instrument": "DummyCam",
306 "id": 1,
307 "full_name": "one",
308 "name_in_raft": "zero",
309 "purpose": "SCIENCE",
310 }
311 self.assertTrue(registry.syncDimensionData(dimensionName3, dimensionValue3))
312 # Sync that again. Note that one field ("raft") is NULL, and that
313 # should be okay.
314 self.assertFalse(registry.syncDimensionData(dimensionName3, dimensionValue3))
315 # Now try that sync with the same primary key but a different value.
316 # This should fail.
317 with self.assertRaises(ConflictingDefinitionError):
318 registry.syncDimensionData(
319 dimensionName3,
320 {
321 "instrument": "DummyCam",
322 "id": 1,
323 "full_name": "one",
324 "name_in_raft": "four",
325 "purpose": "SCIENCE",
326 },
327 )
329 @unittest.skipIf(np is None, "numpy not available.")
330 def testNumpyDataId(self):
331 """Test that we can use a numpy int in a dataId."""
332 registry = self.makeRegistry()
333 dimensionEntries = [
334 ("instrument", {"instrument": "DummyCam"}),
335 ("physical_filter", {"instrument": "DummyCam", "name": "d-r", "band": "R"}),
336 # Using an np.int64 here fails unless Records.fromDict is also
337 # patched to look for numbers.Integral
338 ("visit", {"instrument": "DummyCam", "id": 42, "name": "fortytwo", "physical_filter": "d-r"}),
339 ]
340 for args in dimensionEntries:
341 registry.insertDimensionData(*args)
343 # Try a normal integer and something that looks like an int but
344 # is not.
345 for visit_id in (42, np.int64(42)):
346 with self.subTest(visit_id=visit_id, id_type=type(visit_id).__name__):
347 expanded = registry.expandDataId({"instrument": "DummyCam", "visit": visit_id})
348 self.assertEqual(expanded["visit"], int(visit_id))
349 self.assertIsInstance(expanded["visit"], int)
351 def testDataIdRelationships(self):
352 """Test that `Registry.expandDataId` raises an exception when the given
353 keys are inconsistent.
354 """
355 registry = self.makeRegistry()
356 self.loadData(registry, "base.yaml")
357 # Insert a few more dimension records for the next test.
358 registry.insertDimensionData(
359 "exposure",
360 {"instrument": "Cam1", "id": 1, "obs_id": "one", "physical_filter": "Cam1-G"},
361 )
362 registry.insertDimensionData(
363 "exposure",
364 {"instrument": "Cam1", "id": 2, "obs_id": "two", "physical_filter": "Cam1-G"},
365 )
366 registry.insertDimensionData(
367 "visit_system",
368 {"instrument": "Cam1", "id": 0, "name": "one-to-one"},
369 )
370 registry.insertDimensionData(
371 "visit",
372 {"instrument": "Cam1", "id": 1, "name": "one", "physical_filter": "Cam1-G", "visit_system": 0},
373 )
374 registry.insertDimensionData(
375 "visit_definition",
376 {"instrument": "Cam1", "visit": 1, "exposure": 1, "visit_system": 0},
377 )
378 with self.assertRaises(InconsistentDataIdError):
379 registry.expandDataId(
380 {"instrument": "Cam1", "visit": 1, "exposure": 2},
381 )
383 def testDataset(self):
384 """Basic tests for `Registry.insertDatasets`, `Registry.getDataset`,
385 and `Registry.removeDatasets`.
386 """
387 registry = self.makeRegistry()
388 self.loadData(registry, "base.yaml")
389 run = "tésτ"
390 registry.registerRun(run)
391 datasetType = registry.getDatasetType("bias")
392 dataId = {"instrument": "Cam1", "detector": 2}
393 (ref,) = registry.insertDatasets(datasetType, dataIds=[dataId], run=run)
394 outRef = registry.getDataset(ref.id)
395 self.assertIsNotNone(ref.id)
396 self.assertEqual(ref, outRef)
397 with self.assertRaises(ConflictingDefinitionError):
398 registry.insertDatasets(datasetType, dataIds=[dataId], run=run)
399 registry.removeDatasets([ref])
400 self.assertIsNone(registry.findDataset(datasetType, dataId, collections=[run]))
402 def testFindDataset(self):
403 """Tests for `Registry.findDataset`."""
404 registry = self.makeRegistry()
405 self.loadData(registry, "base.yaml")
406 run = "tésτ"
407 datasetType = registry.getDatasetType("bias")
408 dataId = {"instrument": "Cam1", "detector": 4}
409 registry.registerRun(run)
410 (inputRef,) = registry.insertDatasets(datasetType, dataIds=[dataId], run=run)
411 outputRef = registry.findDataset(datasetType, dataId, collections=[run])
412 self.assertEqual(outputRef, inputRef)
413 # Check that retrieval with invalid dataId raises
414 with self.assertRaises(LookupError):
415 dataId = {"instrument": "Cam1"} # no detector
416 registry.findDataset(datasetType, dataId, collections=run)
417 # Check that different dataIds match to different datasets
418 dataId1 = {"instrument": "Cam1", "detector": 1}
419 (inputRef1,) = registry.insertDatasets(datasetType, dataIds=[dataId1], run=run)
420 dataId2 = {"instrument": "Cam1", "detector": 2}
421 (inputRef2,) = registry.insertDatasets(datasetType, dataIds=[dataId2], run=run)
422 self.assertEqual(registry.findDataset(datasetType, dataId1, collections=run), inputRef1)
423 self.assertEqual(registry.findDataset(datasetType, dataId2, collections=run), inputRef2)
424 self.assertNotEqual(registry.findDataset(datasetType, dataId1, collections=run), inputRef2)
425 self.assertNotEqual(registry.findDataset(datasetType, dataId2, collections=run), inputRef1)
426 # Check that requesting a non-existing dataId returns None
427 nonExistingDataId = {"instrument": "Cam1", "detector": 3}
428 self.assertIsNone(registry.findDataset(datasetType, nonExistingDataId, collections=run))
430 def testRemoveDatasetTypeSuccess(self):
431 """Test that Registry.removeDatasetType works when there are no
432 datasets of that type present.
433 """
434 registry = self.makeRegistry()
435 self.loadData(registry, "base.yaml")
436 registry.removeDatasetType("flat")
437 with self.assertRaises(MissingDatasetTypeError):
438 registry.getDatasetType("flat")
440 def testRemoveDatasetTypeFailure(self):
441 """Test that Registry.removeDatasetType raises when there are datasets
442 of that type present or if the dataset type is for a component.
443 """
444 registry = self.makeRegistry()
445 self.loadData(registry, "base.yaml")
446 self.loadData(registry, "datasets.yaml")
447 with self.assertRaises(OrphanedRecordError):
448 registry.removeDatasetType("flat")
449 with self.assertRaises(ValueError):
450 registry.removeDatasetType(DatasetType.nameWithComponent("flat", "image"))
452 def testImportDatasetsUUID(self):
453 """Test for `Registry._importDatasets` with UUID dataset ID."""
454 if not self.datasetsManager.endswith(".ByDimensionsDatasetRecordStorageManagerUUID"):
455 self.skipTest(f"Unexpected dataset manager {self.datasetsManager}")
457 registry = self.makeRegistry()
458 self.loadData(registry, "base.yaml")
459 for run in range(6):
460 registry.registerRun(f"run{run}")
461 datasetTypeBias = registry.getDatasetType("bias")
462 datasetTypeFlat = registry.getDatasetType("flat")
463 dataIdBias1 = {"instrument": "Cam1", "detector": 1}
464 dataIdBias2 = {"instrument": "Cam1", "detector": 2}
465 dataIdFlat1 = {"instrument": "Cam1", "detector": 1, "physical_filter": "Cam1-G", "band": "g"}
467 dataset_id = uuid.uuid4()
468 ref = DatasetRef(datasetTypeBias, dataIdBias1, id=dataset_id, run="run0")
469 (ref1,) = registry._importDatasets([ref])
470 # UUID is used without change
471 self.assertEqual(ref.id, ref1.id)
473 # All different failure modes
474 refs = (
475 # Importing same DatasetRef with different dataset ID is an error
476 DatasetRef(datasetTypeBias, dataIdBias1, id=uuid.uuid4(), run="run0"),
477 # Same DatasetId but different DataId
478 DatasetRef(datasetTypeBias, dataIdBias2, id=ref1.id, run="run0"),
479 DatasetRef(datasetTypeFlat, dataIdFlat1, id=ref1.id, run="run0"),
480 # Same DatasetRef and DatasetId but different run
481 DatasetRef(datasetTypeBias, dataIdBias1, id=ref1.id, run="run1"),
482 )
483 for ref in refs:
484 with self.assertRaises(ConflictingDefinitionError):
485 registry._importDatasets([ref])
487 # Test for non-unique IDs, they can be re-imported multiple times.
488 for run, idGenMode in ((2, DatasetIdGenEnum.DATAID_TYPE), (4, DatasetIdGenEnum.DATAID_TYPE_RUN)):
489 with self.subTest(idGenMode=idGenMode):
491 # Use integer dataset ID to force UUID calculation in _import
492 ref = DatasetRef(datasetTypeBias, dataIdBias1, id=0, run=f"run{run}")
493 (ref1,) = registry._importDatasets([ref], idGenerationMode=idGenMode)
494 self.assertIsInstance(ref1.id, uuid.UUID)
495 self.assertEqual(ref1.id.version, 5)
497 # Importing it again is OK
498 (ref2,) = registry._importDatasets([ref1])
499 self.assertEqual(ref2.id, ref1.id)
501 # Cannot import to different run with the same ID
502 ref = DatasetRef(datasetTypeBias, dataIdBias1, id=ref1.id, run=f"run{run+1}")
503 with self.assertRaises(ConflictingDefinitionError):
504 registry._importDatasets([ref])
506 ref = DatasetRef(datasetTypeBias, dataIdBias1, id=0, run=f"run{run+1}")
507 if idGenMode is DatasetIdGenEnum.DATAID_TYPE:
508 # Cannot import same DATAID_TYPE ref into a new run
509 with self.assertRaises(ConflictingDefinitionError):
510 (ref2,) = registry._importDatasets([ref], idGenerationMode=idGenMode)
511 else:
512 # DATAID_TYPE_RUN ref can be imported into a new run
513 (ref2,) = registry._importDatasets([ref], idGenerationMode=idGenMode)
515 def testDatasetTypeComponentQueries(self):
516 """Test component options when querying for dataset types.
518 All of the behavior here is deprecated, so many of these tests are
519 currently wrapped in a context to check that we get a warning whenever
520 a component dataset is actually returned.
521 """
522 registry = self.makeRegistry()
523 self.loadData(registry, "base.yaml")
524 self.loadData(registry, "datasets.yaml")
525 # Test querying for dataset types with different inputs.
526 # First query for all dataset types; components should only be included
527 # when components=True.
528 self.assertEqual({"bias", "flat"}, NamedValueSet(registry.queryDatasetTypes()).names)
529 self.assertEqual({"bias", "flat"}, NamedValueSet(registry.queryDatasetTypes(components=False)).names)
530 with self.assertWarns(FutureWarning):
531 self.assertLess(
532 {"bias", "flat", "bias.wcs", "flat.photoCalib"},
533 NamedValueSet(registry.queryDatasetTypes(components=True)).names,
534 )
535 # Use a pattern that can match either parent or components. Again,
536 # components are only returned if components=True.
537 self.assertEqual({"bias"}, NamedValueSet(registry.queryDatasetTypes(re.compile("^bias.*"))).names)
538 self.assertEqual(
539 {"bias"}, NamedValueSet(registry.queryDatasetTypes(re.compile("^bias.*"), components=False)).names
540 )
541 with self.assertWarns(FutureWarning):
542 self.assertLess(
543 {"bias", "bias.wcs"},
544 NamedValueSet(registry.queryDatasetTypes(re.compile("^bias.*"), components=True)).names,
545 )
546 # This pattern matches only a component. In this case we also return
547 # that component dataset type if components=None.
548 with self.assertWarns(FutureWarning):
549 self.assertEqual(
550 {"bias.wcs"}, NamedValueSet(registry.queryDatasetTypes(re.compile(r"^bias\.wcs"))).names
551 )
552 self.assertEqual(
553 set(),
554 NamedValueSet(registry.queryDatasetTypes(re.compile(r"^bias\.wcs"), components=False)).names,
555 )
556 with self.assertWarns(FutureWarning):
557 self.assertEqual(
558 {"bias.wcs"},
559 NamedValueSet(registry.queryDatasetTypes(re.compile(r"^bias\.wcs"), components=True)).names,
560 )
561 # Add a dataset type using a StorageClass that we'll then remove; check
562 # that this does not affect our ability to query for dataset types
563 # (though it will warn).
564 tempStorageClass = StorageClass(
565 name="TempStorageClass",
566 components={
567 "data1": registry.storageClasses.getStorageClass("StructuredDataDict"),
568 "data2": registry.storageClasses.getStorageClass("StructuredDataDict"),
569 },
570 )
571 registry.storageClasses.registerStorageClass(tempStorageClass)
572 datasetType = DatasetType(
573 "temporary",
574 dimensions=["instrument"],
575 storageClass=tempStorageClass,
576 universe=registry.dimensions,
577 )
578 registry.registerDatasetType(datasetType)
579 registry.storageClasses._unregisterStorageClass(tempStorageClass.name)
580 datasetType._storageClass = None
581 del tempStorageClass
582 # Querying for all dataset types, including components, should include
583 # at least all non-component dataset types (and I don't want to
584 # enumerate all of the Exposure components for bias and flat here).
585 with self.assertWarns(FutureWarning):
586 with self.assertLogs("lsst.daf.butler.registry", logging.WARN) as cm:
587 everything = NamedValueSet(registry.queryDatasetTypes(components=True))
588 self.assertIn("TempStorageClass", cm.output[0])
589 self.assertLess({"bias", "flat", "temporary"}, everything.names)
590 # It should not include "temporary.columns", because we tried to remove
591 # the storage class that would tell it about that. So if the next line
592 # fails (i.e. "temporary.columns" _is_ in everything.names), it means
593 # this part of the test isn't doing anything, because the _unregister
594 # call about isn't simulating the real-life case we want it to
595 # simulate, in which different versions of daf_butler in entirely
596 # different Python processes interact with the same repo.
597 self.assertNotIn("temporary.data", everything.names)
598 # Query for dataset types that start with "temp". This should again
599 # not include the component, and also not fail.
600 with self.assertLogs("lsst.daf.butler.registry", logging.WARN) as cm:
601 startsWithTemp = NamedValueSet(registry.queryDatasetTypes(re.compile("temp.*"), components=True))
602 self.assertIn("TempStorageClass", cm.output[0])
603 self.assertEqual({"temporary"}, startsWithTemp.names)
604 # Querying with no components should not warn at all.
605 with self.assertLogs("lsst.daf.butler.registries", logging.WARN) as cm:
606 startsWithTemp = NamedValueSet(registry.queryDatasetTypes(re.compile("temp.*"), components=False))
607 # Must issue a warning of our own to be captured.
608 logging.getLogger("lsst.daf.butler.registries").warning("test message")
609 self.assertEqual(len(cm.output), 1)
610 self.assertIn("test message", cm.output[0])
612 def testComponentLookups(self):
613 """Test searching for component datasets via their parents.
615 All of the behavior here is deprecated, so many of these tests are
616 currently wrapped in a context to check that we get a warning whenever
617 a component dataset is actually returned.
618 """
619 registry = self.makeRegistry()
620 self.loadData(registry, "base.yaml")
621 self.loadData(registry, "datasets.yaml")
622 # Test getting the child dataset type (which does still exist in the
623 # Registry), and check for consistency with
624 # DatasetRef.makeComponentRef.
625 collection = "imported_g"
626 parentType = registry.getDatasetType("bias")
627 childType = registry.getDatasetType("bias.wcs")
628 parentRefResolved = registry.findDataset(
629 parentType, collections=collection, instrument="Cam1", detector=1
630 )
631 self.assertIsInstance(parentRefResolved, DatasetRef)
632 self.assertEqual(childType, parentRefResolved.makeComponentRef("wcs").datasetType)
633 # Search for a single dataset with findDataset.
634 childRef1 = registry.findDataset("bias.wcs", collections=collection, dataId=parentRefResolved.dataId)
635 self.assertEqual(childRef1, parentRefResolved.makeComponentRef("wcs"))
636 # Search for detector data IDs constrained by component dataset
637 # existence with queryDataIds.
638 with self.assertWarns(FutureWarning):
639 dataIds = registry.queryDataIds(
640 ["detector"],
641 datasets=["bias.wcs"],
642 collections=collection,
643 ).toSet()
644 self.assertEqual(
645 dataIds,
646 DataCoordinateSet(
647 {
648 DataCoordinate.standardize(instrument="Cam1", detector=d, graph=parentType.dimensions)
649 for d in (1, 2, 3)
650 },
651 parentType.dimensions,
652 ),
653 )
654 # Search for multiple datasets of a single type with queryDatasets.
655 with self.assertWarns(FutureWarning):
656 childRefs2 = set(
657 registry.queryDatasets(
658 "bias.wcs",
659 collections=collection,
660 )
661 )
662 self.assertEqual(
663 {ref.unresolved() for ref in childRefs2}, {DatasetRef(childType, dataId) for dataId in dataIds}
664 )
666 def testCollections(self):
667 """Tests for registry methods that manage collections."""
668 registry = self.makeRegistry()
669 other_registry = self.makeRegistry(share_repo_with=registry)
670 self.loadData(registry, "base.yaml")
671 self.loadData(registry, "datasets.yaml")
672 run1 = "imported_g"
673 run2 = "imported_r"
674 # Test setting a collection docstring after it has been created.
675 registry.setCollectionDocumentation(run1, "doc for run1")
676 self.assertEqual(registry.getCollectionDocumentation(run1), "doc for run1")
677 registry.setCollectionDocumentation(run1, None)
678 self.assertIsNone(registry.getCollectionDocumentation(run1))
679 datasetType = "bias"
680 # Find some datasets via their run's collection.
681 dataId1 = {"instrument": "Cam1", "detector": 1}
682 ref1 = registry.findDataset(datasetType, dataId1, collections=run1)
683 self.assertIsNotNone(ref1)
684 dataId2 = {"instrument": "Cam1", "detector": 2}
685 ref2 = registry.findDataset(datasetType, dataId2, collections=run1)
686 self.assertIsNotNone(ref2)
687 # Associate those into a new collection, then look for them there.
688 tag1 = "tag1"
689 registry.registerCollection(tag1, type=CollectionType.TAGGED, doc="doc for tag1")
690 # Check that we can query for old and new collections by type.
691 self.assertEqual(set(registry.queryCollections(collectionTypes=CollectionType.RUN)), {run1, run2})
692 self.assertEqual(
693 set(registry.queryCollections(collectionTypes={CollectionType.TAGGED, CollectionType.RUN})),
694 {tag1, run1, run2},
695 )
696 self.assertEqual(registry.getCollectionDocumentation(tag1), "doc for tag1")
697 registry.associate(tag1, [ref1, ref2])
698 self.assertEqual(registry.findDataset(datasetType, dataId1, collections=tag1), ref1)
699 self.assertEqual(registry.findDataset(datasetType, dataId2, collections=tag1), ref2)
700 # Disassociate one and verify that we can't it there anymore...
701 registry.disassociate(tag1, [ref1])
702 self.assertIsNone(registry.findDataset(datasetType, dataId1, collections=tag1))
703 # ...but we can still find ref2 in tag1, and ref1 in the run.
704 self.assertEqual(registry.findDataset(datasetType, dataId1, collections=run1), ref1)
705 self.assertEqual(registry.findDataset(datasetType, dataId2, collections=tag1), ref2)
706 collections = set(registry.queryCollections())
707 self.assertEqual(collections, {run1, run2, tag1})
708 # Associate both refs into tag1 again; ref2 is already there, but that
709 # should be a harmless no-op.
710 registry.associate(tag1, [ref1, ref2])
711 self.assertEqual(registry.findDataset(datasetType, dataId1, collections=tag1), ref1)
712 self.assertEqual(registry.findDataset(datasetType, dataId2, collections=tag1), ref2)
713 # Get a different dataset (from a different run) that has the same
714 # dataset type and data ID as ref2.
715 ref2b = registry.findDataset(datasetType, dataId2, collections=run2)
716 self.assertNotEqual(ref2, ref2b)
717 # Attempting to associate that into tag1 should be an error.
718 with self.assertRaises(ConflictingDefinitionError):
719 registry.associate(tag1, [ref2b])
720 # That error shouldn't have messed up what we had before.
721 self.assertEqual(registry.findDataset(datasetType, dataId1, collections=tag1), ref1)
722 self.assertEqual(registry.findDataset(datasetType, dataId2, collections=tag1), ref2)
723 # Attempt to associate the conflicting dataset again, this time with
724 # a dataset that isn't in the collection and won't cause a conflict.
725 # Should also fail without modifying anything.
726 dataId3 = {"instrument": "Cam1", "detector": 3}
727 ref3 = registry.findDataset(datasetType, dataId3, collections=run1)
728 with self.assertRaises(ConflictingDefinitionError):
729 registry.associate(tag1, [ref3, ref2b])
730 self.assertEqual(registry.findDataset(datasetType, dataId1, collections=tag1), ref1)
731 self.assertEqual(registry.findDataset(datasetType, dataId2, collections=tag1), ref2)
732 self.assertIsNone(registry.findDataset(datasetType, dataId3, collections=tag1))
733 # Register a chained collection that searches [tag1, run2]
734 chain1 = "chain1"
735 registry.registerCollection(chain1, type=CollectionType.CHAINED)
736 self.assertIs(registry.getCollectionType(chain1), CollectionType.CHAINED)
737 # Chained collection exists, but has no collections in it.
738 self.assertFalse(registry.getCollectionChain(chain1))
739 # If we query for all collections, we should get the chained collection
740 # only if we don't ask to flatten it (i.e. yield only its children).
741 self.assertEqual(set(registry.queryCollections(flattenChains=False)), {tag1, run1, run2, chain1})
742 self.assertEqual(set(registry.queryCollections(flattenChains=True)), {tag1, run1, run2})
743 # Attempt to set its child collections to something circular; that
744 # should fail.
745 with self.assertRaises(ValueError):
746 registry.setCollectionChain(chain1, [tag1, chain1])
747 # Add the child collections.
748 registry.setCollectionChain(chain1, [tag1, run2])
749 self.assertEqual(list(registry.getCollectionChain(chain1)), [tag1, run2])
750 self.assertEqual(registry.getCollectionParentChains(tag1), {chain1})
751 self.assertEqual(registry.getCollectionParentChains(run2), {chain1})
752 # Refresh the other registry that points to the same repo, and make
753 # sure it can see the things we've done (note that this does require
754 # an explicit refresh(); that's the documented behavior, because
755 # caching is ~impossible otherwise).
756 if other_registry is not None:
757 other_registry.refresh()
758 self.assertEqual(list(other_registry.getCollectionChain(chain1)), [tag1, run2])
759 self.assertEqual(other_registry.getCollectionParentChains(tag1), {chain1})
760 self.assertEqual(other_registry.getCollectionParentChains(run2), {chain1})
761 # Searching for dataId1 or dataId2 in the chain should return ref1 and
762 # ref2, because both are in tag1.
763 self.assertEqual(registry.findDataset(datasetType, dataId1, collections=chain1), ref1)
764 self.assertEqual(registry.findDataset(datasetType, dataId2, collections=chain1), ref2)
765 # Now disassociate ref2 from tag1. The search (for bias) with
766 # dataId2 in chain1 should then:
767 # 1. not find it in tag1
768 # 2. find a different dataset in run2
769 registry.disassociate(tag1, [ref2])
770 ref2b = registry.findDataset(datasetType, dataId2, collections=chain1)
771 self.assertNotEqual(ref2b, ref2)
772 self.assertEqual(ref2b, registry.findDataset(datasetType, dataId2, collections=run2))
773 # Define a new chain so we can test recursive chains.
774 chain2 = "chain2"
775 registry.registerCollection(chain2, type=CollectionType.CHAINED)
776 registry.setCollectionChain(chain2, [run2, chain1])
777 self.assertEqual(registry.getCollectionParentChains(chain1), {chain2})
778 self.assertEqual(registry.getCollectionParentChains(run2), {chain1, chain2})
779 # Query for collections matching a regex.
780 self.assertCountEqual(
781 list(registry.queryCollections(re.compile("imported_."), flattenChains=False)),
782 ["imported_r", "imported_g"],
783 )
784 # Query for collections matching a regex or an explicit str.
785 self.assertCountEqual(
786 list(registry.queryCollections([re.compile("imported_."), "chain1"], flattenChains=False)),
787 ["imported_r", "imported_g", "chain1"],
788 )
789 # Search for bias with dataId1 should find it via tag1 in chain2,
790 # recursing, because is not in run1.
791 self.assertIsNone(registry.findDataset(datasetType, dataId1, collections=run2))
792 self.assertEqual(registry.findDataset(datasetType, dataId1, collections=chain2), ref1)
793 # Search for bias with dataId2 should find it in run2 (ref2b).
794 self.assertEqual(registry.findDataset(datasetType, dataId2, collections=chain2), ref2b)
795 # Search for a flat that is in run2. That should not be found
796 # at the front of chain2, because of the restriction to bias
797 # on run2 there, but it should be found in at the end of chain1.
798 dataId4 = {"instrument": "Cam1", "detector": 3, "physical_filter": "Cam1-R2"}
799 ref4 = registry.findDataset("flat", dataId4, collections=run2)
800 self.assertIsNotNone(ref4)
801 self.assertEqual(ref4, registry.findDataset("flat", dataId4, collections=chain2))
802 # Deleting a collection that's part of a CHAINED collection is not
803 # allowed, and is exception-safe.
804 with self.assertRaises(Exception):
805 registry.removeCollection(run2)
806 self.assertEqual(registry.getCollectionType(run2), CollectionType.RUN)
807 with self.assertRaises(Exception):
808 registry.removeCollection(chain1)
809 self.assertEqual(registry.getCollectionType(chain1), CollectionType.CHAINED)
810 # Actually remove chain2, test that it's gone by asking for its type.
811 registry.removeCollection(chain2)
812 with self.assertRaises(MissingCollectionError):
813 registry.getCollectionType(chain2)
814 # Actually remove run2 and chain1, which should work now.
815 registry.removeCollection(chain1)
816 registry.removeCollection(run2)
817 with self.assertRaises(MissingCollectionError):
818 registry.getCollectionType(run2)
819 with self.assertRaises(MissingCollectionError):
820 registry.getCollectionType(chain1)
821 # Remove tag1 as well, just to test that we can remove TAGGED
822 # collections.
823 registry.removeCollection(tag1)
824 with self.assertRaises(MissingCollectionError):
825 registry.getCollectionType(tag1)
827 def testCollectionChainFlatten(self):
828 """Test that Registry.setCollectionChain obeys its 'flatten' option."""
829 registry = self.makeRegistry()
830 registry.registerCollection("inner", CollectionType.CHAINED)
831 registry.registerCollection("innermost", CollectionType.RUN)
832 registry.setCollectionChain("inner", ["innermost"])
833 registry.registerCollection("outer", CollectionType.CHAINED)
834 registry.setCollectionChain("outer", ["inner"], flatten=False)
835 self.assertEqual(list(registry.getCollectionChain("outer")), ["inner"])
836 registry.setCollectionChain("outer", ["inner"], flatten=True)
837 self.assertEqual(list(registry.getCollectionChain("outer")), ["innermost"])
839 def testBasicTransaction(self):
840 """Test that all operations within a single transaction block are
841 rolled back if an exception propagates out of the block.
842 """
843 registry = self.makeRegistry()
844 storageClass = StorageClass("testDatasetType")
845 registry.storageClasses.registerStorageClass(storageClass)
846 with registry.transaction():
847 registry.insertDimensionData("instrument", {"name": "Cam1", "class_name": "A"})
848 with self.assertRaises(ValueError):
849 with registry.transaction():
850 registry.insertDimensionData("instrument", {"name": "Cam2"})
851 raise ValueError("Oops, something went wrong")
852 # Cam1 should exist
853 self.assertEqual(registry.expandDataId(instrument="Cam1").records["instrument"].class_name, "A")
854 # But Cam2 and Cam3 should both not exist
855 with self.assertRaises(DataIdValueError):
856 registry.expandDataId(instrument="Cam2")
857 with self.assertRaises(DataIdValueError):
858 registry.expandDataId(instrument="Cam3")
860 def testNestedTransaction(self):
861 """Test that operations within a transaction block are not rolled back
862 if an exception propagates out of an inner transaction block and is
863 then caught.
864 """
865 registry = self.makeRegistry()
866 dimension = registry.dimensions["instrument"]
867 dataId1 = {"instrument": "DummyCam"}
868 dataId2 = {"instrument": "DummyCam2"}
869 checkpointReached = False
870 with registry.transaction():
871 # This should be added and (ultimately) committed.
872 registry.insertDimensionData(dimension, dataId1)
873 with self.assertRaises(sqlalchemy.exc.IntegrityError):
874 with registry.transaction(savepoint=True):
875 # This does not conflict, and should succeed (but not
876 # be committed).
877 registry.insertDimensionData(dimension, dataId2)
878 checkpointReached = True
879 # This should conflict and raise, triggerring a rollback
880 # of the previous insertion within the same transaction
881 # context, but not the original insertion in the outer
882 # block.
883 registry.insertDimensionData(dimension, dataId1)
884 self.assertTrue(checkpointReached)
885 self.assertIsNotNone(registry.expandDataId(dataId1, graph=dimension.graph))
886 with self.assertRaises(DataIdValueError):
887 registry.expandDataId(dataId2, graph=dimension.graph)
889 def testInstrumentDimensions(self):
890 """Test queries involving only instrument dimensions, with no joins to
891 skymap."""
892 registry = self.makeRegistry()
894 # need a bunch of dimensions and datasets for test
895 registry.insertDimensionData(
896 "instrument", dict(name="DummyCam", visit_max=25, exposure_max=300, detector_max=6)
897 )
898 registry.insertDimensionData(
899 "physical_filter",
900 dict(instrument="DummyCam", name="dummy_r", band="r"),
901 dict(instrument="DummyCam", name="dummy_i", band="i"),
902 )
903 registry.insertDimensionData(
904 "detector", *[dict(instrument="DummyCam", id=i, full_name=str(i)) for i in range(1, 6)]
905 )
906 registry.insertDimensionData(
907 "visit_system",
908 dict(instrument="DummyCam", id=1, name="default"),
909 )
910 registry.insertDimensionData(
911 "visit",
912 dict(instrument="DummyCam", id=10, name="ten", physical_filter="dummy_i", visit_system=1),
913 dict(instrument="DummyCam", id=11, name="eleven", physical_filter="dummy_r", visit_system=1),
914 dict(instrument="DummyCam", id=20, name="twelve", physical_filter="dummy_r", visit_system=1),
915 )
916 registry.insertDimensionData(
917 "exposure",
918 dict(instrument="DummyCam", id=100, obs_id="100", physical_filter="dummy_i"),
919 dict(instrument="DummyCam", id=101, obs_id="101", physical_filter="dummy_i"),
920 dict(instrument="DummyCam", id=110, obs_id="110", physical_filter="dummy_r"),
921 dict(instrument="DummyCam", id=111, obs_id="111", physical_filter="dummy_r"),
922 dict(instrument="DummyCam", id=200, obs_id="200", physical_filter="dummy_r"),
923 dict(instrument="DummyCam", id=201, obs_id="201", physical_filter="dummy_r"),
924 )
925 registry.insertDimensionData(
926 "visit_definition",
927 dict(instrument="DummyCam", exposure=100, visit_system=1, visit=10),
928 dict(instrument="DummyCam", exposure=101, visit_system=1, visit=10),
929 dict(instrument="DummyCam", exposure=110, visit_system=1, visit=11),
930 dict(instrument="DummyCam", exposure=111, visit_system=1, visit=11),
931 dict(instrument="DummyCam", exposure=200, visit_system=1, visit=20),
932 dict(instrument="DummyCam", exposure=201, visit_system=1, visit=20),
933 )
934 # dataset types
935 run1 = "test1_r"
936 run2 = "test2_r"
937 tagged2 = "test2_t"
938 registry.registerRun(run1)
939 registry.registerRun(run2)
940 registry.registerCollection(tagged2)
941 storageClass = StorageClass("testDataset")
942 registry.storageClasses.registerStorageClass(storageClass)
943 rawType = DatasetType(
944 name="RAW",
945 dimensions=registry.dimensions.extract(("instrument", "exposure", "detector")),
946 storageClass=storageClass,
947 )
948 registry.registerDatasetType(rawType)
949 calexpType = DatasetType(
950 name="CALEXP",
951 dimensions=registry.dimensions.extract(("instrument", "visit", "detector")),
952 storageClass=storageClass,
953 )
954 registry.registerDatasetType(calexpType)
956 # add pre-existing datasets
957 for exposure in (100, 101, 110, 111):
958 for detector in (1, 2, 3):
959 # note that only 3 of 5 detectors have datasets
960 dataId = dict(instrument="DummyCam", exposure=exposure, detector=detector)
961 (ref,) = registry.insertDatasets(rawType, dataIds=[dataId], run=run1)
962 # exposures 100 and 101 appear in both run1 and tagged2.
963 # 100 has different datasets in the different collections
964 # 101 has the same dataset in both collections.
965 if exposure == 100:
966 (ref,) = registry.insertDatasets(rawType, dataIds=[dataId], run=run2)
967 if exposure in (100, 101):
968 registry.associate(tagged2, [ref])
969 # Add pre-existing datasets to tagged2.
970 for exposure in (200, 201):
971 for detector in (3, 4, 5):
972 # note that only 3 of 5 detectors have datasets
973 dataId = dict(instrument="DummyCam", exposure=exposure, detector=detector)
974 (ref,) = registry.insertDatasets(rawType, dataIds=[dataId], run=run2)
975 registry.associate(tagged2, [ref])
977 dimensions = DimensionGraph(
978 registry.dimensions, dimensions=(rawType.dimensions.required | calexpType.dimensions.required)
979 )
980 # Test that single dim string works as well as list of str
981 rows = registry.queryDataIds("visit", datasets=rawType, collections=run1).expanded().toSet()
982 rowsI = registry.queryDataIds(["visit"], datasets=rawType, collections=run1).expanded().toSet()
983 self.assertEqual(rows, rowsI)
984 # with empty expression
985 rows = registry.queryDataIds(dimensions, datasets=rawType, collections=run1).expanded().toSet()
986 self.assertEqual(len(rows), 4 * 3) # 4 exposures times 3 detectors
987 for dataId in rows:
988 self.assertCountEqual(dataId.keys(), ("instrument", "detector", "exposure", "visit"))
989 packer1 = registry.dimensions.makePacker("visit_detector", dataId)
990 packer2 = registry.dimensions.makePacker("exposure_detector", dataId)
991 self.assertEqual(
992 packer1.unpack(packer1.pack(dataId)),
993 DataCoordinate.standardize(dataId, graph=packer1.dimensions),
994 )
995 self.assertEqual(
996 packer2.unpack(packer2.pack(dataId)),
997 DataCoordinate.standardize(dataId, graph=packer2.dimensions),
998 )
999 self.assertNotEqual(packer1.pack(dataId), packer2.pack(dataId))
1000 self.assertCountEqual(set(dataId["exposure"] for dataId in rows), (100, 101, 110, 111))
1001 self.assertCountEqual(set(dataId["visit"] for dataId in rows), (10, 11))
1002 self.assertCountEqual(set(dataId["detector"] for dataId in rows), (1, 2, 3))
1004 # second collection
1005 rows = registry.queryDataIds(dimensions, datasets=rawType, collections=tagged2).toSet()
1006 self.assertEqual(len(rows), 4 * 3) # 4 exposures times 3 detectors
1007 for dataId in rows:
1008 self.assertCountEqual(dataId.keys(), ("instrument", "detector", "exposure", "visit"))
1009 self.assertCountEqual(set(dataId["exposure"] for dataId in rows), (100, 101, 200, 201))
1010 self.assertCountEqual(set(dataId["visit"] for dataId in rows), (10, 20))
1011 self.assertCountEqual(set(dataId["detector"] for dataId in rows), (1, 2, 3, 4, 5))
1013 # with two input datasets
1014 rows = registry.queryDataIds(dimensions, datasets=rawType, collections=[run1, tagged2]).toSet()
1015 self.assertEqual(len(set(rows)), 6 * 3) # 6 exposures times 3 detectors; set needed to de-dupe
1016 for dataId in rows:
1017 self.assertCountEqual(dataId.keys(), ("instrument", "detector", "exposure", "visit"))
1018 self.assertCountEqual(set(dataId["exposure"] for dataId in rows), (100, 101, 110, 111, 200, 201))
1019 self.assertCountEqual(set(dataId["visit"] for dataId in rows), (10, 11, 20))
1020 self.assertCountEqual(set(dataId["detector"] for dataId in rows), (1, 2, 3, 4, 5))
1022 # limit to single visit
1023 rows = registry.queryDataIds(
1024 dimensions, datasets=rawType, collections=run1, where="visit = 10", instrument="DummyCam"
1025 ).toSet()
1026 self.assertEqual(len(rows), 2 * 3) # 2 exposures times 3 detectors
1027 self.assertCountEqual(set(dataId["exposure"] for dataId in rows), (100, 101))
1028 self.assertCountEqual(set(dataId["visit"] for dataId in rows), (10,))
1029 self.assertCountEqual(set(dataId["detector"] for dataId in rows), (1, 2, 3))
1031 # more limiting expression, using link names instead of Table.column
1032 rows = registry.queryDataIds(
1033 dimensions,
1034 datasets=rawType,
1035 collections=run1,
1036 where="visit = 10 and detector > 1 and 'DummyCam'=instrument",
1037 ).toSet()
1038 self.assertEqual(len(rows), 2 * 2) # 2 exposures times 2 detectors
1039 self.assertCountEqual(set(dataId["exposure"] for dataId in rows), (100, 101))
1040 self.assertCountEqual(set(dataId["visit"] for dataId in rows), (10,))
1041 self.assertCountEqual(set(dataId["detector"] for dataId in rows), (2, 3))
1043 # queryDataIds with only one of `datasets` and `collections` is an
1044 # error.
1045 with self.assertRaises(CollectionError):
1046 registry.queryDataIds(dimensions, datasets=rawType)
1047 with self.assertRaises(ArgumentError):
1048 registry.queryDataIds(dimensions, collections=run1)
1050 # expression excludes everything
1051 rows = registry.queryDataIds(
1052 dimensions, datasets=rawType, collections=run1, where="visit > 1000", instrument="DummyCam"
1053 ).toSet()
1054 self.assertEqual(len(rows), 0)
1056 # Selecting by physical_filter, this is not in the dimensions, but it
1057 # is a part of the full expression so it should work too.
1058 rows = registry.queryDataIds(
1059 dimensions,
1060 datasets=rawType,
1061 collections=run1,
1062 where="physical_filter = 'dummy_r'",
1063 instrument="DummyCam",
1064 ).toSet()
1065 self.assertEqual(len(rows), 2 * 3) # 2 exposures times 3 detectors
1066 self.assertCountEqual(set(dataId["exposure"] for dataId in rows), (110, 111))
1067 self.assertCountEqual(set(dataId["visit"] for dataId in rows), (11,))
1068 self.assertCountEqual(set(dataId["detector"] for dataId in rows), (1, 2, 3))
1070 def testSkyMapDimensions(self):
1071 """Tests involving only skymap dimensions, no joins to instrument."""
1072 registry = self.makeRegistry()
1074 # need a bunch of dimensions and datasets for test, we want
1075 # "band" in the test so also have to add physical_filter
1076 # dimensions
1077 registry.insertDimensionData("instrument", dict(instrument="DummyCam"))
1078 registry.insertDimensionData(
1079 "physical_filter",
1080 dict(instrument="DummyCam", name="dummy_r", band="r"),
1081 dict(instrument="DummyCam", name="dummy_i", band="i"),
1082 )
1083 registry.insertDimensionData("skymap", dict(name="DummyMap", hash="sha!".encode("utf8")))
1084 for tract in range(10):
1085 registry.insertDimensionData("tract", dict(skymap="DummyMap", id=tract))
1086 registry.insertDimensionData(
1087 "patch",
1088 *[dict(skymap="DummyMap", tract=tract, id=patch, cell_x=0, cell_y=0) for patch in range(10)],
1089 )
1091 # dataset types
1092 run = "tésτ"
1093 registry.registerRun(run)
1094 storageClass = StorageClass("testDataset")
1095 registry.storageClasses.registerStorageClass(storageClass)
1096 calexpType = DatasetType(
1097 name="deepCoadd_calexp",
1098 dimensions=registry.dimensions.extract(("skymap", "tract", "patch", "band")),
1099 storageClass=storageClass,
1100 )
1101 registry.registerDatasetType(calexpType)
1102 mergeType = DatasetType(
1103 name="deepCoadd_mergeDet",
1104 dimensions=registry.dimensions.extract(("skymap", "tract", "patch")),
1105 storageClass=storageClass,
1106 )
1107 registry.registerDatasetType(mergeType)
1108 measType = DatasetType(
1109 name="deepCoadd_meas",
1110 dimensions=registry.dimensions.extract(("skymap", "tract", "patch", "band")),
1111 storageClass=storageClass,
1112 )
1113 registry.registerDatasetType(measType)
1115 dimensions = DimensionGraph(
1116 registry.dimensions,
1117 dimensions=(
1118 calexpType.dimensions.required | mergeType.dimensions.required | measType.dimensions.required
1119 ),
1120 )
1122 # add pre-existing datasets
1123 for tract in (1, 3, 5):
1124 for patch in (2, 4, 6, 7):
1125 dataId = dict(skymap="DummyMap", tract=tract, patch=patch)
1126 registry.insertDatasets(mergeType, dataIds=[dataId], run=run)
1127 for aFilter in ("i", "r"):
1128 dataId = dict(skymap="DummyMap", tract=tract, patch=patch, band=aFilter)
1129 registry.insertDatasets(calexpType, dataIds=[dataId], run=run)
1131 # with empty expression
1132 rows = registry.queryDataIds(dimensions, datasets=[calexpType, mergeType], collections=run).toSet()
1133 self.assertEqual(len(rows), 3 * 4 * 2) # 4 tracts x 4 patches x 2 filters
1134 for dataId in rows:
1135 self.assertCountEqual(dataId.keys(), ("skymap", "tract", "patch", "band"))
1136 self.assertCountEqual(set(dataId["tract"] for dataId in rows), (1, 3, 5))
1137 self.assertCountEqual(set(dataId["patch"] for dataId in rows), (2, 4, 6, 7))
1138 self.assertCountEqual(set(dataId["band"] for dataId in rows), ("i", "r"))
1140 # limit to 2 tracts and 2 patches
1141 rows = registry.queryDataIds(
1142 dimensions,
1143 datasets=[calexpType, mergeType],
1144 collections=run,
1145 where="tract IN (1, 5) AND patch IN (2, 7)",
1146 skymap="DummyMap",
1147 ).toSet()
1148 self.assertEqual(len(rows), 2 * 2 * 2) # 2 tracts x 2 patches x 2 filters
1149 self.assertCountEqual(set(dataId["tract"] for dataId in rows), (1, 5))
1150 self.assertCountEqual(set(dataId["patch"] for dataId in rows), (2, 7))
1151 self.assertCountEqual(set(dataId["band"] for dataId in rows), ("i", "r"))
1153 # limit to single filter
1154 rows = registry.queryDataIds(
1155 dimensions, datasets=[calexpType, mergeType], collections=run, where="band = 'i'"
1156 ).toSet()
1157 self.assertEqual(len(rows), 3 * 4 * 1) # 4 tracts x 4 patches x 2 filters
1158 self.assertCountEqual(set(dataId["tract"] for dataId in rows), (1, 3, 5))
1159 self.assertCountEqual(set(dataId["patch"] for dataId in rows), (2, 4, 6, 7))
1160 self.assertCountEqual(set(dataId["band"] for dataId in rows), ("i",))
1162 # Specifying non-existing skymap is an exception
1163 with self.assertRaisesRegex(DataIdValueError, "Unknown values specified for governor dimension"):
1164 rows = registry.queryDataIds(
1165 dimensions, datasets=[calexpType, mergeType], collections=run, where="skymap = 'Mars'"
1166 ).toSet()
1168 def testSpatialJoin(self):
1169 """Test queries that involve spatial overlap joins."""
1170 registry = self.makeRegistry()
1171 self.loadData(registry, "hsc-rc2-subset.yaml")
1173 # Dictionary of spatial DatabaseDimensionElements, keyed by the name of
1174 # the TopologicalFamily they belong to. We'll relate all elements in
1175 # each family to all of the elements in each other family.
1176 families = defaultdict(set)
1177 # Dictionary of {element.name: {dataId: region}}.
1178 regions = {}
1179 for element in registry.dimensions.getDatabaseElements():
1180 if element.spatial is not None:
1181 families[element.spatial.name].add(element)
1182 regions[element.name] = {
1183 record.dataId: record.region for record in registry.queryDimensionRecords(element)
1184 }
1186 # If this check fails, it's not necessarily a problem - it may just be
1187 # a reasonable change to the default dimension definitions - but the
1188 # test below depends on there being more than one family to do anything
1189 # useful.
1190 self.assertEqual(len(families), 2)
1192 # Overlap DatabaseDimensionElements with each other.
1193 for family1, family2 in itertools.combinations(families, 2):
1194 for element1, element2 in itertools.product(families[family1], families[family2]):
1195 graph = DimensionGraph.union(element1.graph, element2.graph)
1196 # Construct expected set of overlapping data IDs via a
1197 # brute-force comparison of the regions we've already fetched.
1198 expected = {
1199 DataCoordinate.standardize({**dataId1.byName(), **dataId2.byName()}, graph=graph)
1200 for (dataId1, region1), (dataId2, region2) in itertools.product(
1201 regions[element1.name].items(), regions[element2.name].items()
1202 )
1203 if not region1.isDisjointFrom(region2)
1204 }
1205 self.assertGreater(len(expected), 2, msg="Test that we aren't just comparing empty sets.")
1206 queried = set(registry.queryDataIds(graph))
1207 self.assertEqual(expected, queried)
1209 # Overlap each DatabaseDimensionElement with the commonSkyPix system.
1210 commonSkyPix = registry.dimensions.commonSkyPix
1211 for elementName, regions in regions.items():
1212 graph = DimensionGraph.union(registry.dimensions[elementName].graph, commonSkyPix.graph)
1213 expected = set()
1214 for dataId, region in regions.items():
1215 for begin, end in commonSkyPix.pixelization.envelope(region):
1216 expected.update(
1217 DataCoordinate.standardize({commonSkyPix.name: index, **dataId.byName()}, graph=graph)
1218 for index in range(begin, end)
1219 )
1220 self.assertGreater(len(expected), 2, msg="Test that we aren't just comparing empty sets.")
1221 queried = set(registry.queryDataIds(graph))
1222 self.assertEqual(expected, queried)
1224 def testAbstractQuery(self):
1225 """Test that we can run a query that just lists the known
1226 bands. This is tricky because band is
1227 backed by a query against physical_filter.
1228 """
1229 registry = self.makeRegistry()
1230 registry.insertDimensionData("instrument", dict(name="DummyCam"))
1231 registry.insertDimensionData(
1232 "physical_filter",
1233 dict(instrument="DummyCam", name="dummy_i", band="i"),
1234 dict(instrument="DummyCam", name="dummy_i2", band="i"),
1235 dict(instrument="DummyCam", name="dummy_r", band="r"),
1236 )
1237 rows = registry.queryDataIds(["band"]).toSet()
1238 self.assertCountEqual(
1239 rows,
1240 [
1241 DataCoordinate.standardize(band="i", universe=registry.dimensions),
1242 DataCoordinate.standardize(band="r", universe=registry.dimensions),
1243 ],
1244 )
1246 def testAttributeManager(self):
1247 """Test basic functionality of attribute manager."""
1248 # number of attributes with schema versions in a fresh database,
1249 # 6 managers with 3 records per manager, plus config for dimensions
1250 VERSION_COUNT = 6 * 3 + 1
1252 registry = self.makeRegistry()
1253 attributes = registry._managers.attributes
1255 # check what get() returns for non-existing key
1256 self.assertIsNone(attributes.get("attr"))
1257 self.assertEqual(attributes.get("attr", ""), "")
1258 self.assertEqual(attributes.get("attr", "Value"), "Value")
1259 self.assertEqual(len(list(attributes.items())), VERSION_COUNT)
1261 # cannot store empty key or value
1262 with self.assertRaises(ValueError):
1263 attributes.set("", "value")
1264 with self.assertRaises(ValueError):
1265 attributes.set("attr", "")
1267 # set value of non-existing key
1268 attributes.set("attr", "value")
1269 self.assertEqual(len(list(attributes.items())), VERSION_COUNT + 1)
1270 self.assertEqual(attributes.get("attr"), "value")
1272 # update value of existing key
1273 with self.assertRaises(ButlerAttributeExistsError):
1274 attributes.set("attr", "value2")
1276 attributes.set("attr", "value2", force=True)
1277 self.assertEqual(len(list(attributes.items())), VERSION_COUNT + 1)
1278 self.assertEqual(attributes.get("attr"), "value2")
1280 # delete existing key
1281 self.assertTrue(attributes.delete("attr"))
1282 self.assertEqual(len(list(attributes.items())), VERSION_COUNT)
1284 # delete non-existing key
1285 self.assertFalse(attributes.delete("non-attr"))
1287 # store bunch of keys and get the list back
1288 data = [
1289 ("version.core", "1.2.3"),
1290 ("version.dimensions", "3.2.1"),
1291 ("config.managers.opaque", "ByNameOpaqueTableStorageManager"),
1292 ]
1293 for key, value in data:
1294 attributes.set(key, value)
1295 items = dict(attributes.items())
1296 for key, value in data:
1297 self.assertEqual(items[key], value)
1299 def testQueryDatasetsDeduplication(self):
1300 """Test that the findFirst option to queryDatasets selects datasets
1301 from collections in the order given".
1302 """
1303 registry = self.makeRegistry()
1304 self.loadData(registry, "base.yaml")
1305 self.loadData(registry, "datasets.yaml")
1306 self.assertCountEqual(
1307 list(registry.queryDatasets("bias", collections=["imported_g", "imported_r"])),
1308 [
1309 registry.findDataset("bias", instrument="Cam1", detector=1, collections="imported_g"),
1310 registry.findDataset("bias", instrument="Cam1", detector=2, collections="imported_g"),
1311 registry.findDataset("bias", instrument="Cam1", detector=3, collections="imported_g"),
1312 registry.findDataset("bias", instrument="Cam1", detector=2, collections="imported_r"),
1313 registry.findDataset("bias", instrument="Cam1", detector=3, collections="imported_r"),
1314 registry.findDataset("bias", instrument="Cam1", detector=4, collections="imported_r"),
1315 ],
1316 )
1317 self.assertCountEqual(
1318 list(registry.queryDatasets("bias", collections=["imported_g", "imported_r"], findFirst=True)),
1319 [
1320 registry.findDataset("bias", instrument="Cam1", detector=1, collections="imported_g"),
1321 registry.findDataset("bias", instrument="Cam1", detector=2, collections="imported_g"),
1322 registry.findDataset("bias", instrument="Cam1", detector=3, collections="imported_g"),
1323 registry.findDataset("bias", instrument="Cam1", detector=4, collections="imported_r"),
1324 ],
1325 )
1326 self.assertCountEqual(
1327 list(registry.queryDatasets("bias", collections=["imported_r", "imported_g"], findFirst=True)),
1328 [
1329 registry.findDataset("bias", instrument="Cam1", detector=1, collections="imported_g"),
1330 registry.findDataset("bias", instrument="Cam1", detector=2, collections="imported_r"),
1331 registry.findDataset("bias", instrument="Cam1", detector=3, collections="imported_r"),
1332 registry.findDataset("bias", instrument="Cam1", detector=4, collections="imported_r"),
1333 ],
1334 )
1336 def testQueryResults(self):
1337 """Test querying for data IDs and then manipulating the QueryResults
1338 object returned to perform other queries.
1339 """
1340 registry = self.makeRegistry()
1341 self.loadData(registry, "base.yaml")
1342 self.loadData(registry, "datasets.yaml")
1343 bias = registry.getDatasetType("bias")
1344 flat = registry.getDatasetType("flat")
1345 # Obtain expected results from methods other than those we're testing
1346 # here. That includes:
1347 # - the dimensions of the data IDs we want to query:
1348 expectedGraph = DimensionGraph(registry.dimensions, names=["detector", "physical_filter"])
1349 # - the dimensions of some other data IDs we'll extract from that:
1350 expectedSubsetGraph = DimensionGraph(registry.dimensions, names=["detector"])
1351 # - the data IDs we expect to obtain from the first queries:
1352 expectedDataIds = DataCoordinateSet(
1353 {
1354 DataCoordinate.standardize(
1355 instrument="Cam1", detector=d, physical_filter=p, universe=registry.dimensions
1356 )
1357 for d, p in itertools.product({1, 2, 3}, {"Cam1-G", "Cam1-R1", "Cam1-R2"})
1358 },
1359 graph=expectedGraph,
1360 hasFull=False,
1361 hasRecords=False,
1362 )
1363 # - the flat datasets we expect to find from those data IDs, in just
1364 # one collection (so deduplication is irrelevant):
1365 expectedFlats = [
1366 registry.findDataset(
1367 flat, instrument="Cam1", detector=1, physical_filter="Cam1-R1", collections="imported_r"
1368 ),
1369 registry.findDataset(
1370 flat, instrument="Cam1", detector=2, physical_filter="Cam1-R1", collections="imported_r"
1371 ),
1372 registry.findDataset(
1373 flat, instrument="Cam1", detector=3, physical_filter="Cam1-R2", collections="imported_r"
1374 ),
1375 ]
1376 # - the data IDs we expect to extract from that:
1377 expectedSubsetDataIds = expectedDataIds.subset(expectedSubsetGraph)
1378 # - the bias datasets we expect to find from those data IDs, after we
1379 # subset-out the physical_filter dimension, both with duplicates:
1380 expectedAllBiases = [
1381 registry.findDataset(bias, instrument="Cam1", detector=1, collections="imported_g"),
1382 registry.findDataset(bias, instrument="Cam1", detector=2, collections="imported_g"),
1383 registry.findDataset(bias, instrument="Cam1", detector=3, collections="imported_g"),
1384 registry.findDataset(bias, instrument="Cam1", detector=2, collections="imported_r"),
1385 registry.findDataset(bias, instrument="Cam1", detector=3, collections="imported_r"),
1386 ]
1387 # - ...and without duplicates:
1388 expectedDeduplicatedBiases = [
1389 registry.findDataset(bias, instrument="Cam1", detector=1, collections="imported_g"),
1390 registry.findDataset(bias, instrument="Cam1", detector=2, collections="imported_r"),
1391 registry.findDataset(bias, instrument="Cam1", detector=3, collections="imported_r"),
1392 ]
1393 # Test against those expected results, using a "lazy" query for the
1394 # data IDs (which re-executes that query each time we use it to do
1395 # something new).
1396 dataIds = registry.queryDataIds(
1397 ["detector", "physical_filter"],
1398 where="detector.purpose = 'SCIENCE'", # this rejects detector=4
1399 instrument="Cam1",
1400 )
1401 self.assertEqual(dataIds.graph, expectedGraph)
1402 self.assertEqual(dataIds.toSet(), expectedDataIds)
1403 self.assertCountEqual(
1404 list(
1405 dataIds.findDatasets(
1406 flat,
1407 collections=["imported_r"],
1408 )
1409 ),
1410 expectedFlats,
1411 )
1412 subsetDataIds = dataIds.subset(expectedSubsetGraph, unique=True)
1413 self.assertEqual(subsetDataIds.graph, expectedSubsetGraph)
1414 self.assertEqual(subsetDataIds.toSet(), expectedSubsetDataIds)
1415 self.assertCountEqual(
1416 list(subsetDataIds.findDatasets(bias, collections=["imported_r", "imported_g"], findFirst=False)),
1417 expectedAllBiases,
1418 )
1419 self.assertCountEqual(
1420 list(subsetDataIds.findDatasets(bias, collections=["imported_r", "imported_g"], findFirst=True)),
1421 expectedDeduplicatedBiases,
1422 )
1424 # Check dimensions match.
1425 with self.assertRaises(ValueError):
1426 subsetDataIds.findDatasets("flat", collections=["imported_r", "imported_g"], findFirst=True)
1428 # Use a component dataset type.
1429 self.assertCountEqual(
1430 [
1431 ref.makeComponentRef("image")
1432 for ref in subsetDataIds.findDatasets(
1433 bias,
1434 collections=["imported_r", "imported_g"],
1435 findFirst=False,
1436 )
1437 ],
1438 [ref.makeComponentRef("image") for ref in expectedAllBiases],
1439 )
1441 # Use a named dataset type that does not exist and a dataset type
1442 # object that does not exist.
1443 unknown_type = DatasetType("not_known", dimensions=bias.dimensions, storageClass="Exposure")
1445 # Test both string name and dataset type object.
1446 test_type: Union[str, DatasetType]
1447 for test_type, test_type_name in (
1448 (unknown_type, unknown_type.name),
1449 (unknown_type.name, unknown_type.name),
1450 ):
1451 with self.assertRaisesRegex(DatasetTypeError, expected_regex=test_type_name):
1452 list(
1453 subsetDataIds.findDatasets(
1454 test_type, collections=["imported_r", "imported_g"], findFirst=True
1455 )
1456 )
1458 # Materialize the bias dataset queries (only) by putting the results
1459 # into temporary tables, then repeat those tests.
1460 with subsetDataIds.findDatasets(
1461 bias, collections=["imported_r", "imported_g"], findFirst=False
1462 ).materialize() as biases:
1463 self.assertCountEqual(list(biases), expectedAllBiases)
1464 with subsetDataIds.findDatasets(
1465 bias, collections=["imported_r", "imported_g"], findFirst=True
1466 ).materialize() as biases:
1467 self.assertCountEqual(list(biases), expectedDeduplicatedBiases)
1468 # Materialize the data ID subset query, but not the dataset queries.
1469 with subsetDataIds.materialize() as subsetDataIds:
1470 self.assertEqual(subsetDataIds.graph, expectedSubsetGraph)
1471 self.assertEqual(subsetDataIds.toSet(), expectedSubsetDataIds)
1472 self.assertCountEqual(
1473 list(
1474 subsetDataIds.findDatasets(
1475 bias, collections=["imported_r", "imported_g"], findFirst=False
1476 )
1477 ),
1478 expectedAllBiases,
1479 )
1480 self.assertCountEqual(
1481 list(
1482 subsetDataIds.findDatasets(bias, collections=["imported_r", "imported_g"], findFirst=True)
1483 ),
1484 expectedDeduplicatedBiases,
1485 )
1486 # Materialize the dataset queries, too.
1487 with subsetDataIds.findDatasets(
1488 bias, collections=["imported_r", "imported_g"], findFirst=False
1489 ).materialize() as biases:
1490 self.assertCountEqual(list(biases), expectedAllBiases)
1491 with subsetDataIds.findDatasets(
1492 bias, collections=["imported_r", "imported_g"], findFirst=True
1493 ).materialize() as biases:
1494 self.assertCountEqual(list(biases), expectedDeduplicatedBiases)
1495 # Materialize the original query, but none of the follow-up queries.
1496 with dataIds.materialize() as dataIds:
1497 self.assertEqual(dataIds.graph, expectedGraph)
1498 self.assertEqual(dataIds.toSet(), expectedDataIds)
1499 self.assertCountEqual(
1500 list(
1501 dataIds.findDatasets(
1502 flat,
1503 collections=["imported_r"],
1504 )
1505 ),
1506 expectedFlats,
1507 )
1508 subsetDataIds = dataIds.subset(expectedSubsetGraph, unique=True)
1509 self.assertEqual(subsetDataIds.graph, expectedSubsetGraph)
1510 self.assertEqual(subsetDataIds.toSet(), expectedSubsetDataIds)
1511 self.assertCountEqual(
1512 list(
1513 subsetDataIds.findDatasets(
1514 bias, collections=["imported_r", "imported_g"], findFirst=False
1515 )
1516 ),
1517 expectedAllBiases,
1518 )
1519 self.assertCountEqual(
1520 list(
1521 subsetDataIds.findDatasets(bias, collections=["imported_r", "imported_g"], findFirst=True)
1522 ),
1523 expectedDeduplicatedBiases,
1524 )
1525 # Materialize just the bias dataset queries.
1526 with subsetDataIds.findDatasets(
1527 bias, collections=["imported_r", "imported_g"], findFirst=False
1528 ).materialize() as biases:
1529 self.assertCountEqual(list(biases), expectedAllBiases)
1530 with subsetDataIds.findDatasets(
1531 bias, collections=["imported_r", "imported_g"], findFirst=True
1532 ).materialize() as biases:
1533 self.assertCountEqual(list(biases), expectedDeduplicatedBiases)
1534 # Materialize the subset data ID query, but not the dataset
1535 # queries.
1536 with subsetDataIds.materialize() as subsetDataIds:
1537 self.assertEqual(subsetDataIds.graph, expectedSubsetGraph)
1538 self.assertEqual(subsetDataIds.toSet(), expectedSubsetDataIds)
1539 self.assertCountEqual(
1540 list(
1541 subsetDataIds.findDatasets(
1542 bias, collections=["imported_r", "imported_g"], findFirst=False
1543 )
1544 ),
1545 expectedAllBiases,
1546 )
1547 self.assertCountEqual(
1548 list(
1549 subsetDataIds.findDatasets(
1550 bias, collections=["imported_r", "imported_g"], findFirst=True
1551 )
1552 ),
1553 expectedDeduplicatedBiases,
1554 )
1555 # Materialize the bias dataset queries, too, so now we're
1556 # materializing every single step.
1557 with subsetDataIds.findDatasets(
1558 bias, collections=["imported_r", "imported_g"], findFirst=False
1559 ).materialize() as biases:
1560 self.assertCountEqual(list(biases), expectedAllBiases)
1561 with subsetDataIds.findDatasets(
1562 bias, collections=["imported_r", "imported_g"], findFirst=True
1563 ).materialize() as biases:
1564 self.assertCountEqual(list(biases), expectedDeduplicatedBiases)
1566 def testStorageClassPropagation(self):
1567 """Test that queries for datasets respect the storage class passed in
1568 as part of a full dataset type.
1569 """
1570 registry = self.makeRegistry()
1571 self.loadData(registry, "base.yaml")
1572 dataset_type_in_registry = DatasetType(
1573 "tbl", dimensions=["instrument"], storageClass="DataFrame", universe=registry.dimensions
1574 )
1575 registry.registerDatasetType(dataset_type_in_registry)
1576 run = "run1"
1577 registry.registerRun(run)
1578 (inserted_ref,) = registry.insertDatasets(
1579 dataset_type_in_registry, [registry.expandDataId(instrument="Cam1")], run=run
1580 )
1581 self.assertEqual(inserted_ref.datasetType, dataset_type_in_registry)
1582 query_dataset_type = DatasetType(
1583 "tbl", dimensions=["instrument"], storageClass="ArrowAstropy", universe=registry.dimensions
1584 )
1585 self.assertNotEqual(dataset_type_in_registry, query_dataset_type)
1586 query_datasets_result = registry.queryDatasets(query_dataset_type, collections=[run])
1587 self.assertEqual(query_datasets_result.parentDatasetType, query_dataset_type) # type: ignore
1588 (query_datasets_ref,) = query_datasets_result
1589 self.assertEqual(query_datasets_ref.datasetType, query_dataset_type)
1590 query_data_ids_find_datasets_result = registry.queryDataIds(["instrument"]).findDatasets(
1591 query_dataset_type, collections=[run]
1592 )
1593 self.assertEqual(query_data_ids_find_datasets_result.parentDatasetType, query_dataset_type)
1594 (query_data_ids_find_datasets_ref,) = query_data_ids_find_datasets_result
1595 self.assertEqual(query_data_ids_find_datasets_ref.datasetType, query_dataset_type)
1596 query_dataset_types_result = registry.queryDatasetTypes(query_dataset_type)
1597 self.assertEqual(list(query_dataset_types_result), [query_dataset_type])
1598 find_dataset_ref = registry.findDataset(query_dataset_type, instrument="Cam1", collections=[run])
1599 self.assertEqual(find_dataset_ref.datasetType, query_dataset_type)
1601 def testEmptyDimensionsQueries(self):
1602 """Test Query and QueryResults objects in the case where there are no
1603 dimensions.
1604 """
1605 # Set up test data: one dataset type, two runs, one dataset in each.
1606 registry = self.makeRegistry()
1607 self.loadData(registry, "base.yaml")
1608 schema = DatasetType("schema", dimensions=registry.dimensions.empty, storageClass="Catalog")
1609 registry.registerDatasetType(schema)
1610 dataId = DataCoordinate.makeEmpty(registry.dimensions)
1611 run1 = "run1"
1612 run2 = "run2"
1613 registry.registerRun(run1)
1614 registry.registerRun(run2)
1615 (dataset1,) = registry.insertDatasets(schema, dataIds=[dataId], run=run1)
1616 (dataset2,) = registry.insertDatasets(schema, dataIds=[dataId], run=run2)
1617 # Query directly for both of the datasets, and each one, one at a time.
1618 self.checkQueryResults(
1619 registry.queryDatasets(schema, collections=[run1, run2], findFirst=False), [dataset1, dataset2]
1620 )
1621 self.checkQueryResults(
1622 registry.queryDatasets(schema, collections=[run1, run2], findFirst=True),
1623 [dataset1],
1624 )
1625 self.checkQueryResults(
1626 registry.queryDatasets(schema, collections=[run2, run1], findFirst=True),
1627 [dataset2],
1628 )
1629 # Query for data IDs with no dimensions.
1630 dataIds = registry.queryDataIds([])
1631 self.checkQueryResults(dataIds, [dataId])
1632 # Use queried data IDs to find the datasets.
1633 self.checkQueryResults(
1634 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=False),
1635 [dataset1, dataset2],
1636 )
1637 self.checkQueryResults(
1638 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=True),
1639 [dataset1],
1640 )
1641 self.checkQueryResults(
1642 dataIds.findDatasets(schema, collections=[run2, run1], findFirst=True),
1643 [dataset2],
1644 )
1645 # Now materialize the data ID query results and repeat those tests.
1646 with dataIds.materialize() as dataIds:
1647 self.checkQueryResults(dataIds, [dataId])
1648 self.checkQueryResults(
1649 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=True),
1650 [dataset1],
1651 )
1652 self.checkQueryResults(
1653 dataIds.findDatasets(schema, collections=[run2, run1], findFirst=True),
1654 [dataset2],
1655 )
1656 # Query for non-empty data IDs, then subset that to get the empty one.
1657 # Repeat the above tests starting from that.
1658 dataIds = registry.queryDataIds(["instrument"]).subset(registry.dimensions.empty, unique=True)
1659 self.checkQueryResults(dataIds, [dataId])
1660 self.checkQueryResults(
1661 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=False),
1662 [dataset1, dataset2],
1663 )
1664 self.checkQueryResults(
1665 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=True),
1666 [dataset1],
1667 )
1668 self.checkQueryResults(
1669 dataIds.findDatasets(schema, collections=[run2, run1], findFirst=True),
1670 [dataset2],
1671 )
1672 with dataIds.materialize() as dataIds:
1673 self.checkQueryResults(dataIds, [dataId])
1674 self.checkQueryResults(
1675 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=False),
1676 [dataset1, dataset2],
1677 )
1678 self.checkQueryResults(
1679 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=True),
1680 [dataset1],
1681 )
1682 self.checkQueryResults(
1683 dataIds.findDatasets(schema, collections=[run2, run1], findFirst=True),
1684 [dataset2],
1685 )
1686 # Query for non-empty data IDs, then materialize, then subset to get
1687 # the empty one. Repeat again.
1688 with registry.queryDataIds(["instrument"]).materialize() as nonEmptyDataIds:
1689 dataIds = nonEmptyDataIds.subset(registry.dimensions.empty, unique=True)
1690 self.checkQueryResults(dataIds, [dataId])
1691 self.checkQueryResults(
1692 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=False),
1693 [dataset1, dataset2],
1694 )
1695 self.checkQueryResults(
1696 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=True),
1697 [dataset1],
1698 )
1699 self.checkQueryResults(
1700 dataIds.findDatasets(schema, collections=[run2, run1], findFirst=True),
1701 [dataset2],
1702 )
1703 with dataIds.materialize() as dataIds:
1704 self.checkQueryResults(dataIds, [dataId])
1705 self.checkQueryResults(
1706 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=False),
1707 [dataset1, dataset2],
1708 )
1709 self.checkQueryResults(
1710 dataIds.findDatasets(schema, collections=[run1, run2], findFirst=True),
1711 [dataset1],
1712 )
1713 self.checkQueryResults(
1714 dataIds.findDatasets(schema, collections=[run2, run1], findFirst=True),
1715 [dataset2],
1716 )
1717 # Query for non-empty data IDs with a constraint on an empty-data-ID
1718 # dataset that exists.
1719 dataIds = registry.queryDataIds(["instrument"], datasets="schema", collections=...)
1720 self.checkQueryResults(
1721 dataIds.subset(unique=True),
1722 [DataCoordinate.standardize(instrument="Cam1", universe=registry.dimensions)],
1723 )
1724 # Again query for non-empty data IDs with a constraint on empty-data-ID
1725 # datasets, but when the datasets don't exist. We delete the existing
1726 # dataset and query just that collection rather than creating a new
1727 # empty collection because this is a bit less likely for our build-time
1728 # logic to shortcut-out (via the collection summaries), and such a
1729 # shortcut would make this test a bit more trivial than we'd like.
1730 registry.removeDatasets([dataset2])
1731 dataIds = registry.queryDataIds(["instrument"], datasets="schema", collections=run2)
1732 self.checkQueryResults(dataIds, [])
1734 def testDimensionDataModifications(self):
1735 """Test that modifying dimension records via:
1736 syncDimensionData(..., update=True) and
1737 insertDimensionData(..., replace=True) works as expected, even in the
1738 presence of datasets using those dimensions and spatial overlap
1739 relationships.
1740 """
1742 def unpack_range_set(ranges: lsst.sphgeom.RangeSet) -> Iterator[int]:
1743 """Unpack a sphgeom.RangeSet into the integers it contains."""
1744 for begin, end in ranges:
1745 yield from range(begin, end)
1747 def range_set_hull(
1748 ranges: lsst.sphgeom.RangeSet,
1749 pixelization: lsst.sphgeom.HtmPixelization,
1750 ) -> lsst.sphgeom.ConvexPolygon:
1751 """Create a ConvexPolygon hull of the region defined by a set of
1752 HTM pixelization index ranges.
1753 """
1754 points = []
1755 for index in unpack_range_set(ranges):
1756 points.extend(pixelization.triangle(index).getVertices())
1757 return lsst.sphgeom.ConvexPolygon(points)
1759 # Use HTM to set up an initial parent region (one arbitrary trixel)
1760 # and four child regions (the trixels within the parent at the next
1761 # level. We'll use the parent as a tract/visit region and the children
1762 # as its patch/visit_detector regions.
1763 registry = self.makeRegistry()
1764 htm6 = registry.dimensions.skypix["htm"][6].pixelization
1765 commonSkyPix = registry.dimensions.commonSkyPix.pixelization
1766 index = 12288
1767 child_ranges_small = lsst.sphgeom.RangeSet(index).scaled(4)
1768 assert htm6.universe().contains(child_ranges_small)
1769 child_regions_small = [htm6.triangle(i) for i in unpack_range_set(child_ranges_small)]
1770 parent_region_small = lsst.sphgeom.ConvexPolygon(
1771 list(itertools.chain.from_iterable(c.getVertices() for c in child_regions_small))
1772 )
1773 assert all(parent_region_small.contains(c) for c in child_regions_small)
1774 # Make a larger version of each child region, defined to be the set of
1775 # htm6 trixels that overlap the original's bounding circle. Make a new
1776 # parent that's the convex hull of the new children.
1777 child_regions_large = [
1778 range_set_hull(htm6.envelope(c.getBoundingCircle()), htm6) for c in child_regions_small
1779 ]
1780 assert all(large.contains(small) for large, small in zip(child_regions_large, child_regions_small))
1781 parent_region_large = lsst.sphgeom.ConvexPolygon(
1782 list(itertools.chain.from_iterable(c.getVertices() for c in child_regions_large))
1783 )
1784 assert all(parent_region_large.contains(c) for c in child_regions_large)
1785 assert parent_region_large.contains(parent_region_small)
1786 assert not parent_region_small.contains(parent_region_large)
1787 assert not all(parent_region_small.contains(c) for c in child_regions_large)
1788 # Find some commonSkyPix indices that overlap the large regions but not
1789 # overlap the small regions. We use commonSkyPix here to make sure the
1790 # real tests later involve what's in the database, not just post-query
1791 # filtering of regions.
1792 child_difference_indices = []
1793 for large, small in zip(child_regions_large, child_regions_small):
1794 difference = list(unpack_range_set(commonSkyPix.envelope(large) - commonSkyPix.envelope(small)))
1795 assert difference, "if this is empty, we can't test anything useful with these regions"
1796 assert all(
1797 not commonSkyPix.triangle(d).isDisjointFrom(large)
1798 and commonSkyPix.triangle(d).isDisjointFrom(small)
1799 for d in difference
1800 )
1801 child_difference_indices.append(difference)
1802 parent_difference_indices = list(
1803 unpack_range_set(
1804 commonSkyPix.envelope(parent_region_large) - commonSkyPix.envelope(parent_region_small)
1805 )
1806 )
1807 assert parent_difference_indices, "if this is empty, we can't test anything useful with these regions"
1808 assert all(
1809 (
1810 not commonSkyPix.triangle(d).isDisjointFrom(parent_region_large)
1811 and commonSkyPix.triangle(d).isDisjointFrom(parent_region_small)
1812 )
1813 for d in parent_difference_indices
1814 )
1815 # Now that we've finally got those regions, we'll insert the large ones
1816 # as tract/patch dimension records.
1817 skymap_name = "testing_v1"
1818 registry.insertDimensionData(
1819 "skymap",
1820 {
1821 "name": skymap_name,
1822 "hash": bytes([42]),
1823 "tract_max": 1,
1824 "patch_nx_max": 2,
1825 "patch_ny_max": 2,
1826 },
1827 )
1828 registry.insertDimensionData("tract", {"skymap": skymap_name, "id": 0, "region": parent_region_large})
1829 registry.insertDimensionData(
1830 "patch",
1831 *[
1832 {"skymap": skymap_name, "tract": 0, "id": n, "cell_x": n % 2, "cell_y": n // 2, "region": c}
1833 for n, c in enumerate(child_regions_large)
1834 ],
1835 )
1836 # Add at dataset that uses these dimensions to make sure that modifying
1837 # them doesn't disrupt foreign keys (need to make sure DB doesn't
1838 # implement insert with replace=True as delete-then-insert).
1839 dataset_type = DatasetType(
1840 "coadd",
1841 dimensions=["tract", "patch"],
1842 universe=registry.dimensions,
1843 storageClass="Exposure",
1844 )
1845 registry.registerDatasetType(dataset_type)
1846 registry.registerCollection("the_run", CollectionType.RUN)
1847 registry.insertDatasets(
1848 dataset_type,
1849 [{"skymap": skymap_name, "tract": 0, "patch": 2}],
1850 run="the_run",
1851 )
1852 # Query for tracts and patches that overlap some "difference" htm9
1853 # pixels; there should be overlaps, because the database has
1854 # the "large" suite of regions.
1855 self.assertEqual(
1856 {0},
1857 {
1858 data_id["tract"]
1859 for data_id in registry.queryDataIds(
1860 ["tract"],
1861 skymap=skymap_name,
1862 dataId={registry.dimensions.commonSkyPix.name: parent_difference_indices[0]},
1863 )
1864 },
1865 )
1866 for patch_id, patch_difference_indices in enumerate(child_difference_indices):
1867 self.assertIn(
1868 patch_id,
1869 {
1870 data_id["patch"]
1871 for data_id in registry.queryDataIds(
1872 ["patch"],
1873 skymap=skymap_name,
1874 dataId={registry.dimensions.commonSkyPix.name: patch_difference_indices[0]},
1875 )
1876 },
1877 )
1878 # Use sync to update the tract region and insert to update the regions
1879 # of the patches, to the "small" suite.
1880 updated = registry.syncDimensionData(
1881 "tract",
1882 {"skymap": skymap_name, "id": 0, "region": parent_region_small},
1883 update=True,
1884 )
1885 self.assertEqual(updated, {"region": parent_region_large})
1886 registry.insertDimensionData(
1887 "patch",
1888 *[
1889 {"skymap": skymap_name, "tract": 0, "id": n, "cell_x": n % 2, "cell_y": n // 2, "region": c}
1890 for n, c in enumerate(child_regions_small)
1891 ],
1892 replace=True,
1893 )
1894 # Query again; there now should be no such overlaps, because the
1895 # database has the "small" suite of regions.
1896 self.assertFalse(
1897 set(
1898 registry.queryDataIds(
1899 ["tract"],
1900 skymap=skymap_name,
1901 dataId={registry.dimensions.commonSkyPix.name: parent_difference_indices[0]},
1902 )
1903 )
1904 )
1905 for patch_id, patch_difference_indices in enumerate(child_difference_indices):
1906 self.assertNotIn(
1907 patch_id,
1908 {
1909 data_id["patch"]
1910 for data_id in registry.queryDataIds(
1911 ["patch"],
1912 skymap=skymap_name,
1913 dataId={registry.dimensions.commonSkyPix.name: patch_difference_indices[0]},
1914 )
1915 },
1916 )
1917 # Update back to the large regions and query one more time.
1918 updated = registry.syncDimensionData(
1919 "tract",
1920 {"skymap": skymap_name, "id": 0, "region": parent_region_large},
1921 update=True,
1922 )
1923 self.assertEqual(updated, {"region": parent_region_small})
1924 registry.insertDimensionData(
1925 "patch",
1926 *[
1927 {"skymap": skymap_name, "tract": 0, "id": n, "cell_x": n % 2, "cell_y": n // 2, "region": c}
1928 for n, c in enumerate(child_regions_large)
1929 ],
1930 replace=True,
1931 )
1932 self.assertEqual(
1933 {0},
1934 {
1935 data_id["tract"]
1936 for data_id in registry.queryDataIds(
1937 ["tract"],
1938 skymap=skymap_name,
1939 dataId={registry.dimensions.commonSkyPix.name: parent_difference_indices[0]},
1940 )
1941 },
1942 )
1943 for patch_id, patch_difference_indices in enumerate(child_difference_indices):
1944 self.assertIn(
1945 patch_id,
1946 {
1947 data_id["patch"]
1948 for data_id in registry.queryDataIds(
1949 ["patch"],
1950 skymap=skymap_name,
1951 dataId={registry.dimensions.commonSkyPix.name: patch_difference_indices[0]},
1952 )
1953 },
1954 )
1956 def testCalibrationCollections(self):
1957 """Test operations on `~CollectionType.CALIBRATION` collections,
1958 including `Registry.certify`, `Registry.decertify`, and
1959 `Registry.findDataset`.
1960 """
1961 # Setup - make a Registry, fill it with some datasets in
1962 # non-calibration collections.
1963 registry = self.makeRegistry()
1964 self.loadData(registry, "base.yaml")
1965 self.loadData(registry, "datasets.yaml")
1966 # Set up some timestamps.
1967 t1 = astropy.time.Time("2020-01-01T01:00:00", format="isot", scale="tai")
1968 t2 = astropy.time.Time("2020-01-01T02:00:00", format="isot", scale="tai")
1969 t3 = astropy.time.Time("2020-01-01T03:00:00", format="isot", scale="tai")
1970 t4 = astropy.time.Time("2020-01-01T04:00:00", format="isot", scale="tai")
1971 t5 = astropy.time.Time("2020-01-01T05:00:00", format="isot", scale="tai")
1972 allTimespans = [
1973 Timespan(a, b) for a, b in itertools.combinations([None, t1, t2, t3, t4, t5, None], r=2)
1974 ]
1975 # Get references to some datasets.
1976 bias2a = registry.findDataset("bias", instrument="Cam1", detector=2, collections="imported_g")
1977 bias3a = registry.findDataset("bias", instrument="Cam1", detector=3, collections="imported_g")
1978 bias2b = registry.findDataset("bias", instrument="Cam1", detector=2, collections="imported_r")
1979 bias3b = registry.findDataset("bias", instrument="Cam1", detector=3, collections="imported_r")
1980 # Register the main calibration collection we'll be working with.
1981 collection = "Cam1/calibs/default"
1982 registry.registerCollection(collection, type=CollectionType.CALIBRATION)
1983 # Cannot associate into a calibration collection (no timespan).
1984 with self.assertRaises(CollectionTypeError):
1985 registry.associate(collection, [bias2a])
1986 # Certify 2a dataset with [t2, t4) validity.
1987 registry.certify(collection, [bias2a], Timespan(begin=t2, end=t4))
1988 # Test that we can query for this dataset via the new collection, both
1989 # on its own and with a RUN collection, as long as we don't try to join
1990 # in temporal dimensions or use findFirst=True.
1991 self.assertEqual(
1992 set(registry.queryDatasets("bias", findFirst=False, collections=collection)),
1993 {bias2a},
1994 )
1995 self.assertEqual(
1996 set(registry.queryDatasets("bias", findFirst=False, collections=[collection, "imported_r"])),
1997 {
1998 bias2a,
1999 bias2b,
2000 bias3b,
2001 registry.findDataset("bias", instrument="Cam1", detector=4, collections="imported_r"),
2002 },
2003 )
2004 self.assertEqual(
2005 set(registry.queryDataIds("detector", datasets="bias", collections=collection)),
2006 {registry.expandDataId(instrument="Cam1", detector=2)},
2007 )
2008 self.assertEqual(
2009 set(registry.queryDataIds("detector", datasets="bias", collections=[collection, "imported_r"])),
2010 {
2011 registry.expandDataId(instrument="Cam1", detector=2),
2012 registry.expandDataId(instrument="Cam1", detector=3),
2013 registry.expandDataId(instrument="Cam1", detector=4),
2014 },
2015 )
2017 # We should not be able to certify 2b with anything overlapping that
2018 # window.
2019 with self.assertRaises(ConflictingDefinitionError):
2020 registry.certify(collection, [bias2b], Timespan(begin=None, end=t3))
2021 with self.assertRaises(ConflictingDefinitionError):
2022 registry.certify(collection, [bias2b], Timespan(begin=None, end=t5))
2023 with self.assertRaises(ConflictingDefinitionError):
2024 registry.certify(collection, [bias2b], Timespan(begin=t1, end=t3))
2025 with self.assertRaises(ConflictingDefinitionError):
2026 registry.certify(collection, [bias2b], Timespan(begin=t1, end=t5))
2027 with self.assertRaises(ConflictingDefinitionError):
2028 registry.certify(collection, [bias2b], Timespan(begin=t1, end=None))
2029 with self.assertRaises(ConflictingDefinitionError):
2030 registry.certify(collection, [bias2b], Timespan(begin=t2, end=t3))
2031 with self.assertRaises(ConflictingDefinitionError):
2032 registry.certify(collection, [bias2b], Timespan(begin=t2, end=t5))
2033 with self.assertRaises(ConflictingDefinitionError):
2034 registry.certify(collection, [bias2b], Timespan(begin=t2, end=None))
2035 # We should be able to certify 3a with a range overlapping that window,
2036 # because it's for a different detector.
2037 # We'll certify 3a over [t1, t3).
2038 registry.certify(collection, [bias3a], Timespan(begin=t1, end=t3))
2039 # Now we'll certify 2b and 3b together over [t4, ∞).
2040 registry.certify(collection, [bias2b, bias3b], Timespan(begin=t4, end=None))
2042 # Fetch all associations and check that they are what we expect.
2043 self.assertCountEqual(
2044 list(
2045 registry.queryDatasetAssociations(
2046 "bias",
2047 collections=[collection, "imported_g", "imported_r"],
2048 )
2049 ),
2050 [
2051 DatasetAssociation(
2052 ref=registry.findDataset("bias", instrument="Cam1", detector=1, collections="imported_g"),
2053 collection="imported_g",
2054 timespan=None,
2055 ),
2056 DatasetAssociation(
2057 ref=registry.findDataset("bias", instrument="Cam1", detector=4, collections="imported_r"),
2058 collection="imported_r",
2059 timespan=None,
2060 ),
2061 DatasetAssociation(ref=bias2a, collection="imported_g", timespan=None),
2062 DatasetAssociation(ref=bias3a, collection="imported_g", timespan=None),
2063 DatasetAssociation(ref=bias2b, collection="imported_r", timespan=None),
2064 DatasetAssociation(ref=bias3b, collection="imported_r", timespan=None),
2065 DatasetAssociation(ref=bias2a, collection=collection, timespan=Timespan(begin=t2, end=t4)),
2066 DatasetAssociation(ref=bias3a, collection=collection, timespan=Timespan(begin=t1, end=t3)),
2067 DatasetAssociation(ref=bias2b, collection=collection, timespan=Timespan(begin=t4, end=None)),
2068 DatasetAssociation(ref=bias3b, collection=collection, timespan=Timespan(begin=t4, end=None)),
2069 ],
2070 )
2072 class Ambiguous:
2073 """Tag class to denote lookups that should be ambiguous."""
2075 pass
2077 def assertLookup(
2078 detector: int, timespan: Timespan, expected: Optional[Union[DatasetRef, Type[Ambiguous]]]
2079 ) -> None:
2080 """Local function that asserts that a bias lookup returns the given
2081 expected result.
2082 """
2083 if expected is Ambiguous:
2084 with self.assertRaises(RuntimeError):
2085 registry.findDataset(
2086 "bias",
2087 collections=collection,
2088 instrument="Cam1",
2089 detector=detector,
2090 timespan=timespan,
2091 )
2092 else:
2093 self.assertEqual(
2094 expected,
2095 registry.findDataset(
2096 "bias",
2097 collections=collection,
2098 instrument="Cam1",
2099 detector=detector,
2100 timespan=timespan,
2101 ),
2102 )
2104 # Systematically test lookups against expected results.
2105 assertLookup(detector=2, timespan=Timespan(None, t1), expected=None)
2106 assertLookup(detector=2, timespan=Timespan(None, t2), expected=None)
2107 assertLookup(detector=2, timespan=Timespan(None, t3), expected=bias2a)
2108 assertLookup(detector=2, timespan=Timespan(None, t4), expected=bias2a)
2109 assertLookup(detector=2, timespan=Timespan(None, t5), expected=Ambiguous)
2110 assertLookup(detector=2, timespan=Timespan(None, None), expected=Ambiguous)
2111 assertLookup(detector=2, timespan=Timespan(t1, t2), expected=None)
2112 assertLookup(detector=2, timespan=Timespan(t1, t3), expected=bias2a)
2113 assertLookup(detector=2, timespan=Timespan(t1, t4), expected=bias2a)
2114 assertLookup(detector=2, timespan=Timespan(t1, t5), expected=Ambiguous)
2115 assertLookup(detector=2, timespan=Timespan(t1, None), expected=Ambiguous)
2116 assertLookup(detector=2, timespan=Timespan(t2, t3), expected=bias2a)
2117 assertLookup(detector=2, timespan=Timespan(t2, t4), expected=bias2a)
2118 assertLookup(detector=2, timespan=Timespan(t2, t5), expected=Ambiguous)
2119 assertLookup(detector=2, timespan=Timespan(t2, None), expected=Ambiguous)
2120 assertLookup(detector=2, timespan=Timespan(t3, t4), expected=bias2a)
2121 assertLookup(detector=2, timespan=Timespan(t3, t5), expected=Ambiguous)
2122 assertLookup(detector=2, timespan=Timespan(t3, None), expected=Ambiguous)
2123 assertLookup(detector=2, timespan=Timespan(t4, t5), expected=bias2b)
2124 assertLookup(detector=2, timespan=Timespan(t4, None), expected=bias2b)
2125 assertLookup(detector=2, timespan=Timespan(t5, None), expected=bias2b)
2126 assertLookup(detector=3, timespan=Timespan(None, t1), expected=None)
2127 assertLookup(detector=3, timespan=Timespan(None, t2), expected=bias3a)
2128 assertLookup(detector=3, timespan=Timespan(None, t3), expected=bias3a)
2129 assertLookup(detector=3, timespan=Timespan(None, t4), expected=bias3a)
2130 assertLookup(detector=3, timespan=Timespan(None, t5), expected=Ambiguous)
2131 assertLookup(detector=3, timespan=Timespan(None, None), expected=Ambiguous)
2132 assertLookup(detector=3, timespan=Timespan(t1, t2), expected=bias3a)
2133 assertLookup(detector=3, timespan=Timespan(t1, t3), expected=bias3a)
2134 assertLookup(detector=3, timespan=Timespan(t1, t4), expected=bias3a)
2135 assertLookup(detector=3, timespan=Timespan(t1, t5), expected=Ambiguous)
2136 assertLookup(detector=3, timespan=Timespan(t1, None), expected=Ambiguous)
2137 assertLookup(detector=3, timespan=Timespan(t2, t3), expected=bias3a)
2138 assertLookup(detector=3, timespan=Timespan(t2, t4), expected=bias3a)
2139 assertLookup(detector=3, timespan=Timespan(t2, t5), expected=Ambiguous)
2140 assertLookup(detector=3, timespan=Timespan(t2, None), expected=Ambiguous)
2141 assertLookup(detector=3, timespan=Timespan(t3, t4), expected=None)
2142 assertLookup(detector=3, timespan=Timespan(t3, t5), expected=bias3b)
2143 assertLookup(detector=3, timespan=Timespan(t3, None), expected=bias3b)
2144 assertLookup(detector=3, timespan=Timespan(t4, t5), expected=bias3b)
2145 assertLookup(detector=3, timespan=Timespan(t4, None), expected=bias3b)
2146 assertLookup(detector=3, timespan=Timespan(t5, None), expected=bias3b)
2148 # Decertify [t3, t5) for all data IDs, and do test lookups again.
2149 # This should truncate bias2a to [t2, t3), leave bias3a unchanged at
2150 # [t1, t3), and truncate bias2b and bias3b to [t5, ∞).
2151 registry.decertify(collection=collection, datasetType="bias", timespan=Timespan(t3, t5))
2152 assertLookup(detector=2, timespan=Timespan(None, t1), expected=None)
2153 assertLookup(detector=2, timespan=Timespan(None, t2), expected=None)
2154 assertLookup(detector=2, timespan=Timespan(None, t3), expected=bias2a)
2155 assertLookup(detector=2, timespan=Timespan(None, t4), expected=bias2a)
2156 assertLookup(detector=2, timespan=Timespan(None, t5), expected=bias2a)
2157 assertLookup(detector=2, timespan=Timespan(None, None), expected=Ambiguous)
2158 assertLookup(detector=2, timespan=Timespan(t1, t2), expected=None)
2159 assertLookup(detector=2, timespan=Timespan(t1, t3), expected=bias2a)
2160 assertLookup(detector=2, timespan=Timespan(t1, t4), expected=bias2a)
2161 assertLookup(detector=2, timespan=Timespan(t1, t5), expected=bias2a)
2162 assertLookup(detector=2, timespan=Timespan(t1, None), expected=Ambiguous)
2163 assertLookup(detector=2, timespan=Timespan(t2, t3), expected=bias2a)
2164 assertLookup(detector=2, timespan=Timespan(t2, t4), expected=bias2a)
2165 assertLookup(detector=2, timespan=Timespan(t2, t5), expected=bias2a)
2166 assertLookup(detector=2, timespan=Timespan(t2, None), expected=Ambiguous)
2167 assertLookup(detector=2, timespan=Timespan(t3, t4), expected=None)
2168 assertLookup(detector=2, timespan=Timespan(t3, t5), expected=None)
2169 assertLookup(detector=2, timespan=Timespan(t3, None), expected=bias2b)
2170 assertLookup(detector=2, timespan=Timespan(t4, t5), expected=None)
2171 assertLookup(detector=2, timespan=Timespan(t4, None), expected=bias2b)
2172 assertLookup(detector=2, timespan=Timespan(t5, None), expected=bias2b)
2173 assertLookup(detector=3, timespan=Timespan(None, t1), expected=None)
2174 assertLookup(detector=3, timespan=Timespan(None, t2), expected=bias3a)
2175 assertLookup(detector=3, timespan=Timespan(None, t3), expected=bias3a)
2176 assertLookup(detector=3, timespan=Timespan(None, t4), expected=bias3a)
2177 assertLookup(detector=3, timespan=Timespan(None, t5), expected=bias3a)
2178 assertLookup(detector=3, timespan=Timespan(None, None), expected=Ambiguous)
2179 assertLookup(detector=3, timespan=Timespan(t1, t2), expected=bias3a)
2180 assertLookup(detector=3, timespan=Timespan(t1, t3), expected=bias3a)
2181 assertLookup(detector=3, timespan=Timespan(t1, t4), expected=bias3a)
2182 assertLookup(detector=3, timespan=Timespan(t1, t5), expected=bias3a)
2183 assertLookup(detector=3, timespan=Timespan(t1, None), expected=Ambiguous)
2184 assertLookup(detector=3, timespan=Timespan(t2, t3), expected=bias3a)
2185 assertLookup(detector=3, timespan=Timespan(t2, t4), expected=bias3a)
2186 assertLookup(detector=3, timespan=Timespan(t2, t5), expected=bias3a)
2187 assertLookup(detector=3, timespan=Timespan(t2, None), expected=Ambiguous)
2188 assertLookup(detector=3, timespan=Timespan(t3, t4), expected=None)
2189 assertLookup(detector=3, timespan=Timespan(t3, t5), expected=None)
2190 assertLookup(detector=3, timespan=Timespan(t3, None), expected=bias3b)
2191 assertLookup(detector=3, timespan=Timespan(t4, t5), expected=None)
2192 assertLookup(detector=3, timespan=Timespan(t4, None), expected=bias3b)
2193 assertLookup(detector=3, timespan=Timespan(t5, None), expected=bias3b)
2195 # Decertify everything, this time with explicit data IDs, then check
2196 # that no lookups succeed.
2197 registry.decertify(
2198 collection,
2199 "bias",
2200 Timespan(None, None),
2201 dataIds=[
2202 dict(instrument="Cam1", detector=2),
2203 dict(instrument="Cam1", detector=3),
2204 ],
2205 )
2206 for detector in (2, 3):
2207 for timespan in allTimespans:
2208 assertLookup(detector=detector, timespan=timespan, expected=None)
2209 # Certify bias2a and bias3a over (-∞, ∞), check that all lookups return
2210 # those.
2211 registry.certify(
2212 collection,
2213 [bias2a, bias3a],
2214 Timespan(None, None),
2215 )
2216 for timespan in allTimespans:
2217 assertLookup(detector=2, timespan=timespan, expected=bias2a)
2218 assertLookup(detector=3, timespan=timespan, expected=bias3a)
2219 # Decertify just bias2 over [t2, t4).
2220 # This should split a single certification row into two (and leave the
2221 # other existing row, for bias3a, alone).
2222 registry.decertify(
2223 collection, "bias", Timespan(t2, t4), dataIds=[dict(instrument="Cam1", detector=2)]
2224 )
2225 for timespan in allTimespans:
2226 assertLookup(detector=3, timespan=timespan, expected=bias3a)
2227 overlapsBefore = timespan.overlaps(Timespan(None, t2))
2228 overlapsAfter = timespan.overlaps(Timespan(t4, None))
2229 if overlapsBefore and overlapsAfter:
2230 expected = Ambiguous
2231 elif overlapsBefore or overlapsAfter:
2232 expected = bias2a
2233 else:
2234 expected = None
2235 assertLookup(detector=2, timespan=timespan, expected=expected)
2237 def testSkipCalibs(self):
2238 """Test how queries handle skipping of calibration collections."""
2239 registry = self.makeRegistry()
2240 self.loadData(registry, "base.yaml")
2241 self.loadData(registry, "datasets.yaml")
2243 coll_calib = "Cam1/calibs/default"
2244 registry.registerCollection(coll_calib, type=CollectionType.CALIBRATION)
2246 # Add all biases to the calibration collection.
2247 # Without this, the logic that prunes dataset subqueries based on
2248 # datasetType-collection summary information will fire before the logic
2249 # we want to test below. This is a good thing (it avoids the dreaded
2250 # NotImplementedError a bit more often) everywhere but here.
2251 registry.certify(coll_calib, registry.queryDatasets("bias", collections=...), Timespan(None, None))
2253 coll_list = [coll_calib, "imported_g", "imported_r"]
2254 chain = "Cam1/chain"
2255 registry.registerCollection(chain, type=CollectionType.CHAINED)
2256 registry.setCollectionChain(chain, coll_list)
2258 # explicit list will raise if findFirst=True or there are temporal
2259 # dimensions
2260 with self.assertRaises(NotImplementedError):
2261 registry.queryDatasets("bias", collections=coll_list, findFirst=True)
2262 with self.assertRaises(NotImplementedError):
2263 registry.queryDataIds(
2264 ["instrument", "detector", "exposure"], datasets="bias", collections=coll_list
2265 ).count()
2267 # chain will skip
2268 datasets = list(registry.queryDatasets("bias", collections=chain))
2269 self.assertGreater(len(datasets), 0)
2271 dataIds = list(registry.queryDataIds(["instrument", "detector"], datasets="bias", collections=chain))
2272 self.assertGreater(len(dataIds), 0)
2274 # glob will skip too
2275 datasets = list(registry.queryDatasets("bias", collections="*d*"))
2276 self.assertGreater(len(datasets), 0)
2278 # regular expression will skip too
2279 pattern = re.compile(".*")
2280 datasets = list(registry.queryDatasets("bias", collections=pattern))
2281 self.assertGreater(len(datasets), 0)
2283 # ellipsis should work as usual
2284 datasets = list(registry.queryDatasets("bias", collections=...))
2285 self.assertGreater(len(datasets), 0)
2287 # few tests with findFirst
2288 datasets = list(registry.queryDatasets("bias", collections=chain, findFirst=True))
2289 self.assertGreater(len(datasets), 0)
2291 def testIngestTimeQuery(self):
2293 registry = self.makeRegistry()
2294 self.loadData(registry, "base.yaml")
2295 dt0 = datetime.utcnow()
2296 self.loadData(registry, "datasets.yaml")
2297 dt1 = datetime.utcnow()
2299 datasets = list(registry.queryDatasets(..., collections=...))
2300 len0 = len(datasets)
2301 self.assertGreater(len0, 0)
2303 where = "ingest_date > T'2000-01-01'"
2304 datasets = list(registry.queryDatasets(..., collections=..., where=where))
2305 len1 = len(datasets)
2306 self.assertEqual(len0, len1)
2308 # no one will ever use this piece of software in 30 years
2309 where = "ingest_date > T'2050-01-01'"
2310 datasets = list(registry.queryDatasets(..., collections=..., where=where))
2311 len2 = len(datasets)
2312 self.assertEqual(len2, 0)
2314 # Check more exact timing to make sure there is no 37 seconds offset
2315 # (after fixing DM-30124). SQLite time precision is 1 second, make
2316 # sure that we don't test with higher precision.
2317 tests = [
2318 # format: (timestamp, operator, expected_len)
2319 (dt0 - timedelta(seconds=1), ">", len0),
2320 (dt0 - timedelta(seconds=1), "<", 0),
2321 (dt1 + timedelta(seconds=1), "<", len0),
2322 (dt1 + timedelta(seconds=1), ">", 0),
2323 ]
2324 for dt, op, expect_len in tests:
2325 dt_str = dt.isoformat(sep=" ")
2327 where = f"ingest_date {op} T'{dt_str}'"
2328 datasets = list(registry.queryDatasets(..., collections=..., where=where))
2329 self.assertEqual(len(datasets), expect_len)
2331 # same with bind using datetime or astropy Time
2332 where = f"ingest_date {op} ingest_time"
2333 datasets = list(
2334 registry.queryDatasets(..., collections=..., where=where, bind={"ingest_time": dt})
2335 )
2336 self.assertEqual(len(datasets), expect_len)
2338 dt_astropy = astropy.time.Time(dt, format="datetime")
2339 datasets = list(
2340 registry.queryDatasets(..., collections=..., where=where, bind={"ingest_time": dt_astropy})
2341 )
2342 self.assertEqual(len(datasets), expect_len)
2344 def testTimespanQueries(self):
2345 """Test query expressions involving timespans."""
2346 registry = self.makeRegistry()
2347 self.loadData(registry, "hsc-rc2-subset.yaml")
2348 # All exposures in the database; mapping from ID to timespan.
2349 visits = {record.id: record.timespan for record in registry.queryDimensionRecords("visit")}
2350 # Just those IDs, sorted (which is also temporal sorting, because HSC
2351 # exposure IDs are monotonically increasing).
2352 ids = sorted(visits.keys())
2353 self.assertGreater(len(ids), 20)
2354 # Pick some quasi-random indexes into `ids` to play with.
2355 i1 = int(len(ids) * 0.1)
2356 i2 = int(len(ids) * 0.3)
2357 i3 = int(len(ids) * 0.6)
2358 i4 = int(len(ids) * 0.8)
2359 # Extract some times from those: just before the beginning of i1 (which
2360 # should be after the end of the exposure before), exactly the
2361 # beginning of i2, just after the beginning of i3 (and before its end),
2362 # and the exact end of i4.
2363 t1 = visits[ids[i1]].begin - astropy.time.TimeDelta(1.0, format="sec")
2364 self.assertGreater(t1, visits[ids[i1 - 1]].end)
2365 t2 = visits[ids[i2]].begin
2366 t3 = visits[ids[i3]].begin + astropy.time.TimeDelta(1.0, format="sec")
2367 self.assertLess(t3, visits[ids[i3]].end)
2368 t4 = visits[ids[i4]].end
2369 # Make sure those are actually in order.
2370 self.assertEqual([t1, t2, t3, t4], sorted([t4, t3, t2, t1]))
2372 bind = {
2373 "t1": t1,
2374 "t2": t2,
2375 "t3": t3,
2376 "t4": t4,
2377 "ts23": Timespan(t2, t3),
2378 }
2380 def query(where):
2381 """Helper function that queries for visit data IDs and returns
2382 results as a sorted, deduplicated list of visit IDs.
2383 """
2384 return sorted(
2385 {
2386 dataId["visit"]
2387 for dataId in registry.queryDataIds("visit", instrument="HSC", bind=bind, where=where)
2388 }
2389 )
2391 # Try a bunch of timespan queries, mixing up the bounds themselves,
2392 # where they appear in the expression, and how we get the timespan into
2393 # the expression.
2395 # t1 is before the start of i1, so this should not include i1.
2396 self.assertEqual(ids[:i1], query("visit.timespan OVERLAPS (null, t1)"))
2397 # t2 is exactly at the start of i2, but ends are exclusive, so these
2398 # should not include i2.
2399 self.assertEqual(ids[i1:i2], query("(t1, t2) OVERLAPS visit.timespan"))
2400 self.assertEqual(ids[:i2], query("visit.timespan < (t2, t4)"))
2401 # t3 is in the middle of i3, so this should include i3.
2402 self.assertEqual(ids[i2 : i3 + 1], query("visit.timespan OVERLAPS ts23"))
2403 # This one should not include t3 by the same reasoning.
2404 self.assertEqual(ids[i3 + 1 :], query("visit.timespan > (t1, t3)"))
2405 # t4 is exactly at the end of i4, so this should include i4.
2406 self.assertEqual(ids[i3 : i4 + 1], query(f"visit.timespan OVERLAPS (T'{t3.tai.isot}', t4)"))
2407 # i4's upper bound of t4 is exclusive so this should not include t4.
2408 self.assertEqual(ids[i4 + 1 :], query("visit.timespan OVERLAPS (t4, NULL)"))
2410 # Now some timespan vs. time scalar queries.
2411 self.assertEqual(ids[:i2], query("visit.timespan < t2"))
2412 self.assertEqual(ids[:i2], query("t2 > visit.timespan"))
2413 self.assertEqual(ids[i3 + 1 :], query("visit.timespan > t3"))
2414 self.assertEqual(ids[i3 + 1 :], query("t3 < visit.timespan"))
2415 self.assertEqual(ids[i3 : i3 + 1], query("visit.timespan OVERLAPS t3"))
2416 self.assertEqual(ids[i3 : i3 + 1], query(f"T'{t3.tai.isot}' OVERLAPS visit.timespan"))
2418 # Empty timespans should not overlap anything.
2419 self.assertEqual([], query("visit.timespan OVERLAPS (t3, t2)"))
2421 def testCollectionSummaries(self):
2422 """Test recording and retrieval of collection summaries."""
2423 self.maxDiff = None
2424 registry = self.makeRegistry()
2425 # Importing datasets from yaml should go through the code path where
2426 # we update collection summaries as we insert datasets.
2427 self.loadData(registry, "base.yaml")
2428 self.loadData(registry, "datasets.yaml")
2429 flat = registry.getDatasetType("flat")
2430 expected1 = CollectionSummary()
2431 expected1.dataset_types.add(registry.getDatasetType("bias"))
2432 expected1.add_data_ids(
2433 flat, [DataCoordinate.standardize(instrument="Cam1", universe=registry.dimensions)]
2434 )
2435 self.assertEqual(registry.getCollectionSummary("imported_g"), expected1)
2436 self.assertEqual(registry.getCollectionSummary("imported_r"), expected1)
2437 # Create a chained collection with both of the imported runs; the
2438 # summary should be the same, because it's a union with itself.
2439 chain = "chain"
2440 registry.registerCollection(chain, CollectionType.CHAINED)
2441 registry.setCollectionChain(chain, ["imported_r", "imported_g"])
2442 self.assertEqual(registry.getCollectionSummary(chain), expected1)
2443 # Associate flats only into a tagged collection and a calibration
2444 # collection to check summaries of those.
2445 tag = "tag"
2446 registry.registerCollection(tag, CollectionType.TAGGED)
2447 registry.associate(tag, registry.queryDatasets(flat, collections="imported_g"))
2448 calibs = "calibs"
2449 registry.registerCollection(calibs, CollectionType.CALIBRATION)
2450 registry.certify(
2451 calibs, registry.queryDatasets(flat, collections="imported_g"), timespan=Timespan(None, None)
2452 )
2453 expected2 = expected1.copy()
2454 expected2.dataset_types.discard("bias")
2455 self.assertEqual(registry.getCollectionSummary(tag), expected2)
2456 self.assertEqual(registry.getCollectionSummary(calibs), expected2)
2457 # Explicitly calling Registry.refresh() should load those same
2458 # summaries, via a totally different code path.
2459 registry.refresh()
2460 self.assertEqual(registry.getCollectionSummary("imported_g"), expected1)
2461 self.assertEqual(registry.getCollectionSummary("imported_r"), expected1)
2462 self.assertEqual(registry.getCollectionSummary(tag), expected2)
2463 self.assertEqual(registry.getCollectionSummary(calibs), expected2)
2465 def testBindInQueryDatasets(self):
2466 """Test that the bind parameter is correctly forwarded in
2467 queryDatasets recursion.
2468 """
2469 registry = self.makeRegistry()
2470 # Importing datasets from yaml should go through the code path where
2471 # we update collection summaries as we insert datasets.
2472 self.loadData(registry, "base.yaml")
2473 self.loadData(registry, "datasets.yaml")
2474 self.assertEqual(
2475 set(registry.queryDatasets("flat", band="r", collections=...)),
2476 set(registry.queryDatasets("flat", where="band=my_band", bind={"my_band": "r"}, collections=...)),
2477 )
2479 def testQueryResultSummaries(self):
2480 """Test summary methods like `count`, `any`, and `explain_no_results`
2481 on `DataCoordinateQueryResults` and `DatasetQueryResults`
2482 """
2483 registry = self.makeRegistry()
2484 self.loadData(registry, "base.yaml")
2485 self.loadData(registry, "datasets.yaml")
2486 self.loadData(registry, "spatial.yaml")
2487 # Default test dataset has two collections, each with both flats and
2488 # biases. Add a new collection with only biases.
2489 registry.registerCollection("biases", CollectionType.TAGGED)
2490 registry.associate("biases", registry.queryDatasets("bias", collections=["imported_g"]))
2491 # First query yields two results, and involves no postprocessing.
2492 query1 = registry.queryDataIds(["physical_filter"], band="r")
2493 self.assertTrue(query1.any(execute=False, exact=False))
2494 self.assertTrue(query1.any(execute=True, exact=False))
2495 self.assertTrue(query1.any(execute=True, exact=True))
2496 self.assertEqual(query1.count(exact=False), 2)
2497 self.assertEqual(query1.count(exact=True), 2)
2498 self.assertFalse(list(query1.explain_no_results()))
2499 # Second query should yield no results, but this isn't detectable
2500 # unless we actually run a query.
2501 query2 = registry.queryDataIds(["physical_filter"], band="h")
2502 self.assertTrue(query2.any(execute=False, exact=False))
2503 self.assertFalse(query2.any(execute=True, exact=False))
2504 self.assertFalse(query2.any(execute=True, exact=True))
2505 self.assertEqual(query2.count(exact=False), 0)
2506 self.assertEqual(query2.count(exact=True), 0)
2507 self.assertFalse(list(query2.explain_no_results()))
2508 # These queries yield no results due to various problems that can be
2509 # spotted prior to execution, yielding helpful diagnostics.
2510 base_query = registry.queryDataIds(["detector", "physical_filter"])
2511 queries_and_snippets = [
2512 (
2513 # Dataset type name doesn't match any existing dataset types.
2514 registry.queryDatasets("nonexistent", collections=...),
2515 ["nonexistent"],
2516 ),
2517 (
2518 # Dataset type object isn't registered.
2519 registry.queryDatasets(
2520 DatasetType(
2521 "nonexistent",
2522 dimensions=["instrument"],
2523 universe=registry.dimensions,
2524 storageClass="Image",
2525 ),
2526 collections=...,
2527 ),
2528 ["nonexistent"],
2529 ),
2530 (
2531 # No datasets of this type in this collection.
2532 registry.queryDatasets("flat", collections=["biases"]),
2533 ["flat", "biases"],
2534 ),
2535 (
2536 # No datasets of this type in this collection.
2537 base_query.findDatasets("flat", collections=["biases"]),
2538 ["flat", "biases"],
2539 ),
2540 (
2541 # No collections matching at all.
2542 registry.queryDatasets("flat", collections=re.compile("potato.+")),
2543 ["potato"],
2544 ),
2545 ]
2546 # The behavior of these additional queries is slated to change in the
2547 # future, so we also check for deprecation warnings.
2548 with self.assertWarns(FutureWarning):
2549 queries_and_snippets.append(
2550 (
2551 # Dataset type name doesn't match any existing dataset
2552 # types.
2553 registry.queryDataIds(["detector"], datasets=["nonexistent"], collections=...),
2554 ["nonexistent"],
2555 )
2556 )
2557 with self.assertWarns(FutureWarning):
2558 queries_and_snippets.append(
2559 (
2560 # Dataset type name doesn't match any existing dataset
2561 # types.
2562 registry.queryDimensionRecords("detector", datasets=["nonexistent"], collections=...),
2563 ["nonexistent"],
2564 )
2565 )
2566 for query, snippets in queries_and_snippets:
2567 self.assertFalse(query.any(execute=False, exact=False))
2568 self.assertFalse(query.any(execute=True, exact=False))
2569 self.assertFalse(query.any(execute=True, exact=True))
2570 self.assertEqual(query.count(exact=False), 0)
2571 self.assertEqual(query.count(exact=True), 0)
2572 messages = list(query.explain_no_results())
2573 self.assertTrue(messages)
2574 # Want all expected snippets to appear in at least one message.
2575 self.assertTrue(
2576 any(
2577 all(snippet in message for snippet in snippets) for message in query.explain_no_results()
2578 ),
2579 messages,
2580 )
2582 # This query does yield results, but should also emit a warning because
2583 # dataset type patterns to queryDataIds is deprecated; just look for
2584 # the warning.
2585 with self.assertWarns(FutureWarning):
2586 registry.queryDataIds(["detector"], datasets=re.compile("^nonexistent$"), collections=...)
2588 # These queries yield no results due to problems that can be identified
2589 # by cheap follow-up queries, yielding helpful diagnostics.
2590 for query, snippets in [
2591 (
2592 # No records for one of the involved dimensions.
2593 registry.queryDataIds(["subfilter"]),
2594 ["dimension records", "subfilter"],
2595 ),
2596 (
2597 # No records for one of the involved dimensions.
2598 registry.queryDimensionRecords("subfilter"),
2599 ["dimension records", "subfilter"],
2600 ),
2601 ]:
2602 self.assertFalse(query.any(execute=True, exact=False))
2603 self.assertFalse(query.any(execute=True, exact=True))
2604 self.assertEqual(query.count(exact=True), 0)
2605 messages = list(query.explain_no_results())
2606 self.assertTrue(messages)
2607 # Want all expected snippets to appear in at least one message.
2608 self.assertTrue(
2609 any(
2610 all(snippet in message for snippet in snippets) for message in query.explain_no_results()
2611 ),
2612 messages,
2613 )
2615 # This query yields four overlaps in the database, but one is filtered
2616 # out in postprocessing. The count queries aren't accurate because
2617 # they don't account for duplication that happens due to an internal
2618 # join against commonSkyPix.
2619 query3 = registry.queryDataIds(["visit", "tract"], instrument="Cam1", skymap="SkyMap1")
2620 self.assertEqual(
2621 {
2622 DataCoordinate.standardize(
2623 instrument="Cam1",
2624 skymap="SkyMap1",
2625 visit=v,
2626 tract=t,
2627 universe=registry.dimensions,
2628 )
2629 for v, t in [(1, 0), (2, 0), (2, 1)]
2630 },
2631 set(query3),
2632 )
2633 self.assertTrue(query3.any(execute=False, exact=False))
2634 self.assertTrue(query3.any(execute=True, exact=False))
2635 self.assertTrue(query3.any(execute=True, exact=True))
2636 self.assertGreaterEqual(query3.count(exact=False), 4)
2637 self.assertGreaterEqual(query3.count(exact=True), 3)
2638 self.assertFalse(list(query3.explain_no_results()))
2639 # This query yields overlaps in the database, but all are filtered
2640 # out in postprocessing. The count queries again aren't very useful.
2641 # We have to use `where=` here to avoid an optimization that
2642 # (currently) skips the spatial postprocess-filtering because it
2643 # recognizes that no spatial join is necessary. That's not ideal, but
2644 # fixing it is out of scope for this ticket.
2645 query4 = registry.queryDataIds(
2646 ["visit", "tract"],
2647 instrument="Cam1",
2648 skymap="SkyMap1",
2649 where="visit=1 AND detector=1 AND tract=0 AND patch=4",
2650 )
2651 self.assertFalse(set(query4))
2652 self.assertTrue(query4.any(execute=False, exact=False))
2653 self.assertTrue(query4.any(execute=True, exact=False))
2654 self.assertFalse(query4.any(execute=True, exact=True))
2655 self.assertGreaterEqual(query4.count(exact=False), 1)
2656 self.assertEqual(query4.count(exact=True), 0)
2657 messages = list(query4.explain_no_results())
2658 self.assertTrue(messages)
2659 self.assertTrue(any("regions did not overlap" in message for message in messages))
2661 # And there are cases when queries make empty results but we do not
2662 # know how to explain that yet (could we just say miracles happen?)
2663 query5 = registry.queryDimensionRecords(
2664 "detector", where="detector.purpose = 'no-purpose'", instrument="Cam1"
2665 )
2666 self.assertEqual(query5.count(exact=True), 0)
2667 messages = list(query5.explain_no_results())
2668 self.assertFalse(messages)
2669 # This query should yield results from one dataset type but not the
2670 # other, which is not registered.
2671 query5 = registry.queryDatasets(["bias", "nonexistent"], collections=["biases"])
2672 self.assertTrue(set(query5))
2673 self.assertTrue(query5.any(execute=False, exact=False))
2674 self.assertTrue(query5.any(execute=True, exact=False))
2675 self.assertTrue(query5.any(execute=True, exact=True))
2676 self.assertGreaterEqual(query5.count(exact=False), 1)
2677 self.assertGreaterEqual(query5.count(exact=True), 1)
2678 self.assertFalse(messages, list(query5.explain_no_results()))
2680 def testQueryDataIdsOrderBy(self):
2681 """Test order_by and limit on result returned by queryDataIds()."""
2682 registry = self.makeRegistry()
2683 self.loadData(registry, "base.yaml")
2684 self.loadData(registry, "datasets.yaml")
2685 self.loadData(registry, "spatial.yaml")
2687 def do_query(dimensions=("visit", "tract"), datasets=None, collections=None):
2688 return registry.queryDataIds(
2689 dimensions, datasets=datasets, collections=collections, instrument="Cam1", skymap="SkyMap1"
2690 )
2692 Test = namedtuple(
2693 "testQueryDataIdsOrderByTest",
2694 ("order_by", "keys", "result", "limit", "datasets", "collections"),
2695 defaults=(None, None, None),
2696 )
2698 test_data = (
2699 Test("tract,visit", "tract,visit", ((0, 1), (0, 1), (0, 2), (0, 2), (1, 2), (1, 2))),
2700 Test("-tract,visit", "tract,visit", ((1, 2), (1, 2), (0, 1), (0, 1), (0, 2), (0, 2))),
2701 Test("tract,-visit", "tract,visit", ((0, 2), (0, 2), (0, 1), (0, 1), (1, 2), (1, 2))),
2702 Test("-tract,-visit", "tract,visit", ((1, 2), (1, 2), (0, 2), (0, 2), (0, 1), (0, 1))),
2703 Test(
2704 "tract.id,visit.id",
2705 "tract,visit",
2706 ((0, 1), (0, 1), (0, 2)),
2707 limit=(3,),
2708 ),
2709 Test("-tract,-visit", "tract,visit", ((1, 2), (1, 2), (0, 2)), limit=(3,)),
2710 Test("tract,visit", "tract,visit", ((0, 2), (1, 2), (1, 2)), limit=(3, 3)),
2711 Test("-tract,-visit", "tract,visit", ((0, 1),), limit=(3, 5)),
2712 Test(
2713 "tract,visit.exposure_time", "tract,visit", ((0, 2), (0, 2), (0, 1), (0, 1), (1, 2), (1, 2))
2714 ),
2715 Test(
2716 "-tract,-visit.exposure_time", "tract,visit", ((1, 2), (1, 2), (0, 1), (0, 1), (0, 2), (0, 2))
2717 ),
2718 Test("tract,-exposure_time", "tract,visit", ((0, 1), (0, 1), (0, 2), (0, 2), (1, 2), (1, 2))),
2719 Test("tract,visit.name", "tract,visit", ((0, 1), (0, 1), (0, 2), (0, 2), (1, 2), (1, 2))),
2720 Test(
2721 "tract,-timespan.begin,timespan.end",
2722 "tract,visit",
2723 ((0, 2), (0, 2), (0, 1), (0, 1), (1, 2), (1, 2)),
2724 ),
2725 Test("visit.day_obs,exposure.day_obs", "visit,exposure", ()),
2726 Test("visit.timespan.begin,-exposure.timespan.begin", "visit,exposure", ()),
2727 Test(
2728 "tract,detector",
2729 "tract,detector",
2730 ((0, 1), (0, 2), (0, 3), (0, 4), (1, 1), (1, 2), (1, 3), (1, 4)),
2731 datasets="flat",
2732 collections="imported_r",
2733 ),
2734 Test(
2735 "tract,detector.full_name",
2736 "tract,detector",
2737 ((0, 1), (0, 2), (0, 3), (0, 4), (1, 1), (1, 2), (1, 3), (1, 4)),
2738 datasets="flat",
2739 collections="imported_r",
2740 ),
2741 Test(
2742 "tract,detector.raft,detector.name_in_raft",
2743 "tract,detector",
2744 ((0, 1), (0, 2), (0, 3), (0, 4), (1, 1), (1, 2), (1, 3), (1, 4)),
2745 datasets="flat",
2746 collections="imported_r",
2747 ),
2748 )
2750 for test in test_data:
2751 order_by = test.order_by.split(",")
2752 keys = test.keys.split(",")
2753 query = do_query(keys, test.datasets, test.collections).order_by(*order_by)
2754 if test.limit is not None:
2755 query = query.limit(*test.limit)
2756 dataIds = tuple(tuple(dataId[k] for k in keys) for dataId in query)
2757 self.assertEqual(dataIds, test.result)
2759 # and materialize
2760 query = do_query(keys).order_by(*order_by)
2761 if test.limit is not None:
2762 query = query.limit(*test.limit)
2763 with query.materialize() as materialized:
2764 dataIds = tuple(tuple(dataId[k] for k in keys) for dataId in materialized)
2765 self.assertEqual(dataIds, test.result)
2767 # errors in a name
2768 for order_by in ("", "-"):
2769 with self.assertRaisesRegex(ValueError, "Empty dimension name in ORDER BY"):
2770 list(do_query().order_by(order_by))
2772 for order_by in ("undimension.name", "-undimension.name"):
2773 with self.assertRaisesRegex(ValueError, "Unknown dimension element name 'undimension'"):
2774 list(do_query().order_by(order_by))
2776 for order_by in ("attract", "-attract"):
2777 with self.assertRaisesRegex(ValueError, "Metadata 'attract' cannot be found in any dimension"):
2778 list(do_query().order_by(order_by))
2780 with self.assertRaisesRegex(ValueError, "Metadata 'exposure_time' exists in more than one dimension"):
2781 list(do_query(("exposure", "visit")).order_by("exposure_time"))
2783 with self.assertRaisesRegex(ValueError, "Timespan exists in more than one dimesion"):
2784 list(do_query(("exposure", "visit")).order_by("timespan.begin"))
2786 with self.assertRaisesRegex(
2787 ValueError, "Cannot find any temporal dimension element for 'timespan.begin'"
2788 ):
2789 list(do_query(("tract")).order_by("timespan.begin"))
2791 with self.assertRaisesRegex(ValueError, "Cannot use 'timespan.begin' with non-temporal element"):
2792 list(do_query(("tract")).order_by("tract.timespan.begin"))
2794 with self.assertRaisesRegex(ValueError, "Field 'name' does not exist in 'tract'."):
2795 list(do_query(("tract")).order_by("tract.name"))
2797 def testQueryDataIdsGovernorExceptions(self):
2798 """Test exceptions raised by queryDataIds() for incorrect governors."""
2799 registry = self.makeRegistry()
2800 self.loadData(registry, "base.yaml")
2801 self.loadData(registry, "datasets.yaml")
2802 self.loadData(registry, "spatial.yaml")
2804 def do_query(dimensions, dataId=None, where=None, bind=None, **kwargs):
2805 return registry.queryDataIds(dimensions, dataId=dataId, where=where, bind=bind, **kwargs)
2807 Test = namedtuple(
2808 "testQueryDataIdExceptionsTest",
2809 ("dimensions", "dataId", "where", "bind", "kwargs", "exception", "count"),
2810 defaults=(None, None, None, {}, None, 0),
2811 )
2813 test_data = (
2814 Test("tract,visit", count=6),
2815 Test("tract,visit", kwargs={"instrument": "Cam1", "skymap": "SkyMap1"}, count=6),
2816 Test(
2817 "tract,visit", kwargs={"instrument": "Cam2", "skymap": "SkyMap1"}, exception=DataIdValueError
2818 ),
2819 Test("tract,visit", dataId={"instrument": "Cam1", "skymap": "SkyMap1"}, count=6),
2820 Test(
2821 "tract,visit", dataId={"instrument": "Cam1", "skymap": "SkyMap2"}, exception=DataIdValueError
2822 ),
2823 Test("tract,visit", where="instrument='Cam1' AND skymap='SkyMap1'", count=6),
2824 Test("tract,visit", where="instrument='Cam1' AND skymap='SkyMap5'", exception=DataIdValueError),
2825 Test(
2826 "tract,visit",
2827 where="instrument=cam AND skymap=map",
2828 bind={"cam": "Cam1", "map": "SkyMap1"},
2829 count=6,
2830 ),
2831 Test(
2832 "tract,visit",
2833 where="instrument=cam AND skymap=map",
2834 bind={"cam": "Cam", "map": "SkyMap"},
2835 exception=DataIdValueError,
2836 ),
2837 )
2839 for test in test_data:
2840 dimensions = test.dimensions.split(",")
2841 if test.exception:
2842 with self.assertRaises(test.exception):
2843 do_query(dimensions, test.dataId, test.where, bind=test.bind, **test.kwargs).count()
2844 else:
2845 query = do_query(dimensions, test.dataId, test.where, bind=test.bind, **test.kwargs)
2846 self.assertEqual(query.count(), test.count)
2848 # and materialize
2849 if test.exception:
2850 with self.assertRaises(test.exception):
2851 query = do_query(dimensions, test.dataId, test.where, bind=test.bind, **test.kwargs)
2852 with query.materialize() as materialized:
2853 materialized.count()
2854 else:
2855 query = do_query(dimensions, test.dataId, test.where, bind=test.bind, **test.kwargs)
2856 with query.materialize() as materialized:
2857 self.assertEqual(materialized.count(), test.count)
2859 def testQueryDimensionRecordsOrderBy(self):
2860 """Test order_by and limit on result returned by
2861 queryDimensionRecords().
2862 """
2863 registry = self.makeRegistry()
2864 self.loadData(registry, "base.yaml")
2865 self.loadData(registry, "datasets.yaml")
2866 self.loadData(registry, "spatial.yaml")
2868 def do_query(element, datasets=None, collections=None):
2869 return registry.queryDimensionRecords(
2870 element, instrument="Cam1", datasets=datasets, collections=collections
2871 )
2873 query = do_query("detector")
2874 self.assertEqual(len(list(query)), 4)
2876 Test = namedtuple(
2877 "testQueryDataIdsOrderByTest",
2878 ("element", "order_by", "result", "limit", "datasets", "collections"),
2879 defaults=(None, None, None),
2880 )
2882 test_data = (
2883 Test("detector", "detector", (1, 2, 3, 4)),
2884 Test("detector", "-detector", (4, 3, 2, 1)),
2885 Test("detector", "raft,-name_in_raft", (2, 1, 4, 3)),
2886 Test("detector", "-detector.purpose", (4,), limit=(1,)),
2887 Test("detector", "-purpose,detector.raft,name_in_raft", (2, 3), limit=(2, 2)),
2888 Test("visit", "visit", (1, 2)),
2889 Test("visit", "-visit.id", (2, 1)),
2890 Test("visit", "zenith_angle", (1, 2)),
2891 Test("visit", "-visit.name", (2, 1)),
2892 Test("visit", "day_obs,-timespan.begin", (2, 1)),
2893 )
2895 for test in test_data:
2896 order_by = test.order_by.split(",")
2897 query = do_query(test.element).order_by(*order_by)
2898 if test.limit is not None:
2899 query = query.limit(*test.limit)
2900 dataIds = tuple(rec.id for rec in query)
2901 self.assertEqual(dataIds, test.result)
2903 # errors in a name
2904 for order_by in ("", "-"):
2905 with self.assertRaisesRegex(ValueError, "Empty dimension name in ORDER BY"):
2906 list(do_query("detector").order_by(order_by))
2908 for order_by in ("undimension.name", "-undimension.name"):
2909 with self.assertRaisesRegex(ValueError, "Element name mismatch: 'undimension'"):
2910 list(do_query("detector").order_by(order_by))
2912 for order_by in ("attract", "-attract"):
2913 with self.assertRaisesRegex(ValueError, "Field 'attract' does not exist in 'detector'."):
2914 list(do_query("detector").order_by(order_by))
2916 def testQueryDimensionRecordsExceptions(self):
2917 """Test exceptions raised by queryDimensionRecords()."""
2918 registry = self.makeRegistry()
2919 self.loadData(registry, "base.yaml")
2920 self.loadData(registry, "datasets.yaml")
2921 self.loadData(registry, "spatial.yaml")
2923 result = registry.queryDimensionRecords("detector")
2924 self.assertEqual(result.count(), 4)
2925 result = registry.queryDimensionRecords("detector", instrument="Cam1")
2926 self.assertEqual(result.count(), 4)
2927 result = registry.queryDimensionRecords("detector", dataId={"instrument": "Cam1"})
2928 self.assertEqual(result.count(), 4)
2929 result = registry.queryDimensionRecords("detector", where="instrument='Cam1'")
2930 self.assertEqual(result.count(), 4)
2931 result = registry.queryDimensionRecords("detector", where="instrument=instr", bind={"instr": "Cam1"})
2932 self.assertEqual(result.count(), 4)
2934 with self.assertRaisesRegex(DataIdValueError, "dimension instrument"):
2935 result = registry.queryDimensionRecords("detector", instrument="NotCam1")
2936 result.count()
2938 with self.assertRaisesRegex(DataIdValueError, "dimension instrument"):
2939 result = registry.queryDimensionRecords("detector", dataId={"instrument": "NotCam1"})
2940 result.count()
2942 with self.assertRaisesRegex(DataIdValueError, "Unknown values specified for governor dimension"):
2943 result = registry.queryDimensionRecords("detector", where="instrument='NotCam1'")
2944 result.count()
2946 with self.assertRaisesRegex(DataIdValueError, "Unknown values specified for governor dimension"):
2947 result = registry.queryDimensionRecords(
2948 "detector", where="instrument=instr", bind={"instr": "NotCam1"}
2949 )
2950 result.count()
2952 def testDatasetConstrainedDimensionRecordQueries(self):
2953 """Test that queryDimensionRecords works even when given a dataset
2954 constraint whose dimensions extend beyond the requested dimension
2955 element's.
2956 """
2957 registry = self.makeRegistry()
2958 self.loadData(registry, "base.yaml")
2959 self.loadData(registry, "datasets.yaml")
2960 # Query for physical_filter dimension records, using a dataset that
2961 # has both physical_filter and dataset dimensions.
2962 records = registry.queryDimensionRecords(
2963 "physical_filter",
2964 datasets=["flat"],
2965 collections="imported_r",
2966 )
2967 self.assertEqual({record.name for record in records}, {"Cam1-R1", "Cam1-R2"})
2968 # Trying to constrain by all dataset types is an error.
2969 with self.assertRaises(TypeError):
2970 list(registry.queryDimensionRecords("physical_filter", datasets=..., collections="imported_r"))
2972 def testSkyPixDatasetQueries(self):
2973 """Test that we can build queries involving skypix dimensions as long
2974 as a dataset type that uses those dimensions is included.
2975 """
2976 registry = self.makeRegistry()
2977 self.loadData(registry, "base.yaml")
2978 dataset_type = DatasetType(
2979 "a", dimensions=["htm7", "instrument"], universe=registry.dimensions, storageClass="int"
2980 )
2981 registry.registerDatasetType(dataset_type)
2982 run = "r"
2983 registry.registerRun(run)
2984 # First try queries where there are no datasets; the concern is whether
2985 # we can even build and execute these queries without raising, even
2986 # when "doomed" query shortcuts are in play.
2987 self.assertFalse(
2988 list(registry.queryDataIds(["htm7", "instrument"], datasets=dataset_type, collections=run))
2989 )
2990 self.assertFalse(list(registry.queryDatasets(dataset_type, collections=run)))
2991 # Now add a dataset and see that we can get it back.
2992 htm7 = registry.dimensions.skypix["htm"][7].pixelization
2993 data_id = registry.expandDataId(instrument="Cam1", htm7=htm7.universe()[0][0])
2994 (ref,) = registry.insertDatasets(dataset_type, [data_id], run=run)
2995 self.assertEqual(
2996 set(registry.queryDataIds(["htm7", "instrument"], datasets=dataset_type, collections=run)),
2997 {data_id},
2998 )
2999 self.assertEqual(set(registry.queryDatasets(dataset_type, collections=run)), {ref})
3001 def testDatasetIdFactory(self):
3002 """Simple test for DatasetIdFactory, mostly to catch potential changes
3003 in its API.
3004 """
3005 registry = self.makeRegistry()
3006 factory = registry.datasetIdFactory
3007 dataset_type = DatasetType(
3008 "datasetType",
3009 dimensions=["detector", "instrument"],
3010 universe=registry.dimensions,
3011 storageClass="int",
3012 )
3013 run = "run"
3014 data_id = DataCoordinate.standardize(instrument="Cam1", detector=1, graph=dataset_type.dimensions)
3016 datasetId = factory.makeDatasetId(run, dataset_type, data_id, DatasetIdGenEnum.UNIQUE)
3017 self.assertIsInstance(datasetId, uuid.UUID)
3018 self.assertEqual(datasetId.version, 4)
3020 datasetId = factory.makeDatasetId(run, dataset_type, data_id, DatasetIdGenEnum.DATAID_TYPE)
3021 self.assertIsInstance(datasetId, uuid.UUID)
3022 self.assertEqual(datasetId.version, 5)
3024 datasetId = factory.makeDatasetId(run, dataset_type, data_id, DatasetIdGenEnum.DATAID_TYPE_RUN)
3025 self.assertIsInstance(datasetId, uuid.UUID)
3026 self.assertEqual(datasetId.version, 5)