Coverage for python / lsst / obs / base / ingest_tests.py: 26%
225 statements
« prev ^ index » next coverage.py v7.13.2, created at 2026-02-04 02:52 -0800
1# This file is part of obs_base.
2#
3# Developed for the LSST Data Management System.
4# This product includes software developed by the LSST Project
5# (https://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 <https://www.gnu.org/licenses/>.
22"""Base class for writing Gen3 raw data ingest tests."""
24__all__ = ("IngestTestBase",)
26import abc
27import os
28import shutil
29import tempfile
30import unittest
32import lsst.afw.cameraGeom
33import lsst.afw.cameraGeom.testUtils # For assertDetectorsEqual
34import lsst.obs.base
35from lsst.daf.butler import Butler, Registry
36from lsst.daf.butler.cli.butler import cli as butlerCli
37from lsst.daf.butler.cli.utils import LogCliRunner
38from lsst.pipe.base import Instrument
39from lsst.resources import ResourcePath
40from lsst.utils import doImportType
42from . import script
45class IngestTestBase(metaclass=abc.ABCMeta):
46 """Base class for tests of gen3 ingest. Subclass from this, then
47 `unittest.TestCase` to get a working test suite.
48 """
50 ingestDir = ""
51 """Root path to ingest files into. Typically `obs_package/tests/`; the
52 actual directory will be a tempdir under this one.
53 """
55 ingestDatasetTypeName = "raw"
56 """The DatasetType to use for the ingest.
58 If this is not an Exposure dataset type the tests will be more limited.
59 """
61 dataIds = []
62 """list of butler data IDs of files that should have been ingested."""
64 file = ""
65 """Full path to a file to ingest in tests."""
67 filterLabel = None
68 """The lsst.afw.image.FilterLabel that should be returned by the above
69 file."""
71 rawIngestTask = "lsst.obs.base.RawIngestTask"
72 """The task to use in the Ingest test."""
74 curatedCalibrationDatasetTypes = None
75 """List or tuple of Datasets types that should be present after calling
76 writeCuratedCalibrations. If `None` writeCuratedCalibrations will
77 not be called and the test will be skipped."""
79 defineVisitsTask = lsst.obs.base.DefineVisitsTask
80 """The task to use to define visits from groups of exposures.
81 This is ignored if ``visits`` is `None`.
82 """
84 visits = {}
85 """A dictionary mapping visit data IDs the lists of exposure data IDs that
86 are associated with them.
87 If this is empty (but not `None`), visit definition will be run but no
88 visits will be expected (e.g. because no exposures are on-sky
89 observations).
90 """
92 seed_config = None
93 """Location of a seed configuration file to pass to butler create.
95 Useful if additional formatters or storage classes need to be defined.
96 """
98 @property
99 @abc.abstractmethod
100 def instrumentClassName(self):
101 """The fully qualified instrument class name.
103 Returns
104 -------
105 `str`
106 The fully qualified instrument class name.
107 """
108 pass
110 @property
111 def instrumentClass(self):
112 """The instrument class."""
113 return doImportType(self.instrumentClassName)
115 @property
116 def instrumentName(self):
117 """The name of the instrument.
119 Returns
120 -------
121 `str`
122 The name of the instrument.
123 """
124 return self.instrumentClass.getName()
126 @classmethod
127 def setUpClass(cls):
128 # Use a temporary working directory.
129 cls.root = tempfile.mkdtemp(dir=cls.ingestDir)
130 cls._createRepo()
132 # Register the instrument and its static metadata.
133 cls._registerInstrument()
135 # Determine the relevant datastore root to use for testing.
136 with Butler.from_config(cls.root) as butler:
137 roots = butler.get_datastore_roots()
138 assert len(roots) == 1 # Only one datastore.
139 cls.datastore_root = list(roots.values())[0]
141 def setUp(self):
142 # Want a unique run name per test.
143 self.outputRun = "raw_ingest_" + self.id()
145 @classmethod
146 def tearDownClass(cls):
147 if os.path.exists(cls.root):
148 shutil.rmtree(cls.root, ignore_errors=True)
150 def verifyIngest(self, files=None, cli=False, fullCheck=False):
151 """
152 Test that RawIngestTask ingested the expected files.
154 Parameters
155 ----------
156 files : `list` [`str`], or None
157 List of files to be ingested, or None to use ``self.file``
158 fullCheck : `bool`, optional
159 If `True`, read the full raw dataset and check component
160 consistency. If `False` check that a component can be read
161 but do not read the entire raw exposure.
163 Notes
164 -----
165 Reading all the ingested test data can be expensive. The code paths
166 for reading the second raw are the same as reading the first so
167 we do not gain anything by doing full checks of everything.
168 Only read full pixel data for first dataset from file.
169 Don't even do that if we are requested not to by the caller.
170 This only really affects files that contain multiple datasets.
171 """
172 butler = Butler.from_config(self.root, run=self.outputRun)
173 self.enterContext(butler)
174 datasets = list(butler.registry.queryDatasets(self.ingestDatasetTypeName, collections=self.outputRun))
175 self.assertEqual(len(datasets), len(self.dataIds))
177 # Can check that the timespan in the day_obs matches the exposure
178 # record.
179 if "day_obs" in butler.dimensions:
180 days = {
181 (rec.instrument, rec.id): rec.timespan
182 for rec in butler.registry.queryDimensionRecords("day_obs")
183 }
185 exp_records = list(butler.registry.queryDimensionRecords("exposure"))
186 for exp in exp_records:
187 day_span = days[exp.instrument, exp.day_obs]
188 if day_span is not None:
189 self.assertTrue(
190 day_span.contains(exp.timespan.begin), f"Timespan mismatch of {exp} and {day_span}"
191 )
193 # Get the URI to the first dataset and check it is inside the
194 # datastore.
195 datasetUri = butler.getURI(datasets[0])
196 self.assertIsNotNone(datasetUri.relative_to(self.datastore_root))
198 # Get the relevant dataset type.
199 datasetType = butler.get_dataset_type(self.ingestDatasetTypeName)
201 for dataId in self.dataIds:
202 # For testing we only read the entire dataset the first time
203 # round if this is an Exposure. If it's not an Exposure
204 # we always read it completely but we don't read components
205 # because for an arbitrary dataset type we can't easily tell
206 # what component to test.
208 if not datasetType.storageClass.name.startswith("Exposure"):
209 exposure = butler.get(self.ingestDatasetTypeName, dataId)
210 # Could be anything so nothing to test by default
211 continue
213 # Check that we can read metadata from a raw.
214 metadata = butler.get(f"{self.ingestDatasetTypeName}.metadata", dataId)
215 if not fullCheck:
216 continue
217 fullCheck = False
218 exposure = butler.get(self.ingestDatasetTypeName, dataId)
220 # Comparing headers will not work directly because of header
221 # fix up provenance.
222 metadata_headers = metadata.toDict()
223 exposure_headers = exposure.getMetadata().toDict()
224 metadata_headers.pop("HIERARCH ASTRO METADATA FIX DATE", None)
225 exposure_headers.pop("HIERARCH ASTRO METADATA FIX DATE", None)
226 self.assertEqual(metadata_headers, exposure_headers)
228 # Since components follow a different code path we check that
229 # WCS match and also we check that at least the shape
230 # of the image is the same (rather than doing per-pixel equality)
231 wcs = butler.get(f"{self.ingestDatasetTypeName}.wcs", dataId)
232 self.assertEqual(wcs, exposure.getWcs())
234 rawImage = butler.get(f"{self.ingestDatasetTypeName}.image", dataId)
235 self.assertEqual(rawImage.getBBox(), exposure.getBBox())
237 # Check that the filter label got the correct band.
238 filterLabel = butler.get(f"{self.ingestDatasetTypeName}.filter", dataId)
239 self.assertEqual(filterLabel, self.filterLabel)
241 # Check that the exposure's Detector is the same as the component
242 # we would read (this is tricky for LSST, which modifies its
243 # detector at read time; for most other cameras it should be
244 # trivially satisfied.
245 detector = butler.get(f"{self.ingestDatasetTypeName}.detector", dataId)
246 self.assertDetectorsEqual(detector, exposure.getDetector(), compareTransforms=False)
248 self.checkRepo(files=files)
250 def checkRepo(self, files=None):
251 """Check the state of the repository after ingest.
253 This is an optional hook provided for subclasses; by default it does
254 nothing.
256 Parameters
257 ----------
258 files : `list` [`str`], or None
259 List of files to be ingested, or None to use ``self.file``
260 """
261 return
263 @classmethod
264 def _createRepo(cls):
265 """Use the Click `testing` module to call the butler command line api
266 to create a repository.
267 """
268 runner = LogCliRunner()
269 args = []
270 if cls.seed_config:
271 args.extend(["--seed-config", cls.seed_config])
272 result = runner.invoke(butlerCli, ["create", cls.root, *args])
273 # Classmethod so assertEqual does not work.
274 assert result.exit_code == 0, f"output: {result.output} exception: {result.exception}"
276 def _ingestRaws(self, transfer, file=None, skip_existing=True):
277 """Use the Click `testing` module to call the butler command line api
278 to ingest raws.
280 Parameters
281 ----------
282 transfer : `str`
283 The external data transfer type.
284 file : `str`
285 Path to a file to ingest instead of the default associated with
286 the object.
287 """
288 if file is None:
289 file = self.file
291 args = [
292 "ingest-raws",
293 self.root,
294 file,
295 "--output-run",
296 self.outputRun,
297 "--transfer",
298 transfer,
299 "--ingest-task",
300 self.rawIngestTask,
301 ]
302 if not skip_existing:
303 args.append("--no-skip-existing")
304 runner = LogCliRunner()
305 result = runner.invoke(butlerCli, args)
306 self.assertEqual(result.exit_code, 0, f"output: {result.output} exception: {result.exception}")
308 @classmethod
309 def _registerInstrument(cls):
310 """Use the Click `testing` module to call the butler command line api
311 to register the instrument.
312 """
313 runner = LogCliRunner()
314 result = runner.invoke(butlerCli, ["register-instrument", cls.root, cls.instrumentClassName])
315 # Classmethod so assertEqual does not work.
316 assert result.exit_code == 0, f"output: {result.output} exception: {result.exception}"
318 def _writeCuratedCalibrations(self):
319 """Use the Click `testing` module to call the butler command line api
320 to write curated calibrations.
321 """
322 runner = LogCliRunner()
323 result = runner.invoke(
324 butlerCli, ["write-curated-calibrations", self.root, self.instrumentName, "test"]
325 )
326 self.assertEqual(result.exit_code, 0, f"output: {result.output} exception: {result.exception}")
328 def testLink(self):
329 self._ingestRaws(transfer="link")
330 self.verifyIngest()
332 def testSymLink(self):
333 self._ingestRaws(transfer="symlink")
334 self.verifyIngest()
336 def testDirect(self):
337 self._ingestRaws(transfer="direct")
339 # Check that it really did have a URI outside of datastore.
340 srcUri = ResourcePath(self.file, forceAbsolute=True)
341 butler = Butler.from_config(self.root, run=self.outputRun)
342 self.enterContext(butler)
343 datasets = list(butler.registry.queryDatasets(self.ingestDatasetTypeName, collections=self.outputRun))
344 datastoreUri = butler.getURI(datasets[0])
345 self.assertEqual(datastoreUri, srcUri)
347 def testCopy(self):
348 self._ingestRaws(transfer="copy")
349 # Only test full read of raws for the copy test. No need to do it
350 # in the other tests since the formatter will be the same in all
351 # cases.
352 self.verifyIngest(fullCheck=True)
354 def testHardLink(self):
355 try:
356 self._ingestRaws(transfer="hardlink")
357 # Running ingest through the Click testing infrastructure causes
358 # the original exception indicating that we can't hard-link
359 # on this filesystem to be turned into a nonzero exit code, which
360 # then trips the test assertion.
361 except (AssertionError, PermissionError) as err:
362 raise unittest.SkipTest(
363 "Skipping hard-link test because input data is on a different filesystem."
364 ) from err
365 self.verifyIngest()
367 def testInPlace(self):
368 """Test that files already in the directory can be added to the
369 registry in-place.
370 """
371 butler = Butler.from_config(self.root, run=self.outputRun)
372 self.enterContext(butler)
374 # If the test uses an index file the index file needs to also
375 # appear in the datastore root along with the file to be ingested.
376 # In that scenario the file name being used for ingest can not
377 # be modified and must have the same name as found in the index
378 # file itself.
379 source_file_uri = ResourcePath(self.file)
380 index_file = source_file_uri.dirname().join("_index.json")
381 pathInStore = source_file_uri.basename()
382 if index_file.exists():
383 os.symlink(index_file.ospath, self.datastore_root.join("_index.json").ospath)
384 else:
385 # No index file so we are free to pick any name.
386 pathInStore = "prefix-" + pathInStore
388 # Create a symlink to the original file so that it looks like it
389 # is now inside the datastore.
390 newPath = self.datastore_root.join(pathInStore)
391 os.symlink(os.path.abspath(self.file), newPath.ospath)
393 # If there is a sidecar file it needs to be linked in as well
394 # since ingest code does not follow symlinks.
395 sidecar_uri = ResourcePath(source_file_uri).updatedExtension(".json")
396 if sidecar_uri.exists():
397 newSidecar = ResourcePath(newPath).updatedExtension(".json")
398 os.symlink(sidecar_uri.ospath, newSidecar.ospath)
400 # Run ingest with auto mode since that should automatically determine
401 # that an in-place ingest is happening.
402 self._ingestRaws(transfer="auto", file=newPath.ospath)
403 self.verifyIngest()
405 # Recreate a butler post-ingest (the earlier one won't see the
406 # ingested files).
407 butler = Butler.from_config(self.root, run=self.outputRun)
408 self.enterContext(butler)
410 # Check that the URI associated with this path is the right one.
411 uri = butler.getURI(self.ingestDatasetTypeName, self.dataIds[0])
412 self.assertEqual(uri.relative_to(self.datastore_root), pathInStore)
414 def testFailOnConflict(self):
415 """Re-ingesting the same data into the repository should fail."""
416 self._ingestRaws(transfer="symlink")
417 self._ingestRaws(transfer="symlink") # Default is to skip.
418 with self.assertRaises(AssertionError):
419 self._ingestRaws(transfer="symlink", skip_existing=False)
421 def testWriteCuratedCalibrations(self):
422 """Test that we can ingest the curated calibrations, and read them
423 with `loadCamera` both before and after.
424 """
425 if self.curatedCalibrationDatasetTypes is None:
426 raise unittest.SkipTest("Class requests disabling of writeCuratedCalibrations test")
428 butler = Butler.from_config(self.root, writeable=False)
429 self.enterContext(butler)
430 collection = self.instrumentClass().makeCalibrationCollectionName("test")
432 # Trying to load a camera with a data ID not known to the registry
433 # is an error, because we can't get any temporal information.
434 with self.assertRaises(LookupError):
435 lsst.obs.base.loadCamera(butler, {"exposure": 0}, collections=collection)
437 # Ingest raws in order to get some exposure records.
438 self._ingestRaws(transfer="auto")
440 # Load camera should returned an unversioned camera because there's
441 # nothing in the repo.
442 camera, isVersioned = lsst.obs.base.loadCamera(butler, self.dataIds[0], collections=collection)
443 self.assertFalse(isVersioned)
444 self.assertIsInstance(camera, lsst.afw.cameraGeom.Camera)
446 self._writeCuratedCalibrations()
448 # Make a new butler instance to make sure we don't have any stale
449 # caches (e.g. of DatasetTypes). Note that we didn't give
450 # _writeCuratedCalibrations the butler instance we had, because it's
451 # trying to test the CLI interface anyway.
452 butler = Butler.from_config(self.root, writeable=False)
453 self.enterContext(butler)
455 instrumentClass = self.instrumentClass()
456 calibration_names = instrumentClass.getCuratedCalibrationNames()
458 for datasetTypeName in self.curatedCalibrationDatasetTypes:
459 with self.subTest(dtype=datasetTypeName):
460 found = list(
461 butler.registry.queryDatasetAssociations(
462 datasetTypeName,
463 collections=collection,
464 )
465 )
466 self.assertGreater(len(found), 0, f"Checking {datasetTypeName}")
467 self.assertIn(datasetTypeName, calibration_names)
469 # Load camera should returned the versioned camera from the repo.
470 camera, isVersioned = lsst.obs.base.loadCamera(butler, self.dataIds[0], collections=collection)
471 self.assertTrue(isVersioned)
472 self.assertIsInstance(camera, lsst.afw.cameraGeom.Camera)
474 def testDefineVisits(self):
475 if self.visits is None:
476 self.skipTest("Expected visits were not defined.")
477 self._ingestRaws(transfer="link")
479 # Check that obscore table (if configured) has correct contents.
480 butler = Butler.from_config(self.root, run=self.outputRun)
481 self.enterContext(butler)
482 self._check_obscore(butler.registry, has_visits=False)
484 # Calling defineVisits tests the implementation of the butler command
485 # line interface "define-visits" subcommand. Functions in the script
486 # folder are generally considered protected and should not be used
487 # as public api.
488 script.defineVisits(
489 self.root,
490 config_file=None,
491 collections=self.outputRun,
492 instrument=self.instrumentName,
493 )
495 # Test that we got the visits we expected.
496 visits = butler.registry.queryDataIds(["visit"]).expanded().toSet()
497 self.assertCountEqual(visits, self.visits.keys())
498 instr = Instrument.from_string(self.instrumentName, butler.registry)
499 camera = instr.getCamera()
500 for foundVisit, (expectedVisit, expectedExposures) in zip(visits, self.visits.items(), strict=True):
501 # Test that this visit is associated with the expected exposures.
502 foundExposures = (
503 butler.registry.queryDataIds(["exposure"], dataId=expectedVisit).expanded().toSet()
504 )
505 self.assertCountEqual(foundExposures, expectedExposures)
506 # Test that we have a visit region, and that it contains all of the
507 # detector+visit regions.
508 self.assertIsNotNone(foundVisit.region)
509 detectorVisitDataIds = (
510 butler.registry.queryDataIds(["visit", "detector"], dataId=expectedVisit).expanded().toSet()
511 )
512 self.assertEqual(len(detectorVisitDataIds), len(camera))
513 for dataId in detectorVisitDataIds:
514 self.assertTrue(foundVisit.region.contains(dataId.region))
516 # Check obscore table again.
517 self._check_obscore(butler.registry, has_visits=True)
519 def _check_obscore(self, registry: Registry, has_visits: bool) -> None:
520 """Verify contents of obscore table."""
521 return