Coverage for python/lsst/daf/butler/transfers/_yaml.py: 14%

208 statements  

« prev     ^ index     » next       coverage.py v7.3.2, created at 2023-12-06 10:53 +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 software is dual licensed under the GNU General Public License and also 

10# under a 3-clause BSD license. Recipients may choose which of these licenses 

11# to use; please see the files gpl-3.0.txt and/or bsd_license.txt, 

12# respectively. If you choose the GPL option then the following text applies 

13# (but note that there is still no warranty even if you opt for BSD instead): 

14# 

15# This program is free software: you can redistribute it and/or modify 

16# it under the terms of the GNU General Public License as published by 

17# the Free Software Foundation, either version 3 of the License, or 

18# (at your option) any later version. 

19# 

20# This program is distributed in the hope that it will be useful, 

21# but WITHOUT ANY WARRANTY; without even the implied warranty of 

22# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 

23# GNU General Public License for more details. 

24# 

25# You should have received a copy of the GNU General Public License 

26# along with this program. If not, see <http://www.gnu.org/licenses/>. 

27 

28from __future__ import annotations 

29 

30__all__ = ["YamlRepoExportBackend", "YamlRepoImportBackend"] 

31 

32import uuid 

33import warnings 

34from collections import UserDict, defaultdict 

35from collections.abc import Iterable, Mapping 

36from datetime import datetime 

37from typing import IO, TYPE_CHECKING, Any 

38 

39import astropy.time 

40import yaml 

41from lsst.resources import ResourcePath 

42from lsst.utils import doImportType 

43from lsst.utils.introspection import find_outside_stacklevel 

44from lsst.utils.iteration import ensure_iterable 

45 

46from .._dataset_association import DatasetAssociation 

47from .._dataset_ref import DatasetId, DatasetRef 

48from .._dataset_type import DatasetType 

49from .._file_dataset import FileDataset 

50from .._named import NamedValueSet 

51from .._timespan import Timespan 

52from ..datastore import Datastore 

53from ..dimensions import DimensionElement, DimensionRecord, DimensionUniverse 

54from ..registry import CollectionType 

55from ..registry.interfaces import ChainedCollectionRecord, CollectionRecord, RunRecord, VersionTuple 

56from ..registry.sql_registry import SqlRegistry 

57from ..registry.versions import IncompatibleVersionError 

58from ._interfaces import RepoExportBackend, RepoImportBackend 

59 

60if TYPE_CHECKING: 

61 from lsst.resources import ResourcePathExpression 

62 

63EXPORT_FORMAT_VERSION = VersionTuple(1, 0, 2) 

64"""Export format version. 

65 

66Files with a different major version or a newer minor version cannot be read by 

67this version of the code. 

68""" 

69 

70 

71class _RefMapper(UserDict[int, uuid.UUID]): 

72 """Create a local dict subclass which creates new deterministic UUID for 

73 missing keys. 

74 """ 

75 

76 _namespace = uuid.UUID("4d4851f4-2890-4d41-8779-5f38a3f5062b") 

77 

78 def __missing__(self, key: int) -> uuid.UUID: 

79 newUUID = uuid.uuid3(namespace=self._namespace, name=str(key)) 

80 self[key] = newUUID 

81 return newUUID 

82 

83 

84_refIntId2UUID = _RefMapper() 

85 

86 

87def _uuid_representer(dumper: yaml.Dumper, data: uuid.UUID) -> yaml.Node: 

88 """Generate YAML representation for UUID. 

89 

90 This produces a scalar node with a tag "!uuid" and value being a regular 

91 string representation of UUID. 

92 """ 

93 return dumper.represent_scalar("!uuid", str(data)) 

94 

95 

96def _uuid_constructor(loader: yaml.Loader, node: yaml.Node) -> uuid.UUID | None: 

97 if node.value is not None: 

98 return uuid.UUID(hex=node.value) 

99 return None 

100 

101 

102yaml.Dumper.add_representer(uuid.UUID, _uuid_representer) 

103yaml.SafeLoader.add_constructor("!uuid", _uuid_constructor) 

104 

105 

106class YamlRepoExportBackend(RepoExportBackend): 

107 """A repository export implementation that saves to a YAML file. 

108 

109 Parameters 

110 ---------- 

111 stream 

112 A writeable file-like object. 

113 """ 

114 

115 def __init__(self, stream: IO, universe: DimensionUniverse): 

116 self.stream = stream 

117 self.universe = universe 

118 self.data: list[dict[str, Any]] = [] 

119 

120 def saveDimensionData(self, element: DimensionElement, *data: DimensionRecord) -> None: 

121 # Docstring inherited from RepoExportBackend.saveDimensionData. 

122 data_dicts = [record.toDict(splitTimespan=True) for record in data] 

123 self.data.append( 

124 { 

125 "type": "dimension", 

126 "element": element.name, 

127 "records": data_dicts, 

128 } 

129 ) 

130 

131 def saveCollection(self, record: CollectionRecord, doc: str | None) -> None: 

132 # Docstring inherited from RepoExportBackend.saveCollections. 

133 data: dict[str, Any] = { 

134 "type": "collection", 

135 "collection_type": record.type.name, 

136 "name": record.name, 

137 } 

138 if doc is not None: 

139 data["doc"] = doc 

140 if isinstance(record, RunRecord): 

141 data["host"] = record.host 

142 data["timespan_begin"] = record.timespan.begin 

143 data["timespan_end"] = record.timespan.end 

144 elif isinstance(record, ChainedCollectionRecord): 

145 data["children"] = list(record.children) 

146 self.data.append(data) 

147 

148 def saveDatasets(self, datasetType: DatasetType, run: str, *datasets: FileDataset) -> None: 

149 # Docstring inherited from RepoExportBackend.saveDatasets. 

150 self.data.append( 

151 { 

152 "type": "dataset_type", 

153 "name": datasetType.name, 

154 "dimensions": list(datasetType.dimensions.names), 

155 "storage_class": datasetType.storageClass_name, 

156 "is_calibration": datasetType.isCalibration(), 

157 } 

158 ) 

159 self.data.append( 

160 { 

161 "type": "dataset", 

162 "dataset_type": datasetType.name, 

163 "run": run, 

164 "records": [ 

165 { 

166 "dataset_id": [ref.id for ref in sorted(dataset.refs)], 

167 "data_id": [dict(ref.dataId.required) for ref in sorted(dataset.refs)], 

168 "path": dataset.path, 

169 "formatter": dataset.formatter, 

170 # TODO: look up and save other collections 

171 } 

172 for dataset in datasets 

173 ], 

174 } 

175 ) 

176 

177 def saveDatasetAssociations( 

178 self, collection: str, collectionType: CollectionType, associations: Iterable[DatasetAssociation] 

179 ) -> None: 

180 # Docstring inherited from RepoExportBackend.saveDatasetAssociations. 

181 if collectionType is CollectionType.TAGGED: 

182 self.data.append( 

183 { 

184 "type": "associations", 

185 "collection": collection, 

186 "collection_type": collectionType.name, 

187 "dataset_ids": [assoc.ref.id for assoc in associations], 

188 } 

189 ) 

190 elif collectionType is CollectionType.CALIBRATION: 

191 idsByTimespan: dict[Timespan, list[DatasetId]] = defaultdict(list) 

192 for association in associations: 

193 assert association.timespan is not None 

194 idsByTimespan[association.timespan].append(association.ref.id) 

195 self.data.append( 

196 { 

197 "type": "associations", 

198 "collection": collection, 

199 "collection_type": collectionType.name, 

200 "validity_ranges": [ 

201 { 

202 "timespan": timespan, 

203 "dataset_ids": dataset_ids, 

204 } 

205 for timespan, dataset_ids in idsByTimespan.items() 

206 ], 

207 } 

208 ) 

209 

210 def finish(self) -> None: 

211 # Docstring inherited from RepoExportBackend. 

212 yaml.dump( 

213 { 

214 "description": "Butler Data Repository Export", 

215 "version": str(EXPORT_FORMAT_VERSION), 

216 "universe_version": self.universe.version, 

217 "universe_namespace": self.universe.namespace, 

218 "data": self.data, 

219 }, 

220 stream=self.stream, 

221 sort_keys=False, 

222 ) 

223 

224 

225class YamlRepoImportBackend(RepoImportBackend): 

226 """A repository import implementation that reads from a YAML file. 

227 

228 Parameters 

229 ---------- 

230 stream 

231 A readable file-like object. 

232 registry : `SqlRegistry` 

233 The registry datasets will be imported into. Only used to retreive 

234 dataset types during construction; all write happen in `register` 

235 and `load`. 

236 """ 

237 

238 def __init__(self, stream: IO, registry: SqlRegistry): 

239 # We read the file fully and convert its contents to Python objects 

240 # instead of loading incrementally so we can spot some problems early; 

241 # because `register` can't be put inside a transaction, we'd rather not 

242 # run that at all if there's going to be problem later in `load`. 

243 wrapper = yaml.safe_load(stream) 

244 if wrapper["version"] == 0: 

245 # Grandfather-in 'version: 0' -> 1.0.0, which is what we wrote 

246 # before we really tried to do versioning here. 

247 fileVersion = VersionTuple(1, 0, 0) 

248 else: 

249 fileVersion = VersionTuple.fromString(wrapper["version"]) 

250 if fileVersion.major != EXPORT_FORMAT_VERSION.major: 

251 raise IncompatibleVersionError( 

252 f"Cannot read repository export file with version={fileVersion} " 

253 f"({EXPORT_FORMAT_VERSION.major}.x.x required)." 

254 ) 

255 if fileVersion.minor > EXPORT_FORMAT_VERSION.minor: 

256 raise IncompatibleVersionError( 

257 f"Cannot read repository export file with version={fileVersion} " 

258 f"< {EXPORT_FORMAT_VERSION.major}.{EXPORT_FORMAT_VERSION.minor}.x required." 

259 ) 

260 self.runs: dict[str, tuple[str | None, Timespan]] = {} 

261 self.chains: dict[str, list[str]] = {} 

262 self.collections: dict[str, CollectionType] = {} 

263 self.collectionDocs: dict[str, str] = {} 

264 self.datasetTypes: NamedValueSet[DatasetType] = NamedValueSet() 

265 self.dimensions: Mapping[DimensionElement, list[DimensionRecord]] = defaultdict(list) 

266 self.tagAssociations: dict[str, list[DatasetId]] = defaultdict(list) 

267 self.calibAssociations: dict[str, dict[Timespan, list[DatasetId]]] = defaultdict(dict) 

268 self.refsByFileId: dict[DatasetId, DatasetRef] = {} 

269 self.registry: SqlRegistry = registry 

270 

271 universe_version = wrapper.get("universe_version", 0) 

272 universe_namespace = wrapper.get("universe_namespace", "daf_butler") 

273 

274 # If this is data exported before the reorganization of visits 

275 # and visit systems and that new schema is in use, some filtering 

276 # will be needed. The entry in the visit dimension record will be 

277 # silently dropped when visit is created but the 

278 # visit_system_membership must be constructed. 

279 migrate_visit_system = False 

280 if ( 

281 universe_version < 2 

282 and universe_namespace == "daf_butler" 

283 and "visit_system_membership" in self.registry.dimensions 

284 ): 

285 migrate_visit_system = True 

286 

287 # Drop "seeing" from visits in files older than version 1. 

288 migrate_visit_seeing = False 

289 if ( 

290 universe_version < 1 

291 and universe_namespace == "daf_butler" 

292 and "visit" in self.registry.dimensions 

293 and "seeing" not in self.registry.dimensions["visit"].metadata 

294 ): 

295 migrate_visit_seeing = True 

296 

297 datasetData = [] 

298 RecordClass: type[DimensionRecord] 

299 for data in wrapper["data"]: 

300 if data["type"] == "dimension": 

301 # convert all datetime values to astropy 

302 for record in data["records"]: 

303 for key in record: 

304 # Some older YAML files were produced with native 

305 # YAML support for datetime, we support reading that 

306 # data back. Newer conversion uses _AstropyTimeToYAML 

307 # class with special YAML tag. 

308 if isinstance(record[key], datetime): 

309 record[key] = astropy.time.Time(record[key], scale="utc") 

310 

311 if data["element"] == "visit": 

312 if migrate_visit_system: 

313 # Must create the visit_system_membership records. 

314 # But first create empty list for visits since other 

315 # logic in this file depends on self.dimensions being 

316 # populated in an order consisteny with primary keys. 

317 self.dimensions[self.registry.dimensions["visit"]] = [] 

318 element = self.registry.dimensions["visit_system_membership"] 

319 RecordClass = element.RecordClass 

320 self.dimensions[element].extend( 

321 RecordClass( 

322 instrument=r["instrument"], visit_system=r.pop("visit_system"), visit=r["id"] 

323 ) 

324 for r in data["records"] 

325 ) 

326 if migrate_visit_seeing: 

327 for record in data["records"]: 

328 record.pop("seeing", None) 

329 

330 element = self.registry.dimensions[data["element"]] 

331 RecordClass = element.RecordClass 

332 self.dimensions[element].extend(RecordClass(**r) for r in data["records"]) 

333 

334 elif data["type"] == "collection": 

335 collectionType = CollectionType.from_name(data["collection_type"]) 

336 if collectionType is CollectionType.RUN: 

337 self.runs[data["name"]] = ( 

338 data["host"], 

339 Timespan(begin=data["timespan_begin"], end=data["timespan_end"]), 

340 ) 

341 elif collectionType is CollectionType.CHAINED: 

342 children = [] 

343 for child in data["children"]: 

344 if not isinstance(child, str): 

345 warnings.warn( 

346 f"CHAINED collection {data['name']} includes restrictions on child " 

347 "collection searches, which are no longer suppored and will be ignored.", 

348 stacklevel=find_outside_stacklevel("lsst.daf.butler"), 

349 ) 

350 # Old form with dataset type restrictions only, 

351 # supported for backwards compatibility. 

352 child, _ = child 

353 children.append(child) 

354 self.chains[data["name"]] = children 

355 else: 

356 self.collections[data["name"]] = collectionType 

357 doc = data.get("doc") 

358 if doc is not None: 

359 self.collectionDocs[data["name"]] = doc 

360 elif data["type"] == "run": 

361 # Also support old form of saving a run with no extra info. 

362 self.runs[data["name"]] = (None, Timespan(None, None)) 

363 elif data["type"] == "dataset_type": 

364 dimensions = data["dimensions"] 

365 if migrate_visit_system and "visit" in dimensions and "visit_system" in dimensions: 

366 dimensions.remove("visit_system") 

367 self.datasetTypes.add( 

368 DatasetType( 

369 data["name"], 

370 dimensions=dimensions, 

371 storageClass=data["storage_class"], 

372 universe=self.registry.dimensions, 

373 isCalibration=data.get("is_calibration", False), 

374 ) 

375 ) 

376 elif data["type"] == "dataset": 

377 # Save raw dataset data for a second loop, so we can ensure we 

378 # know about all dataset types first. 

379 datasetData.append(data) 

380 elif data["type"] == "associations": 

381 collectionType = CollectionType.from_name(data["collection_type"]) 

382 if collectionType is CollectionType.TAGGED: 

383 self.tagAssociations[data["collection"]].extend( 

384 [x if not isinstance(x, int) else _refIntId2UUID[x] for x in data["dataset_ids"]] 

385 ) 

386 elif collectionType is CollectionType.CALIBRATION: 

387 assocsByTimespan = self.calibAssociations[data["collection"]] 

388 for d in data["validity_ranges"]: 

389 if "timespan" in d: 

390 assocsByTimespan[d["timespan"]] = [ 

391 x if not isinstance(x, int) else _refIntId2UUID[x] for x in d["dataset_ids"] 

392 ] 

393 else: 

394 # TODO: this is for backward compatibility, should 

395 # be removed at some point. 

396 assocsByTimespan[Timespan(begin=d["begin"], end=d["end"])] = [ 

397 x if not isinstance(x, int) else _refIntId2UUID[x] for x in d["dataset_ids"] 

398 ] 

399 else: 

400 raise ValueError(f"Unexpected calibration type for association: {collectionType.name}.") 

401 else: 

402 raise ValueError(f"Unexpected dictionary type: {data['type']}.") 

403 # key is (dataset type name, run) 

404 self.datasets: Mapping[tuple[str, str], list[FileDataset]] = defaultdict(list) 

405 for data in datasetData: 

406 datasetType = self.datasetTypes.get(data["dataset_type"]) 

407 if datasetType is None: 

408 datasetType = self.registry.getDatasetType(data["dataset_type"]) 

409 self.datasets[data["dataset_type"], data["run"]].extend( 

410 FileDataset( 

411 d.get("path"), 

412 [ 

413 DatasetRef( 

414 datasetType, 

415 dataId, 

416 run=data["run"], 

417 id=refid if not isinstance(refid, int) else _refIntId2UUID[refid], 

418 ) 

419 for dataId, refid in zip( 

420 ensure_iterable(d["data_id"]), ensure_iterable(d["dataset_id"]), strict=True 

421 ) 

422 ], 

423 formatter=doImportType(d.get("formatter")) if "formatter" in d else None, 

424 ) 

425 for d in data["records"] 

426 ) 

427 

428 def register(self) -> None: 

429 # Docstring inherited from RepoImportBackend.register. 

430 for datasetType in self.datasetTypes: 

431 self.registry.registerDatasetType(datasetType) 

432 for run in self.runs: 

433 self.registry.registerRun(run, doc=self.collectionDocs.get(run)) 

434 # No way to add extra run info to registry yet. 

435 for collection, collection_type in self.collections.items(): 

436 self.registry.registerCollection( 

437 collection, collection_type, doc=self.collectionDocs.get(collection) 

438 ) 

439 for chain, children in self.chains.items(): 

440 self.registry.registerCollection( 

441 chain, CollectionType.CHAINED, doc=self.collectionDocs.get(chain) 

442 ) 

443 self.registry.setCollectionChain(chain, children) 

444 

445 def load( 

446 self, 

447 datastore: Datastore | None, 

448 *, 

449 directory: ResourcePathExpression | None = None, 

450 transfer: str | None = None, 

451 skip_dimensions: set | None = None, 

452 ) -> None: 

453 # Docstring inherited from RepoImportBackend.load. 

454 for element, dimensionRecords in self.dimensions.items(): 

455 if skip_dimensions and element in skip_dimensions: 

456 continue 

457 # Using skip_existing=True here assumes that the records in the 

458 # database are either equivalent or at least preferable to the ones 

459 # being imported. It'd be ideal to check that, but that would mean 

460 # using syncDimensionData, which is not vectorized and is hence 

461 # unacceptably slo. 

462 self.registry.insertDimensionData(element, *dimensionRecords, skip_existing=True) 

463 # FileDatasets to ingest into the datastore (in bulk): 

464 fileDatasets = [] 

465 for records in self.datasets.values(): 

466 # Make a big flattened list of all data IDs and dataset_ids, while 

467 # remembering slices that associate them with the FileDataset 

468 # instances they came from. 

469 datasets: list[DatasetRef] = [] 

470 dataset_ids: list[DatasetId] = [] 

471 slices = [] 

472 for fileDataset in records: 

473 start = len(datasets) 

474 datasets.extend(fileDataset.refs) 

475 dataset_ids.extend(ref.id for ref in fileDataset.refs) 

476 stop = len(datasets) 

477 slices.append(slice(start, stop)) 

478 # Insert all of those DatasetRefs at once. 

479 # For now, we ignore the dataset_id we pulled from the file 

480 # and just insert without one to get a new autoincrement value. 

481 # Eventually (once we have origin in IDs) we'll preserve them. 

482 resolvedRefs = self.registry._importDatasets(datasets) 

483 # Populate our dictionary that maps int dataset_id values from the 

484 # export file to the new DatasetRefs 

485 for fileId, ref in zip(dataset_ids, resolvedRefs, strict=True): 

486 self.refsByFileId[fileId] = ref 

487 # Now iterate over the original records, and install the new 

488 # resolved DatasetRefs to replace the unresolved ones as we 

489 # reorganize the collection information. 

490 for sliceForFileDataset, fileDataset in zip(slices, records, strict=True): 

491 fileDataset.refs = resolvedRefs[sliceForFileDataset] 

492 if directory is not None: 

493 fileDataset.path = ResourcePath(directory, forceDirectory=True).join(fileDataset.path) 

494 fileDatasets.append(fileDataset) 

495 # Ingest everything into the datastore at once. 

496 if datastore is not None and fileDatasets: 

497 datastore.ingest(*fileDatasets, transfer=transfer) 

498 # Associate datasets with tagged collections. 

499 for collection, dataset_ids in self.tagAssociations.items(): 

500 self.registry.associate(collection, [self.refsByFileId[i] for i in dataset_ids]) 

501 # Associate datasets with calibration collections. 

502 for collection, idsByTimespan in self.calibAssociations.items(): 

503 for timespan, dataset_ids in idsByTimespan.items(): 

504 self.registry.certify(collection, [self.refsByFileId[i] for i in dataset_ids], timespan)