Coverage for python/lsst/daf/butler/transfers/_yaml.py: 12%

184 statements  

« prev     ^ index     » next       coverage.py v6.5.0, created at 2023-06-06 09:38 +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/>. 

21 

22from __future__ import annotations 

23 

24__all__ = ["YamlRepoExportBackend", "YamlRepoImportBackend"] 

25 

26import uuid 

27import warnings 

28from collections import defaultdict 

29from datetime import datetime 

30from typing import IO, TYPE_CHECKING, Any, Dict, Iterable, List, Mapping, Optional, Set, Tuple, Type 

31 

32import astropy.time 

33import yaml 

34from lsst.resources import ResourcePath 

35from lsst.utils import doImportType 

36from lsst.utils.iteration import ensure_iterable 

37 

38from ..core import ( 

39 DatasetAssociation, 

40 DatasetId, 

41 DatasetRef, 

42 DatasetType, 

43 Datastore, 

44 DimensionElement, 

45 DimensionRecord, 

46 DimensionUniverse, 

47 FileDataset, 

48 Timespan, 

49) 

50from ..core.named import NamedValueSet 

51from ..registry import CollectionType, Registry 

52from ..registry.interfaces import ChainedCollectionRecord, CollectionRecord, RunRecord, VersionTuple 

53from ..registry.versions import IncompatibleVersionError 

54from ._interfaces import RepoExportBackend, RepoImportBackend 

55 

56if TYPE_CHECKING: 

57 from lsst.resources import ResourcePathExpression 

58 

59EXPORT_FORMAT_VERSION = VersionTuple(1, 0, 2) 

60"""Export format version. 

61 

62Files with a different major version or a newer minor version cannot be read by 

63this version of the code. 

64""" 

65 

66 

67def _uuid_representer(dumper: yaml.Dumper, data: uuid.UUID) -> yaml.Node: 

68 """Generate YAML representation for UUID. 

69 

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

71 string representation of UUID. 

72 """ 

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

74 

75 

76def _uuid_constructor(loader: yaml.Loader, node: yaml.Node) -> Optional[uuid.UUID]: 

77 if node.value is not None: 

78 return uuid.UUID(hex=node.value) 

79 return None 

80 

81 

82yaml.Dumper.add_representer(uuid.UUID, _uuid_representer) 

83yaml.SafeLoader.add_constructor("!uuid", _uuid_constructor) 

84 

85 

86class YamlRepoExportBackend(RepoExportBackend): 

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

88 

89 Parameters 

90 ---------- 

91 stream 

92 A writeable file-like object. 

93 """ 

94 

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

96 self.stream = stream 

97 self.universe = universe 

98 self.data: List[Dict[str, Any]] = [] 

99 

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

101 # Docstring inherited from RepoExportBackend.saveDimensionData. 

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

103 self.data.append( 

104 { 

105 "type": "dimension", 

106 "element": element.name, 

107 "records": data_dicts, 

108 } 

109 ) 

110 

111 def saveCollection(self, record: CollectionRecord, doc: Optional[str]) -> None: 

112 # Docstring inherited from RepoExportBackend.saveCollections. 

113 data: Dict[str, Any] = { 

114 "type": "collection", 

115 "collection_type": record.type.name, 

116 "name": record.name, 

117 } 

118 if doc is not None: 

119 data["doc"] = doc 

120 if isinstance(record, RunRecord): 

121 data["host"] = record.host 

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

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

124 elif isinstance(record, ChainedCollectionRecord): 

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

126 self.data.append(data) 

127 

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

129 # Docstring inherited from RepoExportBackend.saveDatasets. 

130 self.data.append( 

131 { 

132 "type": "dataset_type", 

133 "name": datasetType.name, 

134 "dimensions": [d.name for d in datasetType.dimensions], 

135 "storage_class": datasetType.storageClass_name, 

136 "is_calibration": datasetType.isCalibration(), 

137 } 

138 ) 

139 self.data.append( 

140 { 

141 "type": "dataset", 

142 "dataset_type": datasetType.name, 

143 "run": run, 

144 "records": [ 

145 { 

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

147 "data_id": [ref.dataId.byName() for ref in sorted(dataset.refs)], 

148 "path": dataset.path, 

149 "formatter": dataset.formatter, 

150 # TODO: look up and save other collections 

151 } 

152 for dataset in datasets 

153 ], 

154 } 

155 ) 

156 

157 def saveDatasetAssociations( 

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

159 ) -> None: 

160 # Docstring inherited from RepoExportBackend.saveDatasetAssociations. 

161 if collectionType is CollectionType.TAGGED: 

162 self.data.append( 

163 { 

164 "type": "associations", 

165 "collection": collection, 

166 "collection_type": collectionType.name, 

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

168 } 

169 ) 

170 elif collectionType is CollectionType.CALIBRATION: 

171 idsByTimespan: Dict[Timespan, List[DatasetId]] = defaultdict(list) 

172 for association in associations: 

173 assert association.timespan is not None 

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

175 self.data.append( 

176 { 

177 "type": "associations", 

178 "collection": collection, 

179 "collection_type": collectionType.name, 

180 "validity_ranges": [ 

181 { 

182 "timespan": timespan, 

183 "dataset_ids": dataset_ids, 

184 } 

185 for timespan, dataset_ids in idsByTimespan.items() 

186 ], 

187 } 

188 ) 

189 

190 def finish(self) -> None: 

191 # Docstring inherited from RepoExportBackend. 

192 yaml.dump( 

193 { 

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

195 "version": str(EXPORT_FORMAT_VERSION), 

196 "universe_version": self.universe.version, 

197 "universe_namespace": self.universe.namespace, 

198 "data": self.data, 

199 }, 

200 stream=self.stream, 

201 sort_keys=False, 

202 ) 

203 

204 

205class YamlRepoImportBackend(RepoImportBackend): 

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

207 

208 Parameters 

209 ---------- 

210 stream 

211 A readable file-like object. 

212 registry : `Registry` 

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

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

215 and `load`. 

216 """ 

217 

218 def __init__(self, stream: IO, registry: Registry): 

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

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

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

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

223 wrapper = yaml.safe_load(stream) 

224 if wrapper["version"] == 0: 

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

226 # before we really tried to do versioning here. 

227 fileVersion = VersionTuple(1, 0, 0) 

228 else: 

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

230 if fileVersion.major != EXPORT_FORMAT_VERSION.major: 

231 raise IncompatibleVersionError( 

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

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

234 ) 

235 if fileVersion.minor > EXPORT_FORMAT_VERSION.minor: 

236 raise IncompatibleVersionError( 

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

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

239 ) 

240 self.runs: Dict[str, Tuple[Optional[str], Timespan]] = {} 

241 self.chains: Dict[str, List[str]] = {} 

242 self.collections: Dict[str, CollectionType] = {} 

243 self.collectionDocs: Dict[str, str] = {} 

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

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

246 self.tagAssociations: Dict[str, List[DatasetId]] = defaultdict(list) 

247 self.calibAssociations: Dict[str, Dict[Timespan, List[DatasetId]]] = defaultdict(dict) 

248 self.refsByFileId: Dict[DatasetId, DatasetRef] = {} 

249 self.registry: Registry = registry 

250 

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

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

253 

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

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

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

257 # silently dropped when visit is created but the 

258 # visit_system_membership must be constructed. 

259 migrate_visit_system = False 

260 if ( 

261 universe_version < 2 

262 and universe_namespace == "daf_butler" 

263 and "visit_system_membership" in self.registry.dimensions 

264 ): 

265 migrate_visit_system = True 

266 

267 datasetData = [] 

268 for data in wrapper["data"]: 

269 if data["type"] == "dimension": 

270 # convert all datetime values to astropy 

271 for record in data["records"]: 

272 for key in record: 

273 # Some older YAML files were produced with native 

274 # YAML support for datetime, we support reading that 

275 # data back. Newer conversion uses _AstropyTimeToYAML 

276 # class with special YAML tag. 

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

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

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

280 RecordClass: Type[DimensionRecord] = element.RecordClass 

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

282 

283 if data["element"] == "visit" and migrate_visit_system: 

284 # Must create the visit_system_membership records. 

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

286 RecordClass = element.RecordClass 

287 self.dimensions[element].extend( 

288 RecordClass(instrument=r["instrument"], visit_system=r["visit_system"], visit=r["id"]) 

289 for r in data["records"] 

290 ) 

291 

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

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

294 if collectionType is CollectionType.RUN: 

295 self.runs[data["name"]] = ( 

296 data["host"], 

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

298 ) 

299 elif collectionType is CollectionType.CHAINED: 

300 children = [] 

301 for child in data["children"]: 

302 if not isinstance(child, str): 

303 warnings.warn( 

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

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

306 ) 

307 # Old form with dataset type restrictions only, 

308 # supported for backwards compatibility. 

309 child, _ = child 

310 children.append(child) 

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

312 else: 

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

314 doc = data.get("doc") 

315 if doc is not None: 

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

317 elif data["type"] == "run": 

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

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

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

321 dimensions = data["dimensions"] 

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

323 dimensions.remove("visit_system") 

324 self.datasetTypes.add( 

325 DatasetType( 

326 data["name"], 

327 dimensions=dimensions, 

328 storageClass=data["storage_class"], 

329 universe=self.registry.dimensions, 

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

331 ) 

332 ) 

333 elif data["type"] == "dataset": 

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

335 # know about all dataset types first. 

336 datasetData.append(data) 

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

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

339 if collectionType is CollectionType.TAGGED: 

340 self.tagAssociations[data["collection"]].extend(data["dataset_ids"]) 

341 elif collectionType is CollectionType.CALIBRATION: 

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

343 for d in data["validity_ranges"]: 

344 if "timespan" in d: 

345 assocsByTimespan[d["timespan"]] = d["dataset_ids"] 

346 else: 

347 # TODO: this is for backward compatibility, should 

348 # be removed at some point. 

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

350 else: 

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

352 else: 

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

354 # key is (dataset type name, run) 

355 self.datasets: Mapping[Tuple[str, str], List[FileDataset]] = defaultdict(list) 

356 for data in datasetData: 

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

358 if datasetType is None: 

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

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

361 FileDataset( 

362 d.get("path"), 

363 [ 

364 DatasetRef(datasetType, dataId, run=data["run"], id=refid) 

365 for dataId, refid in zip( 

366 ensure_iterable(d["data_id"]), ensure_iterable(d["dataset_id"]) 

367 ) 

368 ], 

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

370 ) 

371 for d in data["records"] 

372 ) 

373 

374 def register(self) -> None: 

375 # Docstring inherited from RepoImportBackend.register. 

376 for datasetType in self.datasetTypes: 

377 self.registry.registerDatasetType(datasetType) 

378 for run in self.runs: 

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

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

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

382 self.registry.registerCollection( 

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

384 ) 

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

386 self.registry.registerCollection( 

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

388 ) 

389 self.registry.setCollectionChain(chain, children) 

390 

391 def load( 

392 self, 

393 datastore: Optional[Datastore], 

394 *, 

395 directory: ResourcePathExpression | None = None, 

396 transfer: Optional[str] = None, 

397 skip_dimensions: Optional[Set] = None, 

398 ) -> None: 

399 # Docstring inherited from RepoImportBackend.load. 

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

401 if skip_dimensions and element in skip_dimensions: 

402 continue 

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

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

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

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

407 # unacceptably slo. 

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

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

410 fileDatasets = [] 

411 for (datasetTypeName, run), records in self.datasets.items(): 

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

413 # remembering slices that associate them with the FileDataset 

414 # instances they came from. 

415 datasets: List[DatasetRef] = [] 

416 dataset_ids: List[DatasetId] = [] 

417 slices = [] 

418 for fileDataset in records: 

419 start = len(datasets) 

420 datasets.extend(fileDataset.refs) 

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

422 stop = len(datasets) 

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

424 # Insert all of those DatasetRefs at once. 

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

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

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

428 resolvedRefs = self.registry._importDatasets(datasets) 

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

430 # export file to the new DatasetRefs 

431 for fileId, ref in zip(dataset_ids, resolvedRefs): 

432 self.refsByFileId[fileId] = ref 

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

434 # resolved DatasetRefs to replace the unresolved ones as we 

435 # reorganize the collection information. 

436 for sliceForFileDataset, fileDataset in zip(slices, records): 

437 fileDataset.refs = resolvedRefs[sliceForFileDataset] 

438 if directory is not None: 

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

440 fileDatasets.append(fileDataset) 

441 # Ingest everything into the datastore at once. 

442 if datastore is not None and fileDatasets: 

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

444 # Associate datasets with tagged collections. 

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

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

447 # Associate datasets with calibration collections. 

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

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

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