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

115 statements  

« prev     ^ index     » next       coverage.py v7.3.2, created at 2023-10-27 09:44 +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__ = ["RepoExportContext"] 

31 

32from collections import defaultdict 

33from collections.abc import Callable, Iterable, Set 

34from typing import TYPE_CHECKING 

35 

36from .._dataset_association import DatasetAssociation 

37from .._dataset_ref import DatasetId, DatasetRef 

38from .._dataset_type import DatasetType 

39from .._file_dataset import FileDataset 

40from ..datastore import Datastore 

41from ..dimensions import DataCoordinate, DimensionElement, DimensionRecord 

42from ..registry import CollectionType 

43from ..registry.interfaces import ChainedCollectionRecord, CollectionRecord 

44 

45if TYPE_CHECKING: 

46 from lsst.resources import ResourcePathExpression 

47 

48 from ..registry.sql_registry import SqlRegistry 

49 from ._interfaces import RepoExportBackend 

50 

51 

52class RepoExportContext: 

53 """Public interface for exporting a subset of a data repository. 

54 

55 Instances of this class are obtained by calling `Butler.export` as the 

56 value returned by that context manager:: 

57 

58 with butler.export(filename="export.yaml") as export: 

59 export.saveDataIds(...) 

60 export.saveDatasets(...) 

61 

62 Parameters 

63 ---------- 

64 registry : `SqlRegistry` 

65 Registry to export from. 

66 datastore : `Datastore` 

67 Datastore to export from. 

68 backend : `RepoExportBackend` 

69 Implementation class for a particular export file format. 

70 directory : `~lsst.resources.ResourcePathExpression`, optional 

71 Directory to pass to `Datastore.export`. Can be `None` to use 

72 the current working directory. 

73 transfer : `str`, optional 

74 Transfer mode to pass to `Datastore.export`. 

75 """ 

76 

77 def __init__( 

78 self, 

79 registry: SqlRegistry, 

80 datastore: Datastore, 

81 backend: RepoExportBackend, 

82 *, 

83 directory: ResourcePathExpression | None = None, 

84 transfer: str | None = None, 

85 ): 

86 self._registry = registry 

87 self._datastore = datastore 

88 self._backend = backend 

89 self._directory = directory 

90 self._transfer = transfer 

91 self._records: dict[DimensionElement, dict[DataCoordinate, DimensionRecord]] = defaultdict(dict) 

92 self._dataset_ids: set[DatasetId] = set() 

93 self._datasets: dict[DatasetType, dict[str, list[FileDataset]]] = defaultdict( 

94 lambda: defaultdict(list) 

95 ) 

96 self._collections: dict[str, CollectionRecord] = {} 

97 

98 def saveCollection(self, name: str) -> None: 

99 """Export the given collection. 

100 

101 Parameters 

102 ---------- 

103 name: `str` 

104 Name of the collection. 

105 

106 Notes 

107 ----- 

108 `~CollectionType.RUN` collections are also exported automatically when 

109 any dataset referencing them is exported. They may also be explicitly 

110 exported this method to export the collection with no datasets. 

111 Duplicate exports of collections are ignored. 

112 

113 Exporting a `~CollectionType.TAGGED` or `~CollectionType.CALIBRATION` 

114 collection will cause its associations with exported datasets to also 

115 be exported, but it does not export those datasets automatically. 

116 

117 Exporting a `~CollectionType.CHAINED` collection does not automatically 

118 export its child collections; these must be explicitly exported or 

119 already be present in the repository they are being imported into. 

120 """ 

121 self._collections[name] = self._registry._get_collection_record(name) 

122 

123 def saveDimensionData( 

124 self, element: str | DimensionElement, records: Iterable[dict | DimensionRecord] 

125 ) -> None: 

126 """Export the given dimension records associated with one or more data 

127 IDs. 

128 

129 Parameters 

130 ---------- 

131 element : `str` or `DimensionElement` 

132 `DimensionElement` or `str` indicating the logical table these 

133 records are from. 

134 records : `~collections.abc.Iterable` [ `DimensionRecord` or `dict` ] 

135 Records to export, as an iterable containing `DimensionRecord` or 

136 `dict` instances. 

137 """ 

138 if not isinstance(element, DimensionElement): 

139 element = self._registry.dimensions[element] 

140 for record in records: 

141 if not isinstance(record, DimensionRecord): 

142 record = element.RecordClass(**record) 

143 elif record.definition != element: 

144 raise ValueError( 

145 f"Mismatch between element={element.name} and " 

146 f"dimension record with definition={record.definition.name}." 

147 ) 

148 self._records[element].setdefault(record.dataId, record) 

149 

150 def saveDataIds( 

151 self, 

152 dataIds: Iterable[DataCoordinate], 

153 *, 

154 elements: Iterable[str | DimensionElement] | None = None, 

155 ) -> None: 

156 """Export the dimension records associated with one or more data IDs. 

157 

158 Parameters 

159 ---------- 

160 dataIds : iterable of `DataCoordinate`. 

161 Data IDs to export. For large numbers of data IDs obtained by 

162 calls to `Registry.queryDataIds`, it will be much more efficient if 

163 these are expanded to include records (i.e. 

164 `DataCoordinate.hasRecords` returns `True`) prior to the call to 

165 `saveDataIds` via e.g. ``Registry.queryDataIds(...).expanded()``. 

166 elements : iterable of `DimensionElement` or `str`, optional 

167 Dimension elements whose records should be exported. If `None`, 

168 records for all dimensions will be exported. 

169 """ 

170 standardized_elements: Set[DimensionElement] 

171 if elements is None: 

172 standardized_elements = frozenset( 

173 element 

174 for element in self._registry.dimensions.getStaticElements() 

175 if element.hasTable() and element.viewOf is None 

176 ) 

177 else: 

178 standardized_elements = set() 

179 for element in elements: 

180 if not isinstance(element, DimensionElement): 

181 element = self._registry.dimensions[element] 

182 if element.hasTable() and element.viewOf is None: 

183 standardized_elements.add(element) 

184 for dataId in dataIds: 

185 # This is potentially quite slow, because it's approximately 

186 # len(dataId.graph.elements) queries per data ID. But it's a no-op 

187 # if the data ID is already expanded, and DM-26692 will add (or at 

188 # least start to add / unblock) query functionality that should 

189 # let us speed this up internally as well. 

190 dataId = self._registry.expandDataId(dataId) 

191 for record in dataId.records.values(): 

192 if record is not None and record.definition in standardized_elements: 

193 self._records[record.definition].setdefault(record.dataId, record) 

194 

195 def saveDatasets( 

196 self, 

197 refs: Iterable[DatasetRef], 

198 *, 

199 elements: Iterable[str | DimensionElement] | None = None, 

200 rewrite: Callable[[FileDataset], FileDataset] | None = None, 

201 ) -> None: 

202 """Export one or more datasets. 

203 

204 This automatically exports any `DatasetType`, `~CollectionType.RUN` 

205 collections, and dimension records associated with the datasets. 

206 

207 Parameters 

208 ---------- 

209 refs : iterable of `DatasetRef` 

210 References to the datasets to export. Their `DatasetRef.id` 

211 attributes must not be `None`. Duplicates are automatically 

212 ignored. Nested data IDs must have `DataCoordinate.hasRecords` 

213 return `True`. If any reference is to a component dataset, the 

214 parent will be exported instead. 

215 elements : iterable of `DimensionElement` or `str`, optional 

216 Dimension elements whose records should be exported; this is 

217 forwarded to `saveDataIds` when exporting the data IDs of the 

218 given datasets. 

219 rewrite : callable, optional 

220 A callable that takes a single `FileDataset` argument and returns 

221 a modified `FileDataset`. This is typically used to rewrite the 

222 path generated by the datastore. If `None`, the `FileDataset` 

223 returned by `Datastore.export` will be used directly. 

224 

225 Notes 

226 ----- 

227 At present, this only associates datasets with `~CollectionType.RUN` 

228 collections. Other collections will be included in the export in the 

229 future (once `Registry` provides a way to look up that information). 

230 """ 

231 data_ids = set() 

232 refs_to_export = {} 

233 for ref in sorted(refs): 

234 dataset_id = ref.id 

235 # The query interfaces that are often used to generate the refs 

236 # passed here often don't remove duplicates, so do that here for 

237 # convenience. 

238 if dataset_id in self._dataset_ids or dataset_id in refs_to_export: 

239 continue 

240 # Also convert components to composites. 

241 if ref.isComponent(): 

242 ref = ref.makeCompositeRef() 

243 data_ids.add(ref.dataId) 

244 refs_to_export[dataset_id] = ref 

245 # Do a vectorized datastore export, which might be a lot faster than 

246 # one-by-one. 

247 exports = self._datastore.export( 

248 refs_to_export.values(), 

249 directory=self._directory, 

250 transfer=self._transfer, 

251 ) 

252 # Export associated data IDs. 

253 self.saveDataIds(data_ids, elements=elements) 

254 # Rewrite export filenames if desired, and then save them to the 

255 # data structure we'll write in `_finish`. 

256 # If a single exported FileDataset has multiple DatasetRefs, we save 

257 # it with each of them. 

258 for file_dataset in exports: 

259 if rewrite is not None: 

260 file_dataset = rewrite(file_dataset) 

261 for ref in file_dataset.refs: 

262 assert ref.run is not None 

263 self._datasets[ref.datasetType][ref.run].append(file_dataset) 

264 self._dataset_ids.update(refs_to_export.keys()) 

265 

266 def _finish(self) -> None: 

267 """Delegate to the backend to finish the export process. 

268 

269 For use by `Butler.export` only. 

270 """ 

271 for element in self._registry.dimensions.sorted(self._records.keys()): 

272 # To make export deterministic sort the DataCoordinate instances. 

273 r = self._records[element] 

274 self._backend.saveDimensionData(element, *[r[dataId] for dataId in sorted(r.keys())]) 

275 for datasetsByRun in self._datasets.values(): 

276 for run in datasetsByRun: 

277 self._collections[run] = self._registry._get_collection_record(run) 

278 for collectionName in self._computeSortedCollections(): 

279 doc = self._registry.getCollectionDocumentation(collectionName) 

280 self._backend.saveCollection(self._collections[collectionName], doc) 

281 # Sort the dataset types and runs before exporting to ensure 

282 # reproducible order in export file. 

283 for datasetType in sorted(self._datasets.keys()): 

284 for run in sorted(self._datasets[datasetType].keys()): 

285 # Sort the FileDataset 

286 records = sorted(self._datasets[datasetType][run]) 

287 self._backend.saveDatasets(datasetType, run, *records) 

288 # Export associations between datasets and collections. These need to 

289 # be sorted (at two levels; they're dicts) or created more 

290 # deterministically, too, which probably involves more data ID sorting. 

291 datasetAssociations = self._computeDatasetAssociations() 

292 for collection in sorted(datasetAssociations): 

293 self._backend.saveDatasetAssociations( 

294 collection, self._collections[collection].type, sorted(datasetAssociations[collection]) 

295 ) 

296 self._backend.finish() 

297 

298 def _computeSortedCollections(self) -> list[str]: 

299 """Sort collections in a way that is both deterministic and safe 

300 for registering them in a new repo in the presence of nested chains. 

301 

302 This method is intended for internal use by `RepoExportContext` only. 

303 

304 Returns 

305 ------- 

306 names: `List` [ `str` ] 

307 Ordered list of collection names. 

308 """ 

309 # Split collections into CHAINED and everything else, and just 

310 # sort "everything else" lexicographically since there are no 

311 # dependencies. 

312 chains: dict[str, list[str]] = {} 

313 result: list[str] = [] 

314 for record in self._collections.values(): 

315 if record.type is CollectionType.CHAINED: 

316 assert isinstance(record, ChainedCollectionRecord) 

317 chains[record.name] = list(record.children) 

318 else: 

319 result.append(record.name) 

320 result.sort() 

321 # Sort all chains topologically, breaking ties lexicographically. 

322 # Append these to 'result' and remove them from 'chains' as we go. 

323 while chains: 

324 unblocked = { 

325 parent 

326 for parent, children in chains.items() 

327 if not any(child in chains for child in children) 

328 } 

329 if not unblocked: 

330 raise RuntimeError( 

331 f"Apparent cycle in CHAINED collection dependencies involving {unblocked}." 

332 ) 

333 result.extend(sorted(unblocked)) 

334 for name in unblocked: 

335 del chains[name] 

336 return result 

337 

338 def _computeDatasetAssociations(self) -> dict[str, list[DatasetAssociation]]: 

339 """Return datasets-collection associations, grouped by association. 

340 

341 This queries for all associations between exported datasets and 

342 exported TAGGED or CALIBRATION collections and is intended to be run 

343 only by `_finish`, as this ensures all collections and all datasets 

344 have already been exported and hence the order in which they are 

345 exported does not matter. 

346 

347 Returns 

348 ------- 

349 associations : `dict` [ `str`, `list` [ `DatasetAssociation` ] ] 

350 Dictionary keyed by collection name, with values lists of structs 

351 representing an association between that collection and a dataset. 

352 """ 

353 results = defaultdict(list) 

354 for datasetType in self._datasets: 

355 # We query for _all_ datasets of each dataset type we export, in 

356 # the specific collections we are exporting. The worst-case 

357 # efficiency of this is _awful_ (i.e. big repo, exporting a tiny 

358 # subset). But we don't have any better options right now; we need 

359 # a way to query for a _lot_ of explicitly given dataset_ids, and 

360 # the only way to make that scale up is to either upload them to a 

361 # temporary table or recognize when they are already in one because 

362 # the user passed us a QueryResult object. That's blocked by (at 

363 # least) DM-26692. 

364 collectionTypes = {CollectionType.TAGGED} 

365 if datasetType.isCalibration(): 

366 collectionTypes.add(CollectionType.CALIBRATION) 

367 associationIter = self._registry.queryDatasetAssociations( 

368 datasetType, 

369 collections=self._collections.keys(), 

370 collectionTypes=collectionTypes, 

371 flattenChains=False, 

372 ) 

373 for association in associationIter: 

374 if association.ref.id in self._dataset_ids: 

375 results[association.collection].append(association) 

376 return results