Coverage for python/lsst/daf/butler/transfers/_context.py: 10%

110 statements  

« prev     ^ index     » next       coverage.py v6.4.2, created at 2022-07-27 01:57 -0700

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__ = ["RepoExportContext"] 

25 

26from collections import defaultdict 

27from typing import Callable, Dict, Iterable, List, Optional, Set, Union 

28 

29from ..core import ( 

30 DataCoordinate, 

31 DatasetAssociation, 

32 DatasetId, 

33 DatasetRef, 

34 DatasetType, 

35 Datastore, 

36 DimensionElement, 

37 DimensionRecord, 

38 FileDataset, 

39) 

40from ..registry import CollectionType, Registry 

41from ..registry.interfaces import ChainedCollectionRecord, CollectionRecord 

42from ._interfaces import RepoExportBackend 

43 

44 

45class RepoExportContext: 

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

47 

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

49 value returned by that context manager:: 

50 

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

52 export.saveDataIds(...) 

53 export.saveDatasets(...) 

54 

55 Parameters 

56 ---------- 

57 registry : `Registry` 

58 Registry to export from. 

59 datastore : `Datastore` 

60 Datastore to export from. 

61 backend : `RepoExportBackend` 

62 Implementation class for a particular export file format. 

63 directory : `str`, optional 

64 Directory to pass to `Datastore.export`. 

65 transfer : `str`, optional 

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

67 """ 

68 

69 def __init__( 

70 self, 

71 registry: Registry, 

72 datastore: Datastore, 

73 backend: RepoExportBackend, 

74 *, 

75 directory: Optional[str] = None, 

76 transfer: Optional[str] = None, 

77 ): 

78 self._registry = registry 

79 self._datastore = datastore 

80 self._backend = backend 

81 self._directory = directory 

82 self._transfer = transfer 

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

84 self._dataset_ids: Set[DatasetId] = set() 

85 self._datasets: Dict[DatasetType, Dict[str, List[FileDataset]]] = defaultdict( 

86 lambda: defaultdict(list) 

87 ) 

88 self._collections: Dict[str, CollectionRecord] = {} 

89 

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

91 """Export the given collection. 

92 

93 Parameters 

94 ---------- 

95 name: `str` 

96 Name of the collection. 

97 

98 Notes 

99 ----- 

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

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

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

103 Duplicate exports of collections are ignored. 

104 

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

106 collection will cause its associations with exported datasets to also 

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

108 

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

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

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

112 """ 

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

114 

115 def saveDimensionData( 

116 self, element: Union[str, DimensionElement], records: Iterable[Union[dict, DimensionRecord]] 

117 ) -> None: 

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

119 IDs. 

120 

121 Parameters 

122 ---------- 

123 element : `str` or `DimensionElement` 

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

125 records are from. 

126 records : `Iterable` [ `DimensionRecord` or `dict` ] 

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

128 `dict` instances. 

129 """ 

130 if not isinstance(element, DimensionElement): 

131 element = self._registry.dimensions[element] 

132 for record in records: 

133 if not isinstance(record, DimensionRecord): 

134 record = element.RecordClass(**record) 

135 elif record.definition != element: 

136 raise ValueError( 

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

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

139 ) 

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

141 

142 def saveDataIds( 

143 self, 

144 dataIds: Iterable[DataCoordinate], 

145 *, 

146 elements: Optional[Iterable[Union[str, DimensionElement]]] = None, 

147 ) -> None: 

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

149 

150 Parameters 

151 ---------- 

152 dataIds : iterable of `DataCoordinate`. 

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

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

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

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

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

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

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

160 records for all dimensions will be exported. 

161 """ 

162 if elements is None: 

163 elements = frozenset( 

164 element 

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

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

167 ) 

168 else: 

169 elements = set() 

170 for element in elements: 

171 if not isinstance(element, DimensionElement): 

172 element = self._registry.dimensions[element] 

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

174 elements.add(element) 

175 for dataId in dataIds: 

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

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

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

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

180 # let us speed this up internally as well. 

181 dataId = self._registry.expandDataId(dataId) 

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

183 if record is not None and record.definition in elements: 

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

185 

186 def saveDatasets( 

187 self, 

188 refs: Iterable[DatasetRef], 

189 *, 

190 elements: Optional[Iterable[Union[str, DimensionElement]]] = None, 

191 rewrite: Optional[Callable[[FileDataset], FileDataset]] = None, 

192 ) -> None: 

193 """Export one or more datasets. 

194 

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

196 collections, and dimension records associated with the datasets. 

197 

198 Parameters 

199 ---------- 

200 refs : iterable of `DatasetRef` 

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

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

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

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

205 parent will be exported instead. 

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

207 Dimension elements whose records should be exported; this is 

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

209 given datasets. 

210 rewrite : callable, optional 

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

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

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

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

215 

216 Notes 

217 ----- 

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

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

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

221 """ 

222 data_ids = set() 

223 refs_to_export = {} 

224 for ref in sorted(refs): 

225 dataset_id = ref.getCheckedId() 

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

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

228 # convenience. 

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

230 continue 

231 # Also convert components to composites. 

232 if ref.isComponent(): 

233 ref = ref.makeCompositeRef() 

234 data_ids.add(ref.dataId) 

235 refs_to_export[dataset_id] = ref 

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

237 # one-by-one. 

238 exports = self._datastore.export( 

239 refs_to_export.values(), 

240 directory=self._directory, 

241 transfer=self._transfer, 

242 ) 

243 # Export associated data IDs. 

244 self.saveDataIds(data_ids, elements=elements) 

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

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

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

248 # it with each of them. 

249 for file_dataset in exports: 

250 if rewrite is not None: 

251 file_dataset = rewrite(file_dataset) 

252 for ref in file_dataset.refs: 

253 assert ref.run is not None 

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

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

256 

257 def _finish(self) -> None: 

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

259 

260 For use by `Butler.export` only. 

261 """ 

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

263 # To make export deterministic sort the DataCoordinate instances. 

264 r = self._records[element] 

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

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

267 for run in datasetsByRun.keys(): 

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

269 for collectionName in self._computeSortedCollections(): 

270 doc = self._registry.getCollectionDocumentation(collectionName) 

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

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

273 # reproducible order in export file. 

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

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

276 # Sort the FileDataset 

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

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

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

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

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

282 datasetAssociations = self._computeDatasetAssociations() 

283 for collection in sorted(datasetAssociations): 

284 self._backend.saveDatasetAssociations( 

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

286 ) 

287 self._backend.finish() 

288 

289 def _computeSortedCollections(self) -> List[str]: 

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

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

292 

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

294 

295 Returns 

296 ------- 

297 names: `List` [ `str` ] 

298 Ordered list of collection names. 

299 """ 

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

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

302 # dependencies. 

303 chains: Dict[str, List[str]] = {} 

304 result: List[str] = [] 

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

306 if record.type is CollectionType.CHAINED: 

307 assert isinstance(record, ChainedCollectionRecord) 

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

309 else: 

310 result.append(record.name) 

311 result.sort() 

312 # Sort all chains topologically, breaking ties lexicographically. 

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

314 while chains: 

315 unblocked = { 

316 parent 

317 for parent, children in chains.items() 

318 if not any(child in chains.keys() for child in children) 

319 } 

320 if not unblocked: 

321 raise RuntimeError( 

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

323 ) 

324 result.extend(sorted(unblocked)) 

325 for name in unblocked: 

326 del chains[name] 

327 return result 

328 

329 def _computeDatasetAssociations(self) -> Dict[str, List[DatasetAssociation]]: 

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

331 

332 This queries for all associations between exported datasets and 

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

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

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

336 exported does not matter. 

337 

338 Returns 

339 ------- 

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

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

342 representing an association between that collection and a dataset. 

343 """ 

344 results = defaultdict(list) 

345 for datasetType in self._datasets.keys(): 

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

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

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

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

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

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

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

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

354 # least) DM-26692. 

355 collectionTypes = {CollectionType.TAGGED} 

356 if datasetType.isCalibration(): 

357 collectionTypes.add(CollectionType.CALIBRATION) 

358 associationIter = self._registry.queryDatasetAssociations( 

359 datasetType, 

360 collections=self._collections.keys(), 

361 collectionTypes=collectionTypes, 

362 flattenChains=False, 

363 ) 

364 for association in associationIter: 

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

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

367 return results