Coverage for python/lsst/daf/butler/registry/queries/_sql_query_backend.py: 16%

104 statements  

« prev     ^ index     » next       coverage.py v7.3.1, created at 2023-10-02 08:00 +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/>. 

27from __future__ import annotations 

28 

29__all__ = ("SqlQueryBackend",) 

30 

31from collections.abc import Iterable, Mapping, Sequence, Set 

32from typing import TYPE_CHECKING, Any, cast 

33 

34from lsst.daf.relation import ColumnError, ColumnExpression, ColumnTag, Join, Predicate, Relation 

35 

36from ...core import ( 

37 ColumnCategorization, 

38 DataCoordinate, 

39 DatasetType, 

40 DimensionGraph, 

41 DimensionKeyColumnTag, 

42 DimensionRecord, 

43 DimensionRecordColumnTag, 

44 DimensionUniverse, 

45 SkyPixDimension, 

46) 

47from .._collectionType import CollectionType 

48from .._exceptions import DataIdValueError 

49from ..interfaces import CollectionRecord, Database 

50from ._query_backend import QueryBackend 

51from ._sql_query_context import SqlQueryContext 

52 

53if TYPE_CHECKING: 

54 from ..managers import RegistryManagerInstances 

55 

56 

57class SqlQueryBackend(QueryBackend[SqlQueryContext]): 

58 """An implementation of `QueryBackend` for `SqlRegistry`. 

59 

60 Parameters 

61 ---------- 

62 db : `Database` 

63 Object that abstracts the database engine. 

64 managers : `RegistryManagerInstances` 

65 Struct containing the manager objects that back a `SqlRegistry`. 

66 """ 

67 

68 def __init__( 

69 self, 

70 db: Database, 

71 managers: RegistryManagerInstances, 

72 ): 

73 self._db = db 

74 self._managers = managers 

75 

76 @property 

77 def universe(self) -> DimensionUniverse: 

78 # Docstring inherited. 

79 return self._managers.dimensions.universe 

80 

81 def context(self) -> SqlQueryContext: 

82 # Docstring inherited. 

83 return SqlQueryContext(self._db, self._managers.column_types) 

84 

85 def get_collection_name(self, key: Any) -> str: 

86 return self._managers.collections[key].name 

87 

88 def resolve_collection_wildcard( 

89 self, 

90 expression: Any, 

91 *, 

92 collection_types: Set[CollectionType] = CollectionType.all(), 

93 done: set[str] | None = None, 

94 flatten_chains: bool = True, 

95 include_chains: bool | None = None, 

96 ) -> list[CollectionRecord]: 

97 # Docstring inherited. 

98 return self._managers.collections.resolve_wildcard( 

99 expression, 

100 collection_types=collection_types, 

101 done=done, 

102 flatten_chains=flatten_chains, 

103 include_chains=include_chains, 

104 ) 

105 

106 def resolve_dataset_type_wildcard( 

107 self, 

108 expression: Any, 

109 components: bool | None = None, 

110 missing: list[str] | None = None, 

111 explicit_only: bool = False, 

112 components_deprecated: bool = True, 

113 ) -> dict[DatasetType, list[str | None]]: 

114 # Docstring inherited. 

115 return self._managers.datasets.resolve_wildcard( 

116 expression, components, missing, explicit_only, components_deprecated 

117 ) 

118 

119 def filter_dataset_collections( 

120 self, 

121 dataset_types: Iterable[DatasetType], 

122 collections: Sequence[CollectionRecord], 

123 *, 

124 governor_constraints: Mapping[str, Set[str]], 

125 rejections: list[str] | None = None, 

126 ) -> dict[DatasetType, list[CollectionRecord]]: 

127 # Docstring inherited. 

128 result: dict[DatasetType, list[CollectionRecord]] = { 

129 dataset_type: [] for dataset_type in dataset_types 

130 } 

131 for dataset_type, filtered_collections in result.items(): 

132 for collection_record in collections: 

133 if not dataset_type.isCalibration() and collection_record.type is CollectionType.CALIBRATION: 

134 if rejections is not None: 

135 rejections.append( 

136 f"Not searching for non-calibration dataset of type {dataset_type.name!r} " 

137 f"in CALIBRATION collection {collection_record.name!r}." 

138 ) 

139 else: 

140 collection_summary = self._managers.datasets.getCollectionSummary(collection_record) 

141 if collection_summary.is_compatible_with( 

142 dataset_type, 

143 governor_constraints, 

144 rejections=rejections, 

145 name=collection_record.name, 

146 ): 

147 filtered_collections.append(collection_record) 

148 return result 

149 

150 def _make_dataset_query_relation_impl( 

151 self, 

152 dataset_type: DatasetType, 

153 collections: Sequence[CollectionRecord], 

154 columns: Set[str], 

155 context: SqlQueryContext, 

156 ) -> Relation: 

157 # Docstring inherited. 

158 assert len(collections) > 0, ( 

159 "Caller is responsible for handling the case of all collections being rejected (we can't " 

160 "write a good error message without knowing why collections were rejected)." 

161 ) 

162 dataset_storage = self._managers.datasets.find(dataset_type.name) 

163 if dataset_storage is None: 

164 # Unrecognized dataset type means no results. 

165 return self.make_doomed_dataset_relation( 

166 dataset_type, 

167 columns, 

168 messages=[ 

169 f"Dataset type {dataset_type.name!r} is not registered, " 

170 "so no instances of it can exist in any collection." 

171 ], 

172 context=context, 

173 ) 

174 else: 

175 return dataset_storage.make_relation( 

176 *collections, 

177 columns=columns, 

178 context=context, 

179 ) 

180 

181 def make_dimension_relation( 

182 self, 

183 dimensions: DimensionGraph, 

184 columns: Set[ColumnTag], 

185 context: SqlQueryContext, 

186 *, 

187 initial_relation: Relation | None = None, 

188 initial_join_max_columns: frozenset[ColumnTag] | None = None, 

189 initial_dimension_relationships: Set[frozenset[str]] | None = None, 

190 spatial_joins: Iterable[tuple[str, str]] = (), 

191 governor_constraints: Mapping[str, Set[str]], 

192 ) -> Relation: 

193 # Docstring inherited. 

194 

195 default_join = Join(max_columns=initial_join_max_columns) 

196 

197 # Set up the relation variable we'll update as we join more relations 

198 # in, and ensure it is in the SQL engine. 

199 relation = context.make_initial_relation(initial_relation) 

200 

201 if initial_dimension_relationships is None: 

202 relationships = self.extract_dimension_relationships(relation) 

203 else: 

204 relationships = set(initial_dimension_relationships) 

205 

206 # Make a mutable copy of the columns argument. 

207 columns_required = set(columns) 

208 

209 # Sort spatial joins to put those involving the commonSkyPix dimension 

210 # first, since those join subqueries might get reused in implementing 

211 # other joins later. 

212 spatial_joins = list(spatial_joins) 

213 spatial_joins.sort(key=lambda j: self.universe.commonSkyPix.name not in j) 

214 

215 # Next we'll handle spatial joins, since those can require refinement 

216 # predicates that will need region columns to be included in the 

217 # relations we'll join. 

218 predicate: Predicate = Predicate.literal(True) 

219 for element1, element2 in spatial_joins: 

220 (overlaps, needs_refinement) = self._managers.dimensions.make_spatial_join_relation( 

221 element1, 

222 element2, 

223 context=context, 

224 governor_constraints=governor_constraints, 

225 existing_relationships=relationships, 

226 ) 

227 if needs_refinement: 

228 predicate = predicate.logical_and( 

229 context.make_spatial_region_overlap_predicate( 

230 ColumnExpression.reference(DimensionRecordColumnTag(element1, "region")), 

231 ColumnExpression.reference(DimensionRecordColumnTag(element2, "region")), 

232 ) 

233 ) 

234 columns_required.add(DimensionRecordColumnTag(element1, "region")) 

235 columns_required.add(DimensionRecordColumnTag(element2, "region")) 

236 relation = relation.join(overlaps) 

237 relationships.add( 

238 frozenset(self.universe[element1].dimensions.names | self.universe[element2].dimensions.names) 

239 ) 

240 

241 # All skypix columns need to come from either the initial_relation or a 

242 # spatial join, since we need all dimension key columns present in the 

243 # SQL engine and skypix regions are added by postprocessing in the 

244 # native iteration engine. 

245 for dimension in dimensions: 

246 if DimensionKeyColumnTag(dimension.name) not in relation.columns and isinstance( 

247 dimension, SkyPixDimension 

248 ): 

249 raise NotImplementedError( 

250 f"Cannot construct query involving skypix dimension {dimension.name} unless " 

251 "it is part of a dataset subquery, spatial join, or other initial relation." 

252 ) 

253 

254 # Before joining in new tables to provide columns, attempt to restore 

255 # them from the given relation by weakening projections applied to it. 

256 relation, _ = context.restore_columns(relation, columns_required) 

257 

258 # Categorize columns not yet included in the relation to associate them 

259 # with dimension elements and detect bad inputs. 

260 missing_columns = ColumnCategorization.from_iterable(columns_required - relation.columns) 

261 if not (missing_columns.dimension_keys <= dimensions.names): 

262 raise ColumnError( 

263 "Cannot add dimension key column(s) " 

264 f"{{{', '.join(name for name in missing_columns.dimension_keys)}}} " 

265 f"that were not included in the given dimensions {dimensions}." 

266 ) 

267 if missing_columns.datasets: 

268 raise ColumnError( 

269 f"Unexpected dataset columns {missing_columns.datasets} in call to make_dimension_relation; " 

270 "use make_dataset_query_relation or make_dataset_search relation instead, or filter them " 

271 "out if they have already been added or will be added later." 

272 ) 

273 for element_name in missing_columns.dimension_records: 

274 if element_name not in dimensions.elements.names: 

275 raise ColumnError( 

276 f"Cannot join dimension element {element_name} whose dimensions are not a " 

277 f"subset of {dimensions}." 

278 ) 

279 

280 # Iterate over all dimension elements whose relations definitely have 

281 # to be joined in. The order doesn't matter as long as we can assume 

282 # the database query optimizer is going to try to reorder them anyway. 

283 for element in dimensions.elements: 

284 columns_still_needed = missing_columns.dimension_records[element.name] 

285 # Two separate conditions in play here: 

286 # - if we need a record column (not just key columns) from this 

287 # element, we have to join in its relation; 

288 # - if the element establishes a relationship between key columns 

289 # that wasn't already established by the initial relation, we 

290 # always join that element's relation. Any element with 

291 # implied dependencies or the alwaysJoin flag establishes such a 

292 # relationship. 

293 if columns_still_needed or ( 

294 (element.alwaysJoin or element.implied) 

295 and frozenset(element.dimensions.names) not in relationships 

296 ): 

297 storage = self._managers.dimensions[element] 

298 relation = storage.join(relation, default_join, context) 

299 # At this point we've joined in all of the element relations that 

300 # definitely need to be included, but we may not have all of the 

301 # dimension key columns in the query that we want. To fill out that 

302 # set, we iterate over just the given DimensionGraph's dimensions (not 

303 # all dimension *elements*) in reverse topological order. That order 

304 # should reduce the total number of tables we bring in, since each 

305 # dimension will bring in keys for its required dependencies before we 

306 # get to those required dependencies. 

307 for dimension in self.universe.sorted(dimensions, reverse=True): 

308 if DimensionKeyColumnTag(dimension.name) not in relation.columns: 

309 storage = self._managers.dimensions[dimension] 

310 relation = storage.join(relation, default_join, context) 

311 

312 # Add the predicates we constructed earlier, with a transfer to native 

313 # iteration first if necessary. 

314 if not predicate.as_trivial(): 

315 relation = relation.with_rows_satisfying( 

316 predicate, preferred_engine=context.iteration_engine, transfer=True 

317 ) 

318 

319 # Finally project the new relation down to just the columns in the 

320 # initial relation, the dimension key columns, and the new columns 

321 # requested. 

322 columns_kept = set(columns) 

323 if initial_relation is not None: 

324 columns_kept.update(initial_relation.columns) 

325 columns_kept.update(DimensionKeyColumnTag.generate(dimensions.names)) 

326 relation = relation.with_only_columns(columns_kept, preferred_engine=context.preferred_engine) 

327 

328 return relation 

329 

330 def resolve_governor_constraints( 

331 self, dimensions: DimensionGraph, constraints: Mapping[str, Set[str]], context: SqlQueryContext 

332 ) -> Mapping[str, Set[str]]: 

333 # Docstring inherited. 

334 result: dict[str, Set[str]] = {} 

335 for dimension in dimensions.governors: 

336 storage = self._managers.dimensions[dimension] 

337 records = storage.get_record_cache(context) 

338 assert records is not None, "Governor dimensions are always cached." 

339 all_values = {cast(str, data_id[dimension.name]) for data_id in records} 

340 if (constraint_values := constraints.get(dimension.name)) is not None: 

341 if not (constraint_values <= all_values): 

342 raise DataIdValueError( 

343 f"Unknown values specified for governor dimension {dimension.name}: " 

344 f"{constraint_values - all_values}." 

345 ) 

346 result[dimension.name] = constraint_values 

347 else: 

348 result[dimension.name] = all_values 

349 return result 

350 

351 def get_dimension_record_cache( 

352 self, 

353 element_name: str, 

354 context: SqlQueryContext, 

355 ) -> Mapping[DataCoordinate, DimensionRecord] | None: 

356 return self._managers.dimensions[element_name].get_record_cache(context)