Hide keyboard shortcuts

Hot-keys on this page

r m x p   toggle line displays

j k   next/prev highlighted chunk

0   (zero) top of page

1   (one) first highlighted chunk

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/>. 

21from __future__ import annotations 

22 

23__all__ = ["QuerySummary"] # other classes here are local to subpackage 

24 

25from dataclasses import dataclass 

26from typing import Iterator, List, Optional, Union 

27 

28from sqlalchemy.sql import ColumnElement 

29 

30from ...core import ( 

31 DatasetType, 

32 Dimension, 

33 DimensionElement, 

34 DimensionGraph, 

35 DimensionUniverse, 

36 ExpandedDataCoordinate, 

37 NamedKeyDict, 

38 NamedValueSet, 

39 SkyPixDimension, 

40 Timespan, 

41) 

42# We're not trying to add parsing to the lex/yacc parser code, so MyPy 

43# doesn't know about some of these imports. 

44from .exprParser import Node, ParserYacc # type: ignore 

45 

46 

47@dataclass 

48class QueryWhereExpression: 

49 """A struct representing a parsed user-provided WHERE expression. 

50 

51 Parameters 

52 ---------- 

53 universe : `DimensionUniverse` 

54 All known dimensions. 

55 expression : `str`, optional 

56 The string expression to parse. 

57 """ 

58 def __init__(self, universe: DimensionUniverse, expression: Optional[str] = None): 

59 if expression: 

60 from .expressions import InspectionVisitor 

61 try: 

62 parser = ParserYacc() 

63 self.tree = parser.parse(expression) 

64 except Exception as exc: 

65 raise RuntimeError(f"Failed to parse user expression `{expression}'.") from exc 

66 visitor = InspectionVisitor(universe) 

67 assert self.tree is not None 

68 self.tree.visit(visitor) 

69 self.keys = visitor.keys 

70 self.metadata = visitor.metadata 

71 else: 

72 self.tree = None 

73 self.keys = NamedValueSet() 

74 self.metadata = NamedKeyDict() 

75 

76 tree: Optional[Node] 

77 """The parsed user expression tree, if present (`Node` or `None`). 

78 """ 

79 

80 keys: NamedValueSet[Dimension] 

81 """All dimensions whose keys are referenced by the expression 

82 (`NamedValueSet` of `Dimension`). 

83 """ 

84 

85 metadata: NamedKeyDict[DimensionElement, List[str]] 

86 """All dimension elements metadata fields referenced by the expression 

87 (`NamedKeyDict` mapping `DimensionElement` to a `set` of field names). 

88 """ 

89 

90 

91@dataclass 

92class QuerySummary: 

93 """A struct that holds and categorizes the dimensions involved in a query. 

94 

95 A `QuerySummary` instance is necessary to construct a `QueryBuilder`, and 

96 it needs to include all of the dimensions that will be included in the 

97 query (including any needed for querying datasets). 

98 

99 Parameters 

100 ---------- 

101 requested : `DimensionGraph` 

102 The dimensions whose primary keys should be included in the result rows 

103 of the query. 

104 dataId : `ExpandedDataCoordinate`, optional 

105 A fully-expanded data ID identifying dimensions known in advance. If 

106 not provided, will be set to an empty data ID. 

107 expression : `str` or `QueryWhereExpression`, optional 

108 A user-provided string WHERE expression. 

109 """ 

110 def __init__(self, requested: DimensionGraph, *, 

111 dataId: Optional[ExpandedDataCoordinate] = None, 

112 expression: Optional[Union[str, QueryWhereExpression]] = None): 

113 self.requested = requested 

114 self.dataId = dataId if dataId is not None else ExpandedDataCoordinate(requested.universe.empty, (), 

115 records=NamedKeyDict()) 

116 self.expression = (expression if isinstance(expression, QueryWhereExpression) 

117 else QueryWhereExpression(requested.universe, expression)) 

118 

119 requested: DimensionGraph 

120 """Dimensions whose primary keys should be included in the result rows of 

121 the query (`DimensionGraph`). 

122 """ 

123 

124 dataId: ExpandedDataCoordinate 

125 """A data ID identifying dimensions known before query construction 

126 (`ExpandedDataCoordinate`). 

127 """ 

128 

129 expression: QueryWhereExpression 

130 """Information about any parsed user WHERE expression 

131 (`QueryWhereExpression`). 

132 """ 

133 

134 @property 

135 def universe(self) -> DimensionUniverse: 

136 """All known dimensions (`DimensionUniverse`). 

137 """ 

138 return self.requested.universe 

139 

140 @property 

141 def spatial(self) -> NamedValueSet[DimensionElement]: 

142 """Dimension elements whose regions and skypix IDs should be included 

143 in the query (`NamedValueSet` of `DimensionElement`). 

144 """ 

145 # An element may participate spatially in the query if: 

146 # - it's the most precise spatial element for its system in the 

147 # requested dimensions (i.e. in `self.requested.spatial`); 

148 # - it isn't also given at query construction time. 

149 result = NamedValueSet(self.mustHaveKeysJoined.spatial - self.dataId.graph.elements) 

150 if len(result) == 1: 

151 # There's no spatial join, but there might be a WHERE filter based 

152 # on a given region. 

153 if self.dataId.graph.spatial: 

154 # We can only perform those filters against SkyPix dimensions, 

155 # so if what we have isn't one, add the common SkyPix dimension 

156 # to the query; the element we have will be joined to that. 

157 element, = result 

158 if not isinstance(element, SkyPixDimension): 

159 result.add(self.universe.commonSkyPix) 

160 else: 

161 # There is no spatial join or filter in this query. Even 

162 # if this element might be associated with spatial 

163 # information, we don't need it for this query. 

164 return NamedValueSet() 

165 elif len(result) > 1: 

166 # There's a spatial join. Those require the common SkyPix 

167 # system to be included in the query in order to connect them. 

168 result.add(self.universe.commonSkyPix) 

169 return result 

170 

171 @property 

172 def temporal(self) -> NamedValueSet[DimensionElement]: 

173 """Dimension elements whose timespans should be included in the 

174 query (`NamedValueSet` of `DimensionElement`). 

175 """ 

176 # An element may participate temporally in the query if: 

177 # - it's the most precise temporal element for its system in the 

178 # requested dimensions (i.e. in `self.requested.temporal`); 

179 # - it isn't also given at query construction time. 

180 result = NamedValueSet(self.mustHaveKeysJoined.temporal - self.dataId.graph.elements) 

181 if len(result) == 1 and not self.dataId.graph.temporal: 

182 # No temporal join or filter. Even if this element might be 

183 # associated with temporal information, we don't need it for this 

184 # query. 

185 return NamedValueSet() 

186 return result 

187 

188 @property 

189 def mustHaveKeysJoined(self) -> DimensionGraph: 

190 """Dimensions whose primary keys must be used in the JOIN ON clauses 

191 of the query, even if their tables do not appear (`DimensionGraph`). 

192 

193 A `Dimension` primary key can appear in a join clause without its table 

194 via a foreign key column in table of a dependent dimension element or 

195 dataset. 

196 """ 

197 names = set(self.requested.names | self.expression.keys.names) 

198 return DimensionGraph(self.universe, names=names) 

199 

200 @property 

201 def mustHaveTableJoined(self) -> NamedValueSet[DimensionElement]: 

202 """Dimension elements whose associated tables must appear in the 

203 query's FROM clause (`NamedValueSet` of `DimensionElement`). 

204 """ 

205 result = NamedValueSet(self.spatial | self.temporal | self.expression.metadata.keys()) 

206 for dimension in self.mustHaveKeysJoined: 

207 if dimension.implied: 

208 result.add(dimension) 

209 for element in self.mustHaveKeysJoined.union(self.dataId.graph).elements: 

210 if element.alwaysJoin: 

211 result.add(element) 

212 return result 

213 

214 

215@dataclass 

216class DatasetQueryColumns: 

217 """A struct containing the columns used to reconstruct `DatasetRef` 

218 instances from query results. 

219 """ 

220 

221 id: ColumnElement 

222 """Column containing the unique integer ID for this dataset. 

223 """ 

224 

225 runKey: ColumnElement 

226 """Foreign key column to the `~CollectionType.RUN` collection that holds 

227 this dataset. 

228 """ 

229 

230 rank: Optional[ColumnElement] = None 

231 """Column containing the index into the ordered sequence of given 

232 collections for the collection in which this dataset was found. 

233 """ 

234 

235 def __iter__(self) -> Iterator[ColumnElement]: 

236 yield self.id 

237 yield self.runKey 

238 if self.rank is not None: 

239 yield self.rank 

240 

241 

242@dataclass 

243class QueryColumns: 

244 """A struct organizing the columns in an under-construction or currently- 

245 executing query. 

246 

247 Takes no parameters at construction, as expected usage is to add elements 

248 to its container attributes incrementally. 

249 """ 

250 def __init__(self) -> None: 

251 self.keys = NamedKeyDict() 

252 self.timespans = NamedKeyDict() 

253 self.regions = NamedKeyDict() 

254 self.datasets = NamedKeyDict() 

255 

256 keys: NamedKeyDict[Dimension, List[ColumnElement]] 

257 """Columns that correspond to the primary key values of dimensions 

258 (`NamedKeyDict` mapping `Dimension` to a `list` of `ColumnElement`). 

259 

260 Each value list contains columns from multiple tables corresponding to the 

261 same dimension, and the query should constrain the values of those columns 

262 to be the same. 

263 

264 In a `Query`, the keys of this dictionary must include at least the 

265 dimensions in `QuerySummary.requested` and `QuerySummary.dataId.graph`. 

266 """ 

267 

268 timespans: NamedKeyDict[DimensionElement, Timespan[ColumnElement]] 

269 """Columns that correspond to timespans for elements that participate in a 

270 temporal join or filter in the query (`NamedKeyDict` mapping 

271 `DimensionElement` to `Timespan` of `ColumnElement`). 

272 

273 In a `Query`, the keys of this dictionary must be exactly the elements 

274 in `QuerySummary.temporal`. 

275 """ 

276 

277 regions: NamedKeyDict[DimensionElement, ColumnElement] 

278 """Columns that correspond to regions for elements that participate in a 

279 spatial join or filter in the query (`NamedKeyDict` mapping 

280 `DimensionElement` to `ColumnElement`). 

281 

282 In a `Query`, the keys of this dictionary must be exactly the elements 

283 in `QuerySummary.spatial`. 

284 """ 

285 

286 datasets: NamedKeyDict[DatasetType, DatasetQueryColumns] 

287 """Columns that can be used to construct `DatasetRef` instances from query 

288 results, for each `DatasetType` included in the query 

289 (`NamedKeyDict` [ `DatasetType`, `DatasetQueryColumns` ] ). 

290 """ 

291 

292 def getKeyColumn(self, dimension: Dimension) -> ColumnElement: 

293 """ Return one of the columns in self.keys for the given dimension. 

294 

295 The column selected is an implentation detail but is guaranteed to 

296 be deterministic and consistent across multiple calls. 

297 

298 Parameters 

299 ---------- 

300 dimension : `Dimension` 

301 Element for which to obtain a key column. 

302 

303 Returns 

304 ------- 

305 column : `sqlalchemy.sql.ColumnElement` 

306 SQLAlchemy column object. 

307 """ 

308 # Choosing the last element here is entirely for human readers of the 

309 # query (e.g. developers debugging things); it makes it more likely a 

310 # dimension key will be provided by the dimension's own table, or 

311 # failing that, some closely related dimension, which might be less 

312 # surprising to see than e.g. some dataset subquery. From the 

313 # database's perspective this is entirely arbitrary, cause the query 

314 # guarantees they all have equal values. 

315 return self.keys[dimension][-1]