Coverage for python/lsst/daf/butler/registry/collections/synthIntKey.py: 99%

113 statements  

« prev     ^ index     » next       coverage.py v7.4.3, created at 2024-03-12 10:05 +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 

29from ... import ddl 

30 

31__all__ = ["SynthIntKeyCollectionManager"] 

32 

33import logging 

34from collections.abc import Iterable, Mapping 

35from typing import TYPE_CHECKING, Any 

36 

37import sqlalchemy 

38 

39from ...column_spec import COLLECTION_NAME_MAX_LENGTH 

40from ...timespan_database_representation import TimespanDatabaseRepresentation 

41from .._collection_type import CollectionType 

42from ..interfaces import ChainedCollectionRecord, CollectionRecord, RunRecord, VersionTuple 

43from ._base import ( 

44 CollectionTablesTuple, 

45 DefaultCollectionManager, 

46 makeCollectionChainTableSpec, 

47 makeRunTableSpec, 

48) 

49 

50if TYPE_CHECKING: 

51 from .._caching_context import CachingContext 

52 from ..interfaces import Database, StaticTablesContext 

53 

54 

55_KEY_FIELD_SPEC = ddl.FieldSpec( 

56 "collection_id", dtype=sqlalchemy.BigInteger, primaryKey=True, autoincrement=True 

57) 

58 

59 

60# This has to be updated on every schema change 

61_VERSION = VersionTuple(2, 0, 0) 

62 

63_LOG = logging.getLogger(__name__) 

64 

65 

66def _makeTableSpecs(TimespanReprClass: type[TimespanDatabaseRepresentation]) -> CollectionTablesTuple: 

67 return CollectionTablesTuple( 

68 collection=ddl.TableSpec( 

69 fields=[ 

70 _KEY_FIELD_SPEC, 

71 ddl.FieldSpec( 

72 "name", dtype=sqlalchemy.String, length=COLLECTION_NAME_MAX_LENGTH, nullable=False 

73 ), 

74 ddl.FieldSpec("type", dtype=sqlalchemy.SmallInteger, nullable=False), 

75 ddl.FieldSpec("doc", dtype=sqlalchemy.Text, nullable=True), 

76 ], 

77 unique=[("name",)], 

78 ), 

79 run=makeRunTableSpec("collection_id", sqlalchemy.BigInteger, TimespanReprClass), 

80 collection_chain=makeCollectionChainTableSpec("collection_id", sqlalchemy.BigInteger), 

81 ) 

82 

83 

84class SynthIntKeyCollectionManager(DefaultCollectionManager[int]): 

85 """A `CollectionManager` implementation that uses synthetic primary key 

86 (auto-incremented integer) for collections table. 

87 """ 

88 

89 @classmethod 

90 def initialize( 

91 cls, 

92 db: Database, 

93 context: StaticTablesContext, 

94 *, 

95 caching_context: CachingContext, 

96 registry_schema_version: VersionTuple | None = None, 

97 ) -> SynthIntKeyCollectionManager: 

98 # Docstring inherited from CollectionManager. 

99 return cls( 

100 db, 

101 tables=context.addTableTuple(_makeTableSpecs(db.getTimespanRepresentation())), # type: ignore 

102 collectionIdName="collection_id", 

103 caching_context=caching_context, 

104 registry_schema_version=registry_schema_version, 

105 ) 

106 

107 def clone(self, db: Database, caching_context: CachingContext) -> SynthIntKeyCollectionManager: 

108 return SynthIntKeyCollectionManager( 

109 db, 

110 tables=self._tables, 

111 collectionIdName=self._collectionIdName, 

112 caching_context=caching_context, 

113 registry_schema_version=self._registry_schema_version, 

114 ) 

115 

116 @classmethod 

117 def getCollectionForeignKeyName(cls, prefix: str = "collection") -> str: 

118 # Docstring inherited from CollectionManager. 

119 return f"{prefix}_id" 

120 

121 @classmethod 

122 def getRunForeignKeyName(cls, prefix: str = "run") -> str: 

123 # Docstring inherited from CollectionManager. 

124 return f"{prefix}_id" 

125 

126 @classmethod 

127 def addCollectionForeignKey( 

128 cls, 

129 tableSpec: ddl.TableSpec, 

130 *, 

131 prefix: str = "collection", 

132 onDelete: str | None = None, 

133 constraint: bool = True, 

134 **kwargs: Any, 

135 ) -> ddl.FieldSpec: 

136 # Docstring inherited from CollectionManager. 

137 original = _KEY_FIELD_SPEC 

138 copy = ddl.FieldSpec( 

139 cls.getCollectionForeignKeyName(prefix), dtype=original.dtype, autoincrement=False, **kwargs 

140 ) 

141 tableSpec.fields.add(copy) 

142 if constraint: 

143 tableSpec.foreignKeys.append( 

144 ddl.ForeignKeySpec( 

145 "collection", source=(copy.name,), target=(original.name,), onDelete=onDelete 

146 ) 

147 ) 

148 return copy 

149 

150 @classmethod 

151 def addRunForeignKey( 

152 cls, 

153 tableSpec: ddl.TableSpec, 

154 *, 

155 prefix: str = "run", 

156 onDelete: str | None = None, 

157 constraint: bool = True, 

158 **kwargs: Any, 

159 ) -> ddl.FieldSpec: 

160 # Docstring inherited from CollectionManager. 

161 original = _KEY_FIELD_SPEC 

162 copy = ddl.FieldSpec( 

163 cls.getRunForeignKeyName(prefix), dtype=original.dtype, autoincrement=False, **kwargs 

164 ) 

165 tableSpec.fields.add(copy) 

166 if constraint: 166 ↛ 170line 166 didn't jump to line 170, because the condition on line 166 was never false

167 tableSpec.foreignKeys.append( 

168 ddl.ForeignKeySpec("run", source=(copy.name,), target=(original.name,), onDelete=onDelete) 

169 ) 

170 return copy 

171 

172 def getParentChains(self, key: int) -> set[str]: 

173 # Docstring inherited from CollectionManager. 

174 chain = self._tables.collection_chain 

175 collection = self._tables.collection 

176 sql = ( 

177 sqlalchemy.sql.select(collection.columns["name"]) 

178 .select_from(collection) 

179 .join(chain, onclause=collection.columns[self._collectionIdName] == chain.columns["parent"]) 

180 .where(chain.columns["child"] == key) 

181 ) 

182 with self._db.query(sql) as sql_result: 

183 parent_names = set(sql_result.scalars().all()) 

184 return parent_names 

185 

186 def _fetch_by_name(self, names: Iterable[str]) -> list[CollectionRecord[int]]: 

187 # Docstring inherited from base class. 

188 _LOG.debug("Fetching collection records using names %s.", names) 

189 return self._fetch("name", names) 

190 

191 def _fetch_by_key(self, collection_ids: Iterable[int] | None) -> list[CollectionRecord[int]]: 

192 # Docstring inherited from base class. 

193 _LOG.debug("Fetching collection records using IDs %s.", collection_ids) 

194 return self._fetch(self._collectionIdName, collection_ids) 

195 

196 def _fetch( 

197 self, column_name: str, collections: Iterable[int | str] | None 

198 ) -> list[CollectionRecord[int]]: 

199 collection_chain = self._tables.collection_chain 

200 collection = self._tables.collection 

201 sql = sqlalchemy.sql.select(*collection.columns, *self._tables.run.columns).select_from( 

202 collection.join(self._tables.run, isouter=True) 

203 ) 

204 

205 chain_sql = ( 

206 sqlalchemy.sql.select( 

207 collection_chain.columns["parent"], 

208 collection_chain.columns["position"], 

209 collection.columns["name"].label("child_name"), 

210 ) 

211 .select_from(collection_chain) 

212 .join( 

213 collection, 

214 onclause=collection_chain.columns["child"] == collection.columns[self._collectionIdName], 

215 ) 

216 ) 

217 

218 records: list[CollectionRecord[int]] = [] 

219 # We want to keep transactions as short as possible. When we fetch 

220 # everything we want to quickly fetch things into memory and finish 

221 # transaction. When we fetch just few records we need to process first 

222 # query before wi can run second one, 

223 if collections is not None: 

224 sql = sql.where(collection.columns[column_name].in_(collections)) 

225 with self._db.transaction(): 

226 with self._db.query(sql) as sql_result: 

227 sql_rows = sql_result.mappings().fetchall() 

228 

229 records, chained_ids = self._rows_to_records(sql_rows) 

230 

231 if chained_ids: 

232 chain_sql = chain_sql.where(collection_chain.columns["parent"].in_(list(chained_ids))) 

233 

234 with self._db.query(chain_sql) as sql_result: 

235 chain_rows = sql_result.mappings().fetchall() 

236 

237 records += self._rows_to_chains(chain_rows, chained_ids) 

238 

239 else: 

240 with self._db.transaction(): 

241 with self._db.query(sql) as sql_result: 

242 sql_rows = sql_result.mappings().fetchall() 

243 with self._db.query(chain_sql) as sql_result: 

244 chain_rows = sql_result.mappings().fetchall() 

245 

246 records, chained_ids = self._rows_to_records(sql_rows) 

247 records += self._rows_to_chains(chain_rows, chained_ids) 

248 

249 return records 

250 

251 def _rows_to_records(self, rows: Iterable[Mapping]) -> tuple[list[CollectionRecord[int]], dict[int, str]]: 

252 """Convert rows returned from collection query to a list of records 

253 and a dict chained collection names. 

254 """ 

255 records: list[CollectionRecord[int]] = [] 

256 chained_ids: dict[int, str] = {} 

257 TimespanReprClass = self._db.getTimespanRepresentation() 

258 for row in rows: 

259 key: int = row[self._collectionIdName] 

260 name: str = row[self._tables.collection.columns.name] 

261 type = CollectionType(row["type"]) 

262 record: CollectionRecord[int] 

263 if type is CollectionType.RUN: 

264 record = RunRecord[int]( 

265 key=key, 

266 name=name, 

267 host=row[self._tables.run.columns.host], 

268 timespan=TimespanReprClass.extract(row), 

269 ) 

270 records.append(record) 

271 elif type is CollectionType.CHAINED: 

272 # Need to delay chained collection construction until to 

273 # fetch their children names. 

274 chained_ids[key] = name 

275 else: 

276 record = CollectionRecord[int](key=key, name=name, type=type) 

277 records.append(record) 

278 return records, chained_ids 

279 

280 def _rows_to_chains( 

281 self, rows: Iterable[Mapping], chained_ids: dict[int, str] 

282 ) -> list[CollectionRecord[int]]: 

283 """Convert rows returned from collection chain query to a list of 

284 records. 

285 """ 

286 chains_defs: dict[int, list[tuple[int, str]]] = {chain_id: [] for chain_id in chained_ids} 

287 for row in rows: 

288 chains_defs[row["parent"]].append((row["position"], row["child_name"])) 

289 

290 records: list[CollectionRecord[int]] = [] 

291 for key, children in chains_defs.items(): 

292 name = chained_ids[key] 

293 children_names = [child for _, child in sorted(children)] 

294 record = ChainedCollectionRecord[int]( 

295 key=key, 

296 name=name, 

297 children=children_names, 

298 ) 

299 records.append(record) 

300 

301 return records 

302 

303 @classmethod 

304 def currentVersions(cls) -> list[VersionTuple]: 

305 # Docstring inherited from VersionedExtension. 

306 return [_VERSION]