Coverage for python/lsst/daf/butler/registry/wildcards.py: 25%

220 statements  

« prev     ^ index     » next       coverage.py v7.2.7, created at 2023-08-05 01:26 +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 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__ = ( 

24 "CategorizedWildcard", 

25 "CollectionWildcard", 

26 "CollectionSearch", 

27 "DatasetTypeWildcard", 

28) 

29 

30import contextlib 

31import dataclasses 

32import re 

33from collections.abc import Callable, Iterable, Iterator, Mapping, Sequence 

34from types import EllipsisType 

35from typing import Any 

36 

37from deprecated.sphinx import deprecated 

38from lsst.daf.butler._compat import PYDANTIC_V2 

39from lsst.utils.iteration import ensure_iterable 

40 

41from ..core import DatasetType 

42from ..core.utils import globToRegex 

43from ._exceptions import CollectionExpressionError, DatasetTypeExpressionError 

44 

45 

46@dataclasses.dataclass 

47class CategorizedWildcard: 

48 """The results of preprocessing a wildcard expression to separate match 

49 patterns from strings. 

50 

51 The `fromExpression` method should almost always be used to construct 

52 instances, as the regular constructor performs no checking of inputs (and 

53 that can lead to confusing error messages downstream). 

54 """ 

55 

56 @classmethod 

57 def fromExpression( 

58 cls, 

59 expression: Any, 

60 *, 

61 allowAny: bool = True, 

62 allowPatterns: bool = True, 

63 coerceUnrecognized: Callable[[Any], tuple[str, Any] | str] | None = None, 

64 coerceItemValue: Callable[[Any], Any] | None = None, 

65 defaultItemValue: Any | None = None, 

66 ) -> CategorizedWildcard | EllipsisType: 

67 """Categorize a wildcard expression. 

68 

69 Parameters 

70 ---------- 

71 expression 

72 The expression to categorize. May be any of: 

73 - `str` (including glob patterns if ``allowPatterns`` is `True`); 

74 - `re.Pattern` (only if ``allowPatterns`` is `True`); 

75 - objects recognized by ``coerceUnrecognized`` (if provided); 

76 - two-element tuples of (`str`, value) where value is recognized 

77 by ``coerceItemValue`` (if provided); 

78 - a non-`str`, non-mapping iterable containing any of the above; 

79 - the special value `...` (only if ``allowAny`` is `True`), which 

80 matches anything; 

81 - a mapping from `str` to a value are recognized by 

82 ``coerceItemValue`` (if provided); 

83 - a `CategorizedWildcard` instance (passed through unchanged if 

84 it meets the requirements specified by keyword arguments). 

85 allowAny: `bool`, optional 

86 If `False` (`True` is default) raise `TypeError` if `...` is 

87 encountered. 

88 allowPatterns: `bool`, optional 

89 If `False` (`True` is default) raise `TypeError` if a `re.Pattern` 

90 is encountered, or if ``expression`` is a `CategorizedWildcard` 

91 with `patterns` not empty. 

92 coerceUnrecognized: `~collections.abc.Callable`, optional 

93 A callback that takes a single argument of arbitrary type and 

94 returns either a `str` - appended to `strings` - or a `tuple` of 

95 (`str`, `Any`) to be appended to `items`. This will be called on 

96 objects of unrecognized type. Exceptions will be reraised as 

97 `TypeError` (and chained). 

98 coerceItemValue: `~collections.abc.Callable`, optional 

99 If provided, ``expression`` may be a mapping from `str` to any 

100 type that can be passed to this function; the result of that call 

101 will be stored instead as the value in ``self.items``. 

102 defaultItemValue: `Any`, optional 

103 If provided, combine this value with any string values encountered 

104 (including any returned by ``coerceUnrecognized``) to form a 

105 `tuple` and add it to `items`, guaranteeing that `strings` will be 

106 empty. Patterns are never added to `items`. 

107 

108 Returns 

109 ------- 

110 categorized : `CategorizedWildcard` or ``...``. 

111 The struct describing the wildcard. ``...`` is passed through 

112 unchanged. 

113 

114 Raises 

115 ------ 

116 TypeError 

117 Raised if an unsupported type is found in the expression. 

118 """ 

119 assert expression is not None 

120 # See if we were given ...; just return that if we were. 

121 if expression is ...: 

122 if not allowAny: 

123 raise TypeError("This expression may not be unconstrained.") 

124 return ... 

125 if isinstance(expression, cls): 

126 # This is already a CategorizedWildcard. Make sure it meets the 

127 # reqs. implied by the kwargs we got. 

128 if not allowPatterns and expression.patterns: 

129 raise TypeError( 

130 f"Regular expression(s) {expression.patterns} are not allowed in this context." 

131 ) 

132 if defaultItemValue is not None and expression.strings: 

133 if expression.items: 

134 raise TypeError( 

135 "Incompatible preprocessed expression: an ordered sequence of str is " 

136 "needed, but the original order was lost in the preprocessing." 

137 ) 

138 return cls( 

139 strings=[], 

140 patterns=expression.patterns, 

141 items=[(k, defaultItemValue) for k in expression.strings], 

142 ) 

143 elif defaultItemValue is None and expression.items: 

144 if expression.strings: 

145 raise TypeError( 

146 "Incompatible preprocessed expression: an ordered sequence of items is " 

147 "needed, but the original order was lost in the preprocessing." 

148 ) 

149 return cls(strings=[k for k, _ in expression.items], patterns=expression.patterns, items=[]) 

150 else: 

151 # Original expression was created with keyword arguments that 

152 # were at least as restrictive as what we just got; pass it 

153 # through. 

154 return expression 

155 

156 # If we get here, we know we'll be creating a new instance. 

157 # Initialize an empty one now. 

158 self = cls(strings=[], patterns=[], items=[]) 

159 

160 # If mappings are allowed, see if we were given a single mapping by 

161 # trying to get items. 

162 if coerceItemValue is not None: 

163 rawItems = None 

164 with contextlib.suppress(AttributeError): 

165 rawItems = expression.items() 

166 

167 if rawItems is not None: 

168 for k, v in rawItems: 

169 try: 

170 self.items.append((k, coerceItemValue(v))) 

171 except Exception as err: 

172 raise TypeError(f"Could not coerce mapping value '{v}' for key '{k}'.") from err 

173 return self 

174 

175 # Not ..., a CategorizedWildcard instance, or a mapping. Just 

176 # process scalars or an iterable. We put the body of the loop inside 

177 # a local function so we can recurse after coercion. 

178 

179 def process(element: Any, alreadyCoerced: bool = False) -> EllipsisType | None: 

180 if isinstance(element, str): 

181 if defaultItemValue is not None: 

182 self.items.append((element, defaultItemValue)) 

183 return None 

184 else: 

185 # This returns a list but we know we only passed in 

186 # single value. 

187 converted = globToRegex(element) 

188 if converted is ...: 

189 return ... 

190 element = converted[0] 

191 # Let regex and ... go through to the next check 

192 if isinstance(element, str): 

193 self.strings.append(element) 

194 return None 

195 if allowPatterns and isinstance(element, re.Pattern): 

196 self.patterns.append(element) 

197 return None 

198 if alreadyCoerced: 

199 try: 

200 k, v = element 

201 except TypeError: 

202 raise TypeError( 

203 f"Object '{element!r}' returned by coercion function must be `str` or `tuple`." 

204 ) from None 

205 else: 

206 self.items.append((k, v)) 

207 return None 

208 if coerceItemValue is not None: 

209 try: 

210 k, v = element 

211 except TypeError: 

212 pass 

213 else: 

214 if not isinstance(k, str): 

215 raise TypeError(f"Item key '{k}' is not a string.") 

216 try: 

217 v = coerceItemValue(v) 

218 except Exception as err: 

219 raise TypeError(f"Could not coerce tuple item value '{v}' for key '{k}'.") from err 

220 self.items.append((k, v)) 

221 return None 

222 if coerceUnrecognized is not None: 

223 try: 

224 # This should be safe but flake8 cant tell that the 

225 # function will be re-declared next function call 

226 process(coerceUnrecognized(element), alreadyCoerced=True) # noqa: F821 

227 except Exception as err: 

228 raise TypeError(f"Could not coerce expression element '{element!r}'.") from err 

229 else: 

230 extra = "." 

231 if isinstance(element, re.Pattern): 

232 extra = " and patterns are not allowed." 

233 raise TypeError(f"Unsupported object in wildcard expression: '{element!r}'{extra}") 

234 return None 

235 

236 for element in ensure_iterable(expression): 

237 retval = process(element) 

238 if retval is ...: 

239 # One of the globs matched everything 

240 if not allowAny: 

241 raise TypeError("This expression may not be unconstrained.") 

242 return ... 

243 del process 

244 return self 

245 

246 strings: list[str] 

247 """Explicit string values found in the wildcard (`list` [ `str` ]). 

248 """ 

249 

250 patterns: list[re.Pattern] 

251 """Regular expression patterns found in the wildcard 

252 (`list` [ `re.Pattern` ]). 

253 """ 

254 

255 items: list[tuple[str, Any]] 

256 """Two-item tuples that relate string values to other objects 

257 (`list` [ `tuple` [ `str`, `Any` ] ]). 

258 """ 

259 

260 

261if PYDANTIC_V2: 261 ↛ 262line 261 didn't jump to line 262, because the condition on line 261 was never true

262 from pydantic import RootModel # type: ignore 

263 

264 class _CollectionSearch(RootModel): 

265 root: tuple[str, ...] 

266 

267else: 

268 from pydantic import BaseModel 

269 

270 class _CollectionSearch(BaseModel, Sequence[str]): # type: ignore 

271 __root__: tuple[str, ...] 

272 

273 @property 

274 def root(self) -> tuple[str, ...]: 

275 return self.__root__ 

276 

277 

278@deprecated( 

279 reason="Tuples of string collection names are now preferred. Will be removed after v26.", 

280 version="v25.0", 

281 category=FutureWarning, 

282) 

283class CollectionSearch(_CollectionSearch): 

284 """An ordered search path of collections. 

285 

286 The `fromExpression` method should almost always be used to construct 

287 instances, as the regular constructor performs no checking of inputs (and 

288 that can lead to confusing error messages downstream). 

289 

290 Parameters 

291 ---------- 

292 collections : `tuple` [ `str` ] 

293 Tuple of collection names, ordered from the first searched to the last 

294 searched. 

295 

296 Notes 

297 ----- 

298 A `CollectionSearch` is used to find a single dataset (or set of datasets 

299 with different dataset types or data IDs) according to its dataset type and 

300 data ID, giving preference to collections in the order in which they are 

301 specified. A `CollectionWildcard` can be constructed from a broader range 

302 of expressions but does not order the collections to be searched. 

303 

304 `CollectionSearch` is an immutable sequence of `str` collection names. 

305 

306 A `CollectionSearch` instance constructed properly (e.g. via 

307 `fromExpression`) is a unique representation of a particular search path; 

308 it is exactly the same internally and compares as equal to any 

309 `CollectionSearch` constructed from an equivalent expression, regardless of 

310 how different the original expressions appear. 

311 """ 

312 

313 @classmethod 

314 def fromExpression(cls, expression: Any) -> CollectionSearch: 

315 """Process a general expression to construct a `CollectionSearch` 

316 instance. 

317 

318 Parameters 

319 ---------- 

320 expression 

321 May be: 

322 - a `str` collection name; 

323 - an iterable of `str` collection names; 

324 - another `CollectionSearch` instance (passed through 

325 unchanged). 

326 

327 Duplicate entries will be removed (preserving the first appearance 

328 of each collection name). 

329 

330 Returns 

331 ------- 

332 collections : `CollectionSearch` 

333 A `CollectionSearch` instance. 

334 """ 

335 # First see if this is already a CollectionSearch; just pass that 

336 # through unchanged. This lets us standardize expressions (and turn 

337 # single-pass iterators into multi-pass iterables) in advance and pass 

338 # them down to other routines that accept arbitrary expressions. 

339 if isinstance(expression, cls): 

340 return expression 

341 try: 

342 wildcard = CategorizedWildcard.fromExpression( 

343 expression, 

344 allowAny=False, 

345 allowPatterns=False, 

346 ) 

347 except TypeError as err: 

348 raise CollectionExpressionError(str(err)) from None 

349 assert wildcard is not ... 

350 assert not wildcard.patterns 

351 assert not wildcard.items 

352 deduplicated = [] 

353 for name in wildcard.strings: 

354 if name not in deduplicated: 

355 deduplicated.append(name) 

356 if PYDANTIC_V2: 

357 model = cls(tuple(deduplicated)) # type: ignore 

358 else: 

359 model = cls(__root__=tuple(deduplicated)) # type: ignore 

360 return model 

361 

362 def explicitNames(self) -> Iterator[str]: 

363 """Iterate over collection names that were specified explicitly.""" 

364 yield from self.root 

365 

366 def __iter__(self) -> Iterator[str]: # type: ignore 

367 yield from self.root 

368 

369 def __len__(self) -> int: 

370 return len(self.root) 

371 

372 def __getitem__(self, index: Any) -> str: 

373 return self.root[index] 

374 

375 def __eq__(self, other: Any) -> bool: 

376 if isinstance(other, CollectionSearch): 

377 return self.root == other.root 

378 return False 

379 

380 def __str__(self) -> str: 

381 return "[{}]".format(", ".join(self)) 

382 

383 def __repr__(self) -> str: 

384 return f"CollectionSearch({self.root!r})" 

385 

386 

387@dataclasses.dataclass(frozen=True) 

388class CollectionWildcard: 

389 """A validated wildcard for collection names. 

390 

391 The `from_expression` method should almost always be used to construct 

392 instances, as the regular constructor performs no checking of inputs (and 

393 that can lead to confusing error messages downstream). 

394 

395 Notes 

396 ----- 

397 `CollectionWildcard` is expected to be rarely used outside of `Registry` 

398 (which uses it to back several of its "query" methods that take general 

399 expressions for collections), but it may occasionally be useful outside 

400 `Registry` as a way to preprocess expressions that contain single-pass 

401 iterators into a form that can be used to call those `Registry` methods 

402 multiple times. 

403 """ 

404 

405 strings: tuple[str, ...] = () 

406 """An an ordered list of explicitly-named collections. (`tuple` [ `str` ]). 

407 """ 

408 

409 patterns: tuple[re.Pattern, ...] | EllipsisType = ... 

410 """Regular expression patterns to match against collection names, or the 

411 special value ``...`` indicating all collections. 

412 

413 `...` must be accompanied by ``strings=()``. 

414 """ 

415 

416 def __post_init__(self) -> None: 

417 if self.patterns is ... and self.strings: 

418 raise ValueError( 

419 f"Collection wildcard matches any string, but still has explicit strings {self.strings}." 

420 ) 

421 

422 @classmethod 

423 def from_expression(cls, expression: Any, require_ordered: bool = False) -> CollectionWildcard: 

424 """Process a general expression to construct a `CollectionWildcard` 

425 instance. 

426 

427 Parameters 

428 ---------- 

429 expression 

430 May be: 

431 - a `str` collection name; 

432 - an `re.Pattern` instance to match (with `re.Pattern.fullmatch`) 

433 against collection names; 

434 - any iterable containing any of the above; 

435 - another `CollectionWildcard` instance (passed through 

436 unchanged). 

437 

438 Duplicate collection names will be removed (preserving the first 

439 appearance of each collection name). 

440 require_ordered : `bool`, optional 

441 If `True` (`False` is default) require the expression to be 

442 ordered, and raise `CollectionExpressionError` if it is not. 

443 

444 Returns 

445 ------- 

446 wildcard : `CollectionWildcard` 

447 A `CollectionWildcard` instance. 

448 

449 Raises 

450 ------ 

451 CollectionExpressionError 

452 Raised if the patterns has regular expression, glob patterns, or 

453 the ``...`` wildcard, and ``require_ordered=True``. 

454 """ 

455 if isinstance(expression, cls): 

456 return expression 

457 if expression is ...: 

458 return cls() 

459 wildcard = CategorizedWildcard.fromExpression( 

460 expression, 

461 allowAny=True, 

462 allowPatterns=True, 

463 ) 

464 if wildcard is ...: 

465 return cls() 

466 result = cls( 

467 strings=tuple(wildcard.strings), 

468 patterns=tuple(wildcard.patterns), 

469 ) 

470 if require_ordered: 

471 result.require_ordered() 

472 return result 

473 

474 @classmethod 

475 def from_names(cls, names: Iterable[str]) -> CollectionWildcard: 

476 """Construct from an iterable of explicit collection names. 

477 

478 Parameters 

479 ---------- 

480 names : `~collections.abc.Iterable` [ `str` ] 

481 Iterable of collection names. 

482 

483 Returns 

484 ------- 

485 wildcard : ~CollectionWildcard` 

486 A `CollectionWildcard` instance. `require_ordered` is guaranteed 

487 to succeed and return the given names in order. 

488 """ 

489 return cls(strings=tuple(names), patterns=()) 

490 

491 def require_ordered(self) -> tuple[str, ...]: 

492 """Require that this wildcard contains no patterns, and return the 

493 ordered tuple of names that it does hold. 

494 

495 Returns 

496 ------- 

497 names : `tuple` [ `str` ] 

498 Ordered tuple of collection names. 

499 

500 Raises 

501 ------ 

502 CollectionExpressionError 

503 Raised if the patterns has regular expression, glob patterns, or 

504 the ``...`` wildcard. 

505 """ 

506 if self.patterns: 

507 raise CollectionExpressionError( 

508 f"An ordered collection expression is required; got patterns {self.patterns}." 

509 ) 

510 return self.strings 

511 

512 def empty(self) -> bool: 

513 """Return true if both ``strings`` and ``patterns`` are empty.""" 

514 # bool(Ellipsis) is True 

515 return not self.strings and not self.patterns 

516 

517 def __str__(self) -> str: 

518 if self.patterns is ...: 

519 return "..." 

520 else: 

521 terms = list(self.strings) 

522 terms.extend(str(p) for p in self.patterns) 

523 return "[{}]".format(", ".join(terms)) 

524 

525 

526@dataclasses.dataclass 

527class DatasetTypeWildcard: 

528 """A validated expression that resolves to one or more dataset types. 

529 

530 The `from_expression` method should almost always be used to construct 

531 instances, as the regular constructor performs no checking of inputs (and 

532 that can lead to confusing error messages downstream). 

533 """ 

534 

535 values: Mapping[str, DatasetType | None] = dataclasses.field(default_factory=dict) 

536 """A mapping with `str` dataset type name keys and optional `DatasetType` 

537 instances. 

538 """ 

539 

540 patterns: tuple[re.Pattern, ...] | EllipsisType = ... 

541 """Regular expressions to be matched against dataset type names, or the 

542 special value ``...`` indicating all dataset types. 

543 

544 Any pattern matching a dataset type is considered an overall match for 

545 the expression. 

546 """ 

547 

548 @classmethod 

549 def from_expression(cls, expression: Any) -> DatasetTypeWildcard: 

550 """Construct an instance by analyzing the given expression. 

551 

552 Parameters 

553 ---------- 

554 expression 

555 Expression to analyze. May be any of the following: 

556 

557 - a `str` dataset type name; 

558 - a `DatasetType` instance; 

559 - a `re.Pattern` to match against dataset type names; 

560 - an iterable whose elements may be any of the above (any dataset 

561 type matching any element in the list is an overall match); 

562 - an existing `DatasetTypeWildcard` instance; 

563 - the special ``...`` ellipsis object, which matches any dataset 

564 type. 

565 

566 Returns 

567 ------- 

568 query : `DatasetTypeWildcard` 

569 An instance of this class (new unless an existing instance was 

570 passed in). 

571 

572 Raises 

573 ------ 

574 DatasetTypeExpressionError 

575 Raised if the given expression does not have one of the allowed 

576 types. 

577 """ 

578 if isinstance(expression, cls): 

579 return expression 

580 try: 

581 wildcard = CategorizedWildcard.fromExpression( 

582 expression, coerceUnrecognized=lambda d: (d.name, d) 

583 ) 

584 except TypeError as err: 

585 raise DatasetTypeExpressionError(f"Invalid dataset type expression: {expression!r}.") from err 

586 if wildcard is ...: 

587 return cls() 

588 values: dict[str, DatasetType | None] = {} 

589 for name in wildcard.strings: 

590 values[name] = None 

591 for name, item in wildcard.items: 

592 if not isinstance(item, DatasetType): 

593 raise DatasetTypeExpressionError( 

594 f"Invalid value '{item}' of type {type(item)} in dataset type expression; " 

595 "expected str, re.Pattern, DatasetType objects, iterables thereof, or '...'." 

596 ) 

597 values[name] = item 

598 return cls(values, patterns=tuple(wildcard.patterns)) 

599 

600 def __str__(self) -> str: 

601 if self.patterns is ...: 

602 return "..." 

603 else: 

604 terms = list(self.values.keys()) 

605 terms.extend(str(p) for p in self.patterns) 

606 return "[{}]".format(", ".join(terms))