Coverage for python/lsst/obs/base/gen2to3/repoWalker/builders.py : 28%

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 obs_base.
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"""Classes used in `RepoWalker` construction.
23The objects here form a temporary tree that is pruned and then transformed
24into a similar tree of `PathElementHandler` instances. See `BuilderNode`
25method documentation for more information.
26"""
27from __future__ import annotations
29__all__ = ["BuilderSkipInput", "BuilderTargetInput", "BuilderTree"]
31from abc import ABC, abstractmethod
32import os
33import re
34from typing import (
35 Any,
36 Dict,
37 List,
38 Optional,
39 Tuple,
40)
42from lsst.daf.butler import DatasetType, DimensionUniverse, StorageClass, FormatterParameter
43from ..translators import Translator
44from .parser import PathElementParser
45from .scanner import PathElementHandler, DirectoryScanner
46from .handlers import (IgnoreHandler, SubdirectoryHandler, SkipHandler,
47 TargetFileHandler)
50class BuilderNode(ABC):
51 """Abstract interface for nodes in the temporary tree that is used to
52 construct a `RepoWalker`.
53 """
55 @abstractmethod
56 def prune(self) -> Tuple[BuilderNode, List[str], bool]:
57 """Attempt to prune this node and its children from the tree.
59 Returns
60 -------
61 replacement : `BuilderNode`
62 The result of recursively pruning child nodes; often just ``self``.
63 messages : `list` [`str`]
64 Warning messages that should be logged by a parent node when a
65 matching path element is encountered, if this node is pruned.
66 prune : `bool`
67 If `True`, this node may be pruned from the tree (but will not
68 necessarily be - it may correspond to a path element that should
69 be skipped with siblings that should not be).
70 """
71 raise NotImplementedError()
73 @abstractmethod
74 def build(self, parser: PathElementParser, allKeys: Dict[str, type], cumulativeKeys: Dict[str, type], *,
75 fileIgnoreRegEx: Optional[re.Pattern], dirIgnoreRegEx: Optional[re.Pattern]
76 ) -> PathElementHandler:
77 """Transform this node in the build tree into a corresponding
78 `PathElementHandler`, recursing to any children.
80 Must be called after `prune`.
82 Parameters
83 ----------
84 parser : `PathElementParser`
85 An object that matches the path element the new handler is
86 responsible for and extracts a (partial) Gen2 data ID from it.
87 allKeys : `dict` [`str`, `type`]
88 A mapping from Gen2 data ID key to the type of its value. Will
89 contain all keys that may be extracted by the given parser, and
90 possibly others.
91 cumulativeKeys : `dict` [`str`, `type`], optional
92 A dictionary containing key strings and types for Gen2 data ID keys
93 that have been extracted from previous path elements for this
94 template, including those extracted by ``parser``.
96 Returns
97 -------
98 handler : `PathElementHandler`
99 A new handler object.
100 """
101 raise NotImplementedError()
104class BuilderInput(BuilderNode):
105 """An intermediate base for `BuilderNode` classes that are provided as
106 direct inputs to a `RepoWalker`, and generally correspond to exactly one
107 Gen2 dataset type.
109 Parameters
110 ----------
111 template : `str`
112 The complete Gen2 template to be matched (not just the template for
113 one path element).
114 keys : `dict` [`str`, `type`]
115 A mapping from Gen2 data ID key to the type of its value.
116 """
117 def __init__(self, template: str, keys: Dict[str, type]):
118 self.template = template
119 self.keys = keys
120 self.elements = self.template.split(os.path.sep)
122 template: str
123 """The complete Gen2 template to be matched (`str`).
124 """
126 keys: Dict[str, type]
127 """A mapping from Gen2 data ID key to the type of its value
128 (`dict` [`str`, `type`]).
129 """
131 elements: List[str]
132 """The path elements (file or directory levels) of `template`
133 (`list` of `str`).
134 """
137class BuilderSkipInput(BuilderInput):
138 """An input to a `RepoWalker` that indicates that matched files should be
139 skipped, possibly with a warning message.
141 BuilderSkipInputs can be pruned. When they are not pruned, they build
142 `SkipHandler` instances.
144 Parameters
145 ----------
146 template : `str`
147 The complete Gen2 template to be matched (not just the template for
148 one path element).
149 keys : `dict` [`str`, `type`]
150 A mapping from Gen2 data ID key to the type of its value.
151 message : `str`, optional
152 If not `None`, a warning message that should be printed either when a
153 matching file is enountered or a directory that may contain such files
154 is skipped.
155 isForFiles : `bool`, optional
156 If `True` (default), this handler should be run on files. Otherwise it
157 should be run on directories.
158 """
159 def __init__(self, template: str, keys: Dict[str, type], message: Optional[str] = None, *,
160 isForFiles: bool = True):
161 super().__init__(template=template, keys=keys)
162 self._message = message
163 self._isForFiles = isForFiles
165 def build(self, parser: PathElementParser, allKeys: Dict[str, type], cumulativeKeys: Dict[str, type], *,
166 fileIgnoreRegEx: Optional[re.Pattern], dirIgnoreRegEx: Optional[re.Pattern]
167 ) -> PathElementHandler:
168 # Docstring inherited from BuilderNode.
169 return SkipHandler(parser=parser, isForFiles=self._isForFiles, message=self._message)
171 def prune(self) -> Tuple[BuilderNode, List[str], bool]:
172 # Docstring inherited from BuilderNode.
173 return self, [self._message] if self._message is not None else [], True
176class BuilderTargetInput(BuilderInput):
177 """An input to a `RepoWalker` that matches files that correspond to
178 datasets that we want to extract.
180 BuilderTargetInputs can never be pruned, and always build
181 `TargetFileHandler` instances.
183 Parameters
184 ----------
185 datasetTypeName : `str`
186 Name of the dataset type.
187 template : `str`
188 Full Gen2 filename template.
189 keys : `dict` [`str`, `type`]
190 Dictionary that maps Gen2 data ID key to the type of its value.
191 storageClass : `StorageClass`
192 `StorageClass` for the Gen3 dataset type.
193 universe : `DimensionUniverse`
194 All candidate dimensions for the Gen3 dataset type.
195 formatter : `lsst.daf.butler.Formatter` or `str`, optional
196 A Gen 3 formatter class or fully-qualified name.
197 targetHandler : `PathElementHandler`, optional
198 Override target handler for this dataset type.
199 kwargs:
200 Additional keyword arguments are passed to `Translator.makeMatching`,
201 in along with ``datasetTypeName`` and ``keys``.
202 """
203 def __init__(self, *, datasetTypeName: str, template: str, keys: Dict[str, type],
204 storageClass: StorageClass, universe: DimensionUniverse,
205 formatter: FormatterParameter,
206 targetHandler: Optional[PathElementHandler] = None, **kwargs: Any):
207 # strip off [%HDU] identifiers from e.g. DECAM Community Pipeline products
208 template = template.split('[%(')[0]
209 super().__init__(template=template, keys=keys)
210 self._translator = Translator.makeMatching(datasetTypeName, keys, **kwargs)
211 self.datasetType = DatasetType(datasetTypeName, dimensions=self._translator.dimensionNames,
212 storageClass=storageClass, universe=universe)
213 self._formatter = formatter
214 if targetHandler is None:
215 targetHandler = TargetFileHandler
216 self._handler = targetHandler
218 def build(self, parser: PathElementParser, allKeys: Dict[str, type], cumulativeKeys: Dict[str, type], *,
219 fileIgnoreRegEx: Optional[re.Pattern], dirIgnoreRegEx: Optional[re.Pattern]
220 ) -> PathElementHandler:
222 return self._handler(parser=parser, translator=self._translator, datasetType=self.datasetType,
223 formatter=self._formatter)
225 def prune(self) -> Tuple[BuilderNode, List[str], bool]:
226 # Docstring inherited from BuilderNode.
227 return self, [], False
229 datasetType: DatasetType
230 """The Gen3 dataset type extracted by the handler this object builds
231 (`lsst.daf.butler.DatasetType`).
232 """
235class BuilderPrunedTree(BuilderNode):
236 """A `BuilderNode` that represents a subdirectory to be skipped,
237 created by pruning `BuilderTree` that contained only `BuilderSkipInput`
238 instances.
240 BuilderPrunedTrees can be pruned. When they are not pruned, they
241 build `SkipHandler` instances.
243 Parameters
244 ----------
245 messages : `list` [`str`]
246 A list of warning messages to be printed when the handler produced by
247 this builder matches a subdirectory.
248 """
250 def __init__(self, messages: List[str]):
251 self._messages = messages
253 def build(self, parser: PathElementParser, allKeys: Dict[str, type], cumulativeKeys: Dict[str, type], *,
254 fileIgnoreRegEx: Optional[re.Pattern], dirIgnoreRegEx: Optional[re.Pattern]
255 ) -> PathElementHandler:
256 # Docstring inherited from BuilderNode.
257 message = "; ".join(self._messages) if self._messages else None
258 return SkipHandler(parser=parser, isForFiles=False, message=message)
260 def prune(self) -> Tuple[BuilderNode, List[str], bool]:
261 # Docstring inherited from BuilderNode.
262 return self, self._messages, True
265class BuilderDuplicateInputs(BuilderNode):
266 """A `BuilderNode` that represents a collection of `BuilderInput` instances
267 that all have the same template.
268 """
269 def __init__(self, old: BuilderInput, new: BuilderInput):
270 self._children = []
271 if isinstance(old, BuilderDuplicateInputs):
272 self._children.extend(old._children)
273 else:
274 self._children.append(old)
275 self._children.append(new)
276 self._messages = [] # populated in prune()
278 def build(self, parser: PathElementParser, allKeys: Dict[str, type], cumulativeKeys: Dict[str, type], *,
279 fileIgnoreRegEx: Optional[re.Pattern], dirIgnoreRegEx: Optional[re.Pattern]
280 ) -> PathElementHandler:
281 # Docstring inherited from BuilderNode.
282 message = "; ".join(self._messages) if self._messages else None
283 return SkipHandler(parser=parser, isForFiles=False, message=message)
285 def prune(self) -> Tuple[BuilderNode, List[str], bool]:
286 # Docstring inherited from BuilderNode.
287 unprunable = []
288 newChildren = []
289 for child in self._children:
290 newChild, childMessages, toPruneChild = child.prune()
291 if toPruneChild:
292 self._messages.extend(childMessages)
293 else:
294 unprunable.append(newChild)
295 newChildren.append(newChildren)
296 self._children = newChildren
297 if len(unprunable) == 0:
298 # All children are just skips, so we can prune this node if we
299 # remember their messages.
300 return self, self._messages, True
301 elif len(unprunable) == 1 and not self._messages:
302 # Exactly one child is a target, and the others were ignored with
303 # no warning messages. Tell parent node to just use that child,
304 # so if we see any matching files, we just assume they're for that
305 # target.
306 return unprunable[0], [], False
307 else:
308 # Multiple targets or skips with messages, which means we won't
309 # know how to handle any matching files. Replace any messages we
310 # have with a single message that combines them all as well as
311 # any target dataset types that they are ambiguous with.
312 nested = [f"{c.datasetType.name} (target)" for c in unprunable]
313 nested.extend(self._messages)
314 self._messages = [f"ambiguous match: [{', '.join(nested)}]"]
315 return self, self._messages, True
318class BuilderTree(BuilderNode):
319 """A `BuilderNode` that represents a directory.
321 This is the only `BuilderNode` class that is not a leaf node. If all
322 of its children can be pruned, it is replaced by a `BuilderPrunedTree`
323 (which can then be pruned itself). It builds `SubdirectoryHandler`
324 instances when not pruned.
325 """
326 def __init__(self):
327 self._children = {} # Maps template path element to BuilderNode
329 def insert(self, level: int, leaf: BuilderInput):
330 """Insert an input leaf node into the tree, recursively constructing
331 intermediate parents in order to put it at the right level.
333 Parameters
334 ----------
335 level : `int`
336 The level ``self``is at in the larger tree, with zero the
337 repository root. The right level for the leaf is given by the
338 length of ``leaf.elements``.
339 leaf : `BuilderInput`
340 The leaf node to insert.
341 """
342 nextLevel = level + 1
343 element = leaf.elements[level]
344 if nextLevel == len(leaf.elements):
345 conflict = self._children.get(element)
346 if conflict is not None:
347 # Sadly, the Gen2 butler has some actual dataset types that
348 # use the exact same template.
349 leaf = BuilderDuplicateInputs(conflict, leaf)
350 self._children[element] = leaf
351 else:
352 child = self._children.setdefault(element, BuilderTree())
353 child.insert(nextLevel, leaf)
355 def fill(self, scanner: DirectoryScanner, allKeys: Dict[str, type], previousKeys: Dict[str, type], *,
356 fileIgnoreRegEx: Optional[re.Pattern], dirIgnoreRegEx: Optional[re.Pattern]):
357 """Fill a `DirectoryScanner` instance by recursively building all
358 child nodes.
360 Parameters
361 ----------
362 scanner : `DirectoryScanner`
363 Object to populate.
364 allKeys : `dict` [`str`, `type`]
365 Mapping from Gen2 data ID key to its value type, covering all keys
366 that could be used in any child template.
367 previousKeys : `dict` [`str`, `type`], optional
368 A dictionary containing key strings and types for Gen2 data ID keys
369 that have been extracted from previous path elements of the same
370 template.
371 """
372 if fileIgnoreRegEx is not None:
373 scanner.add(IgnoreHandler(fileIgnoreRegEx, isForFiles=True))
374 if dirIgnoreRegEx is not None:
375 scanner.add(IgnoreHandler(dirIgnoreRegEx, isForFiles=False))
376 for template, child in self._children.items():
377 parser = PathElementParser(template, allKeys, previousKeys=previousKeys)
378 cumulativeKeys = previousKeys.copy()
379 cumulativeKeys.update(parser.keys)
380 scanner.add(child.build(parser, allKeys, cumulativeKeys, fileIgnoreRegEx=fileIgnoreRegEx,
381 dirIgnoreRegEx=dirIgnoreRegEx))
383 def prune(self) -> Tuple[BuilderNode, List[str], bool]:
384 # Docstring inherited from BuilderNode.
385 toPruneThis = True
386 newChildren = {}
387 messages = []
388 # Recursively prune children.
389 for template, child in list(self._children.items()):
390 newChild, childMessages, toPruneChild = child.prune()
391 newChildren[template] = newChild
392 messages.extend(childMessages)
393 if not toPruneChild:
394 toPruneThis = False
395 self._children = newChildren
396 if toPruneThis:
397 return BuilderPrunedTree(messages), messages, True
398 else:
399 return self, [], False
401 def build(self, parser: PathElementParser, allKeys: Dict[str, type], cumulativeKeys: Dict[str, type], *,
402 fileIgnoreRegEx: Optional[re.Pattern], dirIgnoreRegEx: Optional[re.Pattern]
403 ) -> PathElementHandler:
404 # Docstring inherited from BuilderNode.
405 built = SubdirectoryHandler(parser)
406 self.fill(built.scanner, allKeys, cumulativeKeys, fileIgnoreRegEx=fileIgnoreRegEx,
407 dirIgnoreRegEx=dirIgnoreRegEx)
408 return built