Coverage for python/lsst/pipe/base/tests/mocks/_storage_class.py: 43%
141 statements
« prev ^ index » next coverage.py v7.2.7, created at 2023-07-23 08:14 +0000
« prev ^ index » next coverage.py v7.2.7, created at 2023-07-23 08:14 +0000
1# This file is part of pipe_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/>.
22from __future__ import annotations
24__all__ = (
25 "MockDataset",
26 "MockStorageClass",
27 "MockDatasetQuantum",
28 "MockStorageClassDelegate",
29 "get_mock_name",
30 "get_original_name",
31 "is_mock_name",
32)
34from collections.abc import Callable, Iterable, Mapping
35from typing import Any, cast
37from lsst.daf.butler import (
38 DatasetComponent,
39 Formatter,
40 FormatterFactory,
41 LookupKey,
42 SerializedDataCoordinate,
43 SerializedDatasetRef,
44 SerializedDatasetType,
45 StorageClass,
46 StorageClassDelegate,
47 StorageClassFactory,
48)
49from lsst.daf.butler._compat import _BaseModelCompat
50from lsst.daf.butler.formatters.json import JsonFormatter
51from lsst.utils.introspection import get_full_type_name
53_NAME_PREFIX: str = "_mock_"
56def get_mock_name(original: str) -> str:
57 """Return the name of the mock storage class, dataset type, or task label
58 for the given original name.
59 """
60 return _NAME_PREFIX + original
63def get_original_name(mock: str) -> str:
64 """Return the name of the original storage class, dataset type, or task
65 label that corresponds to the given mock name.
66 """
67 assert mock.startswith(_NAME_PREFIX)
68 return mock.removeprefix(_NAME_PREFIX)
71def is_mock_name(name: str) -> bool:
72 """Return whether the given name is that of a mock storage class, dataset
73 type, or task label.
74 """
75 return name.startswith(_NAME_PREFIX)
78# Tests for this module are in the ci_middleware package, where we have easy
79# access to complex real storage classes (and their pytypes) to test against.
82class MockDataset(_BaseModelCompat):
83 """The in-memory dataset type used by `MockStorageClass`."""
85 ref: SerializedDatasetRef
86 """Reference used to read and write this dataset.
88 This is a `~lsst.daf.butler.SerializedDatasetRef` instead of a "real" one
89 for two reasons:
91 - the mock dataset may need to be read from disk in a context in which a
92 `~lsst.daf.butler.DimensionUniverse` is unavailable;
93 - we don't want the complexity of having a separate
94 ``SerializedMockDataset``.
96 The downside of this is that we end up effectively reimplementing a few
97 fairly trivial DatasetType/DatasetRef methods that override storage classes
98 and extract components (in `MockStorageClass` and
99 `MockStorageClassDelegate`).
100 """
102 quantum: MockDatasetQuantum | None = None
103 """Description of the quantum that produced this dataset.
104 """
106 output_connection_name: str | None = None
107 """The name of the PipelineTask output connection that produced this
108 dataset.
109 """
111 converted_from: MockDataset | None = None
112 """Another `MockDataset` that underwent a storage class conversion to
113 produce this one.
114 """
116 parent: MockDataset | None = None
117 """Another `MockDataset` from which a component was extract to form this
118 one.
119 """
121 parameters: dict[str, str] | None = None
122 """`repr` of all parameters applied when reading this dataset."""
124 @property
125 def dataset_type(self) -> SerializedDatasetType:
126 return cast(SerializedDatasetType, self.ref.datasetType)
128 @property
129 def storage_class(self) -> str:
130 return cast(str, self.dataset_type.storageClass)
132 def make_derived(self, **kwargs: Any) -> MockDataset:
133 """Return a new MockDataset that represents applying some storage class
134 operation to this one.
136 Keyword arguments are fields of `MockDataset` or
137 `~lsst.daf.butler.SerializedDatasetType` to override in the result.
138 """
139 dataset_type_updates = {
140 k: kwargs.pop(k) for k in list(kwargs) if k in SerializedDatasetType.model_fields # type: ignore
141 }
142 derived_dataset_type = self.dataset_type.copy(update=dataset_type_updates)
143 derived_ref = self.ref.copy(update=dict(datasetType=derived_dataset_type))
144 # Fields below are those that should not be propagated to the derived
145 # dataset, because they're not about the intrinsic on-disk thing.
146 kwargs.setdefault("converted_from", None)
147 kwargs.setdefault("parent", None)
148 kwargs.setdefault("parameters", None)
149 # Also use setdefault on the ref in case caller wants to override that
150 # directly, but this is expected to be rare enough that it's not worth
151 # it to try to optimize out the work above to make derived_ref.
152 kwargs.setdefault("ref", derived_ref)
153 return self.copy(update=kwargs)
156class MockDatasetQuantum(_BaseModelCompat):
157 """Description of the quantum that produced a mock dataset."""
159 task_label: str
160 """Label of the producing PipelineTask in its pipeline."""
162 data_id: SerializedDataCoordinate
163 """Data ID for the quantum."""
165 inputs: dict[str, list[MockDataset]]
166 """Mock datasets provided as input to the quantum."""
169MockDataset.update_forward_refs()
172class MockStorageClassDelegate(StorageClassDelegate):
173 """Implementation of the StorageClassDelegate interface for mock datasets.
175 This class does not implement assembly and disassembly just because it's
176 not needed right now. That could be added in the future with some
177 additional tracking attributes in `MockDataset`.
178 """
180 def assemble(self, components: dict[str, Any], pytype: type | None = None) -> MockDataset:
181 # Docstring inherited.
182 raise NotImplementedError("Mock storage classes do not implement assembly.")
184 def getComponent(self, composite: Any, componentName: str) -> Any:
185 # Docstring inherited.
186 assert isinstance(
187 composite, MockDataset
188 ), f"MockStorageClassDelegate given a non-mock dataset {composite!r}."
189 return composite.make_derived(
190 name=f"{composite.dataset_type.name}.{componentName}",
191 storageClass=self.storageClass.allComponents()[componentName].name,
192 parentStorageClass=self.storageClass.name,
193 parent=composite,
194 )
196 def disassemble(
197 self, composite: Any, subset: Iterable | None = None, override: Any | None = None
198 ) -> dict[str, DatasetComponent]:
199 # Docstring inherited.
200 raise NotImplementedError("Mock storage classes do not implement disassembly.")
202 def handleParameters(self, inMemoryDataset: Any, parameters: Mapping[str, Any] | None = None) -> Any:
203 # Docstring inherited.
204 assert isinstance(
205 inMemoryDataset, MockDataset
206 ), f"MockStorageClassDelegate given a non-mock dataset {inMemoryDataset!r}."
207 if not parameters:
208 return inMemoryDataset
209 return inMemoryDataset.make_derived(parameters={k: repr(v) for k, v in parameters.items()})
212class MockStorageClass(StorageClass):
213 """A reimplementation of `lsst.daf.butler.StorageClass` for mock datasets.
215 Each `MockStorageClass` instance corresponds to a real "original" storage
216 class, with components and conversions that are mocks of the original's
217 components and conversions. The `pytype` for all `MockStorageClass`
218 instances is `MockDataset`.
219 """
221 def __init__(self, original: StorageClass, factory: StorageClassFactory | None = None):
222 name = get_mock_name(original.name)
223 if factory is None:
224 factory = StorageClassFactory()
225 super().__init__(
226 name=name,
227 pytype=MockDataset,
228 components={
229 k: self.get_or_register_mock(v.name, factory) for k, v in original.components.items()
230 },
231 derivedComponents={
232 k: self.get_or_register_mock(v.name, factory) for k, v in original.derivedComponents.items()
233 },
234 parameters=frozenset(original.parameters),
235 delegate=get_full_type_name(MockStorageClassDelegate),
236 # Conversions work differently for mock storage classes, since they
237 # all have the same pytype: we use the original storage class being
238 # mocked to see if we can convert, then just make a new MockDataset
239 # that points back to the original.
240 converters={},
241 )
242 self.original = original
243 # Make certain no one tries to use the converters.
244 self._converters = None # type: ignore
246 def _get_converters_by_type(self) -> dict[type, Callable[[Any], Any]]:
247 # Docstring inherited.
248 raise NotImplementedError("MockStorageClass does not use converters.")
250 @classmethod
251 def get_or_register_mock(
252 cls, original: str, factory: StorageClassFactory | None = None
253 ) -> MockStorageClass:
254 """Return a mock storage class for the given original storage class,
255 creating and registering it if necessary.
257 Parameters
258 ----------
259 original : `str`
260 Name of the original storage class to be mocked.
261 factory : `~lsst.daf.butler.StorageClassFactory`, optional
262 Storage class factory singleton instance.
264 Returns
265 -------
266 mock : `MockStorageClass`
267 New storage class that mocks ``original``.
268 """
269 name = get_mock_name(original)
270 if factory is None:
271 factory = StorageClassFactory()
272 if name in factory:
273 return cast(MockStorageClass, factory.getStorageClass(name))
274 else:
275 result = cls(factory.getStorageClass(original), factory)
276 factory.registerStorageClass(result)
277 return result
279 def allComponents(self) -> Mapping[str, MockStorageClass]:
280 # Docstring inherited.
281 return cast(Mapping[str, MockStorageClass], super().allComponents())
283 @property
284 def components(self) -> Mapping[str, MockStorageClass]:
285 # Docstring inherited.
286 return cast(Mapping[str, MockStorageClass], super().components)
288 @property
289 def derivedComponents(self) -> Mapping[str, MockStorageClass]:
290 # Docstring inherited.
291 return cast(Mapping[str, MockStorageClass], super().derivedComponents)
293 def can_convert(self, other: StorageClass) -> bool:
294 # Docstring inherited.
295 if not isinstance(other, MockStorageClass):
296 return False
297 return self.original.can_convert(other.original)
299 def coerce_type(self, incorrect: Any) -> Any:
300 # Docstring inherited.
301 if not isinstance(incorrect, MockDataset):
302 raise TypeError(
303 f"Mock storage class {self.name!r} can only convert in-memory datasets "
304 f"corresponding to other mock storage classes, not {incorrect!r}."
305 )
306 factory = StorageClassFactory()
307 other_storage_class = factory.getStorageClass(incorrect.storage_class)
308 assert isinstance(other_storage_class, MockStorageClass), "Should not get a MockDataset otherwise."
309 if other_storage_class.name == self.name:
310 return incorrect
311 if not self.can_convert(other_storage_class):
312 raise TypeError(
313 f"Mocked storage class {self.original.name!r} cannot convert from "
314 f"{other_storage_class.original.name!r}."
315 )
316 return incorrect.make_derived(storageClass=self.name, converted_from=incorrect)
319def _monkeypatch_daf_butler() -> None:
320 """Replace methods in daf_butler's StorageClassFactory and FormatterFactory
321 classes to automatically recognize mock storage classes.
323 This monkey-patching is executed when the `lsst.pipe.base.tests.mocks`
324 package is imported, and it affects all butler instances created before or
325 after that imported.
326 """
327 original_get_storage_class = StorageClassFactory.getStorageClass
329 def new_get_storage_class(self: StorageClassFactory, storageClassName: str) -> StorageClass:
330 try:
331 return original_get_storage_class(self, storageClassName)
332 except KeyError:
333 if is_mock_name(storageClassName):
334 return MockStorageClass.get_or_register_mock(get_original_name(storageClassName))
335 raise
337 StorageClassFactory.getStorageClass = new_get_storage_class # type: ignore
339 del new_get_storage_class
341 original_get_formatter_class_with_match = FormatterFactory.getFormatterClassWithMatch
343 def new_get_formatter_class_with_match(
344 self: FormatterFactory, entity: Any
345 ) -> tuple[LookupKey, type[Formatter], dict[str, Any]]:
346 try:
347 return original_get_formatter_class_with_match(self, entity)
348 except KeyError:
349 lookup_keys = (LookupKey(name=entity),) if isinstance(entity, str) else entity._lookupNames()
350 for key in lookup_keys:
351 # This matches mock dataset type names before mock storage
352 # classes, and it would even match some regular dataset types
353 # that are automatic connections (logs, configs, metadata) of
354 # mocked tasks. The latter would be a problem, except that
355 # those should have already matched in the try block above.
356 if is_mock_name(key.name):
357 return (key, JsonFormatter, {})
358 raise
360 FormatterFactory.getFormatterClassWithMatch = new_get_formatter_class_with_match # type: ignore
362 del new_get_formatter_class_with_match
364 original_get_formatter_with_match = FormatterFactory.getFormatterWithMatch
366 def new_get_formatter_with_match(
367 self: FormatterFactory, entity: Any, *args: Any, **kwargs: Any
368 ) -> tuple[LookupKey, Formatter]:
369 try:
370 return original_get_formatter_with_match(self, entity, *args, **kwargs)
371 except KeyError:
372 lookup_keys = (LookupKey(name=entity),) if isinstance(entity, str) else entity._lookupNames()
373 for key in lookup_keys:
374 if is_mock_name(key.name):
375 return (key, JsonFormatter(*args, **kwargs))
376 raise
378 FormatterFactory.getFormatterWithMatch = new_get_formatter_with_match # type: ignore
380 del new_get_formatter_with_match
383_monkeypatch_daf_butler()