Coverage for python/lsst/analysis/tools/interfaces/_analysisTools.py: 23%
139 statements
« prev ^ index » next coverage.py v7.4.0, created at 2024-01-10 14:10 +0000
« prev ^ index » next coverage.py v7.4.0, created at 2024-01-10 14:10 +0000
1# This file is part of analysis_tools.
2#
3# Developed for the LSST Data Management System.
4# This product includes software developed by the LSST Project
5# (https://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 <https://www.gnu.org/licenses/>.
22from __future__ import annotations
24__all__ = ("AnalysisTool",)
26from collections.abc import Mapping
27from functools import wraps
28from typing import Callable, Iterable, Protocol, runtime_checkable
30import lsst.pex.config as pexConfig
31from lsst.pex.config import Field, ListField
32from lsst.pex.config.configurableActions import ConfigurableActionField
33from lsst.verify import Measurement
35from ._actions import AnalysisAction, JointAction, JointResults, NoPlot, PlotAction
36from ._interfaces import KeyedData, KeyedDataSchema, KeyedResults, PlotTypes
37from ._stages import BasePrep, BaseProcess, BaseProduce
40@runtime_checkable
41class _HasOutputNames(Protocol):
42 def getOutputNames(self, config: pexConfig.Config | None = None) -> Iterable[str]:
43 ...
46def _finalizeWrapper(
47 f: Callable[[AnalysisTool], None], cls: type[AnalysisTool]
48) -> Callable[[AnalysisTool], None]:
49 """Wrap a classes finalize function to ensure the base classes special
50 finalize method only fires after the most derived finalize method.
52 Parameters
53 ----------
54 f : `Callable`
55 Function that is being wrapped
56 cls : `type` of `AnalysisTool`
57 The class which is having its function wrapped
59 Returns
60 -------
61 function : `Callable`
62 The new function which wraps the old
63 """
65 @wraps(f)
66 def wrapper(self: AnalysisTool) -> None:
67 # call the wrapped finalize function
68 f(self)
69 # get the method resolution order for the self variable
70 mro = self.__class__.mro()
72 # Find which class in the mro that last defines a finalize method
73 # note that this is in the reverse order from the mro, because the
74 # last class in an inheritance stack is the first in the mro (aka you
75 # walk from the furthest child first.
76 #
77 # Also note that the most derived finalize method need not be the same
78 # as the type of self, as that might inherit from a parent somewhere
79 # between it and the furthest parent.
80 mostDerived: type | None = None
81 for klass in mro:
82 # inspect the classes dictionary to see if it specifically defines
83 # finalize. This is needed because normal lookup will go through
84 # the mro, but this needs to be restricted to each class.
85 if "finalize" in vars(klass):
86 mostDerived = klass
87 break
89 # Find what stage in the MRO walking process the recursive function
90 # call is in.
91 this = super(cls, self).__thisclass__
93 # If the current place in the MRO walking is also the class that
94 # defines the most derived instance of finalize, then call the base
95 # classes private finalize that must be called after everything else.
96 if mostDerived is not None and this == mostDerived:
97 self._baseFinalize()
99 return wrapper
102class AnalysisTool(AnalysisAction):
103 r"""A tool which which calculates a single type of analysis on input data,
104 though it may return more than one result.
106 Although `AnalysisTool`\ s are considered a single type of analysis, the
107 classes themselves can be thought of as a container. `AnalysisTool`\ s
108 are aggregations of `AnalysisAction`\ s to form prep, process, and
109 produce stages. These stages allow better reuse of individual
110 `AnalysisActions` and easier introspection in contexts such as a notebook
111 or interpreter.
113 An `AnalysisTool` can be thought of an an individual configuration that
114 specifies which `AnalysisAction` should run for each stage.
116 The stages themselves are also configurable, allowing control over various
117 aspects of the individual `AnalysisAction`\ s.
118 """
119 prep = ConfigurableActionField[AnalysisAction](doc="Action to run to prepare inputs", default=BasePrep)
120 process = ConfigurableActionField[AnalysisAction](
121 doc="Action to process data into intended form", default=BaseProcess
122 )
123 produce = ConfigurableActionField[AnalysisAction](
124 doc="Action to perform any finalization steps", default=BaseProduce
125 )
126 metric_tags = ListField[str](
127 doc="List of tags which will be associated with metric measurement(s)", default=[]
128 )
130 def __init_subclass__(cls: type[AnalysisTool], **kwargs):
131 super().__init_subclass__(**kwargs)
132 # Wrap all definitions of the finalize method in a special wrapper that
133 # ensures that the bases classes private finalize is called last.
134 if "finalize" in vars(cls):
135 cls.finalize = _finalizeWrapper(cls.finalize, cls)
137 dynamicOutputNames: bool | Field[bool] = False
138 """Determines whether to grant the ``getOutputNames`` method access to
139 config parameters.
140 """
142 parameterizedBand: bool | Field[bool] = True
143 """Specifies if an `AnalysisTool` may parameterize a band within any field
144 in any stage, or if the set of bands is already uniquely determined though
145 configuration. I.e. can this `AnalysisTool` be automatically looped over to
146 produce a result for multiple bands.
147 """
149 def __call__(self, data: KeyedData, **kwargs) -> KeyedResults:
150 bands = kwargs.pop("bands", None)
151 if "plotInfo" in kwargs and kwargs.get("plotInfo") is not None:
152 kwargs["plotInfo"]["plotName"] = self.identity
153 if not self.parameterizedBand or bands is None:
154 if "band" not in kwargs:
155 # Some tasks require a "band" key for naming. This shouldn't
156 # affect the results. DM-35813 should make this unnecessary.
157 kwargs["band"] = "analysisTools"
158 return self._call_single(data, **kwargs)
159 results: KeyedResults = {}
160 for band in bands:
161 kwargs["band"] = band
162 if "plotInfo" in kwargs:
163 kwargs["plotInfo"]["bands"] = band
164 subResult = self._call_single(data, **kwargs)
165 for key, value in subResult.items():
166 match value:
167 case PlotTypes():
168 results[f"{band}_{key}"] = value
169 case Measurement():
170 results[key] = value
171 return results
173 def _call_single(self, data: KeyedData, **kwargs) -> KeyedResults:
174 # create a shallow copy of kwargs
175 kwargs = dict(**kwargs)
176 kwargs["metric_tags"] = list(self.metric_tags or ())
177 prepped: KeyedData = self.prep(data, **kwargs) # type: ignore
178 processed: KeyedData = self.process(prepped, **kwargs) # type: ignore
179 finalized: Mapping[str, PlotTypes] | PlotTypes | Mapping[
180 str, Measurement
181 ] | Measurement | JointResults = self.produce(
182 processed, **kwargs
183 ) # type: ignore
184 return self._process_single_results(finalized)
186 def _getPlotType(self) -> str:
187 match self.produce:
188 case PlotAction():
189 return type(self.produce).__name__
190 case JointAction(plot=NoPlot()):
191 pass
192 case JointAction(plot=plotter):
193 return type(plotter).__name__
195 return ""
197 def _process_single_results(
198 self,
199 results: Mapping[str, PlotTypes] | PlotTypes | Mapping[str, Measurement] | Measurement | JointResults,
200 ) -> KeyedResults:
201 accumulation = {}
202 suffix = self._getPlotType()
203 predicate = f"{self.identity}" if self.identity else ""
204 match results:
205 case Mapping():
206 for key, value in results.items():
207 match value:
208 case PlotTypes():
209 iterable = (predicate, key, suffix)
210 case Measurement():
211 iterable = (predicate, key)
212 refKey = "_".join(x for x in iterable if x)
213 accumulation[refKey] = value
214 case PlotTypes():
215 refKey = "_".join(x for x in (predicate, suffix) if x)
216 accumulation[refKey] = results
217 case Measurement():
218 accumulation[f"{predicate}"] = results
219 case JointResults(plot=plotResults, metric=metricResults):
220 if plotResults is not None:
221 subResult = self._process_single_results(plotResults)
222 accumulation.update(subResult)
223 if metricResults is not None:
224 subResult = self._process_single_results(metricResults)
225 accumulation.update(subResult)
226 return accumulation
228 def getInputSchema(self) -> KeyedDataSchema:
229 return self.prep.getInputSchema()
231 def populatePrepFromProcess(self):
232 """Add additional inputs to the prep stage if supported.
234 If the configured prep action supports adding to it's input schema,
235 attempt to add the required inputs schema from the process stage to the
236 prep stage.
238 This method will be a no-op if the prep action does not support this
239 feature.
240 """
241 self.prep.addInputSchema(self.process.getInputSchema())
243 def getOutputNames(self, config: pexConfig.Config | None = None) -> Iterable[str]:
244 """Return the names of the plots produced by this analysis tool.
246 If there is a `PlotAction` defined in the produce action, these names
247 will either come from the `PlotAction` if it defines a
248 ``getOutputNames`` method (likely if it returns a mapping of figures),
249 or a default value is used and a single figure is assumed.
251 Parameters
252 ----------
253 config : `lsst.pex.config.Config`, optional
254 Configuration of the task. This is only used if the output naming
255 needs to be config-aware.
257 Returns
258 -------
259 result : `tuple` of `str`
260 Names for each plot produced by this action.
261 """
262 match self.produce:
263 case JointAction(plot=NoPlot()):
264 return tuple()
265 case _HasOutputNames():
266 outNames = tuple(self.produce.getOutputNames(config=config))
267 case _:
268 raise ValueError(f"Unsupported Action type {type(self.produce)} for getting output names")
270 results = []
271 suffix = self._getPlotType()
272 if self.parameterizedBand:
273 prefix = "_".join(x for x in ("{band}", self.identity) if x)
274 else:
275 prefix = f"{self.identity}" if self.identity else ""
277 if outNames:
278 for name in outNames:
279 results.append("_".join(x for x in (prefix, name, suffix) if x))
280 else:
281 results.append("_".join(x for x in (prefix, suffix) if x))
282 return results
284 def finalize(self) -> None:
285 """Run any finalization code that depends on configuration being
286 complete.
287 """
288 pass
290 def _baseFinalize(self) -> None:
291 self.populatePrepFromProcess()
293 def freeze(self):
294 if not self.__dict__.get("_finalizeRun"):
295 self.finalize()
296 self.__dict__["_finalizeRun"] = True
297 super().freeze()
300# explicitly wrap the finalize of the base class
301AnalysisTool.finalize = _finalizeWrapper(AnalysisTool.finalize, AnalysisTool)