Coverage for python/lsst/analysis/tools/interfaces/_analysisTools.py: 22%
137 statements
« prev ^ index » next coverage.py v7.2.3, created at 2023-04-28 05:16 -0700
« prev ^ index » next coverage.py v7.2.3, created at 2023-04-28 05:16 -0700
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
30from lsst.pex.config import Field, ListField
31from lsst.pex.config.configurableActions import ConfigurableActionField
32from lsst.verify import Measurement
34from ._actions import AnalysisAction, JointAction, JointResults, NoPlot, PlotAction
35from ._interfaces import KeyedData, KeyedDataSchema, KeyedResults, PlotTypes
36from ._stages import BasePrep, BaseProcess, BaseProduce
39@runtime_checkable
40class _HasOutputNames(Protocol):
41 def getOutputNames(self) -> Iterable[str]:
42 ...
45def _finalizeWrapper(
46 f: Callable[[AnalysisTool], None], cls: type[AnalysisTool]
47) -> Callable[[AnalysisTool], None]:
48 """Wrap a classes finalize function to ensure the base classes special
49 finalize method only fires after the most derived finalize method.
51 Parameters
52 ----------
53 f : `Callable`
54 Function that is being wrapped
55 cls : `type` of `AnalysisTool`
56 The class which is having its function wrapped
58 Returns
59 -------
60 function : `Callable`
61 The new function which wraps the old
62 """
64 @wraps(f)
65 def wrapper(self: AnalysisTool) -> None:
66 # call the wrapped finalize function
67 f(self)
68 # get the method resolution order for the self variable
69 mro = self.__class__.mro()
71 # Find which class in the mro that last defines a finalize method
72 # note that this is in the reverse order from the mro, because the
73 # last class in an inheritance stack is the first in the mro (aka you
74 # walk from the furthest child first.
75 #
76 # Also note that the most derived finalize method need not be the same
77 # as the type of self, as that might inherit from a parent somewhere
78 # between it and the furthest parent.
79 mostDerived: type | None = None
80 for klass in mro:
81 # inspect the classes dictionary to see if it specifically defines
82 # finalize. This is needed because normal lookup will go through
83 # the mro, but this needs to be restricted to each class.
84 if "finalize" in dir(klass):
85 mostDerived = klass
86 break
88 # Find what stage in the MRO walking process the recursive function
89 # call is in.
90 this = super(cls, self).__thisclass__
92 # If the current place in the MRO walking is also the class that
93 # defines the most derived instance of finalize, then call the base
94 # classes private finalize that must be called after everything else.
95 if mostDerived is not None and this == mostDerived:
96 self._baseFinalize()
98 return wrapper
101class AnalysisTool(AnalysisAction):
102 r"""A tool which which calculates a single type of analysis on input data,
103 though it may return more than one result.
105 Although `AnalysisTool`\ s are considered a single type of analysis, the
106 classes themselves can be thought of as a container. `AnalysisTool`\ s
107 are aggregations of `AnalysisAction`\ s to form prep, process, and
108 produce stages. These stages allow better reuse of individual
109 `AnalysisActions` and easier introspection in contexts such as a notebook
110 or interpreter.
112 An `AnalysisTool` can be thought of an an individual configuration that
113 specifies which `AnalysisAction` should run for each stage.
115 The stages themselves are also configurable, allowing control over various
116 aspects of the individual `AnalysisAction`\ s.
117 """
118 prep = ConfigurableActionField[AnalysisAction](doc="Action to run to prepare inputs", default=BasePrep)
119 process = ConfigurableActionField[AnalysisAction](
120 doc="Action to process data into intended form", default=BaseProcess
121 )
122 produce = ConfigurableActionField[AnalysisAction](
123 doc="Action to perform any finalization steps", default=BaseProduce
124 )
125 metric_tags = ListField[str](
126 doc="List of tags which will be associated with metric measurement(s)", default=[]
127 )
128 dataset_identifier = Field[str](doc="An identifier to be associated with output Metrics", optional=True)
129 reference_package = Field[str](
130 doc="A package who's version, at the time of metric upload to a "
131 "time series database, will be converted to a timestamp of when "
132 "that version was produced",
133 default="lsst_distrib",
134 )
135 timestamp_version = Field[str]( 135 ↛ exitline 135 didn't jump to the function exit
136 doc="Which time stamp should be used as the reference timestamp for a "
137 "metric in a time series database, valid values are; "
138 "reference_package_timestamp, run_timestamp, current_timestamp, "
139 "and dataset_timestamp",
140 default="run_timestamp",
141 check=lambda x: x
142 in ("reference_package_timestamp", "run_timestamp", "current_timestamp", "dataset_timestamp"),
143 )
145 def __init_subclass__(cls: type[AnalysisTool], **kwargs):
146 super().__init_subclass__(**kwargs)
147 # Wrap all definitions of the finalize method in a special wrapper that
148 # ensures that the bases classes private finalize is called last.
149 if "finalize" in dir(cls): 149 ↛ exitline 149 didn't return from function '__init_subclass__', because the condition on line 149 was never false
150 cls.finalize = _finalizeWrapper(cls.finalize, cls)
152 parameterizedBand: bool | Field[bool] = True
153 """Specifies if an `AnalysisTool` may parameterize a band within any field
154 in any stage, or if the set of bands is already uniquely determined though
155 configuration. I.e. can this `AnalysisTool` be automatically looped over to
156 produce a result for multiple bands.
157 """
159 def __call__(self, data: KeyedData, **kwargs) -> KeyedResults:
160 bands = kwargs.pop("bands", None)
161 if "plotInfo" in kwargs and kwargs.get("plotInfo") is not None:
162 kwargs["plotInfo"]["plotName"] = self.identity
163 if not self.parameterizedBand or bands is None:
164 if "band" not in kwargs:
165 # Some tasks require a "band" key for naming. This shouldn't
166 # affect the results. DM-35813 should make this unnecessary.
167 kwargs["band"] = "analysisTools"
168 return self._call_single(data, **kwargs)
169 results: KeyedResults = {}
170 for band in bands:
171 kwargs["band"] = band
172 subResult = self._call_single(data, **kwargs)
173 for key, value in subResult.items():
174 match value:
175 case PlotTypes():
176 results[f"{band}_{key}"] = value
177 case Measurement():
178 results[key] = value
179 return results
181 def _call_single(self, data: KeyedData, **kwargs) -> KeyedResults:
182 # create a shallow copy of kwargs
183 kwargs = dict(**kwargs)
184 kwargs["metric_tags"] = list(self.metric_tags or ())
185 prepped: KeyedData = self.prep(data, **kwargs) # type: ignore
186 processed: KeyedData = self.process(prepped, **kwargs) # type: ignore
187 finalized: Mapping[str, PlotTypes] | PlotTypes | Mapping[
188 str, Measurement
189 ] | Measurement | JointResults = self.produce(
190 processed, **kwargs
191 ) # type: ignore
192 return self._process_single_results(finalized)
194 def _getPlotType(self) -> str:
195 match self.produce:
196 case PlotAction():
197 return type(self.produce).__name__
198 case JointAction(plot=NoPlot()):
199 pass
200 case JointAction(plot=plotter):
201 return type(plotter).__name__
203 return ""
205 def _process_single_results(
206 self,
207 results: Mapping[str, PlotTypes] | PlotTypes | Mapping[str, Measurement] | Measurement | JointResults,
208 ) -> KeyedResults:
209 accumulation = {}
210 suffix = self._getPlotType()
211 predicate = f"{self.identity}" if self.identity else ""
212 match results:
213 case Mapping():
214 for key, value in results.items():
215 match value:
216 case PlotTypes():
217 iterable = (predicate, key, suffix)
218 case Measurement():
219 iterable = (predicate, key)
220 refKey = "_".join(x for x in iterable if x)
221 accumulation[refKey] = value
222 case PlotTypes():
223 refKey = "_".join(x for x in (predicate, suffix) if x)
224 accumulation[refKey] = results
225 case Measurement():
226 accumulation[f"{predicate}"] = results
227 case JointResults(plot=plotResults, metric=metricResults):
228 if plotResults is not None:
229 subResult = self._process_single_results(plotResults)
230 accumulation.update(subResult)
231 if metricResults is not None:
232 subResult = self._process_single_results(metricResults)
233 accumulation.update(subResult)
234 return accumulation
236 def getInputSchema(self) -> KeyedDataSchema:
237 return self.prep.getInputSchema()
239 def populatePrepFromProcess(self):
240 """Add additional inputs to the prep stage if supported.
242 If the configured prep action supports adding to it's input schema,
243 attempt to add the required inputs schema from the process stage to the
244 prep stage.
246 This method will be a no-op if the prep action does not support this
247 feature.
248 """
249 self.prep.addInputSchema(self.process.getInputSchema())
251 def getOutputNames(self) -> Iterable[str]:
252 """Return the names of the plots produced by this analysis tool.
254 If there is a `PlotAction` defined in the produce action, these names
255 will either come from the `PlotAction` if it defines a
256 ``getOutputNames`` method (likely if it returns a mapping of figures),
257 or a default value is used and a single figure is assumed.
259 Returns
260 -------
261 result : `tuple` of `str`
262 Names for each plot produced by this action.
263 """
264 match self.produce:
265 case JointAction(plot=NoPlot()):
266 return tuple()
267 case _HasOutputNames():
268 outNames = tuple(self.produce.getOutputNames())
269 case _:
270 raise ValueError(f"Unsupported Action type {type(self.produce)} for getting output names")
272 results = []
273 suffix = self._getPlotType()
274 if self.parameterizedBand:
275 prefix = "_".join(x for x in ("{band}", self.identity) if x)
276 else:
277 prefix = f"{self.identity}" if self.identity else ""
279 if outNames:
280 for name in outNames:
281 results.append("_".join(x for x in (prefix, name, suffix) if x))
282 else:
283 results.append("_".join(x for x in (prefix, suffix) if x))
284 return results
286 def finalize(self) -> None:
287 """Run any finalization code that depends on configuration being
288 complete.
289 """
290 pass
292 def _baseFinalize(self) -> None:
293 self.populatePrepFromProcess()
295 def freeze(self):
296 if not self.__dict__.get("_finalizeRun"):
297 self.finalize()
298 self.__dict__["_finalizeRun"] = True
299 super().freeze()
302# explicitly wrap the finalize of the base class
303AnalysisTool.finalize = _finalizeWrapper(AnalysisTool.finalize, AnalysisTool)