Coverage for python/lsst/meas/base/diaCalculation.py : 19%

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 ap_association.
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 collections import namedtuple
23import numpy as np
24import pandas as pd
26from .catalogCalculation import (CatalogCalculationPluginConfig,
27 CatalogCalculationPlugin,
28 CatalogCalculationConfig,
29 CatalogCalculationTask,
30 CCContext)
31from .pluginsBase import BasePlugin
32from .pluginRegistry import (PluginRegistry, PluginMap)
33import lsst.pipe.base
35# Enforce an error for unsafe column/array value setting in pandas.
36pd.options.mode.chained_assignment = 'raise'
38__all__ = ("DiaObjectCalculationPlugin", "DiaObjectCalculationPluginConfig",
39 "DiaObjectCalculationTask", "DiaObjectCalculationConfig")
42class DiaObjectCalculationPluginConfig(CatalogCalculationPluginConfig):
43 """Default configuration class for DIA catalog calculation plugins.
44 """
45 pass
48class DiaObjectCalculationPlugin(CatalogCalculationPlugin):
49 """Base class for DIA catalog calculation plugins.
51 Task follows CatalogCalculationPlugin with modifications for use in AP.
53 Parameters
54 ----------
55 config : `DiaObjectCalculationPlugin.ConfigClass`
56 Plugin configuration.
57 name : `str`
58 The string the plugin was registered with.
59 metadata : `lsst.daf.base.PropertySet`
60 Plugin metadata that will be attached to the output catalog
61 """
63 ConfigClass = DiaObjectCalculationPluginConfig
65 registry = PluginRegistry(DiaObjectCalculationPluginConfig)
66 """List of available plugins (`lsst.meas.base.PluginRegistry`).
67 """
69 FLUX_MOMENTS_CALCULATED = 5.0
70 """Add order after flux means and stds are calculated.
71 """
73 plugType = 'single'
74 """Does the plugin operate on a single source or the whole catalog (`str`)?
75 If the plugin operates on a single source at a time, this should be set to
76 ``"single"``; if it expects the whoe catalog, to ``"multi"``. If the
77 plugin is of type ``"multi"``, the `fail` method must be implemented to
78 accept the whole catalog. If the plugin is of type ``"single"``, `fail`
79 should accept a single source record.
80 """
82 inputCols = []
83 """DiaObject column names required by the plugin in order to run and
84 complete its calculation. DiaCalculationTask should raise an error is a
85 plugin is instantiated without the needed column available. Input columns
86 should be defined in the DPDD/cat/Apdb schema. Filter dependent columns
87 should be specified without the filter name perpended to them. eg
88 ``PSFluxMean`` instead of ``uPSFluxMean``.
89 """
90 outputCols = []
91 """DiaObject column names output by the plugin. DiaCalculationTask should
92 raise an error if another pluging is run output to the same column.
93 Output columns should be defined in the DPDD/cat/Apdb schema. Filter
94 dependent columns should be specified without the filter name perpended to
95 them. eg ``PSFluxMean`` instead of ``uPSFluxMean``.
96 """
98 needsFilter = True
99 """This plugin requires a filter to be specified. Plugin's using filter
100 names usually deal with fluxes and only a sub-set of the DiaSource
101 catalog. Plugins that to not use the filter name usually run over a value
102 common across all observations/detections such as position.
103 """
105 def __init__(self, config, name, metadata):
106 BasePlugin.__init__(self, config, name)
108 def calculate(self,
109 diaObject,
110 diaSources,
111 filterDiaFluxes=None,
112 filterName=None,
113 **kwargs):
114 """Perform the calculation specified by this plugin.
116 This method can either be used to operate on a single catalog record
117 or a whole catalog, populating it with the output defined by this
118 plugin.
120 Note that results may be added to catalog records as new columns, or
121 may result in changes to existing values.
123 Parameters
124 ----------
125 diaObject : `dict`
126 Summary object to store values in.
127 diaSources : `pandas.DataFrame`
128 DataFrame representing all diaSources associated with this
129 diaObject.
130 filterDiaFluxes : `pandas.DataFrame`
131 DataFrame representing diaSources associated with this
132 diaObject that are observed in the band pass ``filterName``.
133 filterName : `str`
134 Simple name of the filter for the flux being calculated.
135 **kwargs
136 Any additional keyword arguments that may be passed to the plugin.
137 """
138 raise NotImplementedError()
140 def fail(self, diaObject, columns, error=None):
141 """Set diaObject position values to nan.
143 Parameters
144 ----------
145 diaObject : `dict`
146 Summary object to store values in.
147 columns : `list` of `str`
148 List of string names of columns to write a the failed value.
149 error : `BaseException` or `None`
150 Error to pass. Kept for consistency with CatologCalculationPlugin.
151 Unused.
152 """
153 for colName in columns:
154 diaObject[colName] = np.nan
157class DiaObjectCalculationConfig(CatalogCalculationConfig):
158 """Config class for the catalog calculation driver task.
160 Specifies which plugins will execute when the `CatalogCalculationTask`
161 associated with this configuration is run.
162 """
164 plugins = DiaObjectCalculationPlugin.registry.makeField(
165 multi=True,
166 default=["ap_meanPosition",
167 "ap_meanFlux"],
168 doc="Plugins to be run and their configuration")
171class DiaObjectCalculationTask(CatalogCalculationTask):
172 """Run plugins which operate on a catalog of DIA sources.
174 This task facilitates running plugins which will operate on a source
175 catalog. These plugins may do things such as classifying an object based
176 on source record entries inserted during a measurement task.
178 This task differs from CatalogCaculationTask in the following ways:
180 -No multi mode is available for plugins. All plugins are assumed to run
181 in single mode.
183 -Input and output catalog types are assumed to be `pandas.DataFrames` with
184 columns following those used in the Apdb.
186 -No schema argument is passed to the plugins. Each plugin specifies
187 output columns and required inputs.
189 Parameters
190 ----------
191 plugMetaData : `lsst.daf.base.PropertyList` or `None`
192 Will be modified in-place to contain metadata about the plugins being
193 run. If `None`, an empty `~lsst.daf.base.PropertyList` will be
194 created.
195 **kwargs
196 Additional arguments passed to the superclass constructor.
198 Notes
199 -----
200 Plugins may either take an entire catalog to work on at a time, or work on
201 individual records.
202 """
203 ConfigClass = DiaObjectCalculationConfig
204 _DefaultName = "diaObjectCalculation"
206 def __init__(self, plugMetadata=None, **kwargs):
207 lsst.pipe.base.Task.__init__(self, **kwargs)
208 if plugMetadata is None:
209 plugMetadata = lsst.daf.base.PropertyList()
210 self.plugMetadata = plugMetadata
211 self.plugins = PluginMap()
212 self.outputCols = []
214 self.initializePlugins()
216 def initializePlugins(self):
217 """Initialize the plugins according to the configuration.
218 """
220 pluginType = namedtuple('pluginType', 'single multi')
221 self.executionDict = {}
222 # Read the properties for each plugin. Allocate a dictionary entry for
223 # each run level. Verify that the plugins are above the minimum run
224 # level for an catalogCalculation plugin. For each run level, the
225 # plugins are sorted into either single record, or multi record groups
226 # to later be run appropriately
227 for executionOrder, name, config, PluginClass in sorted(self.config.plugins.apply()):
228 if executionOrder not in self.executionDict:
229 self.executionDict[executionOrder] = pluginType(single=[], multi=[])
230 if PluginClass.getExecutionOrder() >= BasePlugin.DEFAULT_CATALOGCALCULATION:
231 plug = PluginClass(config, name, metadata=self.plugMetadata)
233 self._validatePluginCols(plug)
235 self.plugins[name] = plug
236 if plug.plugType == 'single':
237 self.executionDict[executionOrder].single.append(plug)
238 elif plug.plugType == 'multi':
239 self.executionDict[executionOrder].multi.append(plug)
240 else:
241 errorTuple = (PluginClass, PluginClass.getExecutionOrder(),
242 BasePlugin.DEFAULT_CATALOGCALCULATION)
243 raise ValueError("{} has an execution order less than the minimum for an catalogCalculation "
244 "plugin. Value {} : Minimum {}".format(*errorTuple))
246 def _validatePluginCols(self, plug):
247 """Assert that output columns are not duplicated and input columns
248 exist for dependent plugins.
250 Parameters
251 ----------
252 plug : `lsst.ap.association.DiaCalculationPlugin`
253 Plugin to test for output collisions and input needs.
254 """
255 for inputName in plug.inputCols:
256 if inputName not in self.outputCols:
257 errorTuple = (plug.name, plug.getExecutionOrder(),
258 inputName)
259 raise ValueError(
260 "Plugin, {} with execution order {} requires DiaObject "
261 "column {} to exist. Check the execution order of the "
262 "plugin and make sure it runs after a plugin creating "
263 "the column is run.".format(*errorTuple))
264 for outputName in plug.outputCols:
265 if outputName in self.outputCols:
266 errorTuple = (plug.name, plug.getExecutionOrder(),
267 outputName)
268 raise ValueError(
269 "Plugin, {} with execution order {} is attempting to "
270 "output a column {}, however the column is already being "
271 "produced by another plugin. Check other plugins for "
272 "collisions with this one.".format(*errorTuple))
273 else:
274 self.outputCols.append(outputName)
276 @lsst.pipe.base.timeMethod
277 def run(self,
278 diaObjectCat,
279 diaSourceCat,
280 updatedDiaObjectIds,
281 filterNames):
282 """The entry point for the DIA catalog calculation task.
284 Run method both updates the values in the diaObjectCat and appends
285 newly created DiaObjects to the catalog. For catalog column names
286 see the lsst.cat schema definitions for the DiaObject and DiaSource
287 tables (http://github.com/lsst/cat).
289 Parameters
290 ----------
291 diaObjectCat : `pandas.DataFrame`
292 DiaObjects to update values of and append new objects to. DataFrame
293 should be indexed on "diaObjectId"
294 diaSourceCat : `pandas.DataFrame`
295 DiaSources associated with the DiaObjects in diaObjectCat.
296 DataFrame should be indexed on
297 `["diaObjectId", "filterName", "diaSourceId"]`
298 updatedDiaObjectIds : `numpy.ndarray`
299 Integer ids of the DiaObjects to update and create.
300 filterNames : `list` of `str`
301 List of string names of filters to be being processed.
303 Returns
304 -------
305 returnStruct : `lsst.pipe.base.Struct`
306 Struct containing:
308 ``diaObjectCat``
309 Full set of DiaObjects including both un-updated and
310 updated/new DiaObjects (`pandas.DataFrame`).
311 ``updatedDiaObjects``
312 Catalog of DiaObjects that were updated or created by this
313 task (`pandas.DataFrame`).
314 """
315 if diaObjectCat.index.name is None:
316 diaObjectCat.set_index("diaObjectId", inplace=True, drop=False)
317 elif diaObjectCat.index.name != "diaObjectId":
318 self.log.warning(
319 "Input diaObjectCat is indexed on column(s) incompatible with "
320 "this task. Should be indexed on 'diaObjectId'. Trying to set "
321 "index regardless")
322 diaObjectCat.set_index("diaObjectId", inplace=True, drop=False)
324 # ``names`` by default is FrozenList([None]) hence we access the first
325 # element and test for None.
326 if diaSourceCat.index.names[0] is None:
327 diaSourceCat.set_index(
328 ["diaObjectId", "filterName", "diaSourceId"],
329 inplace=True,
330 drop=False)
331 elif (diaSourceCat.index.names
332 != ["diaObjectId", "filterName", "diaSourceId"]):
333 self.log.warning(
334 "Input diaSourceCat is indexed on column(s) incompatible with "
335 "this task. Should be indexed on 'multi-index, "
336 "['diaObjectId', 'filterName', 'diaSourceId']. Trying to set "
337 "index regardless.")
338 diaSourceCat.set_index(
339 ["diaObjectId", "filterName", "diaSourceId"],
340 inplace=True,
341 drop=False)
343 return self.callCompute(diaObjectCat,
344 diaSourceCat,
345 updatedDiaObjectIds,
346 filterNames)
348 @lsst.pipe.base.timeMethod
349 def callCompute(self,
350 diaObjectCat,
351 diaSourceCat,
352 updatedDiaObjectIds,
353 filterNames):
354 """Run each of the plugins on the catalog.
356 For catalog column names see the lsst.cat schema definitions for the
357 DiaObject and DiaSource tables (http://github.com/lsst/cat).
359 Parameters
360 ----------
361 diaObjectCat : `pandas.DataFrame`
362 DiaObjects to update values of and append new objects to. DataFrame
363 should be indexed on "diaObjectId"
364 diaSourceCat : `pandas.DataFrame`
365 DiaSources associated with the DiaObjects in diaObjectCat.
366 DataFrame must be indexed on
367 ["diaObjectId", "filterName", "diaSourceId"]`
368 updatedDiaObjectIds : `numpy.ndarray`
369 Integer ids of the DiaObjects to update and create.
370 filterNames : `list` of `str`
371 List of string names of filters to be being processed.
373 Returns
374 -------
375 returnStruct : `lsst.pipe.base.Struct`
376 Struct containing:
378 ``diaObjectCat``
379 Full set of DiaObjects including both un-updated and
380 updated/new DiaObjects (`pandas.DataFrame`).
381 ``updatedDiaObjects``
382 Catalog of DiaObjects that were updated or created by this
383 task (`pandas.DataFrame`).
385 Raises
386 ------
387 KeyError
388 Raises if `pandas.DataFrame` indexing is not properly set.
389 """
390 # DiaObjects will be updated in place.
391 diaObjectsToUpdate = diaObjectCat.loc[updatedDiaObjectIds, :]
392 self.log.info("Calculating summary stats for %i DiaObjects",
393 len(diaObjectsToUpdate))
395 updatingDiaSources = diaSourceCat.loc[updatedDiaObjectIds, :]
396 diaSourcesGB = updatingDiaSources.groupby(level=0)
397 for runlevel in sorted(self.executionDict):
398 for plug in self.executionDict[runlevel].single:
399 if plug.needsFilter:
400 continue
401 for updatedDiaObjectId in updatedDiaObjectIds:
403 # Sub-select diaSources associated with this diaObject.
404 objDiaSources = updatingDiaSources.loc[updatedDiaObjectId]
406 # Sub-select on diaSources observed in the current filter.
407 with CCContext(plug, updatedDiaObjectId, self.log):
408 # We feed the catalog we need to update and the id
409 # so as to get a few into the catalog and not a copy.
410 # This updates the values in the catalog.
411 plug.calculate(diaObjects=diaObjectsToUpdate,
412 diaObjectId=updatedDiaObjectId,
413 diaSources=objDiaSources,
414 filterDiaSources=None,
415 filterName=None)
416 for plug in self.executionDict[runlevel].multi:
417 if plug.needsFilter:
418 continue
419 with CCContext(plug, diaObjectsToUpdate, self.log):
420 plug.calculate(diaObjects=diaObjectsToUpdate,
421 diaSources=diaSourcesGB,
422 filterDiaSources=None,
423 filterName=None)
425 for filterName in filterNames:
426 try:
427 updatingFilterDiaSources = updatingDiaSources.loc[
428 (slice(None), filterName), :
429 ]
430 except KeyError:
431 self.log.warning("No DiaSource data with fitler=%s. "
432 "Continuing...", filterName)
433 continue
434 # Level=0 here groups by diaObjectId.
435 filterDiaSourcesGB = updatingFilterDiaSources.groupby(level=0)
437 for runlevel in sorted(self.executionDict):
438 for plug in self.executionDict[runlevel].single:
439 if not plug.needsFilter:
440 continue
441 for updatedDiaObjectId in updatedDiaObjectIds:
443 # Sub-select diaSources associated with this diaObject.
444 objDiaSources = updatingDiaSources.loc[updatedDiaObjectId]
446 # Sub-select on diaSources observed in the current filter.
447 try:
448 filterObjDiaSources = objDiaSources.loc[filterName]
449 except KeyError:
450 self.log.warning(
451 "DiaObjectId={updatedDiaObjectId} has no "
452 "DiaSources for filter=%s. "
453 "Continuing...", filterName)
454 with CCContext(plug, updatedDiaObjectId, self.log):
455 # We feed the catalog we need to update and the id
456 # so as to get a few into the catalog and not a copy.
457 # This updates the values in the catalog.
458 plug.calculate(diaObjects=diaObjectsToUpdate,
459 diaObjectId=updatedDiaObjectId,
460 diaSources=objDiaSources,
461 filterDiaSources=filterObjDiaSources,
462 filterName=filterName)
463 for plug in self.executionDict[runlevel].multi:
464 if not plug.needsFilter:
465 continue
466 with CCContext(plug, diaObjectsToUpdate, self.log):
467 plug.calculate(diaObjects=diaObjectsToUpdate,
468 diaSources=diaSourcesGB,
469 filterDiaSources=filterDiaSourcesGB,
470 filterName=filterName)
471 # Need to store the newly updated diaObjects directly as the editing
472 # a view into diaObjectsToUpdate does not update the values of
473 # diaObjectCat.
474 diaObjectCat.loc[updatedDiaObjectIds, :] = diaObjectsToUpdate
475 return lsst.pipe.base.Struct(
476 diaObjectCat=diaObjectCat,
477 updatedDiaObjects=diaObjectsToUpdate)
479 def _initialize_dia_object(self, objId):
480 """Create a new DiaObject with values required to be initialized by the
481 Apdb.
483 Parameters
484 ----------
485 objid : `int`
486 ``diaObjectId`` value for the of the new DiaObject.
488 Returns
489 -------
490 diaObject : `dict`
491 Newly created DiaObject with keys:
493 ``diaObjectId``
494 Unique DiaObjectId (`int`).
495 ``pmParallaxNdata``
496 Number of data points used for parallax calculation (`int`).
497 ``nearbyObj1``
498 Id of the a nearbyObject in the Object table (`int`).
499 ``nearbyObj2``
500 Id of the a nearbyObject in the Object table (`int`).
501 ``nearbyObj3``
502 Id of the a nearbyObject in the Object table (`int`).
503 ``?PSFluxData``
504 Number of data points used to calculate point source flux
505 summary statistics in each bandpass (`int`).
506 """
507 new_dia_object = {"diaObjectId": objId,
508 "pmParallaxNdata": 0,
509 "nearbyObj1": 0,
510 "nearbyObj2": 0,
511 "nearbyObj3": 0}
512 for f in ["u", "g", "r", "i", "z", "y"]:
513 new_dia_object["%sPSFluxNdata" % f] = 0
514 return new_dia_object