Coverage for python/lsst/ap/verify/pipeline_driver.py : 20%

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#
2# This file is part of ap_verify.
3#
4# Developed for the LSST Data Management System.
5# This product includes software developed by the LSST Project
6# (http://www.lsst.org).
7# See the COPYRIGHT file at the top-level directory of this distribution
8# for details of code ownership.
9#
10# This program is free software: you can redistribute it and/or modify
11# it under the terms of the GNU General Public License as published by
12# the Free Software Foundation, either version 3 of the License, or
13# (at your option) any later version.
14#
15# This program is distributed in the hope that it will be useful,
16# but WITHOUT ANY WARRANTY; without even the implied warranty of
17# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18# GNU General Public License for more details.
19#
20# You should have received a copy of the GNU General Public License
21# along with this program. If not, see <http://www.gnu.org/licenses/>.
22#
24"""Interface between `ap_verify` and `ap_pipe`.
26This module handles calling `ap_pipe` and converting any information
27as needed.
28"""
30__all__ = ["ApPipeParser", "runApPipeGen2", "runApPipeGen3"]
32import argparse
33import os
34import re
36import click.testing
38import lsst.log
39from lsst.utils import getPackageDir
40import lsst.pipe.base as pipeBase
41import lsst.obs.base as obsBase
42import lsst.ctrl.mpexec.cli.pipetask
43import lsst.ap.pipe as apPipe
44from lsst.ap.pipe.make_apdb import makeApdb
47class ApPipeParser(argparse.ArgumentParser):
48 """An argument parser for data needed by ``ap_pipe`` activities.
50 This parser is not complete, and is designed to be passed to another parser
51 using the `parent` parameter.
52 """
54 def __init__(self):
55 defaultPipeline = os.path.join(getPackageDir("ap_verify"), "pipelines", "ApVerify.yaml")
57 # Help and documentation will be handled by main program's parser
58 argparse.ArgumentParser.__init__(self, add_help=False)
59 # namespace.dataIds will always be a list of 0 or more nonempty strings, regardless of inputs.
60 # TODO: in Python 3.8+, action='extend' handles nargs='?' more naturally than 'append'.
61 self.add_argument('--id', '-d', '--data-query', dest='dataIds',
62 action=self.AppendOptional, nargs='?', default=[],
63 help='An identifier for the data to process.')
64 self.add_argument("-p", "--pipeline", default=defaultPipeline,
65 help="A custom version of the ap_verify pipeline (e.g., with different metrics).")
66 self.add_argument("--skip-pipeline", action="store_true",
67 help="Do not run the AP pipeline itself. This argument is useful "
68 "for testing metrics on a fixed data set.")
70 class AppendOptional(argparse.Action):
71 """A variant of the built-in "append" action that ignores None values
72 instead of appending them.
73 """
74 # This class can't safely inherit from the built-in "append" action
75 # because there is no public class that implements it.
76 def __call__(self, parser, namespace, values, option_string=None):
77 if values is not None:
78 try:
79 allValues = getattr(namespace, self.dest)
80 allValues.append(values)
81 except AttributeError:
82 setattr(namespace, self.dest, [values])
85def runApPipeGen2(workspace, parsedCmdLine, processes=1):
86 """Run `ap_pipe` on this object's dataset.
88 Parameters
89 ----------
90 workspace : `lsst.ap.verify.workspace.WorkspaceGen2`
91 The abstract location containing input and output repositories.
92 parsedCmdLine : `argparse.Namespace`
93 Command-line arguments, including all arguments supported by `ApPipeParser`.
94 processes : `int`
95 The number of processes with which to call the AP pipeline
97 Returns
98 -------
99 apPipeReturn : `Struct`
100 The `Struct` returned from `~lsst.ap.pipe.ApPipeTask.parseAndRun` with
101 ``doReturnResults=False``. This object is valid even if
102 `~lsst.ap.pipe.ApPipeTask` was never run.
103 """
104 log = lsst.log.Log.getLogger('ap.verify.pipeline_driver.runApPipeGen2')
106 configArgs = _getConfigArguments(workspace)
107 makeApdb(configArgs)
109 pipelineArgs = [workspace.dataRepo,
110 "--output", workspace.outputRepo,
111 "--calib", workspace.calibRepo,
112 "--template", workspace.templateRepo]
113 pipelineArgs.extend(configArgs)
114 if parsedCmdLine.dataIds:
115 for singleId in parsedCmdLine.dataIds:
116 pipelineArgs.extend(["--id", *singleId.split(" ")])
117 else:
118 pipelineArgs.extend(["--id"])
119 pipelineArgs.extend(["--processes", str(processes)])
120 pipelineArgs.extend(["--noExit"])
122 if not parsedCmdLine.skip_pipeline:
123 results = apPipe.ApPipeTask.parseAndRun(pipelineArgs)
124 log.info('Pipeline complete')
125 else:
126 log.info('Skipping AP pipeline entirely.')
127 apPipeParser = apPipe.ApPipeTask._makeArgumentParser()
128 apPipeParsed = apPipeParser.parse_args(config=apPipe.ApPipeTask.ConfigClass(), args=pipelineArgs)
129 results = pipeBase.Struct(
130 argumentParser=apPipeParser,
131 parsedCmd=apPipeParsed,
132 taskRunner=apPipe.ApPipeTask.RunnerClass(TaskClass=apPipe.ApPipeTask, parsedCmd=apPipeParsed),
133 resultList=[],
134 )
136 return results
139def runApPipeGen3(workspace, parsedCmdLine, processes=1):
140 """Run `ap_pipe` on this object's dataset.
142 Parameters
143 ----------
144 workspace : `lsst.ap.verify.workspace.WorkspaceGen3`
145 The abstract location containing input and output repositories.
146 parsedCmdLine : `argparse.Namespace`
147 Command-line arguments, including all arguments supported by `ApPipeParser`.
148 processes : `int`
149 The number of processes with which to call the AP pipeline
150 """
151 log = lsst.log.Log.getLogger('ap.verify.pipeline_driver.runApPipeGen3')
153 # Currently makeApdb has different argument conventions from Gen 3; see DM-22663
154 makeApdb(_getConfigArguments(workspace))
156 pipelineArgs = ["run",
157 "--butler-config", workspace.repo,
158 "--pipeline", parsedCmdLine.pipeline,
159 ]
160 # TODO: collections should be determined exclusively by Workspace.workButler,
161 # but I can't find a way to hook that up to the graph builder. So use the CLI
162 # for now and revisit once DM-26239 is done.
163 pipelineArgs.extend(_getCollectionArguments(workspace))
164 pipelineArgs.extend(_getConfigArgumentsGen3(workspace))
165 if parsedCmdLine.dataIds:
166 for singleId in parsedCmdLine.dataIds:
167 pipelineArgs.extend(["--data-query", singleId])
168 pipelineArgs.extend(["--processes", str(processes)])
169 pipelineArgs.extend(["--register-dataset-types"])
171 if not parsedCmdLine.skip_pipeline:
172 # CliRunner is an unsafe workaround for DM-26239
173 runner = click.testing.CliRunner()
174 # TODO: generalize this code in DM-26028
175 # TODO: work off of workspace.workButler after DM-26239
176 results = runner.invoke(lsst.ctrl.mpexec.cli.pipetask.cli, pipelineArgs)
177 if results.exception:
178 raise RuntimeError("Pipeline failed.") from results.exception
180 log.info('Pipeline complete.')
181 return results.exit_code
182 else:
183 log.info('Skipping AP pipeline entirely.')
186def _getConfigArguments(workspace):
187 """Return the config options for running ApPipeTask on this workspace, as
188 command-line arguments.
190 Parameters
191 ----------
192 workspace : `lsst.ap.verify.workspace.WorkspaceGen2`
193 A Workspace whose config directory may contain an
194 `~lsst.ap.pipe.ApPipeTask` config.
196 Returns
197 -------
198 args : `list` of `str`
199 Command-line arguments calling ``--config`` or ``--configFile``,
200 following the conventions of `sys.argv`.
201 """
202 overrideFile = apPipe.ApPipeTask._DefaultName + ".py"
203 overridePath = os.path.join(workspace.configDir, overrideFile)
205 args = ["--configfile", overridePath]
206 # ApVerify will use the sqlite hooks for the Apdb.
207 args.extend(["--config", "diaPipe.apdb.db_url=sqlite:///" + workspace.dbLocation])
208 args.extend(["--config", "diaPipe.apdb.isolation_level=READ_UNCOMMITTED"])
209 # Put output alerts into the workspace.
210 args.extend(["--config", "diaPipe.alertPackager.alertWriteLocation=" + workspace.alertLocation])
211 args.extend(["--config", "diaPipe.doPackageAlerts=True"])
213 return args
216def _getConfigArgumentsGen3(workspace):
217 """Return the config options for running the Gen 3 AP Pipeline on this
218 workspace, as command-line arguments.
220 Parameters
221 ----------
222 workspace : `lsst.ap.verify.workspace.WorkspaceGen3`
223 A Workspace whose config directory may contain various configs.
225 Returns
226 -------
227 args : `list` of `str`
228 Command-line arguments calling ``--config`` or ``--config-file``,
229 following the conventions of `sys.argv`.
230 """
231 args = [
232 # ApVerify will use the sqlite hooks for the Apdb.
233 "--config", "diaPipe:apdb.db_url=sqlite:///" + workspace.dbLocation,
234 "--config", "diaPipe:apdb.isolation_level=READ_UNCOMMITTED",
235 # Put output alerts into the workspace.
236 "--config", "diaPipe:alertPackager.alertWriteLocation=" + workspace.alertLocation,
237 "--config", "diaPipe:doPackageAlerts=True",
238 # TODO: the configs below should not be needed after DM-26140
239 "--config-file", "calibrate:" + os.path.join(workspace.configDir, "calibrate.py"),
240 "--config-file", "imageDifference:" + os.path.join(workspace.configDir, "imageDifference.py"),
241 ]
242 # TODO: reverse-engineering the instrument should not be needed after DM-26140
243 # pipetask will crash if there is more than one instrument
244 for idRecord in workspace.workButler.registry.queryDataIds("instrument").expanded():
245 className = idRecord.records["instrument"].class_name
246 args.extend(["--instrument", className])
248 return args
251def _getCollectionArguments(workspace):
252 """Return the collections for running the Gen 3 AP Pipeline on this
253 workspace, as command-line arguments.
255 Parameters
256 ----------
257 workspace : `lsst.ap.verify.workspace.WorkspaceGen3`
258 A Workspace with a Gen 3 repository.
260 Returns
261 -------
262 args : `list` of `str`
263 Command-line arguments calling ``--input`` or ``--output``,
264 following the conventions of `sys.argv`.
265 """
266 butler = workspace.workButler
267 # Hard-code the collection names because it's hard to infer the inputs from the Butler
268 inputs = {"skymaps", "refcats"}
269 for dimension in butler.registry.queryDataIds('instrument'):
270 instrument = obsBase.Instrument.fromName(dimension["instrument"], butler.registry)
271 inputs.add(instrument.makeDefaultRawIngestRunName())
272 inputs.add(instrument.makeCalibrationCollectionName())
273 inputs.update(butler.registry.queryCollections(re.compile(r"templates/\w+")))
275 return ["--input", ",".join(inputs),
276 "--output-run", workspace.runName,
277 ]