Coverage for python/lsst/ap/verify/pipeline_driver.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#
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.ctrl.mpexec.execFixupDataId # not part of lsst.ctrl.mpexec
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("--db", "--db_url", default=None,
67 help="A location for the AP database, formatted as if for ApdbConfig.db_url. "
68 "Defaults to an SQLite file in the --output directory.")
69 self.add_argument("--skip-pipeline", action="store_true",
70 help="Do not run the AP pipeline itself. This argument is useful "
71 "for testing metrics on a fixed data set.")
72 self.add_argument("--clean-run", action="store_true",
73 help="Run the pipeline with a new run collection, "
74 "even if one already exists.")
76 class AppendOptional(argparse.Action):
77 """A variant of the built-in "append" action that ignores None values
78 instead of appending them.
79 """
80 # This class can't safely inherit from the built-in "append" action
81 # because there is no public class that implements it.
82 def __call__(self, parser, namespace, values, option_string=None):
83 if values is not None:
84 try:
85 allValues = getattr(namespace, self.dest)
86 allValues.append(values)
87 except AttributeError:
88 setattr(namespace, self.dest, [values])
91def runApPipeGen2(workspace, parsedCmdLine, processes=1):
92 """Run `ap_pipe` on this object's dataset.
94 Parameters
95 ----------
96 workspace : `lsst.ap.verify.workspace.WorkspaceGen2`
97 The abstract location containing input and output repositories.
98 parsedCmdLine : `argparse.Namespace`
99 Command-line arguments, including all arguments supported by `ApPipeParser`.
100 processes : `int`
101 The number of processes with which to call the AP pipeline
103 Returns
104 -------
105 apPipeReturn : `Struct`
106 The `Struct` returned from `~lsst.ap.pipe.ApPipeTask.parseAndRun` with
107 ``doReturnResults=False``. This object is valid even if
108 `~lsst.ap.pipe.ApPipeTask` was never run.
109 """
110 log = lsst.log.Log.getLogger('ap.verify.pipeline_driver.runApPipeGen2')
112 makeApdb(_getApdbArguments(workspace, parsedCmdLine))
114 pipelineArgs = [workspace.dataRepo,
115 "--output", workspace.outputRepo,
116 "--calib", workspace.calibRepo,
117 "--template", workspace.templateRepo]
118 pipelineArgs.extend(_getConfigArguments(workspace, parsedCmdLine))
119 if parsedCmdLine.dataIds:
120 for singleId in parsedCmdLine.dataIds:
121 pipelineArgs.extend(["--id", *singleId.split(" ")])
122 else:
123 pipelineArgs.extend(["--id"])
124 pipelineArgs.extend(["--processes", str(processes)])
125 pipelineArgs.extend(["--noExit"])
127 if not parsedCmdLine.skip_pipeline:
128 results = apPipe.ApPipeTask.parseAndRun(pipelineArgs)
129 log.info('Pipeline complete')
130 else:
131 log.info('Skipping AP pipeline entirely.')
132 apPipeParser = apPipe.ApPipeTask._makeArgumentParser()
133 apPipeParsed = apPipeParser.parse_args(config=apPipe.ApPipeTask.ConfigClass(), args=pipelineArgs)
134 results = pipeBase.Struct(
135 argumentParser=apPipeParser,
136 parsedCmd=apPipeParsed,
137 taskRunner=apPipe.ApPipeTask.RunnerClass(TaskClass=apPipe.ApPipeTask, parsedCmd=apPipeParsed),
138 resultList=[],
139 )
141 return results
144def runApPipeGen3(workspace, parsedCmdLine, processes=1):
145 """Run `ap_pipe` on this object's dataset.
147 Parameters
148 ----------
149 workspace : `lsst.ap.verify.workspace.WorkspaceGen3`
150 The abstract location containing input and output repositories.
151 parsedCmdLine : `argparse.Namespace`
152 Command-line arguments, including all arguments supported by `ApPipeParser`.
153 processes : `int`
154 The number of processes with which to call the AP pipeline
155 """
156 log = lsst.log.Log.getLogger('ap.verify.pipeline_driver.runApPipeGen3')
158 makeApdb(_getApdbArguments(workspace, parsedCmdLine))
160 pipelineArgs = ["run",
161 "--butler-config", workspace.repo,
162 "--pipeline", parsedCmdLine.pipeline,
163 ]
164 # TODO: collections should be determined exclusively by Workspace.workButler,
165 # but I can't find a way to hook that up to the graph builder. So use the CLI
166 # for now and revisit once DM-26239 is done.
167 pipelineArgs.extend(_getCollectionArguments(workspace, reuse=(not parsedCmdLine.clean_run)))
168 pipelineArgs.extend(_getConfigArgumentsGen3(workspace, parsedCmdLine))
169 if parsedCmdLine.dataIds:
170 for singleId in parsedCmdLine.dataIds:
171 pipelineArgs.extend(["--data-query", singleId])
172 pipelineArgs.extend(["--processes", str(processes)])
173 pipelineArgs.extend(["--register-dataset-types"])
174 pipelineArgs.extend(["--graph-fixup", "lsst.ap.verify.pipeline_driver._getExecOrder"])
176 if not parsedCmdLine.skip_pipeline:
177 # CliRunner is an unsafe workaround for DM-26239
178 runner = click.testing.CliRunner()
179 # TODO: generalize this code in DM-26028
180 # TODO: work off of workspace.workButler after DM-26239
181 results = runner.invoke(lsst.ctrl.mpexec.cli.pipetask.cli, pipelineArgs)
182 if results.exception:
183 raise RuntimeError("Pipeline failed.") from results.exception
185 log.info('Pipeline complete.')
186 return results.exit_code
187 else:
188 log.info('Skipping AP pipeline entirely.')
191def _getExecOrder():
192 """Return any constraints on the Gen 3 execution order.
194 The current constraints are that executions of DiaPipelineTask must be
195 ordered by visit ID, but this is subject to change.
197 Returns
198 -------
199 order : `lsst.ctrl.mpexec.ExecutionGraphFixup`
200 An object encoding the desired execution order as an algorithm for
201 modifying inter-quantum dependencies.
203 Notes
204 -----
205 This function must be importable, but need not be public.
206 """
207 # Source association algorithm is not time-symmetric. Force execution of
208 # association (through DiaPipelineTask) in order of ascending visit number.
209 return lsst.ctrl.mpexec.execFixupDataId.ExecFixupDataId(
210 taskLabel="diaPipe", dimensions=["visit", ], reverse=False)
213def _getApdbArguments(workspace, parsed):
214 """Return the config options for running make_apdb.py on this workspace,
215 as command-line arguments.
217 Parameters
218 ----------
219 workspace : `lsst.ap.verify.workspace.Workspace`
220 A Workspace whose config directory may contain an
221 `~lsst.ap.pipe.ApPipeTask` config.
222 parsed : `argparse.Namespace`
223 Command-line arguments, including all arguments supported by `ApPipeParser`.
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 if not parsed.db:
232 parsed.db = "sqlite:///" + workspace.dbLocation
234 args = ["--config", "db_url=" + parsed.db]
235 # Same special-case check as ApdbConfig.validate()
236 if parsed.db.startswith("sqlite"):
237 args.extend(["--config", "isolation_level=READ_UNCOMMITTED"])
239 return args
242def _getConfigArguments(workspace, parsed):
243 """Return the config options for running ApPipeTask on this workspace, as
244 command-line arguments.
246 Parameters
247 ----------
248 workspace : `lsst.ap.verify.workspace.WorkspaceGen2`
249 A Workspace whose config directory may contain an
250 `~lsst.ap.pipe.ApPipeTask` config.
251 parsed : `argparse.Namespace`
252 Command-line arguments, including all arguments supported by `ApPipeParser`.
254 Returns
255 -------
256 args : `list` of `str`
257 Command-line arguments calling ``--config`` or ``--configfile``,
258 following the conventions of `sys.argv`.
259 """
260 overrideFile = apPipe.ApPipeTask._DefaultName + ".py"
261 overridePath = os.path.join(workspace.configDir, overrideFile)
263 args = ["--configfile", overridePath]
264 # Translate APDB-only arguments to work as a sub-config
265 args.extend([("diaPipe.apdb." + arg if arg != "--config" else arg)
266 for arg in _getApdbArguments(workspace, parsed)])
267 # Put output alerts into the workspace.
268 args.extend(["--config", "diaPipe.alertPackager.alertWriteLocation=" + workspace.alertLocation])
269 args.extend(["--config", "diaPipe.doPackageAlerts=True"])
271 return args
274def _getConfigArgumentsGen3(workspace, parsed):
275 """Return the config options for running the Gen 3 AP Pipeline on this
276 workspace, as command-line arguments.
278 Parameters
279 ----------
280 workspace : `lsst.ap.verify.workspace.WorkspaceGen3`
281 A Workspace whose config directory may contain various configs.
282 parsed : `argparse.Namespace`
283 Command-line arguments, including all arguments supported by `ApPipeParser`.
285 Returns
286 -------
287 args : `list` of `str`
288 Command-line arguments calling ``--config`` or ``--config-file``,
289 following the conventions of `sys.argv`.
290 """
291 # Translate APDB-only arguments to work as a sub-config
292 args = [("diaPipe:apdb." + arg if arg != "--config" else arg)
293 for arg in _getApdbArguments(workspace, parsed)]
294 args.extend([
295 # Put output alerts into the workspace.
296 "--config", "diaPipe:alertPackager.alertWriteLocation=" + workspace.alertLocation,
297 "--config", "diaPipe:doPackageAlerts=True",
298 # TODO: the configs below should not be needed after DM-26140
299 "--config-file", "calibrate:" + os.path.join(workspace.configDir, "calibrate.py"),
300 "--config-file", "imageDifference:" + os.path.join(workspace.configDir, "imageDifference.py"),
301 ])
302 # TODO: reverse-engineering the instrument should not be needed after DM-26140
303 # pipetask will crash if there is more than one instrument
304 for idRecord in workspace.workButler.registry.queryDataIds("instrument").expanded():
305 className = idRecord.records["instrument"].class_name
306 args.extend(["--instrument", className])
308 return args
311def _getCollectionArguments(workspace, reuse):
312 """Return the collections for running the Gen 3 AP Pipeline on this
313 workspace, as command-line arguments.
315 Parameters
316 ----------
317 workspace : `lsst.ap.verify.workspace.WorkspaceGen3`
318 A Workspace with a Gen 3 repository.
319 reuse : `bool`
320 If true, use the previous run collection if one exists. Otherwise,
321 create a new run.
323 Returns
324 -------
325 args : `list` of `str`
326 Command-line arguments calling ``--input`` or ``--output``,
327 following the conventions of `sys.argv`.
328 """
329 # workspace.outputName is a chained collection containing all inputs
330 args = ["--output", workspace.outputName,
331 "--clobber-partial-outputs",
332 ]
334 registry = workspace.workButler.registry
335 oldRuns = list(registry.queryCollections(re.compile(workspace.outputName + r"/\d+T\d+Z")))
336 if reuse and oldRuns:
337 args.extend(["--extend-run", "--skip-existing"])
338 return args