Coverage for python/lsst/ap/verify/pipeline_driver.py: 17%
Shortcuts 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
Shortcuts 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
35import subprocess
36import logging
38import lsst.pipe.base as pipeBase
39import lsst.ctrl.mpexec.execFixupDataId # not part of lsst.ctrl.mpexec
40import lsst.ctrl.mpexec.cli.pipetask
41import lsst.ap.pipe as apPipe
42from lsst.ap.pipe.make_apdb import makeApdb
44_LOG = logging.getLogger(__name__)
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 # Help and documentation will be handled by main program's parser
56 argparse.ArgumentParser.__init__(self, add_help=False)
57 # namespace.dataIds will always be a list of 0 or more nonempty strings, regardless of inputs.
58 # TODO: in Python 3.8+, action='extend' handles nargs='?' more naturally than 'append'.
59 self.add_argument('--id', '-d', '--data-query', dest='dataIds',
60 action=self.AppendOptional, nargs='?', default=[],
61 help='An identifier for the data to process.')
62 self.add_argument("-p", "--pipeline", default=None,
63 help="A custom version of the ap_verify pipeline (e.g., with different metrics). "
64 "Defaults to the ApVerify.yaml within --dataset.")
65 self.add_argument("--db", "--db_url", default=None,
66 help="A location for the AP database, formatted as if for ApdbConfig.db_url. "
67 "Defaults to an SQLite file in the --output directory.")
68 self.add_argument("--skip-pipeline", action="store_true",
69 help="Do not run the AP pipeline itself. This argument is useful "
70 "for testing metrics on a fixed data set.")
71 self.add_argument("--clean-run", action="store_true",
72 help="Run the pipeline with a new run collection, "
73 "even if one already exists.")
75 class AppendOptional(argparse.Action):
76 """A variant of the built-in "append" action that ignores None values
77 instead of appending them.
78 """
79 # This class can't safely inherit from the built-in "append" action
80 # because there is no public class that implements it.
81 def __call__(self, parser, namespace, values, option_string=None):
82 if values is not None:
83 try:
84 allValues = getattr(namespace, self.dest)
85 allValues.append(values)
86 except AttributeError:
87 setattr(namespace, self.dest, [values])
90def runApPipeGen2(workspace, parsedCmdLine, processes=1):
91 """Run `ap_pipe` on this object's dataset.
93 Parameters
94 ----------
95 workspace : `lsst.ap.verify.workspace.WorkspaceGen2`
96 The abstract location containing input and output repositories.
97 parsedCmdLine : `argparse.Namespace`
98 Command-line arguments, including all arguments supported by `ApPipeParser`.
99 processes : `int`
100 The number of processes with which to call the AP pipeline
102 Returns
103 -------
104 apPipeReturn : `lsst.pipe.base.Struct`
105 The `~lsst.pipe.base.Struct` returned from
106 `~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 = _LOG.getChild('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
156 Returns
157 -------
158 code : `int`
159 An error code that is zero if the pipeline ran without problems, or
160 nonzero if there were errors. The exact meaning of nonzereo values
161 is an implementation detail.
162 """
163 log = _LOG.getChild('runApPipeGen3')
165 makeApdb(_getApdbArguments(workspace, parsedCmdLine))
167 pipelineFile = _getPipelineFile(workspace, parsedCmdLine)
168 pipelineArgs = ["pipetask", "run",
169 "--butler-config", workspace.repo,
170 "--pipeline", pipelineFile,
171 ]
172 # TODO: workaround for inability to generate crosstalk sources in main
173 # processing pipeline (DM-31492).
174 instruments = {id["instrument"] for id in workspace.workButler.registry.queryDataIds("instrument")}
175 if "DECam" in instruments:
176 crosstalkPipeline = "${AP_PIPE_DIR}/pipelines/DarkEnergyCamera/RunIsrForCrosstalkSources.yaml"
177 crosstalkArgs = ["pipetask", "run",
178 "--butler-config", workspace.repo,
179 "--pipeline", crosstalkPipeline,
180 ]
181 crosstalkArgs.extend(_getCollectionArguments(workspace, reuse=(not parsedCmdLine.clean_run)))
182 if parsedCmdLine.dataIds:
183 for singleId in parsedCmdLine.dataIds:
184 crosstalkArgs.extend(["--data-query", singleId])
185 crosstalkArgs.extend(["--processes", str(processes)])
186 crosstalkArgs.extend(["--register-dataset-types"])
187 subprocess.run(crosstalkArgs, capture_output=False, shell=False, check=False)
189 # Force same output run for crosstalk and main processing.
190 pipelineArgs.extend(_getCollectionArguments(workspace, reuse=True))
191 else:
192 # TODO: collections should be determined exclusively by Workspace.workButler,
193 # but I can't find a way to hook that up to the graph builder. So use the CLI
194 # for now and revisit once DM-26239 is done.
195 pipelineArgs.extend(_getCollectionArguments(workspace, reuse=(not parsedCmdLine.clean_run)))
197 pipelineArgs.extend(_getConfigArgumentsGen3(workspace, parsedCmdLine))
198 if parsedCmdLine.dataIds:
199 for singleId in parsedCmdLine.dataIds:
200 pipelineArgs.extend(["--data-query", singleId])
201 pipelineArgs.extend(["--processes", str(processes)])
202 pipelineArgs.extend(["--register-dataset-types"])
203 pipelineArgs.extend(["--graph-fixup", "lsst.ap.verify.pipeline_driver._getExecOrder"])
205 if not parsedCmdLine.skip_pipeline:
206 # subprocess is an unsafe workaround for DM-26239
207 # TODO: generalize this code in DM-26028
208 # TODO: work off of workspace.workButler after DM-26239
209 results = subprocess.run(pipelineArgs, capture_output=False, shell=False, check=False)
210 log.info('Pipeline complete.')
211 return results.returncode
212 else:
213 log.info('Skipping AP pipeline entirely.')
216def _getExecOrder():
217 """Return any constraints on the Gen 3 execution order.
219 The current constraints are that executions of DiaPipelineTask must be
220 ordered by visit ID, but this is subject to change.
222 Returns
223 -------
224 order : `lsst.ctrl.mpexec.ExecutionGraphFixup`
225 An object encoding the desired execution order as an algorithm for
226 modifying inter-quantum dependencies.
228 Notes
229 -----
230 This function must be importable, but need not be public.
231 """
232 # Source association algorithm is not time-symmetric. Force execution of
233 # association (through DiaPipelineTask) in order of ascending visit number.
234 return lsst.ctrl.mpexec.execFixupDataId.ExecFixupDataId(
235 taskLabel="diaPipe", dimensions=["visit", ], reverse=False)
238def _getPipelineFile(workspace, parsed):
239 """Return the config options for running make_apdb.py on this workspace,
240 as command-line arguments.
242 Parameters
243 ----------
244 workspace : `lsst.ap.verify.workspace.Workspace`
245 A Workspace whose pipeline directory may contain an ApVerify pipeline.
246 parsed : `argparse.Namespace`
247 Command-line arguments, including all arguments supported by `ApPipeParser`.
249 Returns
250 -------
251 pipeline : `str`
252 The location of the pipeline file to use for running ap_verify.
253 """
254 if parsed.pipeline:
255 return parsed.pipeline
256 else:
257 customPipeline = os.path.join(workspace.pipelineDir, "ApVerify.yaml")
258 if os.path.exists(customPipeline):
259 return customPipeline
260 else:
261 return os.path.join("${AP_VERIFY_DIR}", "pipelines", "ApVerify.yaml")
264def _getApdbArguments(workspace, parsed):
265 """Return the config options for running make_apdb.py on this workspace,
266 as command-line arguments.
268 Parameters
269 ----------
270 workspace : `lsst.ap.verify.workspace.Workspace`
271 A Workspace whose config directory may contain an
272 `~lsst.ap.pipe.ApPipeTask` config.
273 parsed : `argparse.Namespace`
274 Command-line arguments, including all arguments supported by `ApPipeParser`.
276 Returns
277 -------
278 args : `list` of `str`
279 Command-line arguments calling ``--config`` or ``--config-file``,
280 following the conventions of `sys.argv`.
281 """
282 if not parsed.db:
283 parsed.db = "sqlite:///" + workspace.dbLocation
285 args = ["--config", "db_url=" + parsed.db]
286 # Same special-case check as ApdbConfig.validate()
287 if parsed.db.startswith("sqlite"):
288 args.extend(["--config", "isolation_level=READ_UNCOMMITTED"])
290 return args
293def _getConfigArguments(workspace, parsed):
294 """Return the config options for running ApPipeTask on this workspace, as
295 command-line arguments.
297 Parameters
298 ----------
299 workspace : `lsst.ap.verify.workspace.WorkspaceGen2`
300 A Workspace whose config directory may contain an
301 `~lsst.ap.pipe.ApPipeTask` config.
302 parsed : `argparse.Namespace`
303 Command-line arguments, including all arguments supported by `ApPipeParser`.
305 Returns
306 -------
307 args : `list` of `str`
308 Command-line arguments calling ``--config`` or ``--configfile``,
309 following the conventions of `sys.argv`.
310 """
311 overrideFile = apPipe.ApPipeTask._DefaultName + ".py"
312 overridePath = os.path.join(workspace.configDir, overrideFile)
314 args = ["--configfile", overridePath]
315 # Translate APDB-only arguments to work as a sub-config
316 args.extend([("diaPipe.apdb." + arg if arg != "--config" else arg)
317 for arg in _getApdbArguments(workspace, parsed)])
318 # Put output alerts into the workspace.
319 args.extend(["--config", "diaPipe.alertPackager.alertWriteLocation=" + workspace.alertLocation])
320 args.extend(["--config", "diaPipe.doPackageAlerts=True"])
322 return args
325def _getConfigArgumentsGen3(workspace, parsed):
326 """Return the config options for running the Gen 3 AP Pipeline on this
327 workspace, as command-line arguments.
329 Parameters
330 ----------
331 workspace : `lsst.ap.verify.workspace.WorkspaceGen3`
332 A Workspace whose config directory may contain various configs.
333 parsed : `argparse.Namespace`
334 Command-line arguments, including all arguments supported by `ApPipeParser`.
336 Returns
337 -------
338 args : `list` of `str`
339 Command-line arguments calling ``--config`` or ``--config-file``,
340 following the conventions of `sys.argv`.
341 """
342 # Translate APDB-only arguments to work as a sub-config
343 args = [("diaPipe:apdb." + arg if arg != "--config" else arg)
344 for arg in _getApdbArguments(workspace, parsed)]
345 args.extend([
346 # Put output alerts into the workspace.
347 "--config", "diaPipe:alertPackager.alertWriteLocation=" + workspace.alertLocation,
348 ])
349 return args
352def _getCollectionArguments(workspace, reuse):
353 """Return the collections for running the Gen 3 AP Pipeline on this
354 workspace, as command-line arguments.
356 Parameters
357 ----------
358 workspace : `lsst.ap.verify.workspace.WorkspaceGen3`
359 A Workspace with a Gen 3 repository.
360 reuse : `bool`
361 If true, use the previous run collection if one exists. Otherwise,
362 create a new run.
364 Returns
365 -------
366 args : `list` of `str`
367 Command-line arguments calling ``--input`` or ``--output``,
368 following the conventions of `sys.argv`.
369 """
370 # workspace.outputName is a chained collection containing all inputs
371 args = ["--output", workspace.outputName,
372 "--clobber-outputs",
373 ]
375 registry = workspace.workButler.registry
376 # Should refresh registry to see crosstalk run from DM-31492, but this
377 # currently leads to a bug involving --skip-existing. The only downside of
378 # the cached registry is that, with two runs for DECam datasets, a rerun of
379 # ap_verify will re-run crosstalk sources in the second run. Using
380 # skip-existing-in would work around that, but would lead to a worse bug in
381 # the case that the user is alternating runs with and without --clean-run.
382 # registry.refresh()
383 oldRuns = list(registry.queryCollections(re.compile(workspace.outputName + r"/\d+T\d+Z")))
384 if reuse and oldRuns:
385 args.extend(["--extend-run", "--skip-existing"])
386 return args