Coverage for python/lsst/ctrl/mpexec/singleQuantumExecutor.py : 13%

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 ctrl_mpexec.
2#
3# Developed for the LSST Data Management System.
4# This product includes software developed by the LSST Project
5# (http://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 <http://www.gnu.org/licenses/>.
22__all__ = ['SingleQuantumExecutor']
24# -------------------------------
25# Imports of standard modules --
26# -------------------------------
27import logging
28from itertools import chain
30# -----------------------------
31# Imports for other modules --
32# -----------------------------
33from .quantumGraphExecutor import QuantumExecutor
34from lsst.log import Log
35from lsst.obs.base import Instrument
36from lsst.pipe.base import ButlerQuantumContext
38# ----------------------------------
39# Local non-exported definitions --
40# ----------------------------------
42_LOG = logging.getLogger(__name__.partition(".")[2])
45class SingleQuantumExecutor(QuantumExecutor):
46 """Executor class which runs one Quantum at a time.
48 Parameters
49 ----------
50 butler : `~lsst.daf.butler.Butler`
51 Data butler.
52 taskFactory : `~lsst.pipe.base.TaskFactory`
53 Instance of a task factory.
54 skipExisting : `bool`, optional
55 If True then quanta with all existing outputs are not executed.
56 clobberPartialOutputs : `bool`, optional
57 If True then delete any partial outputs from quantum execution. If
58 complete outputs exists then exception is raise if ``skipExisting`` is
59 False.
60 enableLsstDebug : `bool`, optional
61 Enable debugging with ``lsstDebug`` facility for a task.
62 """
63 def __init__(self, taskFactory, skipExisting=False, clobberPartialOutputs=False, enableLsstDebug=False):
64 self.taskFactory = taskFactory
65 self.skipExisting = skipExisting
66 self.enableLsstDebug = enableLsstDebug
67 self.clobberPartialOutputs = clobberPartialOutputs
69 def execute(self, taskDef, quantum, butler):
70 # Docstring inherited from QuantumExecutor.execute
71 taskClass, config = taskDef.taskClass, taskDef.config
72 self.setupLogging(taskClass, config, quantum)
74 # check whether to skip or delete old outputs
75 if self.checkExistingOutputs(quantum, butler, taskDef):
76 _LOG.info("Quantum execution skipped due to existing outputs, "
77 f"task={taskClass.__name__} dataId={quantum.dataId}.")
78 return
80 self.updateQuantumInputs(quantum, butler)
82 # enable lsstDebug debugging
83 if self.enableLsstDebug:
84 try:
85 _LOG.debug("Will try to import debug.py")
86 import debug # noqa:F401
87 except ImportError:
88 _LOG.warn("No 'debug' module found.")
90 # initialize global state
91 self.initGlobals(quantum, butler)
93 # Ensure that we are executing a frozen config
94 config.freeze()
96 task = self.makeTask(taskClass, config, butler)
97 self.runQuantum(task, quantum, taskDef, butler)
99 def setupLogging(self, taskClass, config, quantum):
100 """Configure logging system for execution of this task.
102 Ths method can setup logging to attach task- or
103 quantum-specific information to log messages. Potentially this can
104 take into accout some info from task configuration as well.
106 Parameters
107 ----------
108 taskClass : `type`
109 Sub-class of `~lsst.pipe.base.PipelineTask`.
110 config : `~lsst.pipe.base.PipelineTaskConfig`
111 Configuration object for this task
112 quantum : `~lsst.daf.butler.Quantum`
113 Single Quantum instance.
114 """
115 # include input dataIds into MDC
116 dataIds = set(ref.dataId for ref in chain.from_iterable(quantum.predictedInputs.values()))
117 if dataIds:
118 if len(dataIds) == 1:
119 Log.MDC("LABEL", str(dataIds.pop()))
120 else:
121 Log.MDC("LABEL", '[' + ', '.join([str(dataId) for dataId in dataIds]) + ']')
123 def checkExistingOutputs(self, quantum, butler, taskDef):
124 """Decide whether this quantum needs to be executed.
126 If only partial outputs exist then they are removed if
127 ``clobberPartialOutputs`` is True, otherwise an exception is raised.
129 Parameters
130 ----------
131 quantum : `~lsst.daf.butler.Quantum`
132 Quantum to check for existing outputs
133 butler : `~lsst.daf.butler.Butler`
134 Data butler.
135 taskDef : `~lsst.pipe.base.TaskDef`
136 Task definition structure.
138 Returns
139 -------
140 exist : `bool`
141 True if all quantum's outputs exist in a collection and
142 ``skipExisting`` is True, False otherwise.
144 Raises
145 ------
146 RuntimeError
147 Raised if some outputs exist and some not.
148 """
149 collection = butler.run
150 registry = butler.registry
152 existingRefs = []
153 missingRefs = []
154 for datasetRefs in quantum.outputs.values():
155 for datasetRef in datasetRefs:
156 ref = registry.findDataset(datasetRef.datasetType, datasetRef.dataId,
157 collections=butler.run)
158 if ref is None:
159 missingRefs.append(datasetRef)
160 else:
161 existingRefs.append(ref)
162 if existingRefs and missingRefs:
163 # some outputs exist and some don't, either delete existing ones or complain
164 _LOG.debug("Partial outputs exist for task %s dataId=%s collection=%s "
165 "existingRefs=%s missingRefs=%s",
166 taskDef, quantum.dataId, collection, existingRefs, missingRefs)
167 if self.clobberPartialOutputs:
168 _LOG.info("Removing partial outputs for task %s: %s", taskDef, existingRefs)
169 butler.pruneDatasets(existingRefs, disassociate=True, unstore=True, purge=True)
170 return False
171 else:
172 raise RuntimeError(f"Registry inconsistency while checking for existing outputs:"
173 f" collection={collection} existingRefs={existingRefs}"
174 f" missingRefs={missingRefs}")
175 elif existingRefs:
176 # complete outputs exist, this is fine only if skipExisting is set
177 return self.skipExisting
178 else:
179 # no outputs exist
180 return False
182 def makeTask(self, taskClass, config, butler):
183 """Make new task instance.
185 Parameters
186 ----------
187 taskClass : `type`
188 Sub-class of `~lsst.pipe.base.PipelineTask`.
189 config : `~lsst.pipe.base.PipelineTaskConfig`
190 Configuration object for this task
192 Returns
193 -------
194 task : `~lsst.pipe.base.PipelineTask`
195 Instance of ``taskClass`` type.
196 butler : `~lsst.daf.butler.Butler`
197 Data butler.
198 """
199 # call task factory for that
200 return self.taskFactory.makeTask(taskClass, config, None, butler)
202 def updateQuantumInputs(self, quantum, butler):
203 """Update quantum with extra information.
205 Some methods may require input DatasetRefs to have non-None
206 ``dataset_id``, but in case of intermediate dataset it may not be
207 filled during QuantumGraph construction. This method will retrieve
208 missing info from registry.
210 Parameters
211 ----------
212 quantum : `~lsst.daf.butler.Quantum`
213 Single Quantum instance.
214 butler : `~lsst.daf.butler.Butler`
215 Data butler.
216 """
217 for refsForDatasetType in quantum.predictedInputs.values():
218 newRefsForDatasetType = []
219 for ref in refsForDatasetType:
220 if ref.id is None:
221 resolvedRef = butler.registry.findDataset(ref.datasetType, ref.dataId,
222 collections=butler.collections)
223 if resolvedRef is None:
224 raise ValueError(
225 f"Cannot find {ref.datasetType.name} with id {ref.dataId} "
226 f"in collections {butler.collections}."
227 )
228 newRefsForDatasetType.append(resolvedRef)
229 _LOG.debug("Updating dataset ID for %s", ref)
230 else:
231 newRefsForDatasetType.append(ref)
232 refsForDatasetType[:] = newRefsForDatasetType
234 def runQuantum(self, task, quantum, taskDef, butler):
235 """Execute task on a single quantum.
237 Parameters
238 ----------
239 task : `~lsst.pipe.base.PipelineTask`
240 Task object.
241 quantum : `~lsst.daf.butler.Quantum`
242 Single Quantum instance.
243 taskDef : `~lsst.pipe.base.TaskDef`
244 Task definition structure.
245 butler : `~lsst.daf.butler.Butler`
246 Data butler.
247 """
248 # Create a butler that operates in the context of a quantum
249 butlerQC = ButlerQuantumContext(butler, quantum)
251 # Get the input and output references for the task
252 inputRefs, outputRefs = taskDef.connections.buildDatasetRefs(quantum)
254 # Call task runQuantum() method. Any exception thrown by the task
255 # propagates to caller.
256 task.runQuantum(butlerQC, inputRefs, outputRefs)
258 if taskDef.metadataDatasetName is not None:
259 # DatasetRef has to be in the Quantum outputs, can lookup by name
260 try:
261 ref = quantum.outputs[taskDef.metadataDatasetName]
262 except LookupError as exc:
263 raise LookupError(
264 f"Quantum outputs is missing metadata dataset type {taskDef.metadataDatasetName},"
265 f" it could happen due to inconsistent options between Quantum generation"
266 f" and execution") from exc
267 butlerQC.put(task.getFullMetadata(), ref[0])
269 def initGlobals(self, quantum, butler):
270 """Initialize global state needed for task execution.
272 Parameters
273 ----------
274 quantum : `~lsst.daf.butler.Quantum`
275 Single Quantum instance.
276 butler : `~lsst.daf.butler.Butler`
277 Data butler.
279 Notes
280 -----
281 There is an issue with initializing filters singleton which is done
282 by instrument, to avoid requiring tasks to do it in runQuantum()
283 we do it here when any dataId has an instrument dimension. Also for
284 now we only allow single instrument, verify that all instrument
285 names in all dataIds are identical.
287 This will need revision when filter singleton disappears.
288 """
289 oneInstrument = None
290 for datasetRefs in chain(quantum.predictedInputs.values(), quantum.outputs.values()):
291 for datasetRef in datasetRefs:
292 dataId = datasetRef.dataId
293 instrument = dataId.get("instrument")
294 if instrument is not None:
295 if oneInstrument is not None:
296 assert instrument == oneInstrument, \
297 "Currently require that only one instrument is used per graph"
298 else:
299 oneInstrument = instrument
300 Instrument.fromName(instrument, butler.registry)