Merge pull request #328 from lsst/tickets/DM-46098

DM-46098: Add NoWorkFound error handling to metadata metrics

pipelines/apDetectorVisitQualityCore.yaml

@@ -55,36 +55,59 @@ tasks:
       python: |
         from lsst.analysis.tools.atools import *
   diffimTaskCore:
-    class: lsst.analysis.tools.tasks.DiffimDetectorVisitMetricsAnalysisTask
+    class: lsst.analysis.tools.tasks.TaskMetadataAnalysisTask
     config:
-      connections.outputName: diffimMetadata
-
-      atools.diffimMetadataMetric: DiffimMetadataMetricTool
+      connections.inputName: subtractImages_metadata
+      connections.outputName: diffimMetadata  # Will be appended with "_metrics"
+      connections.storageClass: TaskMetadata
+      raiseNoWorkFoundOnEmptyMetadata: true
+      inputDimensions: ["instrument", "visit", "detector"]
+      atools.diffimMetadataMetric: TaskMetadataMetricTool
+      atools.diffimMetadataMetric.taskName: subtractImages
       atools.diffimMetadataMetric.metrics:
         # Format is "metric name in the metadata": units
-        nUnmergedDiaSources: ct
-        nMergedDiaSources: ct
-        nGoodPixels: ct
-        nBadPixels: ct
-        nPixelsDetectedPositive: ct
-        nPixelsDetectedNegative: ct
-        nBadPixelsDetectedPositive: ct
-        nBadPixelsDetectedNegative: ct
         sciencePsfSize: pixel
         templatePsfSize: pixel
         scaleScienceVarianceFactor: ''
         scaleTemplateVarianceFactor: ''
         spatialConditionNum: ''
         spatialKernelSum: ''
         templateCoveragePercent: percent
+      atools.diffimMetadataMetric.subTaskNames:
+        # Format is "metric name in metadata: subtask that produced it"
+        spatialConditionNum: makeKernel
+        spatialKernelSum: makeKernel
+      python: |
+        from lsst.analysis.tools.atools import TaskMetadataMetricTool
+  detectionTaskCore:
+    class: lsst.analysis.tools.tasks.TaskMetadataAnalysisTask
+    config:
+      connections.inputName: detectAndMeasure_metadata
+      connections.outputName: detectAndMeasure_metadata  # Will be appended with "_metrics"
+      connections.storageClass: TaskMetadata
+      raiseNoWorkFoundOnEmptyMetadata: true
+      inputDimensions: ["instrument", "visit", "detector"]
+      atools.diffimMetadataMetric: TaskMetadataMetricTool
+      atools.diffimMetadataMetric.taskName: detectAndMeasure
+      atools.diffimMetadataMetric.metrics:
+        # Format is "metric name in the metadata": units
+        nGoodPixels: ct
+        nBadPixels: ct
+        nPixelsDetectedPositive: ct
+        nPixelsDetectedNegative: ct
+        nBadPixelsDetectedPositive: ct
+        nBadPixelsDetectedNegative: ct
+        nUnmergedDiaSources: ct
+        nMergedDiaSources: ct
       python: |
-        from lsst.analysis.tools.atools import DiffimMetadataMetricTool
+        from lsst.analysis.tools.atools import TaskMetadataMetricTool
   associationCore:
     class: lsst.analysis.tools.tasks.TaskMetadataAnalysisTask
     config:
       connections.inputName: diaPipe_metadata
       connections.outputName: diaPipe_metadata  # Should be appended with "_metrics"
       connections.storageClass: TaskMetadata
+      raiseNoWorkFoundOnEmptyMetadata: true
       inputDimensions: ["instrument", "visit", "detector"]
       atools.diaPipeMetadataMetrics: TaskMetadataMetricTool
       atools.diaPipeMetadataMetrics.taskName: diaPipe
@@ -209,6 +232,7 @@ subsets:
       - analyzeAssocDiaSrcCore
       - analyzeTrailedDiaSrcCore
       - diffimTaskCore
+      - detectionTaskCore
       - associationCore
       - diffimTaskPlots
       - initialPviCore

python/lsst/analysis/tools/atools/diffimMetadataMetrics.py

@@ -22,12 +22,18 @@
 
 __all__ = ("DiffimMetadataMetricTool",)
 
+from deprecated.sphinx import deprecated
 from lsst.pex.config import DictField
 
 from ..actions.scalar import ValueAction
 from ..interfaces import AnalysisTool
 
 
+@deprecated(
+    reason=("This tool is superceded by TaskMetadataMetricTool, which should be used instead."),
+    version="v29.0",
+    category=FutureWarning,
+)
 class DiffimMetadataMetricTool(AnalysisTool):
     """This tool is designed to extract values from diffim task metadata"""
 

python/lsst/analysis/tools/tasks/diffimTaskDetectorVisitMetricsAnalysis.py

@@ -24,6 +24,7 @@
 
 import lsst.pex.config
 import pandas as pd
+from deprecated.sphinx import deprecated
 from lsst.pipe.base import NoWorkFound, connectionTypes
 
 from ..interfaces import AnalysisBaseConfig, AnalysisBaseConnections, AnalysisPipelineTask
@@ -57,6 +58,11 @@ class DiffimDetectorVisitMetricsAnalysisConfig(
     )
 
 
+@deprecated(
+    reason=("TaskMetadataAnalysisTask should be used instead."),
+    version="v29.0",
+    category=FutureWarning,
+)
 class DiffimDetectorVisitMetricsAnalysisTask(AnalysisPipelineTask):
     ConfigClass = DiffimDetectorVisitMetricsAnalysisConfig
     _DefaultName = "DiffimDetectorVisitMetricsAnalysis"
@@ -73,9 +79,8 @@ def runQuantum(self, butlerQC, inputRefs, outputRefs):
         subtractTaskName = inputRefs.metadataSubtract.datasetType.name
         subtractTaskName = subtractTaskName[: subtractTaskName.find("_")]
         metadata |= inputs["metadataSubtract"].metadata[subtractTaskName].to_dict()
-        kernelSubtaskLabel = subtractTaskName + ":" + self.config.kernelSubtaskName
-        metadata |= inputs["metadataSubtract"].metadata[kernelSubtaskLabel].to_dict()
         inputs.pop("metadataSubtract")
+        self.taskName = subtractTaskName
         # Some metadata entries might have different lengths or simply floats.
         # Pass the dict in a list to tell Pandas that this is one row in the
         # dataframe.

python/lsst/analysis/tools/tasks/metadataAnalysis.py

@@ -26,7 +26,7 @@
     "TaskMetadataAnalysisTask",
 )
 
-from lsst.pex.config import ListField
+from lsst.pex.config import Field, ListField
 from lsst.pipe.base import NoWorkFound, connectionTypes
 
 from ..interfaces import AnalysisBaseConfig, AnalysisBaseConnections, AnalysisPipelineTask
@@ -72,6 +72,11 @@ class MetadataAnalysisConfig(
         dtype=str,
         doc="The dimensions of the input dataset.",
     )
+    raiseNoWorkFoundOnEmptyMetadata = Field(
+        dtype=bool,
+        default=False,
+        doc="Raise a NoWorkFound error if none of the configured metrics are in the task metadata.",
+    )
 
 
 class DatasetMetadataAnalysisTask(AnalysisPipelineTask):
@@ -118,9 +123,32 @@ def runQuantum(self, butlerQC, inputRefs, outputRefs):
         inputs = butlerQC.get(inputRefs)
         plotInfo = self.parsePlotInfo(inputs, dataId)
         metadata = inputs["data"].get().to_dict()
+        taskName = inputRefs.data.datasetType.name
+        taskName = taskName[: taskName.find("_")]
         if not metadata:
-            taskName = inputRefs.data.datasetType.name
-            taskName = taskName[: taskName.find("_")]
             raise NoWorkFound(f"No metadata entries for {taskName}.")
+        if self.config.raiseNoWorkFoundOnEmptyMetadata:
+            self.validateMetrics(metadata, taskName)
         outputs = self.run(data=metadata, plotInfo=plotInfo)
         butlerQC.put(outputs, outputRefs)
+
+    def validateMetrics(self, metadata, taskName):
+        """Raise NoWorkFound if there are no metrics in the task metadata.
+
+        Parameters
+        ----------
+        metadata : `dict`
+            The task metadata converted to a dict.
+        taskName : `str`
+            The name of the task to extract metadata from
+
+        Raises
+        ------
+        NoWorkFound
+            If none of the metrics are in the metadata.
+        """
+        for fieldName in self.config.atools.fieldNames:
+            for key in getattr(self.config.atools, fieldName).metrics.keys():
+                if key in metadata[taskName].keys():
+                    return
+        raise NoWorkFound(f"None of the specified metrics were found in the {taskName} metadata")