updated mongo plugin to deal with HDF5 metadata

src/python/dm/common/processing/plugins/mongoDbFileCatalogPlugin.py

@@ -7,15 +7,44 @@ from dm.common.objects.observedFile import ObservedFile
 from dm.common.utility.timeUtility import TimeUtility
 from dm.common.processing.plugins.fileProcessor import FileProcessor
 from dm.common.mongodb.api.fileMongoDbApi import FileMongoDbApi
+from dm.common.utility.dmSubprocess import DmSubprocess
 
 class MongoDbFileCatalogPlugin(FileProcessor):
 
-    def __init__(self, dependsOn=[]):
+    DEFAULT_HDF5_METADATA_COMMAND = None
+
+    def __init__(self, hdf5MetadataCommand=DEFAULT_HDF5_METADATA_COMMAND, dependsOn=[]):
         FileProcessor.__init__(self, dependsOn=dependsOn)
         self.fileMongoDbApi = FileMongoDbApi()
+        self.hdf5MetadataCommand = hdf5MetadataCommand
         self.logger = LoggingManager.getInstance().getLogger(self.__class__.__name__)
 
+    def processHdf5Metadata(self, filePath, fileInfo={}):
+        if not self.hdf5MetadataCommand:
+            return fileInfo
+        experimentName = fileInfo.get('experimentName', '')
+        if not filePath.endswith('.h5'):
+            return fileInfo
+        command = '%s %s' % (self.hdf5MetadataCommand, filePath)
+        subprocess = DmSubprocess.getSubprocess(command)
+        subprocess.run()
+        stdout = subprocess.getStdOut().replace('\n', ';')
+        parts = stdout.split(';')
+        for part in parts:
+            keyValue = part.split('=')
+            key = keyValue[0]
+            if not len(key):
+                continue
+            value = ''
+            if len(keyValue) > 1:
+                value = '='.join(keyValue[1:])
+            if not fileInfo.has_key(key):
+                fileInfo[key] = value
+            else:
+                self.logger.warn('Key %s already exists for file %s (experiment: %s)' % (key, filePath, experimentName))
+
     def processFile(self, fileInfo):
+        filePath = fileInfo.get('filePath')
         experimentFilePath = fileInfo.get('experimentFilePath')
         experimentName = fileInfo.get('experimentName')
         self.logger.debug('Processing file "%s" for experiment %s' % (experimentFilePath, experimentName))
@@ -55,6 +84,7 @@ class MongoDbFileCatalogPlugin(FileProcessor):
             if fileInfo2.has_key(key):
                 del fileInfo2[key]
 
+        self.processHdf5Metadata(filePath, fileInfo2)
         self.logger.debug('File "%s" catalog entry: %s' % (experimentFilePath, str(fileInfo2)))
         self.fileMongoDbApi.updateOrAddExperimentFile(fileInfo2)