Changeset 1821

Timestamp:

05/12/06 17:27:45 (14 years ago)

Author:

selatham

Message:

error trapping

File:

• TI01-discovery/trunk/ingestAutomation/OAIBatch/SpaceTimeIngestFromMOLES.py (modified) (6 diffs)

TI01-discovery/trunk/ingestAutomation/OAIBatch/SpaceTimeIngestFromMOLES.py

@@ -36 +36 @@
 def main(indir):
     if indir == "":
-        sys.exit("Usage: argument 1 = full path of directory where MOLES records reside")
+        sys.exit("USAGE: argument 1 = full path of directory where MOLES records reside")
     else:
         print "INFO: moles records are in %s" %indir
 
 #this is a fix to the  ElementTree namespace problem that namespaces are usually represented as ns0, ns1, ns2 etc.
-    etree._namespace_map.update({'http://ndg.nerc.ac.uk/moles': 'moles', 'http://ndg.nerc.ac.uk/csml' : 'csml', 'http://www.w3.org/1999/xlink':'xlink'})
+    etree._namespace_map.update({'http://ndg.nerc.ac.uk/moles': 'moles', 'http://www.w3.org/1999/xlink':'xlink'})
     numfilesproc = 0
 
@@ -49 +49 @@
                 full_filename = indir + "/" + filename
                 dgMeta=MRW.dgMetadata()
-                dgMeta.fromXML(cElementTree.ElementTree(file=full_filename).getroot())
-                bbox=dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage.dgSpatialCoverage.BoundingBox
-                dates=dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage.dgTemporalCoverage.DateRange
+                try:
+                    dgMeta.fromXML(cElementTree.ElementTree(file=full_filename).getroot())
+                except:
+                    print "WARNING: Cannot parse the XML moles document %s. Will not process" %full_filename
+                    continue
+                try:
+                    bbox=dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage.dgSpatialCoverage.BoundingBox
+                except:
+                    print "INFO: XML moles document %s does not contain a bounding box." %full_filename
+
+                try:
+                    dates=dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage.dgTemporalCoverage.DateRange
+                except:
+                    print "INFO: XML moles document %s does not contain a bounding box." %full_filename
+
                 coverage= [dates.DateRangeEnd, dates.DateRangeStart, bbox.LimitNorth, bbox.LimitSouth,bbox.LimitEast, bbox.LimitWest]
                 #Mid = dgMeta.dgMetadataRecord.dgMetadataID.repositoryIdentifier+"__"+dgMeta.dgMetadataRecord.dgMetadataID.localIdentifier
@@ -69 +81 @@
                     float(west)
                 except:
-                    print "Error: Will not process File %s. Contains incorrect West bounding box limit." %full_filename
+                    print "ERROR:  Will not process File %s. Contains incorrect West bounding box limit." %full_filename
                     continue
                 print "West = %s" %west
@@ -84 +96 @@
                     float(east)
                 except:
-                    print "Error: Will not process File %s. Contains incorrect East bounding box limit." %full_filename
+                    print "ERROR:  Will not process File %s. Contains incorrect East bounding box limit." %full_filename
                     continue
                 print "East = %s" %east
@@ -99 +111 @@
                     float(north)
                 except:
-                    print "Error: Will not process File %s. Contains incorrect North bounding box limit." %full_filename
+                    print "ERROR: Will not process File %s. Contains incorrect North bounding box limit." %full_filename
                     continue
                 print "North = %s" %north
@@ -114 +126 @@
                     float(south)
                 except:
-                    print "Error: Will not process File %s. Contains incorrect North bounding box limit." %full_filename
+                    print "ERROR: Will not process File %s. Contains incorrect North bounding box limit." %full_filename
                     continue
                 print "North = %s" %south
 
                 if id_exists( Mid ):
-                        print "doc %s exists, updating\n" %Mid
-                        do_update( Mid, west, south, east, north, dates.DateRangeStart, dates.DateRangeEnd )
+                    print "INFO: doc %s exists, updating\n" %Mid
+                    do_update( Mid, west, south, east, north, dates.DateRangeStart, dates.DateRangeEnd )
                 else:
-                        print "doc %s does not exist, inserting new record\n" %Mid
-                        do_insert( Mid, west, south, east, north, dates.DateRangeStart, dates.DateRangeEnd )
+                    print "INFO: doc %s does not exist, inserting new record\n" %Mid
+                    do_insert( Mid, west, south, east, north, dates.DateRangeStart, dates.DateRangeEnd )
 
                 numfilesproc += 1
 
+            else:
+                print "WARNING: File %s appears not to be XML. Will not be processed." %filename
 
-    print 'SpaceTimeIngestFromMOLES.py ran to end. files processed= %s' %(numfilesproc)
+    print 'INFO: SpaceTimeIngestFromMOLES.py ran to end. files processed= %s' %(numfilesproc)
 
 if __name__=='__main__':