MCPClient: fix matching originals to manual derivatives

Fixes the matching of original files to their manually created
preservation derivatives in a manualNormalization/preservation/
directory. Previous behaviour would find a match and all other Files
that matched by having a path that was a superfix of the original path.
This would trigger an exception. This fixes the issue by returning the shortest
match. It also adds print statements to better understand what has happened
through the Tasks GUI.

Author: jrwdunham

src/MCPClient/lib/clientScripts/archivematicaCreateMETS2.py

@@ -890,7 +890,12 @@ def createFileSec(directoryPath, parentDiv, baseDirectoryPath,
 
             elif use in ("preservation", "text/ocr"):
                 # Derived files should be in the original file's group
-                d = Derivation.objects.get(derived_file_id=f.uuid)
+                try:
+                    d = Derivation.objects.get(derived_file_id=f.uuid)
+                except Derivation.DoesNotExist:
+                    print('Fatal error: unable to locate a Derivation object'
+                          ' where the derived file is {}'.format(f.uuid))
+                    raise
                 GROUPID = "Group-" + d.source_file_id
 
             elif use == "service":

src/MCPClient/lib/clientScripts/manualNormalizationCreateMetadataAndRestructure.py

@@ -117,6 +117,9 @@
     e = Event.objects.get(event_type="normalization", file_uuid=original_file)
     e.event_outcome_detail = dstR
     e.save()
+    print('Updated the eventOutcomeDetailNote of an existing normalization'
+          ' Event for file {}. Not creating a Derivation object'.format(
+              fileUUID))
 except Event.DoesNotExist:
     # No normalization event was created in normalize.py - probably manually
     # normalized during Ingest
@@ -129,12 +132,16 @@
         eventDetail="manual normalization",
         eventOutcome="",
         eventOutcomeDetailNote=dstR)
+    print('Created a manual normalization Event for file {}.'.format(
+        original_file.uuid))
 
     # Add linking information between files
     # Assuming that if an event already exists, then the derivation does as well
     databaseFunctions.insertIntoDerivations(
         sourceFileUUID=original_file.uuid,
         derivedFileUUID=fileUUID,
         relatedEventUUID=derivationEventUUID)
+    print('Created a Derivation for original file {}, derived file {}, and'
+          ' event {}'.format(original_file.uuid, fileUUID, derivationEventUUID))
 
 exit(0)

src/MCPClient/lib/clientScripts/normalize.py

@@ -147,13 +147,27 @@ def check_manual_normalization(opts):
                             "%SIPDirectory%objects/manualNormalization/access/")
     else:
         return None
-    try:
-        # FIXME: SQL uses removedtime=0. Cannot get Django to express this
-        return File.objects.get(sip=opts.sip_uuid, currentlocation__startswith=path)  # removedtime = 0
-    except (File.DoesNotExist, File.MultipleObjectsReturned):
+
+    # FIXME: SQL uses removedtime=0. Cannot get Django to express this
+    print('Checking for a manually normalized file by trying to get the'
+          ' unique file that matches SIP UUID {} and whose currentlocation'
+          ' value starts with this path: {}.'.format(opts.sip_uuid, path))
+    matches = File.objects.filter(  # removedtime = 0
+        sip=opts.sip_uuid, currentlocation__startswith=path)
+    if not matches:
         # No file with the correct path found, assume not manually normalized
+        print('No such file found.')
         return None
-    return None
+    if len(matches) > 1:
+        # If multiple matches, the shortest one should be the correct one. E.g.,
+        # if original is /a/b/abc.NEF then /a/b/abc.tif and /a/b/abc_1.tif will
+        # both match but /a/b/abc.tif is the correct match.
+        print('Multiple files matching path {} found. Returning the shortest'
+              ' one.')
+        ret = sorted(matches, key=lambda f: f.currentlocation)[0]
+        print('Returning file at {}'.format(ret.currentlocation))
+        return ret
+    return matches[0]
 
 
 def once_normalized(command, opts, replacement_dict):