Nits

arkindex/documents/importer.py

@@ -109,7 +109,7 @@ class ManifestsImporter(ABC):
     Parses JSON manifests and annotation data to import them in the database.
     """
 
-    def __init__(self, imgserv, offline=False, no_annotations=False):
+    def __init__(self, imgserv, offline=False, annotations=True):
         """Initialize a manifest importer
         `imgserv` can be either one ImageServer or a list of ImageServers."""
         if isinstance(imgserv, ImageServer):
@@ -119,7 +119,7 @@ class ManifestsImporter(ABC):
             self.imgserv = imgserv
 
         self.offline = offline
-        self.no_annotations = no_annotations
+        self.annotations = annotations
 
         # This dictionary associates canvas IDs with images and pages
         # Filled by parse_manifest ; used by parse_annotation_list
@@ -158,7 +158,7 @@ class ManifestsImporter(ABC):
                     stream.seek(0)
                     self.parse_manifest(stream)
                     break
-                elif value == 'sc:AnnotationList' and not self.no_annotations:
+                elif value == 'sc:AnnotationList' and not self.annotations:
                     stream.seek(0)
                     self.parse_annotation_list(stream)
                     break

arkindex/documents/management/commands/import_annotations.py

@@ -17,7 +17,7 @@ class Command(BaseCommand):
         parser.add_argument(
             'index_folder',
             help='Folder to recursively search for index files',
-            default='.'
+            default='.',
         )
         parser.add_argument(
             '--dry-run',
@@ -28,13 +28,14 @@ class Command(BaseCommand):
         parser.add_argument(
             '--mask',
             help="""A mask to identify images from the index file path. Cannot be used with --regex.
-                 Syntax: "something<ID>something"
-                 <ID> will be used as the image ID."""
+                 Syntax: "something<PATH>something"
+                 <PATH> will be used as the image path.""",
         )
         parser.add_argument(
             '--regex',
             help="""A regex to use as a mask for more complex cases.
-                 Must have only one capturing group. Cannot be used with --mask."""
+                 Must have only one capturing group. Cannot be used with --mask.""",
+            default=IndexImporter.DEFAULT_MASK,
         )
 
     def handle(self, *args, **options):
@@ -48,14 +49,13 @@ class Command(BaseCommand):
         if options['mask'] is not None and options['regex'] is not None:
             raise CommandError('--mask and --regex cannot be used simultaneously.')
 
-        # If --regex is set, use it, else use the default mask.
-        mask_regex = options['regex'] or IndexImporter.DEFAULT_MASK
+        mask_regex = options['regex']
 
         if options['mask'] is not None:
             mask = options['mask']
-            assert mask.count('<ID>') == 1
-            # Replace <ID> with (.+) and escape the rest
-            mask_regex = '^' + r'(.+)'.join(re.escape(p) for p in mask.split('<ID>')) + '$'
+            assert mask.count('<PATH>') == 1
+            # Replace <PATH> with (.+) and escape the rest
+            mask_regex = '^' + r'(.+)'.join(re.escape(p) for p in mask.split('<PATH>')) + '$'
 
         importer = IndexImporter(options['index_folder'], mask=mask_regex)
         if options['dry_run']:

arkindex/documents/management/commands/import_manifest.py

@@ -56,5 +56,5 @@ class Command(BaseCommand):
             servers,
             options['manifest_folder'],
             offline=options['offline'],
-            no_annotations=options['no_annotations'],
+            annotations=not options['no_annotations'],
         ).run()

arkindex/images/importer.py

@@ -33,7 +33,7 @@ def import_indexes(image, page, index_path, extension='jpg'):
                 continue
             index = REGEX_INDEX.match(line)
             if index is None:
-                logger.warn('Index parsing failed : {}'.format(line))
+                logger.warning('Index parsing failed : {}'.format(line))
                 continue
 
             # Build zone
@@ -194,6 +194,6 @@ class IndexImporter(object):
             except (Image.DoesNotExist, Image.MultipleObjectsReturned):
                 pass
             if image is None:
-                print("{}\tFAIL".format(index_path))
+                logger.warning("{}\tFAIL".format(index_path))
             else:
-                print("{}\t{}".format(index_path, image.path))
+                logger.info("{}\t{}".format(index_path, image.path))