feat: hardening format/media type mapping for rdf graph

ckanext/dcatapchharvest/dcat_helpers.py

@@ -252,6 +252,10 @@ def get_iana_media_type_values():
     records = root.findall('.//ns:record', media_types_namespaces)
     media_type_values = {}
     for record in records:
+        if record.find('ns:file', media_types_namespaces) is None:
+            continue
+        if record.find('ns:name', media_types_namespaces) is None:
+            continue
         name = record.find('ns:name', media_types_namespaces).text
         file_value = record.find('ns:file', media_types_namespaces).text
         media_type_values[name] = media_types_namespaces['ns']+'/'+file_value

ckanext/dcatapchharvest/profiles.py

@@ -893,6 +893,7 @@ def graph_from_dataset(self, dataset_dict, dataset_ref):  # noqa
                 g.add((distribution, FOAF.page, doc))
 
             # Format
+            format_uri = None
             if resource_dict.get('format'):
                 for key, value in valid_formats.items():
                     if resource_dict.get('format') == key:
@@ -904,7 +905,7 @@ def graph_from_dataset(self, dataset_dict, dataset_ref):  # noqa
                         ))
 
             # Set Media Type value if format does not match
-            if not format_uri and resource_dict.get('media_type'):
+            if format_uri is None and resource_dict.get('media_type'):
                     for key, value in valid_media_types.items():
                         if resource_dict.get('media_type') == key:
                             media_type_uri = URIRef(value)