refactor: move URI to label conversion (and vice versa) to SubjectIndex

annif/corpus/subject.py

@@ -78,6 +78,22 @@ def by_label(self, label):
             logger.warning('Unknown subject label "%s"', label)
             return None
 
+    def uris_to_labels(self, uris):
+        """return a list of labels corresponding to the given URIs; unknown
+        URIs are ignored"""
+
+        return [self[subject_id][1]
+                for subject_id in (self.by_uri(uri) for uri in uris)
+                if subject_id is not None]
+
+    def labels_to_uris(self, labels):
+        """return a list of URIs corresponding to the given labels; unknown
+        labels are ignored"""
+
+        return [self[subject_id][0]
+                for subject_id in (self.by_label(label) for label in labels)
+                if subject_id is not None]
+
     def save(self, path):
         """Save this subject index into a file."""
 

annif/corpus/types.py

@@ -29,18 +29,10 @@ def _create_document(self, text, uris, labels):
         information. URIs for labels and vice versa are looked up from the
         subject index, if available."""
 
-        sidx = self._subject_index
-
-        if not uris and labels and sidx:
-            uris = set((sidx[subject_id][0]
-                        for subject_id in (sidx.by_label(label)
-                                           for label in labels)
-                        if subject_id is not None))
-        if not labels and uris and sidx:
-            labels = set((sidx[subject_id][1]
-                          for subject_id in (sidx.by_uri(uri)
-                                             for uri in uris)
-                          if subject_id is not None))
+        if not uris and labels and self._subject_index:
+            uris = set((self._subject_index.labels_to_uris(labels)))
+        if not labels and uris and self._subject_index:
+            labels = set((self._subject_index.uris_to_labels(uris)))
 
         return Document(text=text, uris=uris, labels=labels)