Allow extracting elements holding transcriptions without entities

Closes #254 (closed)

dan/datasets/extract/arkindex.py

@@ -106,9 +106,10 @@ class ArkindexExtractor:
             raise NoTranscriptionError(element.id)
 
         transcription = random.choice(transcriptions)
+        stripped_text = transcription.text.strip()
 
         if not self.tokens:
-            return transcription.text.strip()
+            return stripped_text
 
         entities = get_transcription_entities(
             transcription.id,
@@ -116,6 +117,9 @@ class ArkindexExtractor:
             supported_types=list(self.tokens),
         )
 
+        if not entities.count():
+            return stripped_text
+
         return self.translate(
             entities_to_xml(
                 transcription.text, entities, entity_separators=self.entity_separators