Allow extracting elements holding transcriptions without entities

Closes #254 (closed)

tests/test_extract.py

@@ -425,30 +425,29 @@ def test_empty_transcription(allow_empty, mock_database):
             extractor.extract_transcription(element_no_transcription)
 
 
-def test_extract_transcription_no_tokens(mock_database):
+@pytest.mark.parametrize("tokens", (None, EXTRACTION_DATA_PATH / "tokens.yml"))
+def test_extract_transcription_no_translation(mock_database, tokens):
     extractor = ArkindexExtractor(
         element_type=["text_line"],
         entity_separators=None,
-        # No tokens provided to the extractor
-        tokens=None,
+        tokens=tokens,
     )
-    element = Element.get_by_id("test-page_1-line_1")
-    assert extractor.extract_transcription(element) == "Coupez  Bouis  7.12.14"
 
-
-def test_extract_transcription_without_entities(mock_database):
-    extractor = ArkindexExtractor(
-        element_type=["text_line"],
-        entity_separators=None,
-        tokens=EXTRACTION_DATA_PATH / "tokens.yml",
-    )
     element = Element.get_by_id("test-page_1-line_1")
-    # Deleting all entities on the element transcriptions while leaving the transcriptions intact
-    TranscriptionEntity.delete().where(
-        TranscriptionEntity.transcription.in_(
-            Transcription.select().where(Transcription.element == element)
-        )
-    ).execute()
+    # Deleting one of the two transcriptions from the element
+    Transcription.get(
+        Transcription.element == element,
+        Transcription.worker_version_id == "worker_version_id",
+    ).delete_instance(recursive=True)
+
+    # Deleting all entities on the element remaining transcription while leaving the transcription intact
+    if tokens:
+        TranscriptionEntity.delete().where(
+            TranscriptionEntity.transcription
+            == Transcription.select().where(Transcription.element == element).get()
+        ).execute()
+
+    # Early return with only the element transcription text instead of a translation
     assert extractor.extract_transcription(element) == "Coupez  Bouis  7.12.14"