Merge pull request #585 from xxyzz/zh

Fix 'OGG_url' not defined error in some zh edition pages

src/wiktextract/extractor/share.py

@@ -55,8 +55,8 @@ def split_chinese_variants(text: str) -> Iterable[tuple[Optional[str], str]]:
 def create_audio_url_dict(filename: str) -> dict[str, str]:
     # remove white space and left-to-right mark
     filename = filename.strip(" \u200e")
-    file_extension = filename[filename.rfind(".") + 1 :]
-    if file_extension.lower() == "ogv":
+    file_extension = filename[filename.rfind(".") + 1 :].lower()
+    if file_extension == "ogv":
         # ".ogv" pages are redirected to ".oga" pages in Wikipedia Commons
         filename = filename[:filename.rfind(".")] + ".oga"
         file_extension = "oga"

tests/test_utils.py

@@ -75,6 +75,7 @@ def test_audio_transcode_url(self):
                 "ogg_url": "https://upload.wikimedia.org/wikipedia/commons/transcoded/e/e1/LL-Q150_(fra)-DenisdeShawi-bonjour.wav/LL-Q150_(fra)-DenisdeShawi-bonjour.wav.ogg",
             },
         )
+
         sound_data = create_audio_url_dict("File:Fr-BonjourF.oga")
         self.assertEqual(
             sound_data,
@@ -84,5 +85,16 @@ def test_audio_transcode_url(self):
                 "mp3_url": "https://upload.wikimedia.org/wikipedia/commons/transcoded/b/b9/Fr-BonjourF.oga/Fr-BonjourF.oga.mp3",
             },
         )
+
         sound_data = create_audio_url_dict("Qc-Vancouver.ogv")
         self.assertEqual(sound_data["audio"], "Qc-Vancouver.oga")
+
+        sound_data = create_audio_url_dict("De-Fisch.OGG")
+        self.assertEqual(
+            sound_data,
+            {
+                "audio": "De-Fisch.OGG",
+                "ogg_url": "https://commons.wikimedia.org/wiki/Special:FilePath/De-Fisch.OGG",
+                "mp3_url": "https://upload.wikimedia.org/wikipedia/commons/transcoded/0/0f/De-Fisch.OGG/De-Fisch.OGG.mp3",
+            },
+        )