Adding support for multilingual vtt files

internetarchive · May 16, 2024 · f5df7fe · f5df7fe
1 parent b45bd71
commit f5df7fe
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 7 deletions.
diff --git a/iiify/resolver.py b/iiify/resolver.py
@@ -517,12 +517,12 @@ def create_manifest3(identifier, domain=None, page=None):
                 originals.append(f)
 
             if f['format'] == 'Web Video Text Tracks':
-                # Example: 34C3_-_International_Image_Interoperability_Framework_IIIF_Kulturinstitutionen_schaffen_interop-SvH4fbjOT0A.autogenerated.vtt
-                sourceFilename = f['name'].replace('.autogenerated.vtt', '')
-                # Example: cruz-test.en.vtt
-                sourceFilename = sourceFilename.replace('[a-z][a-z].vtt', '')
-
-                vttfiles[sourceFilename] = [f]    
+                # Example: cruz-test.en.vtt and 34C3_-_International_Image_Interoperability_Framework_IIIF_Kulturinstitutionen_schaffen_interop-SvH4fbjOT0A.autogenerated.vtt
+                sourceFilename = re.sub('\.[a-zA-H-]*\.vtt', '', f['name'])
+                if sourceFilename not in vttfiles:
+                    vttfiles[sourceFilename] = []    
+                    
+                vttfiles[sourceFilename].append(f)    
 
         # create the canvases for each original
         for file in [f for f in originals if f['format'] in ['MPEG4', 'h.264 MPEG4', '512Kb MPEG4', 'HiRes MPEG4', 'MPEG2', 'h.264', 'Matroska', 'Ogg Video', 'Ogg Theora', 'WebM', 'Windows Media', 'Cinepack']]:
@@ -552,7 +552,7 @@ def create_manifest3(identifier, domain=None, page=None):
                         # Assume langauge
                         splitName = vttFile['name'].split(".")
                         lang = splitName[-2]
-                        vtAnno.body.add_label(lang, language=lang)
+                        vtAnno.body.add_label(lang, language="none")
                         vtAnno.body.language = lang
 
                     vttNo += 1

diff --git a/tests/test_video.py b/tests/test_video.py
@@ -46,6 +46,23 @@ def test_vtt_autogenerated(self):
         self.assertFalse("language" in body, "We don't know the language for this item so there shouldn't be a language specified")
         self.assertEqual(body['id'], "https://archive.org/download/youtube-SvH4fbjOT0A/34C3_-_International_Image_Interoperability_Framework_IIIF_Kulturinstitutionen_schaffen_interop-SvH4fbjOT0A.autogenerated.vtt","Unexpected URL for the VTT file")
 
+    def test_vtt_multilingual(self):
+        resp = self.test_app.get("/iiif/3/cruz-test/manifest.json?recache=true")
+        self.assertEqual(resp.status_code, 200)
+        manifest = resp.json
+
+        canvas = manifest['items'][0]
+        self.assertTrue('annotations' in canvas, 'Expected annotations in Canvas')
+        self.assertEqual(len(canvas['annotations']), 1, 'Expected one AnnotationPage')
+        annotations = canvas['annotations'][0]['items']
+        self.assertEqual(len(annotations), 104, 'Expected all 104 langues')
+
+        # Check welsh
+        for item in annotations:
+            self.assertTrue('language' in item['body'], f"All vtt files should have a language: {item}")
+            if item['body']['language'] == 'cy':
+                self.assertEqual(item['body']['id'], 'https://archive.org/download/cruz-test/cruz-test.cy.vtt', 'Unexpected link for the Welsh vtt file')
+
 
 if __name__ == '__main__':
     unittest.main()