Merge pull request #6229 from edx/waheed/tnl527-fix-unicode-error-for-non-utf8-transcript

Fixed unicode problem if transcript file is not UTF-8 encoded.

common/lib/xmodule/xmodule/video_module/video_handlers.py

@@ -300,7 +300,14 @@ class VideoStudioViewHandlers(object):
 
             if request.method == 'POST':
                 subtitles = request.POST['file']
-                save_to_store(subtitles.file.read(), unicode(subtitles.filename), 'application/x-subrip', self.location)
+                try:
+                    file_data = subtitles.file.read()
+                    unicode(file_data, "utf-8", "strict")
+                except UnicodeDecodeError:
+                    log.info("Invalid encoding type for transcript file: {}".format(subtitles.filename))
+                    msg = _("Invalid encoding type, transcripts should be UTF-8 encoded.")
+                    return Response(msg, status=400)
+                save_to_store(file_data, unicode(subtitles.filename), 'application/x-subrip', self.location)
                 generate_sjson_for_all_speeds(self, unicode(subtitles.filename), {}, language)
                 response = {'filename': unicode(subtitles.filename), 'status': 'Success'}
                 return Response(json.dumps(response), status=201)

lms/djangoapps/courseware/tests/test_video_handlers.py

@@ -577,8 +577,10 @@ class TestStudioTranscriptTranslationPostDispatch(TestVideo):
             response = self.item_descriptor.studio_transcript(request=request, dispatch='translation/uk')
 
         request = Request.blank('/translation/uk', POST={'file': ('filename.srt', SRT_content.decode('utf8').encode('cp1251'))})
-        with self.assertRaises(UnicodeDecodeError):  # Non-UTF8 file content encoding.
-            response = self.item_descriptor.studio_transcript(request=request, dispatch='translation/uk')
+        # Non-UTF8 file content encoding.
+        response = self.item_descriptor.studio_transcript(request=request, dispatch='translation/uk')
+        self.assertEqual(response.status_code, 400)
+        self.assertEqual(response.body, "Invalid encoding type, transcripts should be UTF-8 encoded.")
 
         # No language is passed.
         request = Request.blank('/translation', POST={'file': ('filename', SRT_content)})