Deprioritize TranscriptUrl

yt_dlp/extractor/mediasite.py

@@ -7,6 +7,7 @@ from ..utils import (
     ExtractorError,
     determine_ext,
     float_or_none,
+    join_nonempty,
     mimetype2ext,
     smuggle_url,
     str_or_none,
@@ -269,11 +270,11 @@ class MediasiteIE(InfoExtractor):
             formats.extend(stream_formats)
 
         # XXX: Presentation['Presenters']
-        transcripts = presentation.get('Transcripts', {})
+        transcripts = presentation.get('Transcripts', [])
         captions, subtitles = {}, {}
         for transcript in transcripts:
             lang_code = traverse_obj(
-                transcript, (('DetailedLanguageCode', 'LanguageCode'), {str}), get_all=False)
+                transcript, (('DetailedLanguageCode', 'LanguageCode'), {str}), get_all=False) or 'und'
             lang_name = transcript.get('Language')
             t = {
                 'url': transcript.get('CaptionsUrl'),
@@ -283,15 +284,19 @@ class MediasiteIE(InfoExtractor):
                 captions.setdefault(lang_code, []).append(t)
             else:
                 subtitles.setdefault(lang_code, []).append(t)
-        if transcript_url := presentation.get('TranscriptUrl'):
+        if transcript_url := url_or_none(presentation.get('TranscriptUrl')):
+            if 'playbackTicket=' not in transcript_url:
+                transcript_url = join_nonempty(
+                    transcript_url, traverse_obj(presentation, ('Streams', 0, 'SlidePlaybackTicketId', {str})),
+                    delim='?playbackTicket=')
             if determine_ext(transcript_url) != 'txt':
                 if len(transcripts) == 1:
-                    (captions or subtitles).setdefault(lang_code, []).append({
+                    (captions or subtitles)[lang_code].insert(0, {
                         'url': transcript_url,
                         'name': lang_name,
                     })
                 else:
-                    subtitles.setdefault('und', []).append({'url': transcript_url})
+                    subtitles.setdefault('und', []).insert(0, {'url': transcript_url})
 
         return {
             'id': resource_id,