[youtube] Simplify automatic captions URL check (Closes #8287)

[youtube-dl] / youtube_dl / extractor / youtube.py

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 50c1363cd5aa4f0b624240d4278ec77872b3468e..8e8fc14d27b85942c7a0867d125c4a336e878dcf 100644 (file)
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -613,7 +613,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             },
             'params': {
                 'skip_download': 'requires avconv',
-            }
+            },
+            'skip': 'This live event has ended.',
         },
         # Extraction from multiple DASH manifests (https://github.com/rg3/youtube-dl/pull/6097)
         {
@@ -706,6 +707,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         },
         {
             # Title with JS-like syntax "};" (see https://github.com/rg3/youtube-dl/issues/7468)
+            # Also tests cut-off URL expansion in video description (see
+            # https://github.com/rg3/youtube-dl/issues/1892,
+            # https://github.com/rg3/youtube-dl/issues/8164)
             'url': 'https://www.youtube.com/watch?v=lsguqyKfVQg',
             'info_dict': {
                 'id': 'lsguqyKfVQg',
@@ -960,6 +964,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         try:
             args = player_config['args']
             caption_url = args['ttsurl']
+            if not caption_url:
+                self._downloader.report_warning(err_msg)
+                return {}
             timestamp = args['timestamp']
             # We get the available subtitles
             list_params = compat_urllib_parse.urlencode({
@@ -1505,6 +1512,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             for a_format in formats:
                 a_format.setdefault('http_headers', {})['Youtubedl-no-compression'] = 'True'
         else:
+            unavailable_message = self._html_search_regex(
+                r'(?s)<h1[^>]+id="unavailable-message"[^>]*>(.+?)</h1>',
+                video_webpage, 'unavailable message', default=None)
+            if unavailable_message:
+                raise ExtractorError(unavailable_message, expected=True)
             raise ExtractorError('no conn, hlsvp or url_encoded_fmt_stream_map information found in video info')
 
         # Look for the DASH manifest