[youtube] Fix 141 format tests
[youtube-dl] / youtube_dl / extractor / once.py
index 080045d4c1446875cb40fdf85c516e1e4cc419a3..1bf96ea56e29849725db4b9c799ec63330c34224 100644 (file)
@@ -7,7 +7,7 @@ from .common import InfoExtractor
 
 
 class OnceIE(InfoExtractor):
-    _VALID_URL = r'https?://once\.unicornmedia\.com/now/[^/]+/[^/]+/(?P<domain_id>[^/]+)/(?P<application_id>[^/]+)/(?:[^/]+/)?(?P<media_item_id>[^/]+)/content\.(?:once|m3u8|mp4)'
+    _VALID_URL = r'https?://.+?\.unicornmedia\.com/now/[^/]+/[^/]+/(?P<domain_id>[^/]+)/(?P<application_id>[^/]+)/(?:[^/]+/)?(?P<media_item_id>[^/]+)/content\.(?:once|m3u8|mp4)'
     ADAPTIVE_URL_TEMPLATE = 'http://once.unicornmedia.com/now/master/playlist/%s/%s/%s/content.m3u8'
     PROGRESSIVE_URL_TEMPLATE = 'http://once.unicornmedia.com/now/media/progressive/%s/%s/%s/%s/content.mp4'
 
@@ -20,6 +20,10 @@ class OnceIE(InfoExtractor):
             media_item_id, 'mp4', m3u8_id='hls', fatal=False)
         progressive_formats = []
         for adaptive_format in formats:
+            # Prevent advertisement from embedding into m3u8 playlist (see
+            # https://github.com/rg3/youtube-dl/issues/8893#issuecomment-199912684)
+            adaptive_format['url'] = re.sub(
+                r'\badsegmentlength=\d+', r'adsegmentlength=0', adaptive_format['url'])
             rendition_id = self._search_regex(
                 r'/now/media/playlist/[^/]+/[^/]+/([^/]+)',
                 adaptive_format['url'], 'redition id', default=None)