Start moving to ytdl-org

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

diff --git a/youtube_dl/extractor/once.py b/youtube_dl/extractor/once.py
index 403f8c0afcff6e3c0532508dbd060788f400d795..3e44b78290156e5b068fdc5e72b596dfcb61bedf 100644 (file)
--- a/youtube_dl/extractor/once.py
+++ b/youtube_dl/extractor/once.py
@@ -7,20 +7,23 @@ 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/(?:ads/vmap/)?[^/]+/[^/]+/(?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'
 
-    def _extract_once_formats(self, url):
+    def _extract_once_formats(self, url, http_formats_preference=None):
         domain_id, application_id, media_item_id = re.match(
             OnceIE._VALID_URL, url).groups()
-        adaptive_formats = self._extract_m3u8_formats(
+        formats = self._extract_m3u8_formats(
             self.ADAPTIVE_URL_TEMPLATE % (
                 domain_id, application_id, media_item_id),
             media_item_id, 'mp4', m3u8_id='hls', fatal=False)
-        formats = []
-        formats.extend(adaptive_formats)
-        for adaptive_format in adaptive_formats:
+        progressive_formats = []
+        for adaptive_format in formats:
+            # Prevent advertisement from embedding into m3u8 playlist (see
+            # https://github.com/ytdl-org/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)
@@ -32,6 +35,9 @@ class OnceIE(InfoExtractor):
                     'format_id': adaptive_format['format_id'].replace(
                         'hls', 'http'),
                     'protocol': 'http',
+                    'preference': http_formats_preference,
                 })
-                formats.append(progressive_format)
+                progressive_formats.append(progressive_format)
+        self._check_formats(progressive_formats, media_item_id)
+        formats.extend(progressive_formats)
         return formats