Merge pull request #8898 from dstftw/fragment-retries

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

diff --git a/youtube_dl/extractor/once.py b/youtube_dl/extractor/once.py
index 403f8c0afcff6e3c0532508dbd060788f400d795..5db949b176141863a290adfcd1fe3bbda11076b7 100644 (file)
--- a/youtube_dl/extractor/once.py
+++ b/youtube_dl/extractor/once.py
@@ -14,13 +14,16 @@ class OnceIE(InfoExtractor):
     def _extract_once_formats(self, url):
         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/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)
@@ -33,5 +36,7 @@ class OnceIE(InfoExtractor):
                         'hls', 'http'),
                     'protocol': 'http',
                 })
-                formats.append(progressive_format)
+                progressive_formats.append(progressive_format)
+        self._check_formats(progressive_formats, media_item_id)
+        formats.extend(progressive_formats)
         return formats