diff options
| author | Sergey M․ <dstftw@gmail.com> | 2016-11-02 23:34:37 +0700 |
|---|---|---|
| committer | Sergey M․ <dstftw@gmail.com> | 2016-11-02 23:34:37 +0700 |
| commit | 26aae566902251f9674593a2b0f0ca7477b96a56 (patch) | |
| tree | d309bfe791b1e013db35c8854018f06a1e077a9c /youtube_dl | |
| parent | 4f9cd4d36fa88758cdff822f03879c6e0b6aa42d (diff) | |
| download | youtube-dl-26aae566902251f9674593a2b0f0ca7477b96a56.tar.xz | |
[extractor/generic] Improve ISM extraction
Diffstat (limited to 'youtube_dl')
| -rw-r--r-- | youtube_dl/extractor/generic.py | 17 |
1 files changed, 15 insertions, 2 deletions
diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index fc3d01eed..0bb263ce7 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -2453,8 +2453,21 @@ class GenericIE(InfoExtractor): entry_info_dict['formats'] = self._extract_mpd_formats(video_url, video_id) elif ext == 'f4m': entry_info_dict['formats'] = self._extract_f4m_formats(video_url, video_id) - elif re.search(r'(?i)\.ism/manifest', video_url): - entry_info_dict['formats'] = self._extract_ism_formats(video_url, video_id) + elif re.search(r'(?i)\.(?:ism|smil)/manifest', video_url): + # Just matching .ism/manifest is not enough to be reliably sure + # whether it's actually an ISM manifest or some other streaming + # manifest since there are various streaming URL formats + # possible (see [1]) as well as some other shenanigans like + # .smil/manifest URLs that actually serve an ISM (see [2]) and + # so on. + # Thus the most reasonable way to solve this is to delegate + # to generic extractor in order to look into the contents of + # the manifest itself. + # 1. https://azure.microsoft.com/en-us/documentation/articles/media-services-deliver-content-overview/#streaming-url-formats + # 2. https://svs.itworkscdn.net/lbcivod/smil:itwfcdn/lbci/170976.smil/Manifest + entry_info_dict = self.url_result( + smuggle_url(video_url, {'to_generic': True}), + GenericIE.ie_key()) else: entry_info_dict['url'] = video_url |