X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fnba.py;h=862b706bf96719aa071f1f89c73f2a4ef45a20b1;hb=29aef5a33c46ceaa5884a3f074fbb5af4fa44f32;hp=633b42f728489c6e9f9c61a98b8b0b4d38e57be1;hpb=2ff7f8975e98e0e09513b6ba042da23348bacc41;p=youtube-dl diff --git a/youtube_dl/extractor/nba.py b/youtube_dl/extractor/nba.py index 633b42f72..862b706bf 100644 --- a/youtube_dl/extractor/nba.py +++ b/youtube_dl/extractor/nba.py @@ -1,39 +1,47 @@ from __future__ import unicode_literals -import re - from .common import InfoExtractor +from ..utils import ( + remove_end, + parse_duration, +) class NBAIE(InfoExtractor): - _VALID_URL = r'https?://(?:watch\.|www\.)?nba\.com/(?:nba/)?video(?P/[^?]*?)(?:/index\.html)?(?:\?.*)?$' - _TEST = { + _VALID_URL = r'https?://(?:watch\.|www\.)?nba\.com/(?:nba/)?video(?P/[^?]*?)/?(?:/index\.html)?(?:\?.*)?$' + _TESTS = [{ 'url': 'http://www.nba.com/video/games/nets/2012/12/04/0021200253-okc-bkn-recap.nba/index.html', - 'md5': u'c0edcfc37607344e2ff8f13c378c88a4', + 'md5': 'c0edcfc37607344e2ff8f13c378c88a4', 'info_dict': { 'id': '0021200253-okc-bkn-recap.nba', 'ext': 'mp4', - 'description': 'Kevin Durant scores 32 points and dishes out six assists as the Thunder beat the Nets in Brooklyn.', 'title': 'Thunder vs. Nets', + 'description': 'Kevin Durant scores 32 points and dishes out six assists as the Thunder beat the Nets in Brooklyn.', + 'duration': 181, }, - } + }, { + 'url': 'http://www.nba.com/video/games/hornets/2014/12/05/0021400276-nyk-cha-play5.nba/', + 'only_matching': True, + }] def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - video_id = mobj.group('id') - + video_id = self._match_id(url) webpage = self._download_webpage(url, video_id) video_url = 'http://ht-mobile.cdn.turner.com/nba/big' + video_id + '_nba_1280x720.mp4' shortened_video_id = video_id.rpartition('/')[2] - title = self._og_search_title(webpage, default=shortened_video_id).replace('NBA.com: ', '') + title = remove_end( + self._og_search_title(webpage, default=shortened_video_id), ' : NBA.com') - description = self._html_search_regex(r'', webpage, 'description', fatal=False) + description = self._og_search_description(webpage) + duration = parse_duration( + self._html_search_meta('duration', webpage, 'duration')) return { 'id': shortened_video_id, 'url': video_url, 'title': title, 'description': description, + 'duration': duration, }