X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/af478477605bdf3f5d57562035885cfee905f379..a6abd8dc822fb70852437ab5a77ced5f633739f8:/youtube_dl/extractor/nba.py diff --git a/youtube_dl/extractor/nba.py b/youtube_dl/extractor/nba.py index 7e42161..862b706 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(/[^?]*?)(?:/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', - 'file': u'0021200253-okc-bkn-recap.nba.mp4', - 'md5': u'c0edcfc37607344e2ff8f13c378c88a4', + 'md5': 'c0edcfc37607344e2ff8f13c378c88a4', 'info_dict': { - 'description': 'Kevin Durant scores 32 points and dishes out six assists as the Thunder beat the Nets in Brooklyn.', + 'id': '0021200253-okc-bkn-recap.nba', + 'ext': 'mp4', '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(1) - + 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, - 'ext': 'mp4', 'title': title, 'description': description, + 'duration': duration, }