X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/15b1d10671b48df598afd70e17ba21e9e64ac766..76d85602f8a22ca3817c3a86f4f0e8969c0b02a9:/youtube_dl/extractor/nba.py diff --git a/youtube_dl/extractor/nba.py b/youtube_dl/extractor/nba.py index f69fe09..944096e 100644 --- a/youtube_dl/extractor/nba.py +++ b/youtube_dl/extractor/nba.py @@ -1,7 +1,5 @@ from __future__ import unicode_literals -import re - from .common import InfoExtractor from ..utils import ( remove_end, @@ -10,8 +8,8 @@ from ..utils import ( 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': 'c0edcfc37607344e2ff8f13c378c88a4', 'info_dict': { @@ -21,12 +19,25 @@ class NBAIE(InfoExtractor): '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, + }, { + 'url': 'http://watch.nba.com/nba/video/channels/playoffs/2015/05/20/0041400301-cle-atl-recap.nba', + 'info_dict': { + 'id': '0041400301-cle-atl-recap.nba', + 'ext': 'mp4', + 'title': 'NBA GAME TIME | Video: Hawks vs. Cavaliers Game 1', + 'description': 'md5:8094c3498d35a9bd6b1a8c396a071b4d', + 'duration': 228, + }, + 'params': { + 'skip_download': 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' @@ -36,8 +47,12 @@ class NBAIE(InfoExtractor): self._og_search_title(webpage, default=shortened_video_id), ' : NBA.com') description = self._og_search_description(webpage) - duration = parse_duration( - self._html_search_meta('duration', webpage, 'duration', fatal=False)) + duration_str = self._html_search_meta( + 'duration', webpage, 'duration', default=None) + if not duration_str: + duration_str = self._html_search_regex( + r'Duration:\s*(\d+:\d+)', webpage, 'duration', fatal=False) + duration = parse_duration(duration_str) return { 'id': shortened_video_id,