yt-dlp/youtube_dl/extractor/nba.py

64 lines
2.3 KiB
Python
Raw Normal View History

from __future__ import unicode_literals
2013-06-23 21:18:00 +02:00
from .common import InfoExtractor
2014-09-04 15:06:14 +02:00
from ..utils import (
remove_end,
parse_duration,
)
2013-06-23 21:18:00 +02:00
class NBAIE(InfoExtractor):
_VALID_URL = r'https?://(?:watch\.|www\.)?nba\.com/(?:nba/)?video(?P<id>/[^?]*?)/?(?:/index\.html)?(?:\?.*)?$'
2014-12-06 11:26:17 +01:00
_TESTS = [{
'url': 'http://www.nba.com/video/games/nets/2012/12/04/0021200253-okc-bkn-recap.nba/index.html',
2014-09-04 15:06:14 +02:00
'md5': 'c0edcfc37607344e2ff8f13c378c88a4',
'info_dict': {
2014-03-29 14:54:45 +01:00
'id': '0021200253-okc-bkn-recap.nba',
'ext': 'mp4',
'title': 'Thunder vs. Nets',
2014-09-04 15:06:14 +02:00
'description': 'Kevin Durant scores 32 points and dishes out six assists as the Thunder beat the Nets in Brooklyn.',
'duration': 181,
},
2014-12-06 11:26:17 +01:00
}, {
'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,
}
2014-12-06 11:26:17 +01:00
}]
2013-06-23 21:18:00 +02:00
def _real_extract(self, url):
2014-12-06 11:15:25 +01:00
video_id = self._match_id(url)
2013-06-23 21:18:00 +02:00
webpage = self._download_webpage(url, video_id)
video_url = 'http://ht-mobile.cdn.turner.com/nba/big' + video_id + '_nba_1280x720.mp4'
2013-06-23 21:18:00 +02:00
shortened_video_id = video_id.rpartition('/')[2]
2014-09-04 15:06:14 +02:00
title = remove_end(
self._og_search_title(webpage, default=shortened_video_id), ' : NBA.com')
description = self._og_search_description(webpage)
duration_str = self._html_search_meta(
'duration', webpage, 'duration', default=None)
if not duration_str:
duration_str = self._html_search_regex(
r'Duration:</b>\s*(\d+:\d+)', webpage, 'duration', fatal=False)
duration = parse_duration(duration_str)
2013-06-23 21:18:00 +02:00
return {
2013-06-23 21:18:00 +02:00
'id': shortened_video_id,
'url': video_url,
'title': title,
'description': description,
2014-09-04 15:06:14 +02:00
'duration': duration,
2013-06-23 21:18:00 +02:00
}