ytdl/youtube_dl/extractor/nba.py

99 lines
3.7 KiB
Python
Raw Normal View History

from __future__ import unicode_literals
import re
2013-06-23 21:18:00 +02:00
from .common import InfoExtractor
2014-09-04 15:06:14 +02:00
from ..utils import (
parse_duration,
int_or_none,
2015-12-04 07:09:48 +01:00
xpath_text,
xpath_attr,
2014-09-04 15:06:14 +02:00
)
2013-06-23 21:18:00 +02:00
2015-10-03 13:30:05 +02:00
class NBAIE(InfoExtractor):
2015-10-07 13:17:32 +02:00
_VALID_URL = r'https?://(?:watch\.|www\.)?nba\.com/(?P<path>(?:[^/]+/)?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',
2015-10-07 13:17:32 +02:00
'md5': '9e7729d3010a9c71506fd1248f74e4f4',
'info_dict': {
2015-10-03 13:30:05 +02:00
'id': '0021200253-okc-bkn-recap',
2015-10-07 13:17:32 +02:00
'ext': 'flv',
'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,
2015-10-03 13:30:05 +02:00
'timestamp': 1354638466,
'upload_date': '20121204',
},
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,
2015-12-04 07:09:48 +01:00
}, {
2015-10-07 13:17:32 +02:00
'url': 'http://watch.nba.com/video/channels/playoffs/2015/05/20/0041400301-cle-atl-recap.nba',
'md5': 'b2b39b81cf28615ae0c3360a3f9668c4',
'info_dict': {
2015-10-03 13:30:05 +02:00
'id': '0041400301-cle-atl-recap',
'ext': 'mp4',
'title': 'Hawks vs. Cavaliers Game 1',
'description': 'md5:8094c3498d35a9bd6b1a8c396a071b4d',
'duration': 228,
2015-10-03 13:30:05 +02:00
'timestamp': 1432134543,
'upload_date': '20150520',
}
2014-12-06 11:26:17 +01:00
}]
2013-06-23 21:18:00 +02:00
2015-10-03 13:30:05 +02:00
def _real_extract(self, url):
2015-10-07 13:17:32 +02:00
path, video_id = re.match(self._VALID_URL, url).groups()
2015-12-04 16:04:22 +01:00
if path.startswith('nba/'):
path = path[3:]
2015-10-07 13:17:32 +02:00
video_info = self._download_xml('http://www.nba.com/%s.xml' % path, video_id)
2015-12-04 07:09:48 +01:00
video_id = xpath_text(video_info, 'slug')
title = xpath_text(video_info, 'headline')
description = xpath_text(video_info, 'description')
duration = parse_duration(xpath_text(video_info, 'length'))
timestamp = int_or_none(xpath_attr(video_info, 'dateCreated', 'uts'))
2015-10-03 13:30:05 +02:00
thumbnails = []
for image in video_info.find('images'):
thumbnails.append({
'id': image.attrib.get('cut'),
'url': image.text,
'width': int_or_none(image.attrib.get('width')),
'height': int_or_none(image.attrib.get('height')),
})
formats = []
2015-12-04 16:04:22 +01:00
for video_file in video_info.findall('.//file'):
2015-10-03 13:30:05 +02:00
video_url = video_file.text
2015-10-07 07:53:19 +02:00
if video_url.startswith('/'):
continue
2015-10-03 13:30:05 +02:00
if video_url.endswith('.m3u8'):
formats.extend(self._extract_m3u8_formats(video_url, video_id, m3u8_id='hls', fatal=False))
2015-10-03 13:30:05 +02:00
elif video_url.endswith('.f4m'):
formats.extend(self._extract_f4m_formats(video_url + '?hdcore=3.4.1.1', video_id, f4m_id='hds', fatal=False))
2015-10-03 13:30:05 +02:00
else:
key = video_file.attrib.get('bitrate')
2015-12-05 18:47:15 +01:00
format_info = {
2015-10-03 13:30:05 +02:00
'format_id': key,
'url': video_url,
2015-12-05 18:47:15 +01:00
}
mobj = re.search(r'(\d+)x(\d+)(?:_(\d+))?', key)
if mobj:
format_info.update({
2015-12-05 20:58:44 +01:00
'width': int(mobj.group(1)),
'height': int(mobj.group(2)),
2015-12-05 21:04:22 +01:00
'tbr': int_or_none(mobj.group(3)),
2015-12-05 18:47:15 +01:00
})
formats.append(format_info)
2015-10-03 13:30:05 +02:00
self._sort_formats(formats)
return {
2015-10-03 13:30:05 +02:00
'id': video_id,
'title': title,
'description': description,
'duration': duration,
'timestamp': timestamp,
'thumbnails': thumbnails,
'formats': formats,
2013-06-23 21:18:00 +02:00
}