]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/extractor/espn.py
Tidy up cleanup rule and keep lintian happy.
[youtubedl] / youtube_dl / extractor / espn.py
1 from __future__ import unicode_literals
2
3 from .common import InfoExtractor
4 from ..utils import remove_end
5
6
7 class ESPNIE(InfoExtractor):
8 _VALID_URL = r'https?://espn\.go\.com/(?:[^/]+/)*(?P<id>[^/]+)'
9 _TESTS = [{
10 'url': 'http://espn.go.com/video/clip?id=10365079',
11 'md5': '60e5d097a523e767d06479335d1bdc58',
12 'info_dict': {
13 'id': 'FkYWtmazr6Ed8xmvILvKLWjd4QvYZpzG',
14 'ext': 'mp4',
15 'title': '30 for 30 Shorts: Judging Jewell',
16 'description': None,
17 },
18 'params': {
19 'skip_download': True,
20 },
21 'add_ie': ['OoyalaExternal'],
22 }, {
23 # intl video, from http://www.espnfc.us/video/mls-highlights/150/video/2743663/must-see-moments-best-of-the-mls-season
24 'url': 'http://espn.go.com/video/clip?id=2743663',
25 'md5': 'f4ac89b59afc7e2d7dbb049523df6768',
26 'info_dict': {
27 'id': '50NDFkeTqRHB0nXBOK-RGdSG5YQPuxHg',
28 'ext': 'mp4',
29 'title': 'Must-See Moments: Best of the MLS season',
30 },
31 'params': {
32 'skip_download': True,
33 },
34 'add_ie': ['OoyalaExternal'],
35 }, {
36 'url': 'https://espn.go.com/video/iframe/twitter/?cms=espn&id=10365079',
37 'only_matching': True,
38 }, {
39 'url': 'http://espn.go.com/nba/recap?gameId=400793786',
40 'only_matching': True,
41 }, {
42 'url': 'http://espn.go.com/blog/golden-state-warriors/post/_/id/593/how-warriors-rapidly-regained-a-winning-edge',
43 'only_matching': True,
44 }, {
45 'url': 'http://espn.go.com/sports/endurance/story/_/id/12893522/dzhokhar-tsarnaev-sentenced-role-boston-marathon-bombings',
46 'only_matching': True,
47 }, {
48 'url': 'http://espn.go.com/nba/playoffs/2015/story/_/id/12887571/john-wall-washington-wizards-no-swelling-left-hand-wrist-game-5-return',
49 'only_matching': True,
50 }]
51
52 def _real_extract(self, url):
53 video_id = self._match_id(url)
54
55 webpage = self._download_webpage(url, video_id)
56
57 video_id = self._search_regex(
58 r'class=(["\']).*?video-play-button.*?\1[^>]+data-id=["\'](?P<id>\d+)',
59 webpage, 'video id', group='id')
60
61 cms = 'espn'
62 if 'data-source="intl"' in webpage:
63 cms = 'intl'
64 player_url = 'https://espn.go.com/video/iframe/twitter/?id=%s&cms=%s' % (video_id, cms)
65 player = self._download_webpage(
66 player_url, video_id)
67
68 pcode = self._search_regex(
69 r'["\']pcode=([^"\']+)["\']', player, 'pcode')
70
71 title = remove_end(
72 self._og_search_title(webpage),
73 '- ESPN Video').strip()
74
75 return {
76 '_type': 'url_transparent',
77 'url': 'ooyalaexternal:%s:%s:%s' % (cms, video_id, pcode),
78 'ie_key': 'OoyalaExternal',
79 'title': title,
80 }