]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/breakcom.py
1 from __future__
import unicode_literals
4 from .common
import InfoExtractor
5 from .youtube
import YoutubeIE
12 class BreakIE(InfoExtractor
):
13 _VALID_URL
= r
'https?://(?:www\.)?break\.com/video/(?P<display_id>[^/]+?)(?:-(?P<id>\d+))?(?:[/?#&]|$)'
15 'url': 'http://www.break.com/video/when-girls-act-like-guys-2468056',
19 'title': 'When Girls Act Like D-Bags',
24 'url': 'http://www.break.com/video/someone-forgot-boat-brakes-work',
28 'title': 'Whale Watching Boat Crashing Into San Diego Dock',
29 'description': 'md5:afc1b2772f0a8468be51dd80eb021069',
30 'upload_date': '20160331',
31 'uploader': 'Steve Holden',
32 'uploader_id': 'sdholden07',
35 'skip_download': True,
38 'url': 'http://www.break.com/video/ugc/baby-flex-2773063',
39 'only_matching': True,
42 def _real_extract(self
, url
):
43 display_id
, video_id
= self
._match
_valid
_url
(url
).groups()
45 webpage
= self
._download
_webpage
(url
, display_id
)
47 youtube_url
= YoutubeIE
._extract
_url
(webpage
)
49 return self
.url_result(youtube_url
, ie
=YoutubeIE
.ie_key())
51 content
= self
._parse
_json
(
53 r
'(?s)content["\']\s
*:\s
*(\
[.+?\
])\s
*[,\n]', webpage,
59 video_url = url_or_none(video.get('url
'))
62 bitrate = int_or_none(self._search_regex(
63 r'(\d
+)_kbps
', video_url, 'tbr
', default=None))
66 'format_id
': 'http
-%d' % bitrate if bitrate else 'http
',
69 self._sort_formats(formats)
71 title = self._search_regex(
72 (r'title
["\']\s*:\s*(["\'])(?P
<value
>(?
:(?
!\
1).)+)\
1',
73 r'<h1
[^
>]*>(?P
<value
>[^
<]+)'), webpage, 'title
', group='value
')
76 return int_or_none(self._search_regex(
77 r'%s["\']\s*:\s*["\'](\d
+)' % key, webpage, name,
80 age_limit = get('ratings
', 'age limit
')
81 video_id = video_id or get('pid
', 'video
id') or display_id
85 'display_id
': display_id,
87 'thumbnail
': self._og_search_thumbnail(webpage),
88 'age_limit
': age_limit,