]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/porncom.py
3 from .common
import InfoExtractor
4 from ..compat
import compat_urlparse
13 class PornComIE(InfoExtractor
):
14 _VALID_URL
= r
'https?://(?:[a-zA-Z]+\.)?porn\.com/videos/(?:(?P<display_id>[^/]+)-)?(?P<id>\d+)'
16 'url': 'http://www.porn.com/videos/teen-grabs-a-dildo-and-fucks-her-pussy-live-on-1hottie-i-rec-2603339',
17 'md5': '3f30ce76267533cd12ba999263156de7',
20 'display_id': 'teen-grabs-a-dildo-and-fucks-her-pussy-live-on-1hottie-i-rec',
22 'title': 'Teen grabs a dildo and fucks her pussy live on 1hottie, I rec',
23 'thumbnail': r
're:^https?://.*\.jpg$',
31 'url': 'http://se.porn.com/videos/marsha-may-rides-seth-on-top-of-his-thick-cock-2658067',
32 'only_matching': True,
35 def _real_extract(self
, url
):
36 mobj
= self
._match
_valid
_url
(url
)
37 video_id
= mobj
.group('id')
38 display_id
= mobj
.group('display_id') or video_id
40 webpage
= self
._download
_webpage
(url
, display_id
)
42 config
= self
._parse
_json
(
44 (r
'=\s*({.+?})\s*;\s*v1ar\b',
45 r
'=\s*({.+?})\s*,\s*[\da-zA-Z_]+\s*='),
46 webpage
, 'config', default
='{}'),
47 display_id
, transform_source
=js_to_json
, fatal
=False)
50 title
= config
['title']
53 'format_id': stream
.get('id'),
54 'height': int_or_none(self
._search
_regex
(
55 r
'^(\d+)[pP]', stream
.get('id') or '', 'height', default
=None))
56 } for stream
in config
['streams'] if stream
.get('url')]
57 thumbnail
= (compat_urlparse
.urljoin(
58 config
['thumbCDN'], config
['poster'])
59 if config
.get('thumbCDN') and config
.get('poster') else None)
60 duration
= int_or_none(config
.get('length'))
62 title
= self
._search
_regex
(
63 (r
'<title>([^<]+)</title>', r
'<h1[^>]*>([^<]+)</h1>'),
66 'url': compat_urlparse
.urljoin(url
, format_url
),
67 'format_id': '%sp' % height
,
68 'height': int(height
),
69 'filesize_approx': parse_filesize(filesize
),
70 } for format_url
, height
, filesize
in re
.findall(
71 r
'<a[^>]+href="(/download/[^"]+)">[^<]*?(\d+)p<span[^>]*>(\d+\s*[a-zA-Z]+)<',
76 view_count
= str_to_int(self
._search
_regex
(
77 (r
'Views:\s*</span>\s*<span>\s*([\d,.]+)',
78 r
'class=["\']views
["\'][^>]*><p>([\d,.]+)'), webpage,
79 'view count', fatal=False))
81 def extract_list(kind):
82 s = self._search_regex(
83 (r'(?s)%s:\s*</span>\s*<span>(.+?)</span>' % kind.capitalize(),
84 r'(?s)<p[^>]*>%s:(.+?)</p>' % kind.capitalize()),
85 webpage, kind, fatal=False)
86 return re.findall(r'<a[^>]+>([^<]+)</a>', s or '')
90 'display_id': display_id,
92 'thumbnail': thumbnail,
94 'view_count': view_count,
97 'categories': extract_list('categories'),
98 'tags': extract_list('tags'),