1 from __future__
import unicode_literals
5 from .common
import InfoExtractor
13 from ..aes
import aes_decrypt_text
16 class YouPornIE(InfoExtractor
):
17 _VALID_URL
= r
'https?://(?:www\.)?youporn\.com/(?:watch|embed)/(?P<id>\d+)(?:/(?P<display_id>[^/?#&]+))?'
19 'url': 'http://www.youporn.com/watch/505835/sex-ed-is-it-safe-to-masturbate-daily/',
20 'md5': '3744d24c50438cf5b6f6d59feb5055c2',
23 'display_id': 'sex-ed-is-it-safe-to-masturbate-daily',
25 'title': 'Sex Ed: Is It Safe To Masturbate Daily?',
26 'description': 'Love & Sex Answers: http://bit.ly/DanAndJenn -- Is It Unhealthy To Masturbate Daily?',
27 'thumbnail': r
're:^https?://.*\.jpg$',
28 'uploader': 'Ask Dan And Jennifer',
29 'upload_date': '20101217',
30 'average_rating': int,
38 'url': 'http://www.youporn.com/watch/561726/big-tits-awesome-brunette-on-amazing-webcam-show/?from=related3&al=2&from_id=561726&pos=4',
41 'display_id': 'big-tits-awesome-brunette-on-amazing-webcam-show',
43 'title': 'Big Tits Awesome Brunette On amazing webcam show',
44 'description': 'http://sweetlivegirls.com Big Tits Awesome Brunette On amazing webcam show.mp4',
45 'thumbnail': r
're:^https?://.*\.jpg$',
46 'uploader': 'Unknown',
47 'upload_date': '20110418',
48 'average_rating': int,
55 'skip_download': True,
58 'url': 'https://www.youporn.com/embed/505835/sex-ed-is-it-safe-to-masturbate-daily/',
59 'only_matching': True,
61 'url': 'http://www.youporn.com/watch/505835',
62 'only_matching': True,
66 def _extract_urls(webpage
):
68 r
'<iframe[^>]+\bsrc=["\']((?
:https?
:)?
//(?
:www\
.)?youporn\
.com
/embed
/\d
+)',
71 def _real_extract(self, url):
72 mobj = re.match(self._VALID_URL, url)
73 video_id = mobj.group('id')
74 display_id = mobj.group('display_id
') or video_id
76 webpage = self._download_webpage(
77 'http
://www
.youporn
.com
/watch
/%s' % video_id, display_id,
78 headers={'Cookie': 'age_verified=1'})
80 title = self._html_search_regex(
81 r'(?s
)<div
[^
>]+class=["\']watchVideoTitle[^>]+>(.+?)</div>',
82 webpage, 'title', default=None) or self._og_search_title(
83 webpage, default=None) or self._html_search_meta(
84 'title', webpage, fatal=True)
89 definitions = self._parse_json(
91 r'mediaDefinition\s*=\s*(\[.+?\]);', webpage,
92 'media definitions', default='[]'),
93 video_id, fatal=False)
95 for definition in definitions:
96 if not isinstance(definition, dict):
98 video_url = url_or_none(definition.get('videoUrl'))
100 links.append(video_url)
102 # Fallback #1, this also contains extra low quality 180p format
103 for _, link in re.findall(r'<a[^>]+href=(["\'])(http
.+?
)\
1[^
>]+title
=["\']Download [Vv]ideo', webpage):
106 # Fallback #2 (unavailable as at 22.06.2017)
107 sources = self._search_regex(
108 r'(?s)sources\s*:\s*({.+?})', webpage, 'sources', default=None)
110 for _, link in re.findall(r'[^:]+\s*:\s*(["\'])(http
.+?
)\
1', sources):
113 # Fallback #3 (unavailable as at 22.06.2017)
114 for _, link in re.findall(
115 r'(?
:videoSrc|videoIpadUrl|html5PlayerSrc
)\s
*[:=]\s
*(["\'])(http.+?)\1', webpage):
118 # Fallback #4, encrypted links (unavailable as at 22.06.2017)
119 for _, encrypted_link in re.findall(
120 r'encryptedQuality\d{3,4}URL\s*=\s*(["\'])([\da
-zA
-Z
+/=]+)\
1', webpage):
121 links.append(aes_decrypt_text(encrypted_link, title, 32).decode('utf
-8'))
124 for video_url in set(unescapeHTML(link) for link in links):
128 # Video URL's path looks like this
:
129 # /201012/17/505835/720p_1500k_505835/YouPorn%20-%20Sex%20Ed%20Is%20It%20Safe%20To%20Masturbate%20Daily.mp4
130 # /201012/17/505835/vl_240p_240k_505835/YouPorn%20-%20Sex%20Ed%20Is%20It%20Safe%20To%20Masturbate%20Daily.mp4
131 # We will benefit from it by extracting some metadata
132 mobj
= re
.search(r
'(?P<height>\d{3,4})[pP]_(?P<bitrate>\d+)[kK]_\d+/', video_url
)
134 height
= int(mobj
.group('height'))
135 bitrate
= int(mobj
.group('bitrate'))
137 'format_id': '%dp-%dk' % (height
, bitrate
),
142 self
._sort
_formats
(formats
)
144 description
= self
._html
_search
_regex
(
145 r
'(?s)<div[^>]+\bid=["\']description
["\'][^>]*>(.+?)</div>',
146 webpage, 'description',
147 default=None) or self._og_search_description(
148 webpage, default=None)
149 thumbnail = self._search_regex(
150 r'(?:imageurl\s*=|poster\s*:)\s*(["\'])(?P
<thumbnail
>.+?
)\
1',
151 webpage, 'thumbnail
', fatal=False, group='thumbnail
')
153 uploader = self._html_search_regex(
154 r'(?s
)<div
[^
>]+class=["\']submitByLink["\'][^
>]*>(.+?
)</div
>',
155 webpage, 'uploader
', fatal=False)
156 upload_date = unified_strdate(self._html_search_regex(
157 [r'UPLOADED
:\s
*<span
>([^
<]+)',
158 r'Date\s
+[Aa
]dded
:\s
*<span
>([^
<]+)',
159 r'(?s
)<div
[^
>]+class=["\']videoInfo(?:Date|Time)["\'][^
>]*>(.+?
)</div
>'],
160 webpage, 'upload date
', fatal=False))
162 age_limit = self._rta_search(webpage)
164 average_rating = int_or_none(self._search_regex(
165 r'<div
[^
>]+class=["\']videoRatingPercentage["\'][^
>]*>(\d
+)%</div
>',
166 webpage, 'average rating
', fatal=False))
168 view_count = str_to_int(self._search_regex(
169 r'(?s
)<div
[^
>]+class=(["\']).*?\bvideoInfoViews\b.*?\1[^>]*>.*?(?P<count>[\d,.]+)<',
170 webpage, 'view count', fatal=False, group='count'))
171 comment_count = str_to_int(self._search_regex(
172 r'>All [Cc]omments? \(([\d,.]+)\)',
173 webpage, 'comment count', default=None))
175 def extract_tag_box(regex, title):
176 tag_box = self._search_regex(regex, webpage, title, default=None)
179 return re.findall(r'<a[^>]+href=[^>]+>([^<]+)', tag_box)
181 categories = extract_tag_box(
182 r'(?s)Categories:.*?</[^>]+>(.+?)</div>', 'categories')
183 tags = extract_tag_box(
184 r'(?s)Tags:.*?</div>\s*<div[^>]+class=["\']tagBoxContent
["\'][^>]*>(.+?)</div>',
189 'display_id': display_id,
191 'description': description,
192 'thumbnail': thumbnail,
193 'uploader': uploader,
194 'upload_date': upload_date,
195 'average_rating': average_rating,
196 'view_count': view_count,
197 'comment_count': comment_count,
198 'categories': categories,
200 'age_limit': age_limit,