1 from __future__ import unicode_literals
7 from .common import InfoExtractor
10 compat_urllib_parse_unquote,
11 compat_urllib_parse_unquote_plus,
12 compat_urllib_parse_urlparse,
26 class PornHubIE(InfoExtractor):
27 _VALID_URL = r'https?://(?:[a-z]+\.)?pornhub\.com/(?:view_video\.php\?viewkey=|embed/)(?P<id>[0-9a-z]+)'
29 'url': 'http://www.pornhub.com/view_video.php?viewkey=648719015',
30 'md5': '1e19b41231a02eba417839222ac9d58e',
34 'title': 'Seductive Indian beauty strips down and fingers her pink pussy',
44 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph557bbb6676d2d',
45 'only_matching': True,
47 'url': 'http://fr.pornhub.com/view_video.php?viewkey=ph55ca2f9760862',
48 'only_matching': True,
52 def _extract_url(cls, webpage):
54 r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?pornhub\.com/embed/\d+)\1', webpage)
56 return mobj.group('url')
58 def _extract_count(self, pattern, webpage, name):
59 return str_to_int(self._search_regex(
60 pattern, webpage, '%s count' % name, fatal=False))
62 def _real_extract(self, url):
63 video_id = self._match_id(url)
65 req = sanitized_Request(
66 'http://www.pornhub.com/view_video.php?viewkey=%s' % video_id)
67 req.add_header('Cookie', 'age_verified=1')
68 webpage = self._download_webpage(req, video_id)
70 error_msg = self._html_search_regex(
71 r'(?s)<div class="userMessageSection[^"]*".*?>(.*?)</div>',
72 webpage, 'error message', default=None)
74 error_msg = re.sub(r'\s+', ' ', error_msg)
76 'PornHub said: %s' % error_msg,
77 expected=True, video_id=video_id)
79 flashvars = self._parse_json(
81 r'var\s+flashvars_\d+\s*=\s*({.+?});', webpage, 'flashvars', default='{}'),
84 video_title = flashvars.get('video_title')
85 thumbnail = flashvars.get('image_url')
86 duration = int_or_none(flashvars.get('video_duration'))
88 video_title, thumbnail, duration = [None] * 3
91 video_title = self._html_search_regex(r'<h1 [^>]+>([^<]+)', webpage, 'title')
93 video_uploader = self._html_search_regex(
94 r'(?s)From: .+?<(?:a href="/users/|a href="/channels/|span class="username)[^>]+>(.+?)<',
95 webpage, 'uploader', fatal=False)
97 view_count = self._extract_count(
98 r'<span class="count">([\d,\.]+)</span> views', webpage, 'view')
99 like_count = self._extract_count(
100 r'<span class="votesUp">([\d,\.]+)</span>', webpage, 'like')
101 dislike_count = self._extract_count(
102 r'<span class="votesDown">([\d,\.]+)</span>', webpage, 'dislike')
103 comment_count = self._extract_count(
104 r'All Comments\s*<span>\(([\d,.]+)\)', webpage, 'comment')
106 video_urls = list(map(compat_urllib_parse_unquote, re.findall(r"player_quality_[0-9]{3}p\s*=\s*'([^']+)'", webpage)))
107 if webpage.find('"encrypted":true') != -1:
108 password = compat_urllib_parse_unquote_plus(
109 self._search_regex(r'"video_title":"([^"]+)', webpage, 'password'))
110 video_urls = list(map(lambda s: aes_decrypt_text(s, password, 32).decode('utf-8'), video_urls))
113 for video_url in video_urls:
114 path = compat_urllib_parse_urlparse(video_url).path
115 extension = os.path.splitext(path)[1][1:]
116 format = path.split('/')[5].split('_')[:2]
117 format = '-'.join(format)
119 m = re.match(r'^(?P<height>[0-9]+)[pP]-(?P<tbr>[0-9]+)[kK]$', format)
124 height = int(m.group('height'))
125 tbr = int(m.group('tbr'))
135 self._sort_formats(formats)
139 'uploader': video_uploader,
140 'title': video_title,
141 'thumbnail': thumbnail,
142 'duration': duration,
143 'view_count': view_count,
144 'like_count': like_count,
145 'dislike_count': dislike_count,
146 'comment_count': comment_count,
152 class PornHubPlaylistBaseIE(InfoExtractor):
153 def _extract_entries(self, webpage):
156 'http://www.pornhub.com/%s' % video_url,
157 PornHubIE.ie_key(), video_title=title)
158 for video_url, title in orderedSet(re.findall(
159 r'href="/?(view_video\.php\?.*\bviewkey=[\da-z]+[^"]*)"[^>]*\s+title="([^"]+)"',
163 def _real_extract(self, url):
164 playlist_id = self._match_id(url)
166 webpage = self._download_webpage(url, playlist_id)
168 entries = self._extract_entries(webpage)
170 playlist = self._parse_json(
172 r'playlistObject\s*=\s*({.+?});', webpage, 'playlist'),
175 return self.playlist_result(
176 entries, playlist_id, playlist.get('title'), playlist.get('description'))
179 class PornHubPlaylistIE(PornHubPlaylistBaseIE):
180 _VALID_URL = r'https?://(?:www\.)?pornhub\.com/playlist/(?P<id>\d+)'
182 'url': 'http://www.pornhub.com/playlist/6201671',
187 'playlist_mincount': 35,
191 class PornHubUserVideosIE(PornHubPlaylistBaseIE):
192 _VALID_URL = r'https?://(?:www\.)?pornhub\.com/users/(?P<id>[^/]+)/videos'
194 'url': 'http://www.pornhub.com/users/zoe_ph/videos/public',
198 'playlist_mincount': 171,
200 'url': 'http://www.pornhub.com/users/rushandlia/videos',
201 'only_matching': True,
204 def _real_extract(self, url):
205 user_id = self._match_id(url)
208 for page_num in itertools.count(1):
210 webpage = self._download_webpage(
211 url, user_id, 'Downloading page %d' % page_num,
212 query={'page': page_num})
213 except ExtractorError as e:
214 if isinstance(e.cause, compat_HTTPError) and e.cause.code == 404:
216 page_entries = self._extract_entries(webpage)
219 entries.extend(page_entries)
221 return self.playlist_result(entries, user_id)