]>
Commit | Line | Data |
---|---|---|
6c376029 | 1 | # coding: utf-8 |
9933b574 PH |
2 | from __future__ import unicode_literals |
3 | ||
34541395 | 4 | import itertools |
125cfd78 | 5 | import os |
6 | import re | |
7 | ||
8 | from .common import InfoExtractor | |
1cc79574 | 9 | from ..compat import ( |
34541395 | 10 | compat_HTTPError, |
605cbef6 S |
11 | compat_urllib_parse_unquote, |
12 | compat_urllib_parse_unquote_plus, | |
125cfd78 | 13 | compat_urllib_parse_urlparse, |
1cc79574 PH |
14 | ) |
15 | from ..utils import ( | |
50789175 | 16 | ExtractorError, |
ed8648a3 | 17 | int_or_none, |
6bb05b32 | 18 | js_to_json, |
8f9a477e | 19 | orderedSet, |
5c2266df | 20 | sanitized_Request, |
0320ddc1 | 21 | str_to_int, |
125cfd78 | 22 | ) |
23 | from ..aes import ( | |
24 | aes_decrypt_text | |
25 | ) | |
26 | ||
9933b574 | 27 | |
125cfd78 | 28 | class PornHubIE(InfoExtractor): |
bc4b2d75 S |
29 | IE_DESC = 'PornHub and Thumbzilla' |
30 | _VALID_URL = r'''(?x) | |
31 | https?:// | |
32 | (?: | |
33 | (?:[a-z]+\.)?pornhub\.com/(?:view_video\.php\?viewkey=|embed/)| | |
34 | (?:www\.)?thumbzilla\.com/video/ | |
35 | ) | |
b52c9ef1 | 36 | (?P<id>[\da-z]+) |
bc4b2d75 | 37 | ''' |
360075e2 | 38 | _TESTS = [{ |
9933b574 | 39 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=648719015', |
ed8648a3 | 40 | 'md5': '1e19b41231a02eba417839222ac9d58e', |
9933b574 | 41 | 'info_dict': { |
249efaf4 PH |
42 | 'id': '648719015', |
43 | 'ext': 'mp4', | |
611c1dd9 | 44 | 'title': 'Seductive Indian beauty strips down and fingers her pink pussy', |
ed8648a3 S |
45 | 'uploader': 'Babes', |
46 | 'duration': 361, | |
47 | 'view_count': int, | |
48 | 'like_count': int, | |
49 | 'dislike_count': int, | |
50 | 'comment_count': int, | |
51 | 'age_limit': 18, | |
6bb05b32 YCH |
52 | 'tags': list, |
53 | 'categories': list, | |
6c376029 S |
54 | }, |
55 | }, { | |
56 | # non-ASCII title | |
57 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=1331683002', | |
58 | 'info_dict': { | |
59 | 'id': '1331683002', | |
60 | 'ext': 'mp4', | |
61 | 'title': '重庆婷婷女王足交', | |
62 | 'uploader': 'cj397186295', | |
63 | 'duration': 1753, | |
64 | 'view_count': int, | |
65 | 'like_count': int, | |
66 | 'dislike_count': int, | |
67 | 'comment_count': int, | |
68 | 'age_limit': 18, | |
6bb05b32 YCH |
69 | 'tags': list, |
70 | 'categories': list, | |
6c376029 S |
71 | }, |
72 | 'params': { | |
73 | 'skip_download': True, | |
74 | }, | |
360075e2 S |
75 | }, { |
76 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph557bbb6676d2d', | |
77 | 'only_matching': True, | |
272e4db5 | 78 | }, { |
eaaaaec0 | 79 | # removed at the request of cam4.com |
272e4db5 S |
80 | 'url': 'http://fr.pornhub.com/view_video.php?viewkey=ph55ca2f9760862', |
81 | 'only_matching': True, | |
eaaaaec0 S |
82 | }, { |
83 | # removed at the request of the copyright owner | |
84 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=788152859', | |
85 | 'only_matching': True, | |
86 | }, { | |
87 | # removed by uploader | |
88 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph572716d15a111', | |
89 | 'only_matching': True, | |
195f0845 S |
90 | }, { |
91 | # private video | |
92 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph56fd731fce6b7', | |
93 | 'only_matching': True, | |
bc4b2d75 S |
94 | }, { |
95 | 'url': 'https://www.thumbzilla.com/video/ph56c6114abd99a/horny-girlfriend-sex', | |
96 | 'only_matching': True, | |
360075e2 | 97 | }] |
125cfd78 | 98 | |
b52c9ef1 S |
99 | @staticmethod |
100 | def _extract_urls(webpage): | |
101 | return re.findall( | |
102 | r'<iframe[^>]+?src=["\'](?P<url>(?:https?:)?//(?:www\.)?pornhub\.com/embed/[\da-z]+)', | |
103 | webpage) | |
65d161c4 | 104 | |
0320ddc1 | 105 | def _extract_count(self, pattern, webpage, name): |
7700207e S |
106 | return str_to_int(self._search_regex( |
107 | pattern, webpage, '%s count' % name, fatal=False)) | |
0320ddc1 | 108 | |
125cfd78 | 109 | def _real_extract(self, url): |
249efaf4 | 110 | video_id = self._match_id(url) |
125cfd78 | 111 | |
5c2266df | 112 | req = sanitized_Request( |
9fcbd5db | 113 | 'http://www.pornhub.com/view_video.php?viewkey=%s' % video_id) |
125cfd78 | 114 | req.add_header('Cookie', 'age_verified=1') |
115 | webpage = self._download_webpage(req, video_id) | |
116 | ||
50789175 | 117 | error_msg = self._html_search_regex( |
add7d2a0 | 118 | r'(?s)<div[^>]+class=(["\'])(?:(?!\1).)*\b(?:removed|userMessageSection)\b(?:(?!\1).)*\1[^>]*>(?P<error>.+?)</div>', |
3cb3b600 | 119 | webpage, 'error message', default=None, group='error') |
50789175 PH |
120 | if error_msg: |
121 | error_msg = re.sub(r'\s+', ' ', error_msg) | |
122 | raise ExtractorError( | |
123 | 'PornHub said: %s' % error_msg, | |
124 | expected=True, video_id=video_id) | |
125 | ||
6c376029 S |
126 | # video_title from flashvars contains whitespace instead of non-ASCII (see |
127 | # http://www.pornhub.com/view_video.php?viewkey=1331683002), not relying | |
128 | # on that anymore. | |
129 | title = self._html_search_meta( | |
130 | 'twitter:title', webpage, default=None) or self._search_regex( | |
131 | (r'<h1[^>]+class=["\']title["\'][^>]*>(?P<title>[^<]+)', | |
132 | r'<div[^>]+data-video-title=(["\'])(?P<title>.+?)\1', | |
133 | r'shareTitle\s*=\s*(["\'])(?P<title>.+?)\1'), | |
134 | webpage, 'title', group='title') | |
135 | ||
ed8648a3 S |
136 | flashvars = self._parse_json( |
137 | self._search_regex( | |
03442072 | 138 | r'var\s+flashvars_\d+\s*=\s*({.+?});', webpage, 'flashvars', default='{}'), |
ed8648a3 S |
139 | video_id) |
140 | if flashvars: | |
ed8648a3 S |
141 | thumbnail = flashvars.get('image_url') |
142 | duration = int_or_none(flashvars.get('video_duration')) | |
143 | else: | |
6c376029 | 144 | title, thumbnail, duration = [None] * 3 |
ed8648a3 | 145 | |
0320ddc1 | 146 | video_uploader = self._html_search_regex( |
8fc642eb | 147 | r'(?s)From: .+?<(?:a href="/users/|a href="/channels/|span class="username)[^>]+>(.+?)<', |
0320ddc1 | 148 | webpage, 'uploader', fatal=False) |
125cfd78 | 149 | |
7700207e S |
150 | view_count = self._extract_count( |
151 | r'<span class="count">([\d,\.]+)</span> views', webpage, 'view') | |
152 | like_count = self._extract_count( | |
153 | r'<span class="votesUp">([\d,\.]+)</span>', webpage, 'like') | |
154 | dislike_count = self._extract_count( | |
155 | r'<span class="votesDown">([\d,\.]+)</span>', webpage, 'dislike') | |
0320ddc1 | 156 | comment_count = self._extract_count( |
7700207e | 157 | r'All Comments\s*<span>\(([\d,.]+)\)', webpage, 'comment') |
0320ddc1 | 158 | |
f28aeff2 S |
159 | video_urls = [] |
160 | for quote, video_url in re.findall( | |
161 | r'player_quality_[0-9]{3,4}p\s*=\s*(["\'])(.+?)\1;', webpage): | |
162 | video_urls.append(compat_urllib_parse_unquote(re.sub( | |
163 | r'{0}\s*\+\s*{0}'.format(quote), '', video_url))) | |
164 | ||
125cfd78 | 165 | if webpage.find('"encrypted":true') != -1: |
605cbef6 | 166 | password = compat_urllib_parse_unquote_plus( |
7a372b64 | 167 | self._search_regex(r'"video_title":"([^"]+)', webpage, 'password')) |
125cfd78 | 168 | video_urls = list(map(lambda s: aes_decrypt_text(s, password, 32).decode('utf-8'), video_urls)) |
169 | ||
170 | formats = [] | |
171 | for video_url in video_urls: | |
a56f9de1 JMF |
172 | path = compat_urllib_parse_urlparse(video_url).path |
173 | extension = os.path.splitext(path)[1][1:] | |
125cfd78 | 174 | format = path.split('/')[5].split('_')[:2] |
611c1dd9 | 175 | format = '-'.join(format) |
9933b574 | 176 | |
8f5639af | 177 | m = re.match(r'^(?P<height>[0-9]+)[pP]-(?P<tbr>[0-9]+)[kK]$', format) |
9933b574 PH |
178 | if m is None: |
179 | height = None | |
180 | tbr = None | |
181 | else: | |
182 | height = int(m.group('height')) | |
183 | tbr = int(m.group('tbr')) | |
184 | ||
125cfd78 | 185 | formats.append({ |
186 | 'url': video_url, | |
187 | 'ext': extension, | |
188 | 'format': format, | |
189 | 'format_id': format, | |
9933b574 PH |
190 | 'tbr': tbr, |
191 | 'height': height, | |
125cfd78 | 192 | }) |
9933b574 | 193 | self._sort_formats(formats) |
125cfd78 | 194 | |
6bb05b32 YCH |
195 | page_params = self._parse_json(self._search_regex( |
196 | r'page_params\.zoneDetails\[([\'"])[^\'"]+\1\]\s*=\s*(?P<data>{[^}]+})', | |
197 | webpage, 'page parameters', group='data', default='{}'), | |
198 | video_id, transform_source=js_to_json, fatal=False) | |
199 | tags = categories = None | |
200 | if page_params: | |
201 | tags = page_params.get('tags', '').split(',') | |
202 | categories = page_params.get('categories', '').split(',') | |
203 | ||
125cfd78 | 204 | return { |
205 | 'id': video_id, | |
206 | 'uploader': video_uploader, | |
6c376029 | 207 | 'title': title, |
125cfd78 | 208 | 'thumbnail': thumbnail, |
ed8648a3 | 209 | 'duration': duration, |
0320ddc1 S |
210 | 'view_count': view_count, |
211 | 'like_count': like_count, | |
212 | 'dislike_count': dislike_count, | |
213 | 'comment_count': comment_count, | |
125cfd78 | 214 | 'formats': formats, |
750e9833 | 215 | 'age_limit': 18, |
6bb05b32 YCH |
216 | 'tags': tags, |
217 | 'categories': categories, | |
125cfd78 | 218 | } |
e66e1a00 S |
219 | |
220 | ||
40e146aa S |
221 | class PornHubPlaylistBaseIE(InfoExtractor): |
222 | def _extract_entries(self, webpage): | |
223 | return [ | |
3a23bae9 S |
224 | self.url_result( |
225 | 'http://www.pornhub.com/%s' % video_url, | |
226 | PornHubIE.ie_key(), video_title=title) | |
227 | for video_url, title in orderedSet(re.findall( | |
228 | r'href="/?(view_video\.php\?.*\bviewkey=[\da-z]+[^"]*)"[^>]*\s+title="([^"]+)"', | |
229 | webpage)) | |
40e146aa | 230 | ] |
e66e1a00 S |
231 | |
232 | def _real_extract(self, url): | |
233 | playlist_id = self._match_id(url) | |
234 | ||
235 | webpage = self._download_webpage(url, playlist_id) | |
236 | ||
96d315c2 S |
237 | # Only process container div with main playlist content skipping |
238 | # drop-down menu that uses similar pattern for videos (see | |
239 | # https://github.com/rg3/youtube-dl/issues/11594). | |
240 | container = self._search_regex( | |
241 | r'(?s)(<div[^>]+class=["\']container.+)', webpage, | |
242 | 'container', default=webpage) | |
243 | ||
244 | entries = self._extract_entries(container) | |
e66e1a00 S |
245 | |
246 | playlist = self._parse_json( | |
247 | self._search_regex( | |
248 | r'playlistObject\s*=\s*({.+?});', webpage, 'playlist'), | |
249 | playlist_id) | |
250 | ||
251 | return self.playlist_result( | |
252 | entries, playlist_id, playlist.get('title'), playlist.get('description')) | |
40e146aa S |
253 | |
254 | ||
255 | class PornHubPlaylistIE(PornHubPlaylistBaseIE): | |
256 | _VALID_URL = r'https?://(?:www\.)?pornhub\.com/playlist/(?P<id>\d+)' | |
257 | _TESTS = [{ | |
96d315c2 | 258 | 'url': 'http://www.pornhub.com/playlist/4667351', |
40e146aa | 259 | 'info_dict': { |
96d315c2 S |
260 | 'id': '4667351', |
261 | 'title': 'Nataly Hot', | |
40e146aa | 262 | }, |
96d315c2 | 263 | 'playlist_mincount': 2, |
40e146aa S |
264 | }] |
265 | ||
266 | ||
267 | class PornHubUserVideosIE(PornHubPlaylistBaseIE): | |
268 | _VALID_URL = r'https?://(?:www\.)?pornhub\.com/users/(?P<id>[^/]+)/videos' | |
269 | _TESTS = [{ | |
34541395 | 270 | 'url': 'http://www.pornhub.com/users/zoe_ph/videos/public', |
40e146aa | 271 | 'info_dict': { |
34541395 | 272 | 'id': 'zoe_ph', |
40e146aa | 273 | }, |
34541395 S |
274 | 'playlist_mincount': 171, |
275 | }, { | |
276 | 'url': 'http://www.pornhub.com/users/rushandlia/videos', | |
277 | 'only_matching': True, | |
40e146aa S |
278 | }] |
279 | ||
280 | def _real_extract(self, url): | |
281 | user_id = self._match_id(url) | |
282 | ||
34541395 S |
283 | entries = [] |
284 | for page_num in itertools.count(1): | |
285 | try: | |
286 | webpage = self._download_webpage( | |
287 | url, user_id, 'Downloading page %d' % page_num, | |
288 | query={'page': page_num}) | |
289 | except ExtractorError as e: | |
290 | if isinstance(e.cause, compat_HTTPError) and e.cause.code == 404: | |
291 | break | |
292 | page_entries = self._extract_entries(webpage) | |
293 | if not page_entries: | |
294 | break | |
295 | entries.extend(page_entries) | |
296 | ||
297 | return self.playlist_result(entries, user_id) |