]>
Commit | Line | Data |
---|---|---|
6c376029 | 1 | # coding: utf-8 |
9933b574 PH |
2 | from __future__ import unicode_literals |
3 | ||
21fbf0f9 | 4 | import functools |
34541395 | 5 | import itertools |
21fbf0f9 | 6 | import operator |
125cfd78 | 7 | import re |
8 | ||
9 | from .common import InfoExtractor | |
1cc79574 | 10 | from ..compat import ( |
34541395 | 11 | compat_HTTPError, |
79367a98 | 12 | compat_str, |
278d061a | 13 | compat_urllib_request, |
1cc79574 | 14 | ) |
278d061a | 15 | from .openload import PhantomJSwrapper |
1cc79574 | 16 | from ..utils import ( |
b8526c78 | 17 | determine_ext, |
50789175 | 18 | ExtractorError, |
ed8648a3 | 19 | int_or_none, |
0164cd5d | 20 | NO_DEFAULT, |
8f9a477e | 21 | orderedSet, |
e1e35d1a | 22 | remove_quotes, |
0320ddc1 | 23 | str_to_int, |
4938c8d5 | 24 | url_or_none, |
125cfd78 | 25 | ) |
125cfd78 | 26 | |
9933b574 | 27 | |
71a1f617 S |
28 | class PornHubBaseIE(InfoExtractor): |
29 | def _download_webpage_handle(self, *args, **kwargs): | |
30 | def dl(*args, **kwargs): | |
31 | return super(PornHubBaseIE, self)._download_webpage_handle(*args, **kwargs) | |
32 | ||
33 | webpage, urlh = dl(*args, **kwargs) | |
34 | ||
35 | if any(re.search(p, webpage) for p in ( | |
36 | r'<body\b[^>]+\bonload=["\']go\(\)', | |
37 | r'document\.cookie\s*=\s*["\']RNKEY=', | |
38 | r'document\.location\.reload\(true\)')): | |
39 | url_or_request = args[0] | |
40 | url = (url_or_request.get_full_url() | |
41 | if isinstance(url_or_request, compat_urllib_request.Request) | |
42 | else url_or_request) | |
43 | phantom = PhantomJSwrapper(self, required_version='2.0') | |
44 | phantom.get(url, html=webpage) | |
45 | webpage, urlh = dl(*args, **kwargs) | |
46 | ||
47 | return webpage, urlh | |
48 | ||
49 | ||
50 | class PornHubIE(PornHubBaseIE): | |
bc4b2d75 S |
51 | IE_DESC = 'PornHub and Thumbzilla' |
52 | _VALID_URL = r'''(?x) | |
53 | https?:// | |
54 | (?: | |
fa9b8c66 | 55 | (?:[^/]+\.)?(?P<host>pornhub(?:premium)?\.(?:com|net))/(?:(?:view_video\.php|video/show)\?viewkey=|embed/)| |
bc4b2d75 S |
56 | (?:www\.)?thumbzilla\.com/video/ |
57 | ) | |
b52c9ef1 | 58 | (?P<id>[\da-z]+) |
bc4b2d75 | 59 | ''' |
360075e2 | 60 | _TESTS = [{ |
9933b574 | 61 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=648719015', |
ed8648a3 | 62 | 'md5': '1e19b41231a02eba417839222ac9d58e', |
9933b574 | 63 | 'info_dict': { |
249efaf4 PH |
64 | 'id': '648719015', |
65 | 'ext': 'mp4', | |
611c1dd9 | 66 | 'title': 'Seductive Indian beauty strips down and fingers her pink pussy', |
ed8648a3 | 67 | 'uploader': 'Babes', |
3661ebf2 | 68 | 'upload_date': '20130628', |
ed8648a3 S |
69 | 'duration': 361, |
70 | 'view_count': int, | |
71 | 'like_count': int, | |
72 | 'dislike_count': int, | |
73 | 'comment_count': int, | |
74 | 'age_limit': 18, | |
6bb05b32 YCH |
75 | 'tags': list, |
76 | 'categories': list, | |
6c376029 S |
77 | }, |
78 | }, { | |
79 | # non-ASCII title | |
80 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=1331683002', | |
81 | 'info_dict': { | |
82 | 'id': '1331683002', | |
83 | 'ext': 'mp4', | |
84 | 'title': '重庆婷婷女王足交', | |
79367a98 | 85 | 'uploader': 'Unknown', |
3661ebf2 | 86 | 'upload_date': '20150213', |
6c376029 S |
87 | 'duration': 1753, |
88 | 'view_count': int, | |
89 | 'like_count': int, | |
90 | 'dislike_count': int, | |
91 | 'comment_count': int, | |
92 | 'age_limit': 18, | |
6bb05b32 YCH |
93 | 'tags': list, |
94 | 'categories': list, | |
6c376029 S |
95 | }, |
96 | 'params': { | |
97 | 'skip_download': True, | |
98 | }, | |
4938c8d5 GF |
99 | }, { |
100 | # subtitles | |
101 | 'url': 'https://www.pornhub.com/view_video.php?viewkey=ph5af5fef7c2aa7', | |
102 | 'info_dict': { | |
103 | 'id': 'ph5af5fef7c2aa7', | |
104 | 'ext': 'mp4', | |
105 | 'title': 'BFFS - Cute Teen Girls Share Cock On the Floor', | |
106 | 'uploader': 'BFFs', | |
107 | 'duration': 622, | |
108 | 'view_count': int, | |
109 | 'like_count': int, | |
110 | 'dislike_count': int, | |
111 | 'comment_count': int, | |
112 | 'age_limit': 18, | |
113 | 'tags': list, | |
114 | 'categories': list, | |
115 | 'subtitles': { | |
116 | 'en': [{ | |
117 | "ext": 'srt' | |
118 | }] | |
119 | }, | |
120 | }, | |
121 | 'params': { | |
122 | 'skip_download': True, | |
123 | }, | |
360075e2 S |
124 | }, { |
125 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph557bbb6676d2d', | |
126 | 'only_matching': True, | |
272e4db5 | 127 | }, { |
eaaaaec0 | 128 | # removed at the request of cam4.com |
272e4db5 S |
129 | 'url': 'http://fr.pornhub.com/view_video.php?viewkey=ph55ca2f9760862', |
130 | 'only_matching': True, | |
eaaaaec0 S |
131 | }, { |
132 | # removed at the request of the copyright owner | |
133 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=788152859', | |
134 | 'only_matching': True, | |
135 | }, { | |
136 | # removed by uploader | |
137 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph572716d15a111', | |
138 | 'only_matching': True, | |
195f0845 S |
139 | }, { |
140 | # private video | |
141 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph56fd731fce6b7', | |
142 | 'only_matching': True, | |
bc4b2d75 S |
143 | }, { |
144 | 'url': 'https://www.thumbzilla.com/video/ph56c6114abd99a/horny-girlfriend-sex', | |
145 | 'only_matching': True, | |
a99cc4ca S |
146 | }, { |
147 | 'url': 'http://www.pornhub.com/video/show?viewkey=648719015', | |
148 | 'only_matching': True, | |
f97c0991 S |
149 | }, { |
150 | 'url': 'https://www.pornhub.net/view_video.php?viewkey=203640933', | |
151 | 'only_matching': True, | |
fa9b8c66 TW |
152 | }, { |
153 | 'url': 'https://www.pornhubpremium.com/view_video.php?viewkey=ph5e4acdae54a82', | |
154 | 'only_matching': True, | |
360075e2 | 155 | }] |
125cfd78 | 156 | |
b52c9ef1 S |
157 | @staticmethod |
158 | def _extract_urls(webpage): | |
159 | return re.findall( | |
3430ff9b | 160 | r'<iframe[^>]+?src=["\'](?P<url>(?:https?:)?//(?:www\.)?pornhub\.(?:com|net)/embed/[\da-z]+)', |
b52c9ef1 | 161 | webpage) |
65d161c4 | 162 | |
0320ddc1 | 163 | def _extract_count(self, pattern, webpage, name): |
7700207e S |
164 | return str_to_int(self._search_regex( |
165 | pattern, webpage, '%s count' % name, fatal=False)) | |
0320ddc1 | 166 | |
125cfd78 | 167 | def _real_extract(self, url): |
3430ff9b S |
168 | mobj = re.match(self._VALID_URL, url) |
169 | host = mobj.group('host') or 'pornhub.com' | |
170 | video_id = mobj.group('id') | |
7399ca1f | 171 | |
fa9b8c66 TW |
172 | if 'premium' in host: |
173 | if not self._downloader.params.get('cookiefile'): | |
174 | raise ExtractorError( | |
175 | 'PornHub Premium requires authentication.' | |
176 | ' You may want to use --cookies.', | |
177 | expected=True) | |
178 | ||
3430ff9b | 179 | self._set_cookie(host, 'age_verified', '1') |
125cfd78 | 180 | |
9a372f14 | 181 | def dl_webpage(platform): |
3430ff9b | 182 | self._set_cookie(host, 'platform', platform) |
9a372f14 | 183 | return self._download_webpage( |
2c53c0eb | 184 | 'https://www.%s/view_video.php?viewkey=%s' % (host, video_id), |
79367a98 | 185 | video_id, 'Downloading %s webpage' % platform) |
9a372f14 S |
186 | |
187 | webpage = dl_webpage('pc') | |
125cfd78 | 188 | |
50789175 | 189 | error_msg = self._html_search_regex( |
add7d2a0 | 190 | r'(?s)<div[^>]+class=(["\'])(?:(?!\1).)*\b(?:removed|userMessageSection)\b(?:(?!\1).)*\1[^>]*>(?P<error>.+?)</div>', |
3cb3b600 | 191 | webpage, 'error message', default=None, group='error') |
50789175 PH |
192 | if error_msg: |
193 | error_msg = re.sub(r'\s+', ' ', error_msg) | |
194 | raise ExtractorError( | |
195 | 'PornHub said: %s' % error_msg, | |
196 | expected=True, video_id=video_id) | |
197 | ||
6c376029 S |
198 | # video_title from flashvars contains whitespace instead of non-ASCII (see |
199 | # http://www.pornhub.com/view_video.php?viewkey=1331683002), not relying | |
200 | # on that anymore. | |
79367a98 | 201 | title = self._html_search_meta( |
46cc54ca S |
202 | 'twitter:title', webpage, default=None) or self._html_search_regex( |
203 | (r'(?s)<h1[^>]+class=["\']title["\'][^>]*>(?P<title>.+?)</h1>', | |
204 | r'<div[^>]+data-video-title=(["\'])(?P<title>(?:(?!\1).)+)\1', | |
205 | r'shareTitle["\']\s*[=:]\s*(["\'])(?P<title>(?:(?!\1).)+)\1'), | |
6c376029 S |
206 | webpage, 'title', group='title') |
207 | ||
79367a98 S |
208 | video_urls = [] |
209 | video_urls_set = set() | |
4938c8d5 | 210 | subtitles = {} |
79367a98 | 211 | |
ed8648a3 S |
212 | flashvars = self._parse_json( |
213 | self._search_regex( | |
03442072 | 214 | r'var\s+flashvars_\d+\s*=\s*({.+?});', webpage, 'flashvars', default='{}'), |
ed8648a3 S |
215 | video_id) |
216 | if flashvars: | |
4938c8d5 GF |
217 | subtitle_url = url_or_none(flashvars.get('closedCaptionsFile')) |
218 | if subtitle_url: | |
219 | subtitles.setdefault('en', []).append({ | |
220 | 'url': subtitle_url, | |
221 | 'ext': 'srt', | |
222 | }) | |
ed8648a3 S |
223 | thumbnail = flashvars.get('image_url') |
224 | duration = int_or_none(flashvars.get('video_duration')) | |
79367a98 S |
225 | media_definitions = flashvars.get('mediaDefinitions') |
226 | if isinstance(media_definitions, list): | |
227 | for definition in media_definitions: | |
228 | if not isinstance(definition, dict): | |
229 | continue | |
230 | video_url = definition.get('videoUrl') | |
231 | if not video_url or not isinstance(video_url, compat_str): | |
232 | continue | |
233 | if video_url in video_urls_set: | |
234 | continue | |
235 | video_urls_set.add(video_url) | |
236 | video_urls.append( | |
237 | (video_url, int_or_none(definition.get('quality')))) | |
ed8648a3 | 238 | else: |
79367a98 S |
239 | thumbnail, duration = [None] * 2 |
240 | ||
0164cd5d | 241 | def extract_js_vars(webpage, pattern, default=NO_DEFAULT): |
79367a98 | 242 | assignments = self._search_regex( |
0164cd5d | 243 | pattern, webpage, 'encoded url', default=default) |
f4134726 S |
244 | if not assignments: |
245 | return {} | |
246 | ||
247 | assignments = assignments.split(';') | |
79367a98 S |
248 | |
249 | js_vars = {} | |
250 | ||
251 | def parse_js_value(inp): | |
252 | inp = re.sub(r'/\*(?:(?!\*/).)*?\*/', '', inp) | |
253 | if '+' in inp: | |
254 | inps = inp.split('+') | |
255 | return functools.reduce( | |
256 | operator.concat, map(parse_js_value, inps)) | |
257 | inp = inp.strip() | |
258 | if inp in js_vars: | |
259 | return js_vars[inp] | |
260 | return remove_quotes(inp) | |
261 | ||
262 | for assn in assignments: | |
263 | assn = assn.strip() | |
264 | if not assn: | |
265 | continue | |
266 | assn = re.sub(r'var\s+', '', assn) | |
267 | vname, value = assn.split('=', 1) | |
268 | js_vars[vname] = parse_js_value(value) | |
f4134726 | 269 | return js_vars |
79367a98 | 270 | |
f4134726 S |
271 | def add_video_url(video_url): |
272 | v_url = url_or_none(video_url) | |
273 | if not v_url: | |
274 | return | |
275 | if v_url in video_urls_set: | |
276 | return | |
277 | video_urls.append((v_url, None)) | |
278 | video_urls_set.add(v_url) | |
279 | ||
280 | if not video_urls: | |
281 | FORMAT_PREFIXES = ('media', 'quality') | |
282 | js_vars = extract_js_vars( | |
283 | webpage, r'(var\s+(?:%s)_.+)' % '|'.join(FORMAT_PREFIXES), | |
0164cd5d | 284 | default=None) |
f4134726 S |
285 | if js_vars: |
286 | for key, format_url in js_vars.items(): | |
287 | if any(key.startswith(p) for p in FORMAT_PREFIXES): | |
288 | add_video_url(format_url) | |
0164cd5d S |
289 | if not video_urls and re.search( |
290 | r'<[^>]+\bid=["\']lockedPlayer', webpage): | |
291 | raise ExtractorError( | |
292 | 'Video %s is locked' % video_id, expected=True) | |
f4134726 S |
293 | |
294 | if not video_urls: | |
295 | js_vars = extract_js_vars( | |
296 | dl_webpage('tv'), r'(var.+?mediastring.+?)</script>') | |
297 | add_video_url(js_vars['mediastring']) | |
79367a98 S |
298 | |
299 | for mobj in re.finditer( | |
300 | r'<a[^>]+\bclass=["\']downloadBtn\b[^>]+\bhref=(["\'])(?P<url>(?:(?!\1).)+)\1', | |
301 | webpage): | |
302 | video_url = mobj.group('url') | |
303 | if video_url not in video_urls_set: | |
304 | video_urls.append((video_url, None)) | |
305 | video_urls_set.add(video_url) | |
306 | ||
3661ebf2 | 307 | upload_date = None |
79367a98 S |
308 | formats = [] |
309 | for video_url, height in video_urls: | |
3661ebf2 S |
310 | if not upload_date: |
311 | upload_date = self._search_regex( | |
312 | r'/(\d{6}/\d{2})/', video_url, 'upload data', default=None) | |
313 | if upload_date: | |
314 | upload_date = upload_date.replace('/', '') | |
f4134726 S |
315 | ext = determine_ext(video_url) |
316 | if ext == 'mpd': | |
b8526c78 S |
317 | formats.extend(self._extract_mpd_formats( |
318 | video_url, video_id, mpd_id='dash', fatal=False)) | |
319 | continue | |
f4134726 S |
320 | elif ext == 'm3u8': |
321 | formats.extend(self._extract_m3u8_formats( | |
322 | video_url, video_id, 'mp4', entry_protocol='m3u8_native', | |
323 | m3u8_id='hls', fatal=False)) | |
324 | continue | |
79367a98 S |
325 | tbr = None |
326 | mobj = re.search(r'(?P<height>\d+)[pP]?_(?P<tbr>\d+)[kK]', video_url) | |
327 | if mobj: | |
328 | if not height: | |
329 | height = int(mobj.group('height')) | |
330 | tbr = int(mobj.group('tbr')) | |
331 | formats.append({ | |
332 | 'url': video_url, | |
333 | 'format_id': '%dp' % height if height else None, | |
334 | 'height': height, | |
335 | 'tbr': tbr, | |
336 | }) | |
337 | self._sort_formats(formats) | |
ed8648a3 | 338 | |
0320ddc1 | 339 | video_uploader = self._html_search_regex( |
2d4fe594 | 340 | r'(?s)From: .+?<(?:a\b[^>]+\bhref=["\']/(?:(?:user|channel)s|model|pornstar)/|span\b[^>]+\bclass=["\']username)[^>]+>(.+?)<', |
0320ddc1 | 341 | webpage, 'uploader', fatal=False) |
125cfd78 | 342 | |
7700207e | 343 | view_count = self._extract_count( |
540b9f51 | 344 | r'<span class="count">([\d,\.]+)</span> [Vv]iews', webpage, 'view') |
7700207e S |
345 | like_count = self._extract_count( |
346 | r'<span class="votesUp">([\d,\.]+)</span>', webpage, 'like') | |
347 | dislike_count = self._extract_count( | |
348 | r'<span class="votesDown">([\d,\.]+)</span>', webpage, 'dislike') | |
0320ddc1 | 349 | comment_count = self._extract_count( |
7700207e | 350 | r'All Comments\s*<span>\(([\d,.]+)\)', webpage, 'comment') |
0320ddc1 | 351 | |
5dda1ede | 352 | def extract_list(meta_key): |
d2d970d0 | 353 | div = self._search_regex( |
5dda1ede S |
354 | r'(?s)<div[^>]+\bclass=["\'].*?\b%sWrapper[^>]*>(.+?)</div>' |
355 | % meta_key, webpage, meta_key, default=None) | |
d2d970d0 | 356 | if div: |
5dda1ede | 357 | return re.findall(r'<a[^>]+\bhref=[^>]+>([^<]+)', div) |
6bb05b32 | 358 | |
125cfd78 | 359 | return { |
360 | 'id': video_id, | |
361 | 'uploader': video_uploader, | |
3661ebf2 | 362 | 'upload_date': upload_date, |
6c376029 | 363 | 'title': title, |
125cfd78 | 364 | 'thumbnail': thumbnail, |
ed8648a3 | 365 | 'duration': duration, |
0320ddc1 S |
366 | 'view_count': view_count, |
367 | 'like_count': like_count, | |
368 | 'dislike_count': dislike_count, | |
369 | 'comment_count': comment_count, | |
79367a98 | 370 | 'formats': formats, |
750e9833 | 371 | 'age_limit': 18, |
5dda1ede S |
372 | 'tags': extract_list('tags'), |
373 | 'categories': extract_list('categories'), | |
4938c8d5 | 374 | 'subtitles': subtitles, |
125cfd78 | 375 | } |
e66e1a00 S |
376 | |
377 | ||
71a1f617 | 378 | class PornHubPlaylistBaseIE(PornHubBaseIE): |
3430ff9b | 379 | def _extract_entries(self, webpage, host): |
475bcb22 S |
380 | # Only process container div with main playlist content skipping |
381 | # drop-down menu that uses similar pattern for videos (see | |
067aa17e | 382 | # https://github.com/ytdl-org/youtube-dl/issues/11594). |
475bcb22 S |
383 | container = self._search_regex( |
384 | r'(?s)(<div[^>]+class=["\']container.+)', webpage, | |
385 | 'container', default=webpage) | |
386 | ||
40e146aa | 387 | return [ |
3a23bae9 | 388 | self.url_result( |
3430ff9b | 389 | 'http://www.%s/%s' % (host, video_url), |
3a23bae9 S |
390 | PornHubIE.ie_key(), video_title=title) |
391 | for video_url, title in orderedSet(re.findall( | |
392 | r'href="/?(view_video\.php\?.*\bviewkey=[\da-z]+[^"]*)"[^>]*\s+title="([^"]+)"', | |
475bcb22 | 393 | container)) |
40e146aa | 394 | ] |
e66e1a00 S |
395 | |
396 | def _real_extract(self, url): | |
3430ff9b S |
397 | mobj = re.match(self._VALID_URL, url) |
398 | host = mobj.group('host') | |
399 | playlist_id = mobj.group('id') | |
e66e1a00 S |
400 | |
401 | webpage = self._download_webpage(url, playlist_id) | |
402 | ||
3430ff9b | 403 | entries = self._extract_entries(webpage, host) |
e66e1a00 S |
404 | |
405 | playlist = self._parse_json( | |
406 | self._search_regex( | |
cd138d8b S |
407 | r'(?:playlistObject|PLAYLIST_VIEW)\s*=\s*({.+?});', webpage, |
408 | 'playlist', default='{}'), | |
409 | playlist_id, fatal=False) | |
410 | title = playlist.get('title') or self._search_regex( | |
411 | r'>Videos\s+in\s+(.+?)\s+[Pp]laylist<', webpage, 'title', fatal=False) | |
e66e1a00 S |
412 | |
413 | return self.playlist_result( | |
cd138d8b | 414 | entries, playlist_id, title, playlist.get('description')) |
40e146aa S |
415 | |
416 | ||
21b08463 | 417 | class PornHubUserIE(PornHubPlaylistBaseIE): |
fa9b8c66 | 418 | _VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?(?P<host>pornhub(?:premium)?\.(?:com|net))/(?:(?:user|channel)s|model|pornstar)/(?P<id>[^/?#&]+))(?:[?#&]|/(?!videos)|$)' |
40e146aa | 419 | _TESTS = [{ |
21b08463 S |
420 | 'url': 'https://www.pornhub.com/model/zoe_ph', |
421 | 'playlist_mincount': 118, | |
422 | }, { | |
423 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious', | |
40e146aa | 424 | 'info_dict': { |
21b08463 | 425 | 'id': 'liz-vicious', |
40e146aa | 426 | }, |
21b08463 S |
427 | 'playlist_mincount': 118, |
428 | }, { | |
429 | 'url': 'https://www.pornhub.com/users/russianveet69', | |
9634de17 | 430 | 'only_matching': True, |
21b08463 S |
431 | }, { |
432 | 'url': 'https://www.pornhub.com/channels/povd', | |
9634de17 S |
433 | 'only_matching': True, |
434 | }, { | |
435 | 'url': 'https://www.pornhub.com/model/zoe_ph?abc=1', | |
436 | 'only_matching': True, | |
21b08463 S |
437 | }] |
438 | ||
21b08463 S |
439 | def _real_extract(self, url): |
440 | mobj = re.match(self._VALID_URL, url) | |
441 | user_id = mobj.group('id') | |
442 | return self.url_result( | |
9634de17 | 443 | '%s/videos' % mobj.group('url'), ie=PornHubPagedVideoListIE.ie_key(), |
21b08463 S |
444 | video_id=user_id) |
445 | ||
446 | ||
447 | class PornHubPagedPlaylistBaseIE(PornHubPlaylistBaseIE): | |
4bf568d3 S |
448 | @staticmethod |
449 | def _has_more(webpage): | |
450 | return re.search( | |
451 | r'''(?x) | |
452 | <li[^>]+\bclass=["\']page_next| | |
453 | <link[^>]+\brel=["\']next| | |
454 | <button[^>]+\bid=["\']moreDataBtn | |
455 | ''', webpage) is not None | |
456 | ||
21b08463 S |
457 | def _real_extract(self, url): |
458 | mobj = re.match(self._VALID_URL, url) | |
459 | host = mobj.group('host') | |
9634de17 | 460 | item_id = mobj.group('id') |
21b08463 | 461 | |
1f7a563a S |
462 | page = int_or_none(self._search_regex( |
463 | r'\bpage=(\d+)', url, 'page', default=None)) | |
464 | ||
21b08463 | 465 | entries = [] |
1f7a563a | 466 | for page_num in (page, ) if page is not None else itertools.count(1): |
21b08463 S |
467 | try: |
468 | webpage = self._download_webpage( | |
4bf568d3 | 469 | url, item_id, 'Downloading page %d' % page_num, |
21b08463 S |
470 | query={'page': page_num}) |
471 | except ExtractorError as e: | |
472 | if isinstance(e.cause, compat_HTTPError) and e.cause.code == 404: | |
473 | break | |
474 | raise | |
475 | page_entries = self._extract_entries(webpage, host) | |
476 | if not page_entries: | |
477 | break | |
478 | entries.extend(page_entries) | |
479 | if not self._has_more(webpage): | |
480 | break | |
481 | ||
9634de17 | 482 | return self.playlist_result(orderedSet(entries), item_id) |
21b08463 S |
483 | |
484 | ||
9634de17 | 485 | class PornHubPagedVideoListIE(PornHubPagedPlaylistBaseIE): |
fa9b8c66 | 486 | _VALID_URL = r'https?://(?:[^/]+\.)?(?P<host>pornhub(?:premium)?\.(?:com|net))/(?P<id>(?:[^/]+/)*[^/?#&]+)' |
21b08463 | 487 | _TESTS = [{ |
1f7a563a | 488 | 'url': 'https://www.pornhub.com/model/zoe_ph/videos', |
21b08463 | 489 | 'only_matching': True, |
34541395 S |
490 | }, { |
491 | 'url': 'http://www.pornhub.com/users/rushandlia/videos', | |
492 | 'only_matching': True, | |
21b08463 S |
493 | }, { |
494 | 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos', | |
495 | 'info_dict': { | |
9634de17 | 496 | 'id': 'pornstar/jenny-blighe/videos', |
21b08463 S |
497 | }, |
498 | 'playlist_mincount': 149, | |
1f7a563a S |
499 | }, { |
500 | 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos?page=3', | |
501 | 'info_dict': { | |
9634de17 | 502 | 'id': 'pornstar/jenny-blighe/videos', |
1f7a563a S |
503 | }, |
504 | 'playlist_mincount': 40, | |
f66df20c PV |
505 | }, { |
506 | # default sorting as Top Rated Videos | |
507 | 'url': 'https://www.pornhub.com/channels/povd/videos', | |
508 | 'info_dict': { | |
9634de17 | 509 | 'id': 'channels/povd/videos', |
f66df20c PV |
510 | }, |
511 | 'playlist_mincount': 293, | |
512 | }, { | |
513 | # Top Rated Videos | |
514 | 'url': 'https://www.pornhub.com/channels/povd/videos?o=ra', | |
515 | 'only_matching': True, | |
516 | }, { | |
517 | # Most Recent Videos | |
518 | 'url': 'https://www.pornhub.com/channels/povd/videos?o=da', | |
519 | 'only_matching': True, | |
520 | }, { | |
521 | # Most Viewed Videos | |
522 | 'url': 'https://www.pornhub.com/channels/povd/videos?o=vi', | |
523 | 'only_matching': True, | |
92ded33a S |
524 | }, { |
525 | 'url': 'http://www.pornhub.com/users/zoe_ph/videos/public', | |
526 | 'only_matching': True, | |
21b08463 S |
527 | }, { |
528 | # Most Viewed Videos | |
529 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=mv', | |
530 | 'only_matching': True, | |
531 | }, { | |
532 | # Top Rated Videos | |
533 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=tr', | |
534 | 'only_matching': True, | |
535 | }, { | |
536 | # Longest Videos | |
537 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=lg', | |
538 | 'only_matching': True, | |
539 | }, { | |
540 | # Newest Videos | |
541 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=cm', | |
542 | 'only_matching': True, | |
21b08463 S |
543 | }, { |
544 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos/paid', | |
545 | 'only_matching': True, | |
546 | }, { | |
547 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos/fanonly', | |
548 | 'only_matching': True, | |
9634de17 S |
549 | }, { |
550 | 'url': 'https://www.pornhub.com/video', | |
551 | 'only_matching': True, | |
552 | }, { | |
553 | 'url': 'https://www.pornhub.com/video?page=3', | |
554 | 'only_matching': True, | |
555 | }, { | |
556 | 'url': 'https://www.pornhub.com/video/search?search=123', | |
557 | 'only_matching': True, | |
558 | }, { | |
559 | 'url': 'https://www.pornhub.com/categories/teen', | |
560 | 'only_matching': True, | |
561 | }, { | |
562 | 'url': 'https://www.pornhub.com/categories/teen?page=3', | |
563 | 'only_matching': True, | |
564 | }, { | |
565 | 'url': 'https://www.pornhub.com/hd', | |
566 | 'only_matching': True, | |
567 | }, { | |
568 | 'url': 'https://www.pornhub.com/hd?page=3', | |
569 | 'only_matching': True, | |
570 | }, { | |
571 | 'url': 'https://www.pornhub.com/described-video', | |
572 | 'only_matching': True, | |
573 | }, { | |
574 | 'url': 'https://www.pornhub.com/described-video?page=2', | |
575 | 'only_matching': True, | |
576 | }, { | |
577 | 'url': 'https://www.pornhub.com/video/incategories/60fps-1/hd-porn', | |
578 | 'only_matching': True, | |
579 | }, { | |
580 | 'url': 'https://www.pornhub.com/playlist/44121572', | |
581 | 'info_dict': { | |
582 | 'id': 'playlist/44121572', | |
583 | }, | |
584 | 'playlist_mincount': 132, | |
585 | }, { | |
586 | 'url': 'https://www.pornhub.com/playlist/4667351', | |
587 | 'only_matching': True, | |
588 | }, { | |
589 | 'url': 'https://de.pornhub.com/playlist/4667351', | |
590 | 'only_matching': True, | |
40e146aa S |
591 | }] |
592 | ||
21b08463 S |
593 | @classmethod |
594 | def suitable(cls, url): | |
595 | return (False | |
9634de17 S |
596 | if PornHubIE.suitable(url) or PornHubUserIE.suitable(url) or PornHubUserVideosUploadIE.suitable(url) |
597 | else super(PornHubPagedVideoListIE, cls).suitable(url)) | |
21b08463 | 598 | |
34541395 | 599 | |
21b08463 | 600 | class PornHubUserVideosUploadIE(PornHubPagedPlaylistBaseIE): |
fa9b8c66 | 601 | _VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?(?P<host>pornhub(?:premium)?\.(?:com|net))/(?:(?:user|channel)s|model|pornstar)/(?P<id>[^/]+)/videos/upload)' |
21b08463 S |
602 | _TESTS = [{ |
603 | 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos/upload', | |
604 | 'info_dict': { | |
605 | 'id': 'jenny-blighe', | |
606 | }, | |
607 | 'playlist_mincount': 129, | |
608 | }, { | |
609 | 'url': 'https://www.pornhub.com/model/zoe_ph/videos/upload', | |
610 | 'only_matching': True, | |
611 | }] |