2 from __future__
import unicode_literals
9 from .common
import InfoExtractor
10 from ..compat
import (
13 compat_urllib_request
,
15 from .openload
import PhantomJSwrapper
31 class PornHubBaseIE(InfoExtractor
):
32 _NETRC_MACHINE
= 'pornhub'
34 def _download_webpage_handle(self
, *args
, **kwargs
):
35 def dl(*args
, **kwargs
):
36 return super(PornHubBaseIE
, self
)._download
_webpage
_handle
(*args
, **kwargs
)
38 ret
= dl(*args
, **kwargs
)
45 if any(re
.search(p
, webpage
) for p
in (
46 r
'<body\b[^>]+\bonload=["\']go\
(\
)',
47 r'document\
.cookie\s
*=\s
*["\']RNKEY=',
48 r'document\.location\.reload\(true\)')):
49 url_or_request = args[0]
50 url = (url_or_request.get_full_url()
51 if isinstance(url_or_request, compat_urllib_request.Request)
53 phantom = PhantomJSwrapper(self, required_version='2.0')
54 phantom.get(url, html=webpage)
55 webpage, urlh = dl(*args, **kwargs)
59 def _real_initialize(self):
60 self._logged_in = False
62 def _login(self, host):
66 site = host.split('.')[0]
68 # Both sites pornhub and pornhubpremium have separate accounts
69 # so there should be an option to provide credentials for both.
70 # At the same time some videos are available under the same video id
71 # on both sites so that we have to identify them as the same video.
72 # For that purpose we have to keep both in the same extractor
73 # but under different netrc machines.
74 username, password = self._get_login_info(netrc_machine=site)
78 login_url = 'https://www.%s/%slogin' % (host, 'premium/' if 'premium' in host else '')
79 login_page = self._download_webpage(
80 login_url, None, 'Downloading %s login page' % site)
82 def is_logged(webpage):
83 return any(re.search(p, webpage) for p in (
84 r'class=["\']signOut
',
85 r'>Sign\s
+[Oo
]ut\s
*<'))
87 if is_logged(login_page):
88 self._logged_in = True
91 login_form = self._hidden_inputs(login_page)
98 response = self._download_json(
99 'https
://www
.%s/front
/authenticate
' % host, None,
100 'Logging
in to
%s' % site,
101 data=urlencode_postdata(login_form),
103 'Content
-Type
': 'application
/x
-www
-form
-urlencoded
; charset
=UTF
-8',
104 'Referer
': login_url,
105 'X
-Requested
-With
': 'XMLHttpRequest
',
108 if response.get('success
') == '1':
109 self._logged_in = True
112 message = response.get('message
')
113 if message is not None:
114 raise ExtractorError(
115 'Unable to login
: %s' % message, expected=True)
117 raise ExtractorError('Unable to log
in')
120 class PornHubIE(PornHubBaseIE):
121 IE_DESC = 'PornHub
and Thumbzilla
'
122 _VALID_URL = r'''(?x)
125 (?:[^/]+\.)?(?P<host>pornhub(?:premium)?\.(?:com|net|org))/(?:(?:view_video\.php|video/show)\?viewkey=|embed/)|
126 (?:www\.)?thumbzilla\.com/video/
131 'url
': 'http
://www
.pornhub
.com
/view_video
.php?viewkey
=648719015',
132 'md5
': 'a6391306d050e4547f62b3f485dd9ba9
',
136 'title
': 'Seductive Indian beauty strips down
and fingers her pink pussy
',
138 'upload_date
': '20130628',
139 'timestamp
': 1372447216,
143 'dislike_count
': int,
144 'comment_count
': int,
151 'url
': 'http
://www
.pornhub
.com
/view_video
.php?viewkey
=1331683002',
156 'upload_date
': '20150213',
157 'timestamp
': 1423804862,
161 'dislike_count
': int,
162 'comment_count
': int,
168 'skip_download
': True,
172 'url
': 'https
://www
.pornhub
.com
/view_video
.php?viewkey
=ph5af5fef7c2aa7
',
174 'id': 'ph5af5fef7c2aa7
',
176 'title
': 'BFFS
- Cute Teen Girls Share Cock On the Floor
',
181 'dislike_count
': int,
182 'comment_count
': int,
193 'skip_download
': True,
195 'skip
': 'This video has been disabled
',
197 'url
': 'http
://www
.pornhub
.com
/view_video
.php?viewkey
=ph557bbb6676d2d
',
198 'only_matching
': True,
200 # removed at the request of cam4.com
201 'url
': 'http
://fr
.pornhub
.com
/view_video
.php?viewkey
=ph55ca2f9760862
',
202 'only_matching
': True,
204 # removed at the request of the copyright owner
205 'url
': 'http
://www
.pornhub
.com
/view_video
.php?viewkey
=788152859',
206 'only_matching
': True,
208 # removed by uploader
209 'url
': 'http
://www
.pornhub
.com
/view_video
.php?viewkey
=ph572716d15a111
',
210 'only_matching
': True,
213 'url
': 'http
://www
.pornhub
.com
/view_video
.php?viewkey
=ph56fd731fce6b7
',
214 'only_matching
': True,
216 'url
': 'https
://www
.thumbzilla
.com
/video
/ph56c6114abd99a
/horny
-girlfriend
-sex
',
217 'only_matching
': True,
219 'url
': 'http
://www
.pornhub
.com
/video
/show?viewkey
=648719015',
220 'only_matching
': True,
222 'url
': 'https
://www
.pornhub
.net
/view_video
.php?viewkey
=203640933',
223 'only_matching
': True,
225 'url
': 'https
://www
.pornhub
.org
/view_video
.php?viewkey
=203640933',
226 'only_matching
': True,
228 'url
': 'https
://www
.pornhubpremium
.com
/view_video
.php?viewkey
=ph5e4acdae54a82
',
229 'only_matching
': True,
231 # Some videos are available with the same id on both premium
232 # and non-premium sites (e.g. this and the following test)
233 'url
': 'https
://www
.pornhub
.com
/view_video
.php?viewkey
=ph5f75b0f4b18e3
',
234 'only_matching
': True,
236 'url
': 'https
://www
.pornhubpremium
.com
/view_video
.php?viewkey
=ph5f75b0f4b18e3
',
237 'only_matching
': True,
241 def _extract_urls(webpage):
243 r'<iframe
[^
>]+?src
=["\'](?P<url>(?:https?:)?//(?:www\.)?pornhub(?:premium)?\.(?:com|net|org)/embed/[\da-z]+)',
246 def _extract_count(self, pattern, webpage, name):
247 return str_to_int(self._search_regex(
248 pattern, webpage, '%s count' % name, fatal=False))
250 def _real_extract(self, url):
251 mobj = re.match(self._VALID_URL, url)
252 host = mobj.group('host') or 'pornhub.com'
253 video_id = mobj.group('id')
257 self._set_cookie(host, 'age_verified', '1')
259 def dl_webpage(platform):
260 self._set_cookie(host, 'platform', platform)
261 return self._download_webpage(
262 'https://www.%s/view_video.php?viewkey=%s' % (host, video_id),
263 video_id, 'Downloading %s webpage' % platform)
265 webpage = dl_webpage('pc')
267 error_msg = self._html_search_regex(
268 r'(?s)<div[^>]+class=(["\'])(?
:(?
!\
1).)*\b(?
:removed|userMessageSection
)\b(?
:(?
!\
1).)*\
1[^
>]*>(?P
<error
>.+?
)</div
>',
269 webpage, 'error message
', default=None, group='error
')
271 error_msg = re.sub(r'\s
+', ' ', error_msg)
272 raise ExtractorError(
273 'PornHub said
: %s' % error_msg,
274 expected=True, video_id=video_id)
276 # video_title from flashvars contains whitespace instead of non-ASCII (see
277 # http://www.pornhub.com/view_video.php?viewkey=1331683002), not relying
279 title = self._html_search_meta(
280 'twitter
:title
', webpage, default=None) or self._html_search_regex(
281 (r'(?s
)<h1
[^
>]+class=["\']title["\'][^
>]*>(?P
<title
>.+?
)</h1
>',
282 r'<div
[^
>]+data
-video
-title
=(["\'])(?P<title>(?:(?!\1).)+)\1',
283 r'shareTitle["\']\s
*[=:]\s
*(["\'])(?P<title>(?:(?!\1).)+)\1'),
284 webpage, 'title', group='title')
287 video_urls_set = set()
290 flashvars = self._parse_json(
292 r'var\s+flashvars_\d+\s*=\s*({.+?});', webpage, 'flashvars', default='{}'),
295 subtitle_url = url_or_none(flashvars.get('closedCaptionsFile'))
297 subtitles.setdefault('en', []).append({
301 thumbnail = flashvars.get('image_url')
302 duration = int_or_none(flashvars.get('video_duration'))
303 media_definitions = flashvars.get('mediaDefinitions')
304 if isinstance(media_definitions, list):
305 for definition in media_definitions:
306 if not isinstance(definition, dict):
308 video_url = definition.get('videoUrl')
309 if not video_url or not isinstance(video_url, compat_str):
311 if video_url in video_urls_set:
313 video_urls_set.add(video_url)
315 (video_url, int_or_none(definition.get('quality'))))
317 thumbnail, duration = [None] * 2
319 def extract_js_vars(webpage, pattern, default=NO_DEFAULT):
320 assignments = self._search_regex(
321 pattern, webpage, 'encoded url', default=default)
325 assignments = assignments.split(';')
329 def parse_js_value(inp):
330 inp = re.sub(r'/\*(?:(?!\*/).)*?\*/', '', inp)
332 inps = inp.split('+')
333 return functools.reduce(
334 operator.concat, map(parse_js_value, inps))
338 return remove_quotes(inp)
340 for assn in assignments:
344 assn = re.sub(r'var\s+', '', assn)
345 vname, value = assn.split('=', 1)
346 js_vars[vname] = parse_js_value(value)
349 def add_video_url(video_url):
350 v_url = url_or_none(video_url)
353 if v_url in video_urls_set:
355 video_urls.append((v_url, None))
356 video_urls_set.add(v_url)
358 def parse_quality_items(quality_items):
359 q_items = self._parse_json(quality_items, video_id, fatal=False)
360 if not isinstance(q_items, list):
363 if isinstance(item, dict):
364 add_video_url(item.get('url'))
367 FORMAT_PREFIXES = ('media', 'quality', 'qualityItems')
368 js_vars = extract_js_vars(
369 webpage, r'(var\s+(?:%s)_.+)' % '|'.join(FORMAT_PREFIXES),
372 for key, format_url in js_vars.items():
373 if key.startswith(FORMAT_PREFIXES[-1]):
374 parse_quality_items(format_url)
375 elif any(key.startswith(p) for p in FORMAT_PREFIXES[:2]):
376 add_video_url(format_url)
377 if not video_urls and re.search(
378 r'<[^>]+\bid=["\']lockedPlayer
', webpage):
379 raise ExtractorError(
380 'Video
%s is locked
' % video_id, expected=True)
383 js_vars = extract_js_vars(
384 dl_webpage('tv
'), r'(var
.+?mediastring
.+?
)</script
>')
385 add_video_url(js_vars['mediastring
'])
387 for mobj in re.finditer(
388 r'<a
[^
>]+\bclass
=["\']downloadBtn\b[^>]+\bhref=(["\'])(?P
<url
>(?
:(?
!\
1).)+)\
1',
390 video_url = mobj.group('url
')
391 if video_url not in video_urls_set:
392 video_urls.append((video_url, None))
393 video_urls_set.add(video_url)
397 for video_url, height in video_urls:
399 upload_date = self._search_regex(
400 r'/(\d{6}
/\d{2}
)/', video_url, 'upload data
', default=None)
402 upload_date = upload_date.replace('/', '')
403 ext = determine_ext(video_url)
405 formats.extend(self._extract_mpd_formats(
406 video_url, video_id, mpd_id='dash
', fatal=False))
409 formats.extend(self._extract_m3u8_formats(
410 video_url, video_id, 'mp4
', entry_protocol='m3u8_native
',
411 m3u8_id='hls
', fatal=False))
414 mobj = re.search(r'(?P
<height
>\d
+)[pP
]?
_(?P
<tbr
>\d
+)[kK
]', video_url)
417 height = int(mobj.group('height
'))
418 tbr = int(mobj.group('tbr
'))
421 'format_id
': '%dp
' % height if height else None,
425 self._sort_formats(formats)
427 video_uploader = self._html_search_regex(
428 r'(?s
)From
: 
;.+?
<(?
:a
\b[^
>]+\bhref
=["\']/(?:(?:user|channel)s|model|pornstar)/|span\b[^>]+\bclass=["\']username
)[^
>]+>(.+?
)<',
429 webpage, 'uploader
', default=None)
431 def extract_vote_count(kind, name):
432 return self._extract_count(
433 (r'<span
[^
>]+\bclass
="votes%s"[^
>]*>([\d
,\
.]+)</span
>' % kind,
434 r'<span
[^
>]+\bclass
=["\']votes%s["\'][^
>]*\bdata
-rating
=["\'](\d+)' % kind),
437 view_count = self._extract_count(
438 r'<span class="count
">([\d,\.]+)</span> [Vv]iews', webpage, 'view')
439 like_count = extract_vote_count('Up', 'like')
440 dislike_count = extract_vote_count('Down', 'dislike')
441 comment_count = self._extract_count(
442 r'All Comments\s*<span>\(([\d,.]+)\)', webpage, 'comment')
444 def extract_list(meta_key):
445 div = self._search_regex(
446 r'(?s)<div[^>]+\bclass=["\'].*?
\b%sWrapper
[^
>]*>(.+?
)</div
>'
447 % meta_key, webpage, meta_key, default=None)
449 return re.findall(r'<a
[^
>]+\bhref
=[^
>]+>([^
<]+)', div)
451 info = self._search_json_ld(webpage, video_id, default={})
452 # description provided in JSON-LD is irrelevant
453 info['description
'] = None
457 'uploader
': video_uploader,
458 'upload_date
': upload_date,
460 'thumbnail
': thumbnail,
461 'duration
': duration,
462 'view_count
': view_count,
463 'like_count
': like_count,
464 'dislike_count
': dislike_count,
465 'comment_count
': comment_count,
468 'tags
': extract_list('tags
'),
469 'categories
': extract_list('categories
'),
470 'subtitles
': subtitles,
474 class PornHubPlaylistBaseIE(PornHubBaseIE):
475 def _extract_page(self, url):
476 return int_or_none(self._search_regex(
477 r'\bpage
=(\d
+)', url, 'page
', default=None))
479 def _extract_entries(self, webpage, host):
480 # Only process container div with main playlist content skipping
481 # drop-down menu that uses similar pattern for videos (see
482 # https://github.com/ytdl-org/youtube-dl/issues/11594).
483 container = self._search_regex(
484 r'(?s
)(<div
[^
>]+class=["\']container.+)', webpage,
485 'container', default=webpage)
489 'http://www.%s/%s' % (host, video_url),
490 PornHubIE.ie_key(), video_title=title)
491 for video_url, title in orderedSet(re.findall(
492 r'href="/?
(view_video\
.php
\?.*\bviewkey
=[\da
-z
]+[^
"]*)"[^
>]*\s
+title
="([^"]+)"',
497 class PornHubUserIE(PornHubPlaylistBaseIE):
498 _VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?(?P<host>pornhub(?:premium)?\.(?:com|net|org))/(?:(?:user|channel)s|model|pornstar)/(?P<id>[^/?#&]+))(?:[?#&]|/(?!videos)|$)'
500 'url': 'https://www.pornhub.com/model/zoe_ph',
501 'playlist_mincount': 118,
503 'url': 'https://www.pornhub.com/pornstar/liz-vicious',
507 'playlist_mincount': 118,
509 'url': 'https://www.pornhub.com/users/russianveet69',
510 'only_matching': True,
512 'url': 'https://www.pornhub.com/channels/povd',
513 'only_matching': True,
515 'url': 'https://www.pornhub.com/model/zoe_ph?abc=1',
516 'only_matching': True,
518 # Unavailable via /videos page, but available with direct pagination
519 # on pornstar page (see [1]), requires premium
520 # 1. https://github.com/ytdl-org/youtube-dl/issues/27853
521 'url': 'https://www.pornhubpremium.com/pornstar/sienna-west',
522 'only_matching': True,
524 # Same as before, multi page
525 'url': 'https://www.pornhubpremium.com/pornstar/lily-labeau',
526 'only_matching': True,
529 def _real_extract(self, url):
530 mobj = re.match(self._VALID_URL, url)
531 user_id = mobj.group('id')
532 videos_url = '%s/videos' % mobj.group('url')
533 page = self._extract_page(url)
535 videos_url = update_url_query(videos_url, {'page': page})
536 return self.url_result(
537 videos_url, ie=PornHubPagedVideoListIE.ie_key(), video_id=user_id)
540 class PornHubPagedPlaylistBaseIE(PornHubPlaylistBaseIE):
542 def _has_more(webpage):
545 <li[^>]+\bclass=["\']page_next|
546 <link
[^
>]+\brel
=["\']next|
547 <button[^>]+\bid=["\']moreDataBtn
548 ''', webpage) is not None
550 def _entries(self, url, host, item_id):
551 page = self._extract_page(url)
555 def download_page(base_url, num, fallback=False):
556 note = 'Downloading page %d%s' % (num, ' (switch to fallback)' if fallback else '')
557 return self._download_webpage(
558 base_url, item_id, note, query={'page': num})
561 return isinstance(e.cause, compat_HTTPError) and e.cause.code == 404
564 has_page = page is not None
565 first_page = page if has_page else 1
566 for page_num in (first_page, ) if has_page else itertools.count(first_page):
569 webpage = download_page(base_url, page_num)
570 except ExtractorError as e:
571 # Some sources may not be available via /videos page,
572 # trying to fallback to main page pagination (see [1])
573 # 1. https://github.com/ytdl-org/youtube-dl/issues/27853
574 if is_404(e) and page_num == first_page and VIDEOS in base_url:
575 base_url = base_url.replace(VIDEOS, '')
576 webpage = download_page(base_url, page_num, fallback=True)
579 except ExtractorError as e:
580 if is_404(e) and page_num != first_page:
583 page_entries = self._extract_entries(webpage, host)
586 for e in page_entries:
588 if not self._has_more(webpage):
591 def _real_extract(self, url):
592 mobj = re.match(self._VALID_URL, url)
593 host = mobj.group('host')
594 item_id = mobj.group('id')
598 return self.playlist_result(self._entries(url, host, item_id), item_id)
601 class PornHubPagedVideoListIE(PornHubPagedPlaylistBaseIE):
602 _VALID_URL = r'https?://(?:[^/]+\.)?(?P<host>pornhub(?:premium)?\.(?:com|net|org))/(?P<id>(?:[^/]+/)*[^/?#&]+)'
604 'url': 'https://www.pornhub.com/model/zoe_ph/videos',
605 'only_matching': True,
607 'url': 'http://www.pornhub.com/users/rushandlia/videos',
608 'only_matching': True,
610 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos',
612 'id': 'pornstar/jenny-blighe/videos',
614 'playlist_mincount': 149,
616 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos?page=3',
618 'id': 'pornstar/jenny-blighe/videos',
620 'playlist_mincount': 40,
622 # default sorting as Top Rated Videos
623 'url': 'https://www.pornhub.com/channels/povd/videos',
625 'id': 'channels/povd/videos',
627 'playlist_mincount': 293,
630 'url': 'https://www.pornhub.com/channels/povd/videos?o=ra',
631 'only_matching': True,
634 'url': 'https://www.pornhub.com/channels/povd/videos?o=da',
635 'only_matching': True,
638 'url': 'https://www.pornhub.com/channels/povd/videos?o=vi',
639 'only_matching': True,
641 'url': 'http://www.pornhub.com/users/zoe_ph/videos/public',
642 'only_matching': True,
645 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=mv',
646 'only_matching': True,
649 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=tr',
650 'only_matching': True,
653 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=lg',
654 'only_matching': True,
657 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=cm',
658 'only_matching': True,
660 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos/paid',
661 'only_matching': True,
663 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos/fanonly',
664 'only_matching': True,
666 'url': 'https://www.pornhub.com/video',
667 'only_matching': True,
669 'url': 'https://www.pornhub.com/video?page=3',
670 'only_matching': True,
672 'url': 'https://www.pornhub.com/video/search?search=123',
673 'only_matching': True,
675 'url': 'https://www.pornhub.com/categories/teen',
676 'only_matching': True,
678 'url': 'https://www.pornhub.com/categories/teen?page=3',
679 'only_matching': True,
681 'url': 'https://www.pornhub.com/hd',
682 'only_matching': True,
684 'url': 'https://www.pornhub.com/hd?page=3',
685 'only_matching': True,
687 'url': 'https://www.pornhub.com/described-video',
688 'only_matching': True,
690 'url': 'https://www.pornhub.com/described-video?page=2',
691 'only_matching': True,
693 'url': 'https://www.pornhub.com/video/incategories/60fps-1/hd-porn',
694 'only_matching': True,
696 'url': 'https://www.pornhub.com/playlist/44121572',
698 'id': 'playlist/44121572',
700 'playlist_mincount': 132,
702 'url': 'https://www.pornhub.com/playlist/4667351',
703 'only_matching': True,
705 'url': 'https://de.pornhub.com/playlist/4667351',
706 'only_matching': True,
710 def suitable(cls, url):
712 if PornHubIE.suitable(url) or PornHubUserIE.suitable(url) or PornHubUserVideosUploadIE.suitable(url)
713 else super(PornHubPagedVideoListIE, cls).suitable(url))
716 class PornHubUserVideosUploadIE(PornHubPagedPlaylistBaseIE):
717 _VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?(?P<host>pornhub(?:premium)?\.(?:com|net|org))/(?:(?:user|channel)s|model|pornstar)/(?P<id>[^/]+)/videos/upload)'
719 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos/upload',
721 'id': 'jenny-blighe',
723 'playlist_mincount': 129,
725 'url': 'https://www.pornhub.com/model/zoe_ph/videos/upload',
726 'only_matching': True,