8 from .common
import InfoExtractor
9 from .openload
import PhantomJSwrapper
10 from ..compat
import compat_HTTPError
, compat_str
29 class PornHubBaseIE(InfoExtractor
):
30 _NETRC_MACHINE
= 'pornhub'
31 _PORNHUB_HOST_RE
= r
'(?:(?P<host>pornhub(?:premium)?\.(?:com|net|org))|pornhubvybmsymdol4iibwgwtkpwmeyd6luq2gxajgjzfjvotyt5zhyd\.onion)'
33 def _download_webpage_handle(self
, *args
, **kwargs
):
34 def dl(*args
, **kwargs
):
35 return super(PornHubBaseIE
, self
)._download
_webpage
_handle
(*args
, **kwargs
)
37 ret
= dl(*args
, **kwargs
)
44 if any(re
.search(p
, webpage
) for p
in (
45 r
'<body\b[^>]+\bonload=["\']go\
(\
)',
46 r'document\
.cookie\s
*=\s
*["\']RNKEY=',
47 r'document\.location\.reload\(true\)')):
48 url_or_request = args[0]
49 url = (url_or_request.get_full_url()
50 if isinstance(url_or_request, urllib.request.Request)
52 phantom = PhantomJSwrapper(self, required_version='2.0')
53 phantom.get(url, html=webpage)
54 webpage, urlh = dl(*args, **kwargs)
58 def _real_initialize(self):
59 self._logged_in = False
61 def _set_age_cookies(self, host):
62 self._set_cookie(host, 'age_verified', '1')
63 self._set_cookie(host, 'accessAgeDisclaimerPH', '1')
64 self._set_cookie(host, 'accessPH', '1')
66 def _login(self, host):
70 site = host.split('.')[0]
72 # Both sites pornhub and pornhubpremium have separate accounts
73 # so there should be an option to provide credentials for both.
74 # At the same time some videos are available under the same video id
75 # on both sites so that we have to identify them as the same video.
76 # For that purpose we have to keep both in the same extractor
77 # but under different netrc machines.
78 username, password = self._get_login_info(netrc_machine=site)
82 login_url = 'https://www.%s/%slogin' % (host, 'premium/' if 'premium' in host else '')
83 login_page = self._download_webpage(
84 login_url, None, 'Downloading %s login page' % site)
86 def is_logged(webpage):
87 return any(re.search(p, webpage) for p in (
88 r'class=["\']signOut
',
89 r'>Sign\s
+[Oo
]ut\s
*<'))
91 if is_logged(login_page):
92 self._logged_in = True
95 login_form = self._hidden_inputs(login_page)
102 response = self._download_json(
103 'https
://www
.%s/front
/authenticate
' % host, None,
104 'Logging
in to
%s' % site,
105 data=urlencode_postdata(login_form),
107 'Content
-Type
': 'application
/x
-www
-form
-urlencoded
; charset
=UTF
-8',
108 'Referer
': login_url,
109 'X
-Requested
-With
': 'XMLHttpRequest
',
112 if response.get('success
') == '1':
113 self._logged_in = True
116 message = response.get('message
')
117 if message is not None:
118 raise ExtractorError(
119 'Unable to login
: %s' % message, expected=True)
121 raise ExtractorError('Unable to log
in')
124 class PornHubIE(PornHubBaseIE):
125 IE_DESC = 'PornHub
and Thumbzilla
'
126 _VALID_URL = r'''(?x)
131 /(?:(?:view_video\.php|video/show)\?viewkey=|embed/)|
132 (?:www\.)?thumbzilla\.com/video/
135 ''' % PornHubBaseIE._PORNHUB_HOST_RE
136 _EMBED_REGEX = [r'<iframe
[^
>]+?src
=["\'](?P<url>(?:https?:)?//(?:www\.)?pornhub(?:premium)?\.(?:com|net|org)/embed/[\da-z]+)']
138 'url': 'http://www.pornhub.com/view_video.php?viewkey=648719015',
139 'md5': 'a6391306d050e4547f62b3f485dd9ba9',
143 'title': 'Seductive Indian beauty strips down and fingers her pink pussy',
145 'upload_date': '20130628',
146 'timestamp': 1372447216,
150 'dislike_count': int,
151 'comment_count': int,
159 'url': 'http://www.pornhub.com/view_video.php?viewkey=1331683002',
164 'upload_date': '20150213',
165 'timestamp': 1423804862,
169 'dislike_count': int,
170 'comment_count': int,
176 'skip_download': True,
178 'skip': 'Video has been flagged for verification in accordance with our trust and safety policy',
181 'url': 'https://www.pornhub.com/view_video.php?viewkey=ph5af5fef7c2aa7',
183 'id': 'ph5af5fef7c2aa7',
185 'title': 'BFFS - Cute Teen Girls Share Cock On the Floor',
190 'dislike_count': int,
191 'comment_count': int,
202 'skip_download': True,
204 'skip': 'This video has been disabled',
206 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph601dc30bae19a',
208 'id': 'ph601dc30bae19a',
209 'uploader': 'Projekt Melody',
210 'uploader_id': 'projekt-melody',
211 'upload_date': '20210205',
212 'title': '"Welcome to My Pussy Mansion
" - CB Stream (02/03/21)',
213 'thumbnail': r're:https?://.+',
216 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph557bbb6676d2d',
217 'only_matching': True,
219 # removed at the request of cam4.com
220 'url': 'http://fr.pornhub.com/view_video.php?viewkey=ph55ca2f9760862',
221 'only_matching': True,
223 # removed at the request of the copyright owner
224 'url': 'http://www.pornhub.com/view_video.php?viewkey=788152859',
225 'only_matching': True,
227 # removed by uploader
228 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph572716d15a111',
229 'only_matching': True,
232 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph56fd731fce6b7',
233 'only_matching': True,
235 'url': 'https://www.thumbzilla.com/video/ph56c6114abd99a/horny-girlfriend-sex',
236 'only_matching': True,
238 'url': 'http://www.pornhub.com/video/show?viewkey=648719015',
239 'only_matching': True,
241 'url': 'https://www.pornhub.net/view_video.php?viewkey=203640933',
242 'only_matching': True,
244 'url': 'https://www.pornhub.org/view_video.php?viewkey=203640933',
245 'only_matching': True,
247 'url': 'https://www.pornhubpremium.com/view_video.php?viewkey=ph5e4acdae54a82',
248 'only_matching': True,
250 # Some videos are available with the same id on both premium
251 # and non-premium sites (e.g. this and the following test)
252 'url': 'https://www.pornhub.com/view_video.php?viewkey=ph5f75b0f4b18e3',
253 'only_matching': True,
255 'url': 'https://www.pornhubpremium.com/view_video.php?viewkey=ph5f75b0f4b18e3',
256 'only_matching': True,
259 'url': 'https://www.pornhub.com/view_video.php?viewkey=ph5a9813bfa7156',
260 'only_matching': True,
262 'url': 'http://pornhubvybmsymdol4iibwgwtkpwmeyd6luq2gxajgjzfjvotyt5zhyd.onion/view_video.php?viewkey=ph5a9813bfa7156',
263 'only_matching': True,
266 def _extract_count(self, pattern, webpage, name):
267 return str_to_int(self._search_regex(pattern, webpage, '%s count' % name, default=None))
269 def _real_extract(self, url):
270 mobj = self._match_valid_url(url)
271 host = mobj.group('host') or 'pornhub.com'
272 video_id = mobj.group('id')
275 self._set_age_cookies(host)
277 def dl_webpage(platform):
278 self._set_cookie(host, 'platform', platform)
279 return self._download_webpage(
280 'https://www.%s/view_video.php?viewkey=%s' % (host, video_id),
281 video_id, 'Downloading %s webpage' % platform)
283 webpage = dl_webpage('pc')
285 error_msg = self._html_search_regex(
286 (r'(?s)<div[^>]+class=(["\'])(?
:(?
!\
1).)*\b(?
:removed|userMessageSection
)\b(?
:(?
!\
1).)*\
1[^
>]*>(?P
<error
>.+?
)</div
>',
287 r'(?s
)<section
[^
>]+class=["\']noVideo["\'][^
>]*>(?P
<error
>.+?
)</section
>'),
288 webpage, 'error message
', default=None, group='error
')
290 error_msg = re.sub(r'\s
+', ' ', error_msg)
291 raise ExtractorError(
292 'PornHub said
: %s' % error_msg,
293 expected=True, video_id=video_id)
295 if any(re.search(p, webpage) for p in (
296 r'class=["\']geoBlocked["\']',
297 r'>\s
*This content
is unavailable
in your country
')):
298 self.raise_geo_restricted()
300 # video_title from flashvars contains whitespace instead of non-ASCII (see
301 # http://www.pornhub.com/view_video.php?viewkey=1331683002), not relying
303 title = self._html_search_meta(
304 'twitter
:title
', webpage, default=None) or self._html_search_regex(
305 (r'(?s
)<h1
[^
>]+class=["\']title["\'][^
>]*>(?P
<title
>.+?
)</h1
>',
306 r'<div
[^
>]+data
-video
-title
=(["\'])(?P<title>(?:(?!\1).)+)\1',
307 r'shareTitle["\']\s
*[=:]\s
*(["\'])(?P<title>(?:(?!\1).)+)\1'),
308 webpage, 'title', group='title')
311 video_urls_set = set()
314 flashvars = self._parse_json(
316 r'var\s+flashvars_\d+\s*=\s*({.+?});', webpage, 'flashvars', default='{}'),
319 subtitle_url = url_or_none(flashvars.get('closedCaptionsFile'))
321 subtitles.setdefault('en', []).append({
325 thumbnail = flashvars.get('image_url')
326 duration = int_or_none(flashvars.get('video_duration'))
327 media_definitions = flashvars.get('mediaDefinitions')
328 if isinstance(media_definitions, list):
329 for definition in media_definitions:
330 if not isinstance(definition, dict):
332 video_url = definition.get('videoUrl')
333 if not video_url or not isinstance(video_url, compat_str):
335 if video_url in video_urls_set:
337 video_urls_set.add(video_url)
339 (video_url, int_or_none(definition.get('quality'))))
341 thumbnail, duration = [None] * 2
343 def extract_js_vars(webpage, pattern, default=NO_DEFAULT):
344 assignments = self._search_regex(
345 pattern, webpage, 'encoded url', default=default)
349 assignments = assignments.split(';')
353 def parse_js_value(inp):
354 inp = re.sub(r'/\*(?:(?!\*/).)*?\*/', '', inp)
356 inps = inp.split('+')
357 return functools.reduce(
358 operator.concat, map(parse_js_value, inps))
362 return remove_quotes(inp)
364 for assn in assignments:
368 assn = re.sub(r'var\s+', '', assn)
369 vname, value = assn.split('=', 1)
370 js_vars[vname] = parse_js_value(value)
373 def add_video_url(video_url):
374 v_url = url_or_none(video_url)
377 if v_url in video_urls_set:
379 video_urls.append((v_url, None))
380 video_urls_set.add(v_url)
382 def parse_quality_items(quality_items):
383 q_items = self._parse_json(quality_items, video_id, fatal=False)
384 if not isinstance(q_items, list):
387 if isinstance(item, dict):
388 add_video_url(item.get('url'))
391 FORMAT_PREFIXES = ('media', 'quality', 'qualityItems')
392 js_vars = extract_js_vars(
393 webpage, r'(var\s+(?:%s)_.+)' % '|'.join(FORMAT_PREFIXES),
396 for key, format_url in js_vars.items():
397 if key.startswith(FORMAT_PREFIXES[-1]):
398 parse_quality_items(format_url)
399 elif any(key.startswith(p) for p in FORMAT_PREFIXES[:2]):
400 add_video_url(format_url)
401 if not video_urls and re.search(
402 r'<[^>]+\bid=["\']lockedPlayer
', webpage):
403 raise ExtractorError(
404 'Video
%s is locked
' % video_id, expected=True)
407 js_vars = extract_js_vars(
408 dl_webpage('tv
'), r'(var
.+?mediastring
.+?
)</script
>')
409 add_video_url(js_vars['mediastring
'])
411 for mobj in re.finditer(
412 r'<a
[^
>]+\bclass
=["\']downloadBtn\b[^>]+\bhref=(["\'])(?P
<url
>(?
:(?
!\
1).)+)\
1',
414 video_url = mobj.group('url
')
415 if video_url not in video_urls_set:
416 video_urls.append((video_url, None))
417 video_urls_set.add(video_url)
422 def add_format(format_url, height=None):
423 ext = determine_ext(format_url)
425 formats.extend(self._extract_mpd_formats(
426 format_url, video_id, mpd_id='dash
', fatal=False))
429 formats.extend(self._extract_m3u8_formats(
430 format_url, video_id, 'mp4
', entry_protocol='m3u8_native
',
431 m3u8_id='hls
', fatal=False))
434 height = int_or_none(self._search_regex(
435 r'(?P
<height
>\d
+)[pP
]?_\d
+[kK
]', format_url, 'height
',
439 'format_id
': format_field(height, None, '%dp
'),
443 for video_url, height in video_urls:
445 upload_date = self._search_regex(
446 r'/(\d{6}
/\d{2}
)/', video_url, 'upload data
', default=None)
448 upload_date = upload_date.replace('/', '')
449 if '/video
/get_media
' in video_url:
450 medias = self._download_json(video_url, video_id, fatal=False)
451 if isinstance(medias, list):
453 if not isinstance(media, dict):
455 video_url = url_or_none(media.get('videoUrl
'))
458 height = int_or_none(media.get('quality
'))
459 add_format(video_url, height)
461 add_format(video_url)
463 model_profile = self._search_json(
464 r'var\s
+MODEL_PROFILE\s
*=', webpage, 'model profile
', video_id, fatal=False)
465 video_uploader = self._html_search_regex(
466 r'(?s
)From
: 
;.+?
<(?
:a
\b[^
>]+\bhref
=["\']/(?:(?:user|channel)s|model|pornstar)/|span\b[^>]+\bclass=["\']username
)[^
>]+>(.+?
)<',
467 webpage, 'uploader
', default=None) or model_profile.get('username
')
469 def extract_vote_count(kind, name):
470 return self._extract_count(
471 (r'<span
[^
>]+\bclass
="votes%s"[^
>]*>([\d
,\
.]+)</span
>' % kind,
472 r'<span
[^
>]+\bclass
=["\']votes%s["\'][^
>]*\bdata
-rating
=["\'](\d+)' % kind),
475 view_count = self._extract_count(
476 r'<span class="count
">([\d,\.]+)</span> [Vv]iews', webpage, 'view')
477 like_count = extract_vote_count('Up', 'like')
478 dislike_count = extract_vote_count('Down', 'dislike')
479 comment_count = self._extract_count(
480 r'All Comments\s*<span>\(([\d,.]+)\)', webpage, 'comment')
482 def extract_list(meta_key):
483 div = self._search_regex(
484 r'(?s)<div[^>]+\bclass=["\'].*?
\b%sWrapper
[^
>]*>(.+?
)</div
>'
485 % meta_key, webpage, meta_key, default=None)
487 return [clean_html(x).strip() for x in re.findall(r'(?s
)<a
[^
>]+\bhref
=[^
>]+>.+?
</a
>', div)]
489 info = self._search_json_ld(webpage, video_id, default={})
490 # description provided in JSON-LD is irrelevant
491 info['description
'] = None
495 'uploader
': video_uploader,
496 'uploader_id
': remove_start(model_profile.get('modelProfileLink
'), '/model
/'),
497 'upload_date
': upload_date,
499 'thumbnail
': thumbnail,
500 'duration
': duration,
501 'view_count
': view_count,
502 'like_count
': like_count,
503 'dislike_count
': dislike_count,
504 'comment_count
': comment_count,
507 'tags
': extract_list('tags
'),
508 'categories
': extract_list('categories
'),
509 'cast
': extract_list('pornstars
'),
510 'subtitles
': subtitles,
514 class PornHubPlaylistBaseIE(PornHubBaseIE):
515 def _extract_page(self, url):
516 return int_or_none(self._search_regex(
517 r'\bpage
=(\d
+)', url, 'page
', default=None))
519 def _extract_entries(self, webpage, host):
520 # Only process container div with main playlist content skipping
521 # drop-down menu that uses similar pattern for videos (see
522 # https://github.com/ytdl-org/youtube-dl/issues/11594).
523 container = self._search_regex(
524 r'(?s
)(<div
[^
>]+class=["\']container.+)', webpage,
525 'container', default=webpage)
529 'http://www.%s/%s' % (host, video_url),
530 PornHubIE.ie_key(), video_title=title)
531 for video_url, title in orderedSet(re.findall(
532 r'href="/?
(view_video\
.php
\?.*\bviewkey
=[\da
-z
]+[^
"]*)"[^
>]*\s
+title
="([^"]+)"',
537 class PornHubUserIE(PornHubPlaylistBaseIE):
538 _VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?%s/(?:(?:user|channel)s|model|pornstar)/(?P<id>[^/?#&]+))(?:[?#&]|/(?!videos)|$)' % PornHubBaseIE._PORNHUB_HOST_RE
540 'url': 'https://www.pornhub.com/model/zoe_ph',
541 'playlist_mincount': 118,
543 'url': 'https://www.pornhub.com/pornstar/liz-vicious',
547 'playlist_mincount': 118,
549 'url': 'https://www.pornhub.com/users/russianveet69',
550 'only_matching': True,
552 'url': 'https://www.pornhub.com/channels/povd',
553 'only_matching': True,
555 'url': 'https://www.pornhub.com/model/zoe_ph?abc=1',
556 'only_matching': True,
558 # Unavailable via /videos page, but available with direct pagination
559 # on pornstar page (see [1]), requires premium
560 # 1. https://github.com/ytdl-org/youtube-dl/issues/27853
561 'url': 'https://www.pornhubpremium.com/pornstar/sienna-west',
562 'only_matching': True,
564 # Same as before, multi page
565 'url': 'https://www.pornhubpremium.com/pornstar/lily-labeau',
566 'only_matching': True,
568 'url': 'https://pornhubvybmsymdol4iibwgwtkpwmeyd6luq2gxajgjzfjvotyt5zhyd.onion/model/zoe_ph',
569 'only_matching': True,
572 def _real_extract(self, url):
573 mobj = self._match_valid_url(url)
574 user_id = mobj.group('id')
575 videos_url = '%s/videos' % mobj.group('url')
576 self._set_age_cookies(mobj.group('host'))
577 page = self._extract_page(url)
579 videos_url = update_url_query(videos_url, {'page': page})
580 return self.url_result(
581 videos_url, ie=PornHubPagedVideoListIE.ie_key(), video_id=user_id)
584 class PornHubPagedPlaylistBaseIE(PornHubPlaylistBaseIE):
586 def _has_more(webpage):
589 <li[^>]+\bclass=["\']page_next|
590 <link
[^
>]+\brel
=["\']next|
591 <button[^>]+\bid=["\']moreDataBtn
592 ''', webpage) is not None
594 def _entries(self, url, host, item_id):
595 page = self._extract_page(url)
599 def download_page(base_url, num, fallback=False):
600 note = 'Downloading page %d%s' % (num, ' (switch to fallback)' if fallback else '')
601 return self._download_webpage(
602 base_url, item_id, note, query={'page': num})
605 return isinstance(e.cause, compat_HTTPError) and e.cause.code == 404
608 has_page = page is not None
609 first_page = page if has_page else 1
610 for page_num in (first_page, ) if has_page else itertools.count(first_page):
613 webpage = download_page(base_url, page_num)
614 except ExtractorError as e:
615 # Some sources may not be available via /videos page,
616 # trying to fallback to main page pagination (see [1])
617 # 1. https://github.com/ytdl-org/youtube-dl/issues/27853
618 if is_404(e) and page_num == first_page and VIDEOS in base_url:
619 base_url = base_url.replace(VIDEOS, '')
620 webpage = download_page(base_url, page_num, fallback=True)
623 except ExtractorError as e:
624 if is_404(e) and page_num != first_page:
627 page_entries = self._extract_entries(webpage, host)
630 for e in page_entries:
632 if not self._has_more(webpage):
635 def _real_extract(self, url):
636 mobj = self._match_valid_url(url)
637 host = mobj.group('host')
638 item_id = mobj.group('id')
641 self._set_age_cookies(host)
643 return self.playlist_result(self._entries(url, host, item_id), item_id)
646 class PornHubPagedVideoListIE(PornHubPagedPlaylistBaseIE):
647 _VALID_URL = r'https?://(?:[^/]+\.)?%s/(?!playlist/)(?P<id>(?:[^/]+/)*[^/?#&]+)' % PornHubBaseIE._PORNHUB_HOST_RE
649 'url': 'https://www.pornhub.com/model/zoe_ph/videos',
650 'only_matching': True,
652 'url': 'http://www.pornhub.com/users/rushandlia/videos',
653 'only_matching': True,
655 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos',
657 'id': 'pornstar/jenny-blighe/videos',
659 'playlist_mincount': 149,
661 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos?page=3',
663 'id': 'pornstar/jenny-blighe/videos',
665 'playlist_mincount': 40,
667 # default sorting as Top Rated Videos
668 'url': 'https://www.pornhub.com/channels/povd/videos',
670 'id': 'channels/povd/videos',
672 'playlist_mincount': 293,
675 'url': 'https://www.pornhub.com/channels/povd/videos?o=ra',
676 'only_matching': True,
679 'url': 'https://www.pornhub.com/channels/povd/videos?o=da',
680 'only_matching': True,
683 'url': 'https://www.pornhub.com/channels/povd/videos?o=vi',
684 'only_matching': True,
686 'url': 'http://www.pornhub.com/users/zoe_ph/videos/public',
687 'only_matching': True,
690 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=mv',
691 'only_matching': True,
694 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=tr',
695 'only_matching': True,
698 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=lg',
699 'only_matching': True,
702 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=cm',
703 'only_matching': True,
705 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos/paid',
706 'only_matching': True,
708 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos/fanonly',
709 'only_matching': True,
711 'url': 'https://www.pornhub.com/video',
712 'only_matching': True,
714 'url': 'https://www.pornhub.com/video?page=3',
715 'only_matching': True,
717 'url': 'https://www.pornhub.com/video/search?search=123',
718 'only_matching': True,
720 'url': 'https://www.pornhub.com/categories/teen',
721 'only_matching': True,
723 'url': 'https://www.pornhub.com/categories/teen?page=3',
724 'only_matching': True,
726 'url': 'https://www.pornhub.com/hd',
727 'only_matching': True,
729 'url': 'https://www.pornhub.com/hd?page=3',
730 'only_matching': True,
732 'url': 'https://www.pornhub.com/described-video',
733 'only_matching': True,
735 'url': 'https://www.pornhub.com/described-video?page=2',
736 'only_matching': True,
738 'url': 'https://www.pornhub.com/video/incategories/60fps-1/hd-porn',
739 'only_matching': True,
741 'url': 'https://pornhubvybmsymdol4iibwgwtkpwmeyd6luq2gxajgjzfjvotyt5zhyd.onion/model/zoe_ph/videos',
742 'only_matching': True,
746 def suitable(cls, url):
748 if PornHubIE.suitable(url) or PornHubUserIE.suitable(url) or PornHubUserVideosUploadIE.suitable(url)
749 else super(PornHubPagedVideoListIE, cls).suitable(url))
752 class PornHubUserVideosUploadIE(PornHubPagedPlaylistBaseIE):
753 _VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?%s/(?:(?:user|channel)s|model|pornstar)/(?P<id>[^/]+)/videos/upload)' % PornHubBaseIE._PORNHUB_HOST_RE
755 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos/upload',
757 'id': 'jenny-blighe',
759 'playlist_mincount': 129,
761 'url': 'https://www.pornhub.com/model/zoe_ph/videos/upload',
762 'only_matching': True,
764 'url': 'http://pornhubvybmsymdol4iibwgwtkpwmeyd6luq2gxajgjzfjvotyt5zhyd.onion/pornstar/jenny-blighe/videos/upload',
765 'only_matching': True,
769 class PornHubPlaylistIE(PornHubPlaylistBaseIE):
770 _VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?%s/playlist/(?P<id>[^/?#&]+))' % PornHubBaseIE._PORNHUB_HOST_RE
772 'url': 'https://www.pornhub.com/playlist/44121572',
776 'playlist_count': 77,
778 'url': 'https://www.pornhub.com/playlist/4667351',
779 'only_matching': True,
781 'url': 'https://de.pornhub.com/playlist/4667351',
782 'only_matching': True,
784 'url': 'https://de.pornhub.com/playlist/4667351?page=2',
785 'only_matching': True,
788 def _entries(self, url, host, item_id):
789 webpage = self._download_webpage(url, item_id, 'Downloading page 1')
790 playlist_id = self._search_regex(r'var\s+playlistId\s*=\s*"([^"]+)"', webpage, 'playlist_id')
791 video_count = int_or_none(
792 self._search_regex(r'var\s+itemsCount\s*=\s*([0-9]+)\s*\|\|', webpage, 'video_count'))
793 token = self._search_regex(r'var\s+token\s*=\s*"([^"]+)"', webpage, 'token')
794 page_count = math.ceil((video_count - 36) / 40.) + 1
795 page_entries = self._extract_entries(webpage, host)
797 def download_page(page_num):
798 note = 'Downloading page {}'.format(page_num)
799 page_url = 'https://www.{}/playlist/viewChunked'.format(host)
800 return self._download_webpage(page_url, item_id, note, query={
806 for page_num in range(1, page_count + 1):
808 webpage = download_page(page_num)
809 page_entries = self._extract_entries(webpage, host)
812 for e in page_entries:
815 def _real_extract(self, url):
816 mobj = self._match_valid_url(url)
817 host = mobj.group('host')
818 item_id = mobj.group('id')
821 self._set_age_cookies(host)
823 return self.playlist_result(self._entries(mobj.group('url'), host, item_id), item_id)