2016-06-13 16:57:59 -05:00
|
|
|
# coding: utf-8
|
2014-01-07 03:25:34 -06:00
|
|
|
from __future__ import unicode_literals
|
|
|
|
|
2017-03-20 18:29:39 -05:00
|
|
|
import functools
|
2016-03-27 13:50:46 -05:00
|
|
|
import itertools
|
2017-03-20 18:29:39 -05:00
|
|
|
import operator
|
2013-10-26 18:04:22 -05:00
|
|
|
import re
|
|
|
|
|
|
|
|
from .common import InfoExtractor
|
2014-12-13 05:24:42 -06:00
|
|
|
from ..compat import (
|
2016-03-27 13:50:46 -05:00
|
|
|
compat_HTTPError,
|
2018-07-14 06:05:06 -05:00
|
|
|
compat_str,
|
2019-01-22 14:51:29 -06:00
|
|
|
compat_urllib_request,
|
2014-12-13 05:24:42 -06:00
|
|
|
)
|
2019-01-22 14:51:29 -06:00
|
|
|
from .openload import PhantomJSwrapper
|
2014-12-13 05:24:42 -06:00
|
|
|
from ..utils import (
|
2019-03-22 13:09:33 -05:00
|
|
|
determine_ext,
|
2015-01-22 16:48:58 -06:00
|
|
|
ExtractorError,
|
2016-02-19 09:42:46 -06:00
|
|
|
int_or_none,
|
2016-03-27 13:21:08 -05:00
|
|
|
orderedSet,
|
2017-03-21 13:59:27 -05:00
|
|
|
remove_quotes,
|
2014-03-22 09:29:01 -05:00
|
|
|
str_to_int,
|
2018-07-28 18:24:10 -05:00
|
|
|
url_or_none,
|
2013-10-26 18:04:22 -05:00
|
|
|
)
|
|
|
|
|
2014-01-07 03:25:34 -06:00
|
|
|
|
2019-01-22 15:12:06 -06:00
|
|
|
class PornHubBaseIE(InfoExtractor):
|
|
|
|
def _download_webpage_handle(self, *args, **kwargs):
|
|
|
|
def dl(*args, **kwargs):
|
|
|
|
return super(PornHubBaseIE, self)._download_webpage_handle(*args, **kwargs)
|
|
|
|
|
|
|
|
webpage, urlh = dl(*args, **kwargs)
|
|
|
|
|
|
|
|
if any(re.search(p, webpage) for p in (
|
|
|
|
r'<body\b[^>]+\bonload=["\']go\(\)',
|
|
|
|
r'document\.cookie\s*=\s*["\']RNKEY=',
|
|
|
|
r'document\.location\.reload\(true\)')):
|
|
|
|
url_or_request = args[0]
|
|
|
|
url = (url_or_request.get_full_url()
|
|
|
|
if isinstance(url_or_request, compat_urllib_request.Request)
|
|
|
|
else url_or_request)
|
|
|
|
phantom = PhantomJSwrapper(self, required_version='2.0')
|
|
|
|
phantom.get(url, html=webpage)
|
|
|
|
webpage, urlh = dl(*args, **kwargs)
|
|
|
|
|
|
|
|
return webpage, urlh
|
|
|
|
|
|
|
|
|
|
|
|
class PornHubIE(PornHubBaseIE):
|
2016-07-01 14:11:07 -05:00
|
|
|
IE_DESC = 'PornHub and Thumbzilla'
|
|
|
|
_VALID_URL = r'''(?x)
|
|
|
|
https?://
|
|
|
|
(?:
|
2018-12-01 03:45:51 -06:00
|
|
|
(?:[^/]+\.)?(?P<host>pornhub\.(?:com|net))/(?:(?:view_video\.php|video/show)\?viewkey=|embed/)|
|
2016-07-01 14:11:07 -05:00
|
|
|
(?:www\.)?thumbzilla\.com/video/
|
|
|
|
)
|
2016-11-06 08:52:00 -06:00
|
|
|
(?P<id>[\da-z]+)
|
2016-07-01 14:11:07 -05:00
|
|
|
'''
|
2015-06-18 11:26:17 -05:00
|
|
|
_TESTS = [{
|
2014-01-07 03:25:34 -06:00
|
|
|
'url': 'http://www.pornhub.com/view_video.php?viewkey=648719015',
|
2016-02-19 09:42:46 -06:00
|
|
|
'md5': '1e19b41231a02eba417839222ac9d58e',
|
2014-01-07 03:25:34 -06:00
|
|
|
'info_dict': {
|
2014-10-26 18:33:35 -05:00
|
|
|
'id': '648719015',
|
|
|
|
'ext': 'mp4',
|
2016-02-14 03:37:17 -06:00
|
|
|
'title': 'Seductive Indian beauty strips down and fingers her pink pussy',
|
2016-02-19 09:42:46 -06:00
|
|
|
'uploader': 'Babes',
|
2018-09-14 14:04:43 -05:00
|
|
|
'upload_date': '20130628',
|
2016-02-19 09:42:46 -06:00
|
|
|
'duration': 361,
|
|
|
|
'view_count': int,
|
|
|
|
'like_count': int,
|
|
|
|
'dislike_count': int,
|
|
|
|
'comment_count': int,
|
|
|
|
'age_limit': 18,
|
2016-09-11 06:22:51 -05:00
|
|
|
'tags': list,
|
|
|
|
'categories': list,
|
2016-06-13 16:57:59 -05:00
|
|
|
},
|
|
|
|
}, {
|
|
|
|
# non-ASCII title
|
|
|
|
'url': 'http://www.pornhub.com/view_video.php?viewkey=1331683002',
|
|
|
|
'info_dict': {
|
|
|
|
'id': '1331683002',
|
|
|
|
'ext': 'mp4',
|
|
|
|
'title': '重庆婷婷女王足交',
|
2018-07-14 06:05:06 -05:00
|
|
|
'uploader': 'Unknown',
|
2018-09-14 14:04:43 -05:00
|
|
|
'upload_date': '20150213',
|
2016-06-13 16:57:59 -05:00
|
|
|
'duration': 1753,
|
|
|
|
'view_count': int,
|
|
|
|
'like_count': int,
|
|
|
|
'dislike_count': int,
|
|
|
|
'comment_count': int,
|
|
|
|
'age_limit': 18,
|
2016-09-11 06:22:51 -05:00
|
|
|
'tags': list,
|
|
|
|
'categories': list,
|
2016-06-13 16:57:59 -05:00
|
|
|
},
|
|
|
|
'params': {
|
|
|
|
'skip_download': True,
|
|
|
|
},
|
2018-07-28 18:24:10 -05:00
|
|
|
}, {
|
|
|
|
# subtitles
|
|
|
|
'url': 'https://www.pornhub.com/view_video.php?viewkey=ph5af5fef7c2aa7',
|
|
|
|
'info_dict': {
|
|
|
|
'id': 'ph5af5fef7c2aa7',
|
|
|
|
'ext': 'mp4',
|
|
|
|
'title': 'BFFS - Cute Teen Girls Share Cock On the Floor',
|
|
|
|
'uploader': 'BFFs',
|
|
|
|
'duration': 622,
|
|
|
|
'view_count': int,
|
|
|
|
'like_count': int,
|
|
|
|
'dislike_count': int,
|
|
|
|
'comment_count': int,
|
|
|
|
'age_limit': 18,
|
|
|
|
'tags': list,
|
|
|
|
'categories': list,
|
|
|
|
'subtitles': {
|
|
|
|
'en': [{
|
|
|
|
"ext": 'srt'
|
|
|
|
}]
|
|
|
|
},
|
|
|
|
},
|
|
|
|
'params': {
|
|
|
|
'skip_download': True,
|
|
|
|
},
|
2015-06-18 11:26:17 -05:00
|
|
|
}, {
|
|
|
|
'url': 'http://www.pornhub.com/view_video.php?viewkey=ph557bbb6676d2d',
|
|
|
|
'only_matching': True,
|
2015-09-15 10:24:01 -05:00
|
|
|
}, {
|
2016-06-30 15:18:27 -05:00
|
|
|
# removed at the request of cam4.com
|
2015-09-15 10:24:01 -05:00
|
|
|
'url': 'http://fr.pornhub.com/view_video.php?viewkey=ph55ca2f9760862',
|
|
|
|
'only_matching': True,
|
2016-06-30 15:18:27 -05:00
|
|
|
}, {
|
|
|
|
# removed at the request of the copyright owner
|
|
|
|
'url': 'http://www.pornhub.com/view_video.php?viewkey=788152859',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
# removed by uploader
|
|
|
|
'url': 'http://www.pornhub.com/view_video.php?viewkey=ph572716d15a111',
|
|
|
|
'only_matching': True,
|
2016-07-03 15:27:00 -05:00
|
|
|
}, {
|
|
|
|
# private video
|
|
|
|
'url': 'http://www.pornhub.com/view_video.php?viewkey=ph56fd731fce6b7',
|
|
|
|
'only_matching': True,
|
2016-07-01 14:11:07 -05:00
|
|
|
}, {
|
|
|
|
'url': 'https://www.thumbzilla.com/video/ph56c6114abd99a/horny-girlfriend-sex',
|
|
|
|
'only_matching': True,
|
2017-05-05 14:46:37 -05:00
|
|
|
}, {
|
|
|
|
'url': 'http://www.pornhub.com/video/show?viewkey=648719015',
|
|
|
|
'only_matching': True,
|
2018-11-17 22:14:46 -06:00
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.net/view_video.php?viewkey=203640933',
|
|
|
|
'only_matching': True,
|
2015-06-18 11:26:17 -05:00
|
|
|
}]
|
2013-10-26 18:04:22 -05:00
|
|
|
|
2016-11-06 08:52:00 -06:00
|
|
|
@staticmethod
|
|
|
|
def _extract_urls(webpage):
|
|
|
|
return re.findall(
|
2018-12-01 03:45:51 -06:00
|
|
|
r'<iframe[^>]+?src=["\'](?P<url>(?:https?:)?//(?:www\.)?pornhub\.(?:com|net)/embed/[\da-z]+)',
|
2016-11-06 08:52:00 -06:00
|
|
|
webpage)
|
2015-06-12 16:36:16 -05:00
|
|
|
|
2014-03-22 09:29:01 -05:00
|
|
|
def _extract_count(self, pattern, webpage, name):
|
2015-03-30 08:41:04 -05:00
|
|
|
return str_to_int(self._search_regex(
|
|
|
|
pattern, webpage, '%s count' % name, fatal=False))
|
2014-03-22 09:29:01 -05:00
|
|
|
|
2013-10-26 18:04:22 -05:00
|
|
|
def _real_extract(self, url):
|
2018-12-01 03:45:51 -06:00
|
|
|
mobj = re.match(self._VALID_URL, url)
|
|
|
|
host = mobj.group('host') or 'pornhub.com'
|
|
|
|
video_id = mobj.group('id')
|
2018-03-10 10:14:16 -06:00
|
|
|
|
2018-12-01 03:45:51 -06:00
|
|
|
self._set_cookie(host, 'age_verified', '1')
|
2013-10-26 18:04:22 -05:00
|
|
|
|
2017-02-14 10:52:41 -06:00
|
|
|
def dl_webpage(platform):
|
2018-12-01 03:45:51 -06:00
|
|
|
self._set_cookie(host, 'platform', platform)
|
2017-02-14 10:52:41 -06:00
|
|
|
return self._download_webpage(
|
2019-05-22 15:56:54 -05:00
|
|
|
'https://www.%s/view_video.php?viewkey=%s' % (host, video_id),
|
2018-07-14 06:05:06 -05:00
|
|
|
video_id, 'Downloading %s webpage' % platform)
|
2017-02-14 10:52:41 -06:00
|
|
|
|
|
|
|
webpage = dl_webpage('pc')
|
2013-10-26 18:04:22 -05:00
|
|
|
|
2015-01-22 16:48:58 -06:00
|
|
|
error_msg = self._html_search_regex(
|
2016-07-22 09:24:09 -05:00
|
|
|
r'(?s)<div[^>]+class=(["\'])(?:(?!\1).)*\b(?:removed|userMessageSection)\b(?:(?!\1).)*\1[^>]*>(?P<error>.+?)</div>',
|
2016-06-30 15:14:23 -05:00
|
|
|
webpage, 'error message', default=None, group='error')
|
2015-01-22 16:48:58 -06:00
|
|
|
if error_msg:
|
|
|
|
error_msg = re.sub(r'\s+', ' ', error_msg)
|
|
|
|
raise ExtractorError(
|
|
|
|
'PornHub said: %s' % error_msg,
|
|
|
|
expected=True, video_id=video_id)
|
|
|
|
|
2016-06-13 16:57:59 -05:00
|
|
|
# video_title from flashvars contains whitespace instead of non-ASCII (see
|
|
|
|
# http://www.pornhub.com/view_video.php?viewkey=1331683002), not relying
|
|
|
|
# on that anymore.
|
2018-07-14 06:05:06 -05:00
|
|
|
title = self._html_search_meta(
|
2016-06-13 16:57:59 -05:00
|
|
|
'twitter:title', webpage, default=None) or self._search_regex(
|
|
|
|
(r'<h1[^>]+class=["\']title["\'][^>]*>(?P<title>[^<]+)',
|
|
|
|
r'<div[^>]+data-video-title=(["\'])(?P<title>.+?)\1',
|
|
|
|
r'shareTitle\s*=\s*(["\'])(?P<title>.+?)\1'),
|
|
|
|
webpage, 'title', group='title')
|
|
|
|
|
2018-07-14 06:05:06 -05:00
|
|
|
video_urls = []
|
|
|
|
video_urls_set = set()
|
2018-07-28 18:24:10 -05:00
|
|
|
subtitles = {}
|
2018-07-14 06:05:06 -05:00
|
|
|
|
2016-02-19 09:42:46 -06:00
|
|
|
flashvars = self._parse_json(
|
|
|
|
self._search_regex(
|
2016-03-27 14:21:44 -05:00
|
|
|
r'var\s+flashvars_\d+\s*=\s*({.+?});', webpage, 'flashvars', default='{}'),
|
2016-02-19 09:42:46 -06:00
|
|
|
video_id)
|
|
|
|
if flashvars:
|
2018-07-28 18:24:10 -05:00
|
|
|
subtitle_url = url_or_none(flashvars.get('closedCaptionsFile'))
|
|
|
|
if subtitle_url:
|
|
|
|
subtitles.setdefault('en', []).append({
|
|
|
|
'url': subtitle_url,
|
|
|
|
'ext': 'srt',
|
|
|
|
})
|
2016-02-19 09:42:46 -06:00
|
|
|
thumbnail = flashvars.get('image_url')
|
|
|
|
duration = int_or_none(flashvars.get('video_duration'))
|
2018-07-14 06:05:06 -05:00
|
|
|
media_definitions = flashvars.get('mediaDefinitions')
|
|
|
|
if isinstance(media_definitions, list):
|
|
|
|
for definition in media_definitions:
|
|
|
|
if not isinstance(definition, dict):
|
|
|
|
continue
|
|
|
|
video_url = definition.get('videoUrl')
|
|
|
|
if not video_url or not isinstance(video_url, compat_str):
|
|
|
|
continue
|
|
|
|
if video_url in video_urls_set:
|
|
|
|
continue
|
|
|
|
video_urls_set.add(video_url)
|
|
|
|
video_urls.append(
|
|
|
|
(video_url, int_or_none(definition.get('quality'))))
|
2016-02-19 09:42:46 -06:00
|
|
|
else:
|
2018-07-14 06:05:06 -05:00
|
|
|
thumbnail, duration = [None] * 2
|
|
|
|
|
|
|
|
if not video_urls:
|
|
|
|
tv_webpage = dl_webpage('tv')
|
|
|
|
|
|
|
|
assignments = self._search_regex(
|
|
|
|
r'(var.+?mediastring.+?)</script>', tv_webpage,
|
|
|
|
'encoded url').split(';')
|
|
|
|
|
|
|
|
js_vars = {}
|
|
|
|
|
|
|
|
def parse_js_value(inp):
|
|
|
|
inp = re.sub(r'/\*(?:(?!\*/).)*?\*/', '', inp)
|
|
|
|
if '+' in inp:
|
|
|
|
inps = inp.split('+')
|
|
|
|
return functools.reduce(
|
|
|
|
operator.concat, map(parse_js_value, inps))
|
|
|
|
inp = inp.strip()
|
|
|
|
if inp in js_vars:
|
|
|
|
return js_vars[inp]
|
|
|
|
return remove_quotes(inp)
|
|
|
|
|
|
|
|
for assn in assignments:
|
|
|
|
assn = assn.strip()
|
|
|
|
if not assn:
|
|
|
|
continue
|
|
|
|
assn = re.sub(r'var\s+', '', assn)
|
|
|
|
vname, value = assn.split('=', 1)
|
|
|
|
js_vars[vname] = parse_js_value(value)
|
|
|
|
|
|
|
|
video_url = js_vars['mediastring']
|
|
|
|
if video_url not in video_urls_set:
|
|
|
|
video_urls.append((video_url, None))
|
|
|
|
video_urls_set.add(video_url)
|
|
|
|
|
|
|
|
for mobj in re.finditer(
|
|
|
|
r'<a[^>]+\bclass=["\']downloadBtn\b[^>]+\bhref=(["\'])(?P<url>(?:(?!\1).)+)\1',
|
|
|
|
webpage):
|
|
|
|
video_url = mobj.group('url')
|
|
|
|
if video_url not in video_urls_set:
|
|
|
|
video_urls.append((video_url, None))
|
|
|
|
video_urls_set.add(video_url)
|
|
|
|
|
2018-09-14 14:04:43 -05:00
|
|
|
upload_date = None
|
2018-07-14 06:05:06 -05:00
|
|
|
formats = []
|
|
|
|
for video_url, height in video_urls:
|
2018-09-14 14:04:43 -05:00
|
|
|
if not upload_date:
|
|
|
|
upload_date = self._search_regex(
|
|
|
|
r'/(\d{6}/\d{2})/', video_url, 'upload data', default=None)
|
|
|
|
if upload_date:
|
|
|
|
upload_date = upload_date.replace('/', '')
|
2019-03-22 13:09:33 -05:00
|
|
|
if determine_ext(video_url) == 'mpd':
|
|
|
|
formats.extend(self._extract_mpd_formats(
|
|
|
|
video_url, video_id, mpd_id='dash', fatal=False))
|
|
|
|
continue
|
2018-07-14 06:05:06 -05:00
|
|
|
tbr = None
|
|
|
|
mobj = re.search(r'(?P<height>\d+)[pP]?_(?P<tbr>\d+)[kK]', video_url)
|
|
|
|
if mobj:
|
|
|
|
if not height:
|
|
|
|
height = int(mobj.group('height'))
|
|
|
|
tbr = int(mobj.group('tbr'))
|
|
|
|
formats.append({
|
|
|
|
'url': video_url,
|
|
|
|
'format_id': '%dp' % height if height else None,
|
|
|
|
'height': height,
|
|
|
|
'tbr': tbr,
|
|
|
|
})
|
|
|
|
self._sort_formats(formats)
|
2016-02-19 09:42:46 -06:00
|
|
|
|
2014-03-22 09:29:01 -05:00
|
|
|
video_uploader = self._html_search_regex(
|
2018-09-05 12:51:20 -05:00
|
|
|
r'(?s)From: .+?<(?:a\b[^>]+\bhref=["\']/(?:(?:user|channel)s|model|pornstar)/|span\b[^>]+\bclass=["\']username)[^>]+>(.+?)<',
|
2014-03-22 09:29:01 -05:00
|
|
|
webpage, 'uploader', fatal=False)
|
2013-10-26 18:04:22 -05:00
|
|
|
|
2015-03-30 08:41:04 -05:00
|
|
|
view_count = self._extract_count(
|
|
|
|
r'<span class="count">([\d,\.]+)</span> views', webpage, 'view')
|
|
|
|
like_count = self._extract_count(
|
|
|
|
r'<span class="votesUp">([\d,\.]+)</span>', webpage, 'like')
|
|
|
|
dislike_count = self._extract_count(
|
|
|
|
r'<span class="votesDown">([\d,\.]+)</span>', webpage, 'dislike')
|
2014-03-22 09:29:01 -05:00
|
|
|
comment_count = self._extract_count(
|
2015-03-30 08:41:04 -05:00
|
|
|
r'All Comments\s*<span>\(([\d,.]+)\)', webpage, 'comment')
|
2014-03-22 09:29:01 -05:00
|
|
|
|
2019-02-05 10:06:55 -06:00
|
|
|
def extract_list(meta_key):
|
2019-02-04 11:33:54 -06:00
|
|
|
div = self._search_regex(
|
2019-02-05 10:06:55 -06:00
|
|
|
r'(?s)<div[^>]+\bclass=["\'].*?\b%sWrapper[^>]*>(.+?)</div>'
|
|
|
|
% meta_key, webpage, meta_key, default=None)
|
2019-02-04 11:33:54 -06:00
|
|
|
if div:
|
2019-02-05 10:06:55 -06:00
|
|
|
return re.findall(r'<a[^>]+\bhref=[^>]+>([^<]+)', div)
|
2016-09-11 06:22:51 -05:00
|
|
|
|
2013-10-26 18:04:22 -05:00
|
|
|
return {
|
|
|
|
'id': video_id,
|
|
|
|
'uploader': video_uploader,
|
2018-09-14 14:04:43 -05:00
|
|
|
'upload_date': upload_date,
|
2016-06-13 16:57:59 -05:00
|
|
|
'title': title,
|
2013-10-26 18:04:22 -05:00
|
|
|
'thumbnail': thumbnail,
|
2016-02-19 09:42:46 -06:00
|
|
|
'duration': duration,
|
2014-03-22 09:29:01 -05:00
|
|
|
'view_count': view_count,
|
|
|
|
'like_count': like_count,
|
|
|
|
'dislike_count': dislike_count,
|
|
|
|
'comment_count': comment_count,
|
2018-07-14 06:05:06 -05:00
|
|
|
'formats': formats,
|
2013-10-28 00:50:17 -05:00
|
|
|
'age_limit': 18,
|
2019-02-05 10:06:55 -06:00
|
|
|
'tags': extract_list('tags'),
|
|
|
|
'categories': extract_list('categories'),
|
2018-07-28 18:24:10 -05:00
|
|
|
'subtitles': subtitles,
|
2013-10-26 18:04:22 -05:00
|
|
|
}
|
2015-02-19 10:15:19 -06:00
|
|
|
|
|
|
|
|
2019-01-22 15:12:06 -06:00
|
|
|
class PornHubPlaylistBaseIE(PornHubBaseIE):
|
2018-12-01 03:45:51 -06:00
|
|
|
def _extract_entries(self, webpage, host):
|
2017-08-12 19:53:02 -05:00
|
|
|
# Only process container div with main playlist content skipping
|
|
|
|
# drop-down menu that uses similar pattern for videos (see
|
2019-03-09 06:14:41 -06:00
|
|
|
# https://github.com/ytdl-org/youtube-dl/issues/11594).
|
2017-08-12 19:53:02 -05:00
|
|
|
container = self._search_regex(
|
|
|
|
r'(?s)(<div[^>]+class=["\']container.+)', webpage,
|
|
|
|
'container', default=webpage)
|
|
|
|
|
2016-02-18 10:29:17 -06:00
|
|
|
return [
|
2016-03-27 13:32:57 -05:00
|
|
|
self.url_result(
|
2018-12-01 03:45:51 -06:00
|
|
|
'http://www.%s/%s' % (host, video_url),
|
2016-03-27 13:32:57 -05:00
|
|
|
PornHubIE.ie_key(), video_title=title)
|
|
|
|
for video_url, title in orderedSet(re.findall(
|
|
|
|
r'href="/?(view_video\.php\?.*\bviewkey=[\da-z]+[^"]*)"[^>]*\s+title="([^"]+)"',
|
2017-08-12 19:53:02 -05:00
|
|
|
container))
|
2016-02-18 10:29:17 -06:00
|
|
|
]
|
2015-02-19 10:15:19 -06:00
|
|
|
|
|
|
|
def _real_extract(self, url):
|
2018-12-01 03:45:51 -06:00
|
|
|
mobj = re.match(self._VALID_URL, url)
|
|
|
|
host = mobj.group('host')
|
|
|
|
playlist_id = mobj.group('id')
|
2015-02-19 10:15:19 -06:00
|
|
|
|
|
|
|
webpage = self._download_webpage(url, playlist_id)
|
|
|
|
|
2018-12-01 03:45:51 -06:00
|
|
|
entries = self._extract_entries(webpage, host)
|
2015-02-19 10:15:19 -06:00
|
|
|
|
|
|
|
playlist = self._parse_json(
|
|
|
|
self._search_regex(
|
2017-06-04 03:54:19 -05:00
|
|
|
r'(?:playlistObject|PLAYLIST_VIEW)\s*=\s*({.+?});', webpage,
|
|
|
|
'playlist', default='{}'),
|
|
|
|
playlist_id, fatal=False)
|
|
|
|
title = playlist.get('title') or self._search_regex(
|
|
|
|
r'>Videos\s+in\s+(.+?)\s+[Pp]laylist<', webpage, 'title', fatal=False)
|
2015-02-19 10:15:19 -06:00
|
|
|
|
|
|
|
return self.playlist_result(
|
2017-06-04 03:54:19 -05:00
|
|
|
entries, playlist_id, title, playlist.get('description'))
|
2016-02-18 10:29:17 -06:00
|
|
|
|
|
|
|
|
2019-06-21 17:34:46 -05:00
|
|
|
class PornHubUserIE(PornHubPlaylistBaseIE):
|
2019-06-21 20:37:07 -05:00
|
|
|
_VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?pornhub\.(?:com|net)/(?:(?:user|channel)s|model|pornstar)/(?P<id>[^/?#&]+))(?:[?#&]|/(?!videos)|$)'
|
2016-02-18 10:29:17 -06:00
|
|
|
_TESTS = [{
|
2019-06-21 17:34:46 -05:00
|
|
|
'url': 'https://www.pornhub.com/model/zoe_ph',
|
|
|
|
'playlist_mincount': 118,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/pornstar/liz-vicious',
|
2016-02-18 10:29:17 -06:00
|
|
|
'info_dict': {
|
2019-06-21 17:34:46 -05:00
|
|
|
'id': 'liz-vicious',
|
2016-02-18 10:29:17 -06:00
|
|
|
},
|
2019-06-21 17:34:46 -05:00
|
|
|
'playlist_mincount': 118,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/users/russianveet69',
|
2019-06-21 20:37:07 -05:00
|
|
|
'only_matching': True,
|
2019-06-21 17:34:46 -05:00
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/channels/povd',
|
2019-06-21 20:37:07 -05:00
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/model/zoe_ph?abc=1',
|
|
|
|
'only_matching': True,
|
2019-06-21 17:34:46 -05:00
|
|
|
}]
|
|
|
|
|
|
|
|
def _real_extract(self, url):
|
|
|
|
mobj = re.match(self._VALID_URL, url)
|
|
|
|
user_id = mobj.group('id')
|
|
|
|
return self.url_result(
|
2019-06-21 20:37:07 -05:00
|
|
|
'%s/videos' % mobj.group('url'), ie=PornHubPagedVideoListIE.ie_key(),
|
2019-06-21 17:34:46 -05:00
|
|
|
video_id=user_id)
|
|
|
|
|
|
|
|
|
|
|
|
class PornHubPagedPlaylistBaseIE(PornHubPlaylistBaseIE):
|
|
|
|
def _real_extract(self, url):
|
|
|
|
mobj = re.match(self._VALID_URL, url)
|
|
|
|
host = mobj.group('host')
|
2019-06-21 20:37:07 -05:00
|
|
|
item_id = mobj.group('id')
|
2019-06-21 17:34:46 -05:00
|
|
|
|
2019-06-21 18:01:43 -05:00
|
|
|
page = int_or_none(self._search_regex(
|
|
|
|
r'\bpage=(\d+)', url, 'page', default=None))
|
|
|
|
|
2019-06-21 17:34:46 -05:00
|
|
|
page_url = self._make_page_url(url)
|
|
|
|
|
|
|
|
entries = []
|
2019-06-21 18:01:43 -05:00
|
|
|
for page_num in (page, ) if page is not None else itertools.count(1):
|
2019-06-21 17:34:46 -05:00
|
|
|
try:
|
|
|
|
webpage = self._download_webpage(
|
2019-06-21 20:37:07 -05:00
|
|
|
page_url, item_id, 'Downloading page %d' % page_num,
|
2019-06-21 17:34:46 -05:00
|
|
|
query={'page': page_num})
|
|
|
|
except ExtractorError as e:
|
|
|
|
if isinstance(e.cause, compat_HTTPError) and e.cause.code == 404:
|
|
|
|
break
|
|
|
|
raise
|
|
|
|
page_entries = self._extract_entries(webpage, host)
|
|
|
|
if not page_entries:
|
|
|
|
break
|
|
|
|
entries.extend(page_entries)
|
|
|
|
if not self._has_more(webpage):
|
|
|
|
break
|
|
|
|
|
2019-06-21 20:37:07 -05:00
|
|
|
return self.playlist_result(orderedSet(entries), item_id)
|
2019-06-21 17:34:46 -05:00
|
|
|
|
|
|
|
|
2019-06-21 20:37:07 -05:00
|
|
|
class PornHubPagedVideoListIE(PornHubPagedPlaylistBaseIE):
|
|
|
|
_VALID_URL = r'https?://(?:[^/]+\.)?(?P<host>pornhub\.(?:com|net))/(?P<id>(?:[^/]+/)*[^/?#&]+)'
|
2019-06-21 17:34:46 -05:00
|
|
|
_TESTS = [{
|
2019-06-21 18:01:43 -05:00
|
|
|
'url': 'https://www.pornhub.com/model/zoe_ph/videos',
|
2019-06-21 17:34:46 -05:00
|
|
|
'only_matching': True,
|
2016-03-27 13:50:46 -05:00
|
|
|
}, {
|
|
|
|
'url': 'http://www.pornhub.com/users/rushandlia/videos',
|
|
|
|
'only_matching': True,
|
2019-06-21 17:34:46 -05:00
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos',
|
|
|
|
'info_dict': {
|
2019-06-21 20:37:07 -05:00
|
|
|
'id': 'pornstar/jenny-blighe/videos',
|
2019-06-21 17:34:46 -05:00
|
|
|
},
|
|
|
|
'playlist_mincount': 149,
|
2019-06-21 18:01:43 -05:00
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos?page=3',
|
|
|
|
'info_dict': {
|
2019-06-21 20:37:07 -05:00
|
|
|
'id': 'pornstar/jenny-blighe/videos',
|
2019-06-21 18:01:43 -05:00
|
|
|
},
|
|
|
|
'playlist_mincount': 40,
|
2018-02-16 12:17:06 -06:00
|
|
|
}, {
|
|
|
|
# default sorting as Top Rated Videos
|
|
|
|
'url': 'https://www.pornhub.com/channels/povd/videos',
|
|
|
|
'info_dict': {
|
2019-06-21 20:37:07 -05:00
|
|
|
'id': 'channels/povd/videos',
|
2018-02-16 12:17:06 -06:00
|
|
|
},
|
|
|
|
'playlist_mincount': 293,
|
|
|
|
}, {
|
|
|
|
# Top Rated Videos
|
|
|
|
'url': 'https://www.pornhub.com/channels/povd/videos?o=ra',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
# Most Recent Videos
|
|
|
|
'url': 'https://www.pornhub.com/channels/povd/videos?o=da',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
# Most Viewed Videos
|
|
|
|
'url': 'https://www.pornhub.com/channels/povd/videos?o=vi',
|
|
|
|
'only_matching': True,
|
2018-04-11 16:53:45 -05:00
|
|
|
}, {
|
|
|
|
'url': 'http://www.pornhub.com/users/zoe_ph/videos/public',
|
|
|
|
'only_matching': True,
|
2019-06-21 17:34:46 -05:00
|
|
|
}, {
|
|
|
|
# Most Viewed Videos
|
|
|
|
'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=mv',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
# Top Rated Videos
|
|
|
|
'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=tr',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
# Longest Videos
|
|
|
|
'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=lg',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
# Newest Videos
|
|
|
|
'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=cm',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos/paid',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos/fanonly',
|
|
|
|
'only_matching': True,
|
2019-06-21 20:37:07 -05:00
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/video',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/video?page=3',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/video/search?search=123',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/categories/teen',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/categories/teen?page=3',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/hd',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/hd?page=3',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/described-video',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/described-video?page=2',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/video/incategories/60fps-1/hd-porn',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/playlist/44121572',
|
|
|
|
'info_dict': {
|
|
|
|
'id': 'playlist/44121572',
|
|
|
|
},
|
|
|
|
'playlist_mincount': 132,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/playlist/4667351',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'https://de.pornhub.com/playlist/4667351',
|
|
|
|
'only_matching': True,
|
2016-02-18 10:29:17 -06:00
|
|
|
}]
|
|
|
|
|
2019-06-21 17:34:46 -05:00
|
|
|
@classmethod
|
|
|
|
def suitable(cls, url):
|
|
|
|
return (False
|
2019-06-21 20:37:07 -05:00
|
|
|
if PornHubIE.suitable(url) or PornHubUserIE.suitable(url) or PornHubUserVideosUploadIE.suitable(url)
|
|
|
|
else super(PornHubPagedVideoListIE, cls).suitable(url))
|
2019-06-21 17:34:46 -05:00
|
|
|
|
|
|
|
def _make_page_url(self, url):
|
|
|
|
return url
|
|
|
|
|
|
|
|
@staticmethod
|
|
|
|
def _has_more(webpage):
|
|
|
|
return re.search(
|
|
|
|
r'''(?x)
|
|
|
|
<li[^>]+\bclass=["\']page_next|
|
|
|
|
<link[^>]+\brel=["\']next|
|
|
|
|
<button[^>]+\bid=["\']moreDataBtn
|
|
|
|
''', webpage) is not None
|
2016-02-18 10:29:17 -06:00
|
|
|
|
2016-03-27 13:50:46 -05:00
|
|
|
|
2019-06-21 17:34:46 -05:00
|
|
|
class PornHubUserVideosUploadIE(PornHubPagedPlaylistBaseIE):
|
|
|
|
_VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?(?P<host>pornhub\.(?:com|net))/(?:(?:user|channel)s|model|pornstar)/(?P<id>[^/]+)/videos/upload)'
|
|
|
|
_TESTS = [{
|
|
|
|
'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos/upload',
|
|
|
|
'info_dict': {
|
|
|
|
'id': 'jenny-blighe',
|
|
|
|
},
|
|
|
|
'playlist_mincount': 129,
|
|
|
|
}, {
|
|
|
|
'url': 'https://www.pornhub.com/model/zoe_ph/videos/upload',
|
|
|
|
'only_matching': True,
|
|
|
|
}]
|
|
|
|
|
|
|
|
def _make_page_url(self, url):
|
|
|
|
mobj = re.match(self._VALID_URL, url)
|
|
|
|
return '%s/ajax' % mobj.group('url')
|
|
|
|
|
|
|
|
@staticmethod
|
|
|
|
def _has_more(webpage):
|
|
|
|
return True
|