-# coding: utf-8
-from __future__ import unicode_literals
-
import functools
import itertools
import math
import re
from .common import InfoExtractor
-from ..compat import (
- compat_HTTPError,
- compat_str,
- compat_urllib_request,
-)
from .openload import PhantomJSwrapper
+from ..compat import compat_str
+from ..networking import Request
+from ..networking.exceptions import HTTPError
from ..utils import (
+ NO_DEFAULT,
+ ExtractorError,
clean_html,
determine_ext,
- ExtractorError,
format_field,
int_or_none,
merge_dicts,
- NO_DEFAULT,
orderedSet,
remove_quotes,
+ remove_start,
str_to_int,
update_url_query,
- urlencode_postdata,
url_or_none,
+ urlencode_postdata,
)
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)
+ url = (url_or_request.url
+ if isinstance(url_or_request, Request)
else url_or_request)
phantom = PhantomJSwrapper(self, required_version='2.0')
phantom.get(url, html=webpage)
def _real_initialize(self):
self._logged_in = False
+ def _set_age_cookies(self, host):
+ self._set_cookie(host, 'age_verified', '1')
+ self._set_cookie(host, 'accessAgeDisclaimerPH', '1')
+ self._set_cookie(host, 'accessAgeDisclaimerUK', '1')
+ self._set_cookie(host, 'accessPH', '1')
+
def _login(self, host):
if self._logged_in:
return
def is_logged(webpage):
return any(re.search(p, webpage) for p in (
- r'class=["\']signOut',
- r'>Sign\s+[Oo]ut\s*<'))
+ r'id="profileMenuDropdown"',
+ r'class="ph-icon-logout"'))
if is_logged(login_page):
self._logged_in = True
login_form = self._hidden_inputs(login_page)
login_form.update({
- 'username': username,
+ 'email': username,
'password': password,
})
)
(?P<id>[\da-z]+)
''' % PornHubBaseIE._PORNHUB_HOST_RE
+ _EMBED_REGEX = [r'<iframe[^>]+?src=["\'](?P<url>(?:https?:)?//(?:www\.)?pornhub(?:premium)?\.(?:com|net|org)/embed/[\da-z]+)']
_TESTS = [{
'url': 'http://www.pornhub.com/view_video.php?viewkey=648719015',
'md5': 'a6391306d050e4547f62b3f485dd9ba9',
'skip_download': True,
},
'skip': 'This video has been disabled',
+ }, {
+ 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph601dc30bae19a',
+ 'info_dict': {
+ 'id': 'ph601dc30bae19a',
+ 'uploader': 'Projekt Melody',
+ 'uploader_id': 'projekt-melody',
+ 'upload_date': '20210205',
+ 'title': '"Welcome to My Pussy Mansion" - CB Stream (02/03/21)',
+ 'thumbnail': r're:https?://.+',
+ },
}, {
'url': 'http://www.pornhub.com/view_video.php?viewkey=ph557bbb6676d2d',
'only_matching': True,
'only_matching': True,
}]
- @staticmethod
- def _extract_urls(webpage):
- return re.findall(
- r'<iframe[^>]+?src=["\'](?P<url>(?:https?:)?//(?:www\.)?pornhub(?:premium)?\.(?:com|net|org)/embed/[\da-z]+)',
- webpage)
-
def _extract_count(self, pattern, webpage, name):
return str_to_int(self._search_regex(pattern, webpage, '%s count' % name, default=None))
video_id = mobj.group('id')
self._login(host)
-
- self._set_cookie(host, 'age_verified', '1')
+ self._set_age_cookies(host)
def dl_webpage(platform):
self._set_cookie(host, 'platform', platform)
default=None))
formats.append({
'url': format_url,
- 'format_id': format_field(height, template='%dp'),
+ 'format_id': format_field(height, None, '%dp'),
'height': height,
})
continue
add_format(video_url)
- # field_preference is unnecessary here, but kept for code-similarity with youtube-dl
- self._sort_formats(
- formats, field_preference=('height', 'width', 'fps', 'format_id'))
-
+ model_profile = self._search_json(
+ r'var\s+MODEL_PROFILE\s*=', webpage, 'model profile', video_id, fatal=False)
video_uploader = self._html_search_regex(
r'(?s)From: .+?<(?:a\b[^>]+\bhref=["\']/(?:(?:user|channel)s|model|pornstar)/|span\b[^>]+\bclass=["\']username)[^>]+>(.+?)<',
- webpage, 'uploader', default=None)
+ webpage, 'uploader', default=None) or model_profile.get('username')
def extract_vote_count(kind, name):
return self._extract_count(
return merge_dicts({
'id': video_id,
'uploader': video_uploader,
+ 'uploader_id': remove_start(model_profile.get('modelProfileLink'), '/model/'),
'upload_date': upload_date,
'title': title,
'thumbnail': thumbnail,
mobj = self._match_valid_url(url)
user_id = mobj.group('id')
videos_url = '%s/videos' % mobj.group('url')
+ self._set_age_cookies(mobj.group('host'))
page = self._extract_page(url)
if page:
videos_url = update_url_query(videos_url, {'page': page})
base_url, item_id, note, query={'page': num})
def is_404(e):
- return isinstance(e.cause, compat_HTTPError) and e.cause.code == 404
+ return isinstance(e.cause, HTTPError) and e.cause.status == 404
base_url = url
has_page = page is not None
item_id = mobj.group('id')
self._login(host)
+ self._set_age_cookies(host)
return self.playlist_result(self._entries(url, host, item_id), item_id)
item_id = mobj.group('id')
self._login(host)
+ self._set_age_cookies(host)
return self.playlist_result(self._entries(mobj.group('url'), host, item_id), item_id)