[YouPorn] Incorporate yt-dlp PR 8827

* from https://github.com/yt-dlp/yt-dlp/pull/8827
* extract from webpage instead of broken API URL
* thx The-MAGI
This commit is contained in:
dirkf 2024-03-01 14:48:30 +00:00
parent 820fae3b3a
commit e1a57763d7

View File

@ -8,6 +8,9 @@ from ..utils import (
extract_attributes, extract_attributes,
int_or_none, int_or_none,
str_to_int, str_to_int,
merge_dicts,
T,
traverse_obj,
unified_strdate, unified_strdate,
url_or_none, url_or_none,
) )
@ -15,6 +18,7 @@ from ..utils import (
class YouPornIE(InfoExtractor): class YouPornIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?youporn\.com/(?:watch|embed)/(?P<id>\d+)(?:/(?P<display_id>[^/?#&]+))?' _VALID_URL = r'https?://(?:www\.)?youporn\.com/(?:watch|embed)/(?P<id>\d+)(?:/(?P<display_id>[^/?#&]+))?'
_EMBED_REGEX = [r'<iframe[^>]+\bsrc=["\'](?P<url>(?:https?:)?//(?:www\.)?youporn\.com/embed/\d+)']
_TESTS = [{ _TESTS = [{
'url': 'http://www.youporn.com/watch/505835/sex-ed-is-it-safe-to-masturbate-daily/', 'url': 'http://www.youporn.com/watch/505835/sex-ed-is-it-safe-to-masturbate-daily/',
'md5': '3744d24c50438cf5b6f6d59feb5055c2', 'md5': '3744d24c50438cf5b6f6d59feb5055c2',
@ -66,57 +70,91 @@ class YouPornIE(InfoExtractor):
}, { }, {
'url': 'https://www.youporn.com/watch/13922959/femdom-principal/', 'url': 'https://www.youporn.com/watch/13922959/femdom-principal/',
'only_matching': True, 'only_matching': True,
}, {
'url': 'https://www.youporn.com/watch/16290308/tinderspecial-trailer1/',
'info_dict': {
'id': '46949121',
'age_limit': 18,
'categories': [],
'description': None, # SEO spam using title removed
'display_id': 'tinderspecial-trailer1',
'duration': 298.0,
'ext': 'mp4',
'upload_date': '20201123',
'uploader': 'Ersties',
'tags': [],
'thumbnail': 'https://fi1.ypncdn.com/m=eaSaaTbWx/202011/23/16290308/original/3.jpg',
'timestamp': 1606147564,
'title': 'Tinder In Real Life',
'view_count': int,
}
}] }]
@staticmethod @classmethod
def _extract_urls(webpage): def _extract_urls(cls, webpage):
return re.findall( def yield_urls():
r'<iframe[^>]+\bsrc=["\']((?:https?:)?//(?:www\.)?youporn\.com/embed/\d+)', for p in cls._EMBED_REGEX:
webpage) for m in re.finditer(p, webpage):
yield m.group('url')
return list(yield_urls())
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) display_id = self._match_valid_url(url).group('id', 'display_id')
video_id = mobj.group('id') url = 'http://www.youporn.com/watch/%s' % (display_id[0],)
display_id = mobj.group('display_id') or video_id display_id = display_id[1] or display_id[0]
webpage = self._download_webpage(
url, display_id, headers={'Cookie': 'age_verified=1'})
video_id = display_id[0]
definitions = self._download_json( playervars = self._search_json(
'https://www.youporn.com/api/video/media_definitions/%s/' % video_id, r'\bplayervars\s*:', webpage, 'playervars', display_id)
display_id)
def get_fmt(x):
v_url = url_or_none(x.get('videoUrl'))
if v_url:
x['videoUrl'] = v_url
return (x['format'], x)
defs_by_format = dict(traverse_obj(playervars, (
'mediaDefinitions', lambda _, v: v.get('format'), T(get_fmt))))
def get_format_data(f):
if f not in defs_by_format:
return []
return self._download_json(
defs_by_format[f]['videoUrl'], display_id, '{0}-formats'.format(f))
formats = [] formats = []
for definition in definitions: # Try to extract only the actual master m3u8 first, avoiding the duplicate single resolution "master" m3u8s
if not isinstance(definition, dict): for hls_url in traverse_obj(
continue get_format_data('hls'),
video_url = url_or_none(definition.get('videoUrl')) (lambda _, v: not isinstance(v['defaultQuality'], bool), 'videoUrl'),
if not video_url: (Ellipsis, 'videoUrl')):
continue formats.extend(self._extract_m3u8_formats(
f = { hls_url, video_id, 'mp4', fatal=False, m3u8_id='hls',
'url': video_url, entry_protocol='m3u8_native'))
'filesize': int_or_none(definition.get('videoSize')),
} for f in traverse_obj(get_format_data('mp4'), (
height = int_or_none(definition.get('quality')) lambda _, v: v.get('videoUrl'), {
'url': ('videoUrl', T(url_or_none)),
'filesize': ('videoSize', T(int_or_none)),
'height': ('quality', T(int_or_none)),
}, T(lambda x: x.get('videoUrl') and x))):
# Video URL's path looks like this: # Video URL's path looks like this:
# /201012/17/505835/720p_1500k_505835/YouPorn%20-%20Sex%20Ed%20Is%20It%20Safe%20To%20Masturbate%20Daily.mp4 # /201012/17/505835/720p_1500k_505835/YouPorn%20-%20Sex%20Ed%20Is%20It%20Safe%20To%20Masturbate%20Daily.mp4
# /201012/17/505835/vl_240p_240k_505835/YouPorn%20-%20Sex%20Ed%20Is%20It%20Safe%20To%20Masturbate%20Daily.mp4 # /201012/17/505835/vl_240p_240k_505835/YouPorn%20-%20Sex%20Ed%20Is%20It%20Safe%20To%20Masturbate%20Daily.mp4
# /videos/201703/11/109285532/1080P_4000K_109285532.mp4 # /videos/201703/11/109285532/1080P_4000K_109285532.mp4
# We will benefit from it by extracting some metadata # We will benefit from it by extracting some metadata
mobj = re.search(r'(?P<height>\d{3,4})[pP]_(?P<bitrate>\d+)[kK]_\d+', video_url) mobj = re.search(r'(?P<height>\d{3,4})[pP]_(?P<bitrate>\d+)[kK]_\d+', f['videoUrl'])
if mobj: if mobj:
if not height: if not f.get('height'):
height = int(mobj.group('height')) f['height'] = int(mobj.group('height'))
bitrate = int(mobj.group('bitrate')) f['tbr'] = int(mobj.group('bitrate'))
f.update({ f['format_id'] = '%dp-%dk' % (f['height'], f['tbr'])
'format_id': '%dp-%dk' % (height, bitrate),
'tbr': bitrate,
})
f['height'] = height
formats.append(f) formats.append(f)
self._sort_formats(formats) self._sort_formats(formats)
webpage = self._download_webpage(
'http://www.youporn.com/watch/%s' % video_id, display_id,
headers={'Cookie': 'age_verified=1'})
title = self._html_search_regex( title = self._html_search_regex(
r'(?s)<div[^>]+class=["\']watchVideoTitle[^>]+>(.+?)</div>', r'(?s)<div[^>]+class=["\']watchVideoTitle[^>]+>(.+?)</div>',
webpage, 'title', default=None) or self._og_search_title( webpage, 'title', default=None) or self._og_search_title(
@ -168,9 +206,12 @@ class YouPornIE(InfoExtractor):
r'(?s)Tags:.*?</div>\s*<div[^>]+class=["\']tagBoxContent["\'][^>]*>(.+?)</div>', r'(?s)Tags:.*?</div>\s*<div[^>]+class=["\']tagBoxContent["\'][^>]*>(.+?)</div>',
'tags') 'tags')
return { data = self._search_json_ld(webpage, video_id, expected_type='VideoObject', fatal=False) or {}
data.pop('url', None)
result = merge_dicts(data, {
'id': video_id, 'id': video_id,
'display_id': display_id, 'display_id': display_id if display_id != video_id else None,
'title': title, 'title': title,
'description': description, 'description': description,
'thumbnail': thumbnail, 'thumbnail': thumbnail,
@ -183,4 +224,5 @@ class YouPornIE(InfoExtractor):
'tags': tags, 'tags': tags,
'age_limit': age_limit, 'age_limit': age_limit,
'formats': formats, 'formats': formats,
} })
return result