X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/e76f531201cd41dfc0ce00be28bcc5c575c7acc5..a7fd25993be110e616bb9bf6d602949b54df1a7d:/youtube_dl/extractor/fourtube.py diff --git a/youtube_dl/extractor/fourtube.py b/youtube_dl/extractor/fourtube.py index fb6d108..a9a1f91 100644 --- a/youtube_dl/extractor/fourtube.py +++ b/youtube_dl/extractor/fourtube.py @@ -4,40 +4,51 @@ import re from .common import InfoExtractor from ..compat import ( - compat_urllib_request, + compat_b64decode, + compat_str, + compat_urllib_parse_unquote, + compat_urlparse, ) from ..utils import ( + int_or_none, parse_duration, parse_iso8601, + str_or_none, str_to_int, + try_get, + unified_timestamp, + url_or_none, ) -class FourTubeIE(InfoExtractor): - IE_NAME = '4tube' - _VALID_URL = r'https?://(?:www\.)?4tube\.com/videos/(?P\d+)' +class FourTubeBaseIE(InfoExtractor): + _TKN_HOST = 'tkn.kodicdn.com' - _TEST = { - 'url': 'http://www.4tube.com/videos/209733/hot-babe-holly-michaels-gets-her-ass-stuffed-by-black', - 'md5': '6516c8ac63b03de06bc8eac14362db4f', - 'info_dict': { - 'id': '209733', - 'ext': 'mp4', - 'title': 'Hot Babe Holly Michaels gets her ass stuffed by black', - 'uploader': 'WCP Club', - 'uploader_id': 'wcp-club', - 'upload_date': '20131031', - 'timestamp': 1383263892, - 'duration': 583, - 'view_count': int, - 'like_count': int, - 'categories': list, - 'age_limit': 18, - } - } + def _extract_formats(self, url, video_id, media_id, sources): + token_url = 'https://%s/%s/desktop/%s' % ( + self._TKN_HOST, media_id, '+'.join(sources)) + + parsed_url = compat_urlparse.urlparse(url) + tokens = self._download_json(token_url, video_id, data=b'', headers={ + 'Origin': '%s://%s' % (parsed_url.scheme, parsed_url.hostname), + 'Referer': url, + }) + formats = [{ + 'url': tokens[format]['token'], + 'format_id': format + 'p', + 'resolution': format + 'p', + 'quality': int(format), + } for format in sources] + self._sort_formats(formats) + return formats def _real_extract(self, url): - video_id = self._match_id(url) + mobj = re.match(self._VALID_URL, url) + kind, video_id, display_id = mobj.group('kind', 'id', 'display_id') + + if kind == 'm' or not display_id: + url = self._URL_TEMPLATE % video_id + webpage = self._download_webpage(url, video_id) title = self._html_search_meta('name', webpage) @@ -45,14 +56,14 @@ class FourTubeIE(InfoExtractor): 'uploadDate', webpage)) thumbnail = self._html_search_meta('thumbnailUrl', webpage) uploader_id = self._html_search_regex( - r'', + r'', webpage, 'uploader id', fatal=False) uploader = self._html_search_regex( - r'', + r'', webpage, 'uploader', fatal=False) categories_html = self._search_regex( - r'(?s)>\s*Categories / Tags\s*.*?', + r'(?s)>\s*Categories / Tags\s*.*?', webpage, 'categories', fatal=False) categories = None if categories_html: @@ -61,11 +72,11 @@ class FourTubeIE(InfoExtractor): r'(?s)
  • (.*?)', categories_html)] view_count = str_to_int(self._search_regex( - r'', - webpage, 'view count', fatal=False)) + r']+itemprop="interactionCount"[^>]+content="UserPlays:([0-9,]+)">', + webpage, 'view count', default=None)) like_count = str_to_int(self._search_regex( - r'', - webpage, 'like count', fatal=False)) + r']+itemprop="interactionCount"[^>]+content="UserLikes:([0-9,]+)">', + webpage, 'like count', default=None)) duration = parse_duration(self._html_search_meta('duration', webpage)) media_id = self._search_regex( @@ -87,21 +98,7 @@ class FourTubeIE(InfoExtractor): media_id = params[0] sources = ['%s' % p for p in params[2]] - token_url = 'http://tkn.4tube.com/{0}/desktop/{1}'.format( - media_id, '+'.join(sources)) - headers = { - b'Content-Type': b'application/x-www-form-urlencoded', - b'Origin': b'http://www.4tube.com', - } - token_req = compat_urllib_request.Request(token_url, b'{}', headers) - tokens = self._download_json(token_req, video_id) - formats = [{ - 'url': tokens[format]['token'], - 'format_id': format + 'p', - 'resolution': format + 'p', - 'quality': int(format), - } for format in sources] - self._sort_formats(formats) + formats = self._extract_formats(url, video_id, media_id, sources) return { 'id': video_id, @@ -117,3 +114,195 @@ class FourTubeIE(InfoExtractor): 'duration': duration, 'age_limit': 18, } + + +class FourTubeIE(FourTubeBaseIE): + IE_NAME = '4tube' + _VALID_URL = r'https?://(?:(?Pwww|m)\.)?4tube\.com/(?:videos|embed)/(?P\d+)(?:/(?P[^/?#&]+))?' + _URL_TEMPLATE = 'https://www.4tube.com/videos/%s/video' + _TESTS = [{ + 'url': 'http://www.4tube.com/videos/209733/hot-babe-holly-michaels-gets-her-ass-stuffed-by-black', + 'md5': '6516c8ac63b03de06bc8eac14362db4f', + 'info_dict': { + 'id': '209733', + 'ext': 'mp4', + 'title': 'Hot Babe Holly Michaels gets her ass stuffed by black', + 'uploader': 'WCP Club', + 'uploader_id': 'wcp-club', + 'upload_date': '20131031', + 'timestamp': 1383263892, + 'duration': 583, + 'view_count': int, + 'like_count': int, + 'categories': list, + 'age_limit': 18, + }, + }, { + 'url': 'http://www.4tube.com/embed/209733', + 'only_matching': True, + }, { + 'url': 'http://m.4tube.com/videos/209733/hot-babe-holly-michaels-gets-her-ass-stuffed-by-black', + 'only_matching': True, + }] + + +class FuxIE(FourTubeBaseIE): + _VALID_URL = r'https?://(?:(?Pwww|m)\.)?fux\.com/(?:video|embed)/(?P\d+)(?:/(?P[^/?#&]+))?' + _URL_TEMPLATE = 'https://www.fux.com/video/%s/video' + _TESTS = [{ + 'url': 'https://www.fux.com/video/195359/awesome-fucking-kitchen-ends-cum-swallow', + 'info_dict': { + 'id': '195359', + 'ext': 'mp4', + 'title': 'Awesome fucking in the kitchen ends with cum swallow', + 'uploader': 'alenci2342', + 'uploader_id': 'alenci2342', + 'upload_date': '20131230', + 'timestamp': 1388361660, + 'duration': 289, + 'view_count': int, + 'like_count': int, + 'categories': list, + 'age_limit': 18, + }, + 'params': { + 'skip_download': True, + }, + }, { + 'url': 'https://www.fux.com/embed/195359', + 'only_matching': True, + }, { + 'url': 'https://www.fux.com/video/195359/awesome-fucking-kitchen-ends-cum-swallow', + 'only_matching': True, + }] + + +class PornTubeIE(FourTubeBaseIE): + _VALID_URL = r'https?://(?:(?Pwww|m)\.)?porntube\.com/(?:videos/(?P[^/]+)_|embed/)(?P\d+)' + _URL_TEMPLATE = 'https://www.porntube.com/videos/video_%s' + _TKN_HOST = 'tkn.porntube.com' + _TESTS = [{ + 'url': 'https://www.porntube.com/videos/teen-couple-doing-anal_7089759', + 'info_dict': { + 'id': '7089759', + 'ext': 'mp4', + 'title': 'Teen couple doing anal', + 'uploader': 'Alexy', + 'uploader_id': '91488', + 'upload_date': '20150606', + 'timestamp': 1433595647, + 'duration': 5052, + 'view_count': int, + 'like_count': int, + 'age_limit': 18, + }, + 'params': { + 'skip_download': True, + }, + }, { + 'url': 'https://www.porntube.com/videos/squirting-teen-ballerina-ecg_1331406', + 'info_dict': { + 'id': '1331406', + 'ext': 'mp4', + 'title': 'Squirting Teen Ballerina on ECG', + 'uploader': 'Exploited College Girls', + 'uploader_id': '665', + 'channel': 'Exploited College Girls', + 'channel_id': '665', + 'upload_date': '20130920', + 'timestamp': 1379685485, + 'duration': 851, + 'view_count': int, + 'like_count': int, + 'age_limit': 18, + }, + 'params': { + 'skip_download': True, + }, + }, { + 'url': 'https://www.porntube.com/embed/7089759', + 'only_matching': True, + }, { + 'url': 'https://m.porntube.com/videos/teen-couple-doing-anal_7089759', + 'only_matching': True, + }] + + def _real_extract(self, url): + mobj = re.match(self._VALID_URL, url) + video_id, display_id = mobj.group('id', 'display_id') + + webpage = self._download_webpage(url, display_id) + + video = self._parse_json( + self._search_regex( + r'INITIALSTATE\s*=\s*(["\'])(?P(?:(?!\1).)+)\1', + webpage, 'data', group='value'), video_id, + transform_source=lambda x: compat_urllib_parse_unquote( + compat_b64decode(x).decode('utf-8')))['page']['video'] + + title = video['title'] + media_id = video['mediaId'] + sources = [compat_str(e['height']) + for e in video['encodings'] if e.get('height')] + formats = self._extract_formats(url, video_id, media_id, sources) + + thumbnail = url_or_none(video.get('masterThumb')) + uploader = try_get(video, lambda x: x['user']['username'], compat_str) + uploader_id = str_or_none(try_get( + video, lambda x: x['user']['id'], int)) + channel = try_get(video, lambda x: x['channel']['name'], compat_str) + channel_id = str_or_none(try_get( + video, lambda x: x['channel']['id'], int)) + like_count = int_or_none(video.get('likes')) + dislike_count = int_or_none(video.get('dislikes')) + view_count = int_or_none(video.get('playsQty')) + duration = int_or_none(video.get('durationInSeconds')) + timestamp = unified_timestamp(video.get('publishedAt')) + + return { + 'id': video_id, + 'title': title, + 'formats': formats, + 'thumbnail': thumbnail, + 'uploader': uploader or channel, + 'uploader_id': uploader_id or channel_id, + 'channel': channel, + 'channel_id': channel_id, + 'timestamp': timestamp, + 'like_count': like_count, + 'dislike_count': dislike_count, + 'view_count': view_count, + 'duration': duration, + 'age_limit': 18, + } + + +class PornerBrosIE(FourTubeBaseIE): + _VALID_URL = r'https?://(?:(?Pwww|m)\.)?pornerbros\.com/(?:videos/(?P[^/]+)_|embed/)(?P\d+)' + _URL_TEMPLATE = 'https://www.pornerbros.com/videos/video_%s' + _TESTS = [{ + 'url': 'https://www.pornerbros.com/videos/skinny-brunette-takes-big-cock-down-her-anal-hole_181369', + 'md5': '6516c8ac63b03de06bc8eac14362db4f', + 'info_dict': { + 'id': '181369', + 'ext': 'mp4', + 'title': 'Skinny brunette takes big cock down her anal hole', + 'uploader': 'PornerBros HD', + 'uploader_id': 'pornerbros-hd', + 'upload_date': '20130130', + 'timestamp': 1359527401, + 'duration': 1224, + 'view_count': int, + 'categories': list, + 'age_limit': 18, + }, + 'params': { + 'skip_download': True, + }, + }, { + 'url': 'https://www.pornerbros.com/embed/181369', + 'only_matching': True, + }, { + 'url': 'https://m.pornerbros.com/videos/skinny-brunette-takes-big-cock-down-her-anal-hole_181369', + 'only_matching': True, + }]