X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/87a0165ca7e39af4dacb7ec637063b2cd35ae40b..fe6cc1b2aebf52f3fcdcc22ca8e846b67a406cc0:/youtube_dl/extractor/tnaflix.py diff --git a/youtube_dl/extractor/tnaflix.py b/youtube_dl/extractor/tnaflix.py index 77d56b8..b3573c6 100644 --- a/youtube_dl/extractor/tnaflix.py +++ b/youtube_dl/extractor/tnaflix.py @@ -18,9 +18,12 @@ from ..utils import ( class TNAFlixNetworkBaseIE(InfoExtractor): # May be overridden in descendants if necessary _CONFIG_REGEX = [ - r'flashvars\.config\s*=\s*escape\("([^"]+)"', - r']+name="config\d?" value="([^"]+)"', + r'flashvars\.config\s*=\s*escape\("(?P[^"]+)"', + r']+name="config\d?" value="(?P[^"]+)"', + r'config\s*=\s*(["\'])(?P(?:https?:)?//(?:(?!\1).)+)\1', ] + _HOST = 'tna' + _VKEY_SUFFIX = '' _TITLE_REGEX = r']+name="title" value="([^"]+)"' _DESCRIPTION_REGEX = r']+name="description" value="([^"]+)"' _UPLOADER_REGEX = r']+name="username" value="([^"]+)"' @@ -72,26 +75,34 @@ class TNAFlixNetworkBaseIE(InfoExtractor): def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) video_id = mobj.group('id') - display_id = mobj.group('display_id') if 'display_id' in mobj.groupdict() else video_id + for display_id_key in ('display_id', 'display_id_2'): + if display_id_key in mobj.groupdict(): + display_id = mobj.group(display_id_key) + if display_id: + break + else: + display_id = video_id webpage = self._download_webpage(url, display_id) cfg_url = self._proto_relative_url(self._html_search_regex( - self._CONFIG_REGEX, webpage, 'flashvars.config', default=None), 'http:') + self._CONFIG_REGEX, webpage, 'flashvars.config', default=None, + group='url'), 'http:') if not cfg_url: inputs = self._hidden_inputs(webpage) - cfg_url = ('https://cdn-fck.tnaflix.com/tnaflix/%s.fid?key=%s&VID=%s&premium=1&vip=1&alpha' - % (inputs['vkey'], inputs['nkey'], video_id)) + cfg_url = ('https://cdn-fck.%sflix.com/%sflix/%s%s.fid?key=%s&VID=%s&premium=1&vip=1&alpha' + % (self._HOST, self._HOST, inputs['vkey'], self._VKEY_SUFFIX, inputs['nkey'], video_id)) cfg_xml = self._download_xml( cfg_url, display_id, 'Downloading metadata', - transform_source=fix_xml_ampersands) + transform_source=fix_xml_ampersands, headers={'Referer': url}) formats = [] def extract_video_url(vl): - return re.sub('speed=\d+', 'speed=', unescapeHTML(vl.text)) + # Any URL modification now results in HTTP Error 403: Forbidden + return unescapeHTML(vl.text) video_link = cfg_xml.find('./videoLink') if video_link is not None: @@ -174,7 +185,7 @@ class TNAFlixNetworkEmbedIE(TNAFlixNetworkBaseIE): 'display_id': '6538', 'ext': 'mp4', 'title': 'Educational xxx video', - 'thumbnail': 're:https?://.*\.jpg$', + 'thumbnail': r're:https?://.*\.jpg$', 'age_limit': 18, }, 'params': { @@ -192,24 +203,27 @@ class TNAFlixNetworkEmbedIE(TNAFlixNetworkBaseIE): webpage)] -class TNAFlixIE(TNAFlixNetworkBaseIE): +class TNAEMPFlixBaseIE(TNAFlixNetworkBaseIE): + _DESCRIPTION_REGEX = r'(?s)>Description:]+>(.+?)<' + _UPLOADER_REGEX = r'by\s*]+\bhref=["\']/profile/[^>]+>([^<]+)<' + _CATEGORIES_REGEX = r'(?s)]*>Categories:(.+?)' + + +class TNAFlixIE(TNAEMPFlixBaseIE): _VALID_URL = r'https?://(?:www\.)?tnaflix\.com/[^/]+/(?P[^/]+)/video(?P\d+)' _TITLE_REGEX = r'(.+?) - (?:TNAFlix Porn Videos|TNAFlix\.com)' - _DESCRIPTION_REGEX = r'(?s)>Description:]+>(.+?)<' - _UPLOADER_REGEX = r'\s*Verified Member\s*\s*(.+?)<' - _CATEGORIES_REGEX = r'(?s)]*>Categories:(.+?)' _TESTS = [{ # anonymous uploader, no categories 'url': 'http://www.tnaflix.com/porn-stars/Carmella-Decesare-striptease/video553878', - 'md5': 'ecf3498417d09216374fc5907f9c6ec0', + 'md5': '7e569419fe6d69543d01e6be22f5f7c4', 'info_dict': { 'id': '553878', 'display_id': 'Carmella-Decesare-striptease', 'ext': 'mp4', 'title': 'Carmella Decesare - striptease', - 'thumbnail': 're:https?://.*\.jpg$', + 'thumbnail': r're:https?://.*\.jpg$', 'duration': 91, 'age_limit': 18, 'categories': ['Porn Stars'], @@ -224,11 +238,11 @@ class TNAFlixIE(TNAFlixNetworkBaseIE): 'ext': 'mp4', 'title': 'Educational xxx video', 'description': 'md5:b4fab8f88a8621c8fabd361a173fe5b8', - 'thumbnail': 're:https?://.*\.jpg$', + 'thumbnail': r're:https?://.*\.jpg$', 'duration': 164, 'age_limit': 18, 'uploader': 'bobwhite39', - 'categories': ['Amateur Porn', 'Squirting Videos', 'Teen Girls 18+'], + 'categories': list, } }, { 'url': 'https://www.tnaflix.com/amateur-porn/bunzHD-Ms.Donk/video358632', @@ -236,21 +250,22 @@ class TNAFlixIE(TNAFlixNetworkBaseIE): }] -class EMPFlixIE(TNAFlixNetworkBaseIE): - _VALID_URL = r'https?://(?:www\.)?empflix\.com/videos/(?P.+?)-(?P[0-9]+)\.html' +class EMPFlixIE(TNAEMPFlixBaseIE): + _VALID_URL = r'https?://(?:www\.)?empflix\.com/(?:videos/(?P.+?)-|[^/]+/(?P[^/]+)/video)(?P[0-9]+)' - _UPLOADER_REGEX = r']+class="infoTitle"[^>]*>Uploaded By:(.+?)' + _HOST = 'emp' + _VKEY_SUFFIX = '-1' _TESTS = [{ 'url': 'http://www.empflix.com/videos/Amateur-Finger-Fuck-33051.html', - 'md5': 'b1bc15b6412d33902d6e5952035fcabc', + 'md5': 'bc30d48b91a7179448a0bda465114676', 'info_dict': { 'id': '33051', 'display_id': 'Amateur-Finger-Fuck', 'ext': 'mp4', 'title': 'Amateur Finger Fuck', 'description': 'Amateur solo finger fucking.', - 'thumbnail': 're:https?://.*\.jpg$', + 'thumbnail': r're:https?://.*\.jpg$', 'duration': 83, 'age_limit': 18, 'uploader': 'cwbike', @@ -259,6 +274,9 @@ class EMPFlixIE(TNAFlixNetworkBaseIE): }, { 'url': 'http://www.empflix.com/videos/[AROMA][ARMD-718]-Aoi-Yoshino-Sawa-25826.html', 'only_matching': True, + }, { + 'url': 'https://www.empflix.com/amateur-porn/Amateur-Finger-Fuck/video33051', + 'only_matching': True, }] @@ -280,7 +298,7 @@ class MovieFapIE(TNAFlixNetworkBaseIE): 'ext': 'mp4', 'title': 'Experienced MILF Amazing Handjob', 'description': 'Experienced MILF giving an Amazing Handjob', - 'thumbnail': 're:https?://.*\.jpg$', + 'thumbnail': r're:https?://.*\.jpg$', 'age_limit': 18, 'uploader': 'darvinfred06', 'view_count': int, @@ -298,7 +316,7 @@ class MovieFapIE(TNAFlixNetworkBaseIE): 'ext': 'flv', 'title': 'Jeune Couple Russe', 'description': 'Amateur', - 'thumbnail': 're:https?://.*\.jpg$', + 'thumbnail': r're:https?://.*\.jpg$', 'age_limit': 18, 'uploader': 'whiskeyjar', 'view_count': int,