X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;ds=sidebyside;f=youtube_dl%2Fextractor%2Fsohu.py;h=49e5d09ae450d11bb567a2fe95ecba55998c8b42;hb=c02ec7d4300d3e2607f48fe73011fd8caa38f90c;hp=7db5b2f13367749aebef75e0785bc842bdba19fd;hpb=cd65491c306f644d7bb3c7ad98795a3f8660be49;p=youtube-dl.git diff --git a/youtube_dl/extractor/sohu.py b/youtube_dl/extractor/sohu.py index 7db5b2f13..49e5d09ae 100644 --- a/youtube_dl/extractor/sohu.py +++ b/youtube_dl/extractor/sohu.py @@ -6,9 +6,12 @@ import re from .common import InfoExtractor from ..compat import ( compat_str, - compat_urllib_request + compat_urllib_parse_urlencode, +) +from ..utils import ( + ExtractorError, + sanitized_Request, ) -from ..utils import url_sanitize_consecutive_slashes class SohuIE(InfoExtractor): @@ -23,9 +26,7 @@ class SohuIE(InfoExtractor): 'ext': 'mp4', 'title': 'MVï¼Far East MovementãThe Illestã', }, - 'params': { - 'cn_verification_proxy': 'proxy.uku.im:8888' - } + 'skip': 'On available in China', }, { 'url': 'http://tv.sohu.com/20150305/n409385080.shtml', 'md5': '699060e75cf58858dd47fb9c03c42cfb', @@ -47,6 +48,7 @@ class SohuIE(InfoExtractor): 'url': 'http://my.tv.sohu.com/pl/8384802/78910339.shtml', 'info_dict': { 'id': '78910339', + 'title': 'ãç¥æ¢èå®æç§ç±ã第13æ æäºä¹å½± 赫å¡éå§', }, 'playlist': [{ 'md5': 'bdbfb8f39924725e6589c146bc1883ad', @@ -73,6 +75,17 @@ class SohuIE(InfoExtractor): 'title': 'ãç¥æ¢èå®æç§ç±ã第13æ æäºä¹å½± 赫å¡éå§', } }] + }, { + 'note': 'Video with title containing dash', + 'url': 'http://my.tv.sohu.com/us/249884221/78932792.shtml', + 'info_dict': { + 'id': '78932792', + 'ext': 'mp4', + 'title': 'youtube-dl testing video', + }, + 'params': { + 'skip_download': True + } }] def _real_extract(self, url): @@ -83,29 +96,37 @@ class SohuIE(InfoExtractor): else: base_data_url = 'http://hot.vrs.sohu.com/vrs_flash.action?vid=' - req = compat_urllib_request.Request(base_data_url + vid_id) + req = sanitized_Request(base_data_url + vid_id) cn_verification_proxy = self._downloader.params.get('cn_verification_proxy') if cn_verification_proxy: req.add_header('Ytdl-request-proxy', cn_verification_proxy) - return self._download_json(req, video_id, - 'Downloading JSON data for %s' % vid_id) + return self._download_json( + req, video_id, + 'Downloading JSON data for %s' % vid_id) mobj = re.match(self._VALID_URL, url) video_id = mobj.group('id') mytv = mobj.group('mytv') is not None webpage = self._download_webpage(url, video_id) - raw_title = self._html_search_regex( - r'(?s)