From 6d2d21f713614141cba09310cb60d2edd76c79ca Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Mon, 9 Sep 2013 19:56:16 +0200 Subject: [PATCH] [sohu] add support for my.tv.sohu.com urls (fixes #1398) --- youtube_dl/extractor/sohu.py | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/youtube_dl/extractor/sohu.py b/youtube_dl/extractor/sohu.py index 77bb0a8dc..2b9bf0cb7 100644 --- a/youtube_dl/extractor/sohu.py +++ b/youtube_dl/extractor/sohu.py @@ -8,7 +8,7 @@ from ..utils import ExtractorError class SohuIE(InfoExtractor): - _VALID_URL = r'https?://tv\.sohu\.com/\d+?/n(?P\d+)\.shtml.*?' + _VALID_URL = r'https?://(?Pmy\.)?tv\.sohu\.com/.+?/(?(mytv)|n)(?P\d+)\.shtml.*?' _TEST = { u'url': u'http://tv.sohu.com/20130724/n382479172.shtml#super', @@ -21,8 +21,11 @@ class SohuIE(InfoExtractor): def _real_extract(self, url): - def _fetch_data(vid_id): - base_data_url = u'http://hot.vrs.sohu.com/vrs_flash.action?vid=' + def _fetch_data(vid_id, mytv=False): + if mytv: + base_data_url = 'http://my.tv.sohu.com/play/videonew.do?vid=' + else: + base_data_url = u'http://hot.vrs.sohu.com/vrs_flash.action?vid=' data_url = base_data_url + str(vid_id) data_json = self._download_webpage( data_url, video_id, @@ -31,15 +34,16 @@ class SohuIE(InfoExtractor): 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)(.+?)', webpage, u'video title') title = raw_title.partition('-')[0].strip() - vid = self._html_search_regex(r'var vid="(\d+)"', webpage, + vid = self._html_search_regex(r'var vid ?= ?["\'](\d+)["\']', webpage, u'video path') - data = _fetch_data(vid) + data = _fetch_data(vid, mytv) QUALITIES = ('ori', 'super', 'high', 'nor') vid_ids = [data['data'][q + 'Vid'] @@ -51,7 +55,7 @@ class SohuIE(InfoExtractor): # For now, we just pick the highest available quality vid_id = vid_ids[-1] - format_data = data if vid == vid_id else _fetch_data(vid_id) + format_data = data if vid == vid_id else _fetch_data(vid_id, mytv) part_count = format_data['data']['totalBlocks'] allot = format_data['allot'] prot = format_data['prot']