From 56c7366547462ecec0536df58971249a8a870ddd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?=
\n*?(.+?)\n*?
'
+ compiled = re.compile(pattern, re.DOTALL)
+ title = self._search_regex(compiled, webpage, u'video title').strip('\t\n')
+ title = self._clearn_html(title)
+ pattern = re.compile(r'var vid="(\d+)"')
+ result = re.search(pattern, webpage)
+ if not result:
+ logging.info('[Sohu] could not get vid')
+ return None
+ vid = result.group(1)
+ logging.info('vid: %s' % vid)
+ base_url_1 = 'http://hot.vrs.sohu.com/vrs_flash.action?vid='
+ url_1 = base_url_1 + vid
+ logging.info('json url: %s' % url_1)
+ json_1 = json.loads(urllib2.urlopen(url_1).read())
+ # get the highest definition video vid and json infomation.
+ vids = []
+ qualities = ('oriVid', 'superVid', 'highVid', 'norVid')
+ for vid_name in qualities:
+ vids.append(json_1['data'][vid_name])
+ clearest_vid = 0
+ for i, v in enumerate(vids):
+ if v != 0:
+ clearest_vid = v
+ logging.info('quality definition: %s' % qualities[i][:-3])
+ break
+ if not clearest_vid:
+ logging.warning('could not find valid clearest_vid')
+ return None
+ if vid != clearest_vid:
+ url_1 = '%s%d' % (base_url_1, clearest_vid)
+ logging.info('highest definition json url: %s' % url_1)
+ json_1 = json.loads(urllib2.urlopen(url_1).read())
+ allot = json_1['allot']
+ prot = json_1['prot']
+ clipsURL = json_1['data']['clipsURL']
+ su = json_1['data']['su']
+ num_of_parts = json_1['data']['totalBlocks']
+ logging.info('Total parts: %d' % num_of_parts)
+ base_url_3 = 'http://allot/?prot=prot&file=clipsURL[i]&new=su[i]'
+ files_info = []
+ for i in range(num_of_parts):
+ middle_url = 'http://%s/?prot=%s&file=%s&new=%s' % (allot, prot, clipsURL[i], su[i])
+ logging.info('middle url part %d: %s' % (i, middle_url))
+ middle_info = urllib2.urlopen(middle_url).read().split('|')
+ middle_part_1 = middle_info[0]
+ download_url = '%s%s?key=%s' % (middle_info[0], su[i], middle_info[3])
+
+ info = {
+ 'id': '%s_part%02d' % (video_id, i + 1),
+ 'title': title,
+ 'url': download_url,
+ 'ext': 'mp4',
+ }
+ files_info.append(info)
+ time.sleep(1)
+
+ return files_info
--
cgit v1.2.3
From 4ec929dc9b55a2588b4a27e64871c5bfa900bf37 Mon Sep 17 00:00:00 2001
From: huohuarong \n*?(.+?)\n*?
'
compiled = re.compile(pattern, re.DOTALL)
title = self._search_regex(compiled, webpage, u'video title').strip('\t\n')
- title = self._clearn_html(title)
+ title = clean_html(title)
pattern = re.compile(r'var vid="(\d+)"')
result = re.search(pattern, webpage)
if not result:
@@ -93,5 +83,8 @@ class SohuIE(InfoExtractor):
}
files_info.append(info)
time.sleep(1)
-
+ if num_of_parts == 1:
+ info = files_info[0]
+ info['id'] = video_id
+ return info
return files_info
--
cgit v1.2.3
From 968b5e0112a83f2a4637226d4d743b394ebed038 Mon Sep 17 00:00:00 2001
From: Dan Church \n*?(.+?)\n*?
'
+ pattern = r'