you-get/src/you_get/extractor/sohu.py
2013-10-08 14:19:33 +02:00

62 lines
2.3 KiB
Python

#!/usr/bin/env python
__all__ = ['sohu_download']
from ..common import *
import json
def real_url(host, prot, file, new):
url = 'http://%s/?prot=%s&file=%s&new=%s' % (host, prot, file, new)
start, _, host, key = get_html(url).split('|')[:4]
return '%s%s?key=%s' % (start[:-1], new, key)
def sohu_download(url, output_dir = '.', merge = True, info_only = False):
html = get_html(url)
vid = r1('vid\s*=\s*"(\d+)"', html)
if not vid:
vid = r1('vid\s*:\s*"(\d+)"', html)
if vid:
data = json.loads(get_decoded_html('http://hot.vrs.sohu.com/vrs_flash.action?vid=%s' % vid))
for qtyp in ["oriVid","superVid","highVid" ,"norVid","relativeId"]:
hqvid = data['data'][qtyp]
if hqvid != 0 and hqvid != vid :
data = json.loads(get_decoded_html('http://hot.vrs.sohu.com/vrs_flash.action?vid=%s' % hqvid))
break
host = data['allot']
prot = data['prot']
urls = []
data = data['data']
title = data['tvName']
size = sum(data['clipsBytes'])
assert len(data['clipsURL']) == len(data['clipsBytes']) == len(data['su'])
for file, new in zip(data['clipsURL'], data['su']):
urls.append(real_url(host, prot, file, new))
assert data['clipsURL'][0].endswith('.mp4')
else:
if re.match(r'http://share.vrs.sohu.com', url):
vid = r1('id=(\d+)', url)
else:
vid = r1('vid\s*=\s*\'(\d+)\'', get_html(url))
data = json.loads(get_decoded_html('http://my.tv.sohu.com/play/videonew.do?vid=%s&referer=http://my.tv.sohu.com' % vid))
host = data['allot']
prot = data['prot']
urls = []
data = data['data']
title = data['tvName']
size = sum([int(clipsBytes) for clipsBytes in data['clipsBytes']])
assert len(data['clipsURL']) == len(data['clipsBytes']) == len(data['su'])
for file, new in zip(data['clipsURL'], data['su']):
urls.append(real_url(host, prot, file, new))
assert data['clipsURL'][0].endswith('.mp4')
print_info(site_info, title, 'mp4', size)
if not info_only:
download_urls(urls, title, 'mp4', size, output_dir, refer = url, merge = merge)
site_info = "Sohu.com"
download = sohu_download
download_playlist = playlist_not_supported('sohu')