[c56] Add suppot for multiple formats

This commit is contained in:
Philipp Hagemeister 2014-01-07 10:19:15 +01:00
parent 42154ad5bc
commit 84c92dc00f
1 changed files with 22 additions and 18 deletions

View File

@ -1,21 +1,21 @@
# coding: utf-8 # coding: utf-8
from __future__ import unicode_literals
import re import re
import json import json
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import determine_ext
class C56IE(InfoExtractor): class C56IE(InfoExtractor):
_VALID_URL = r'https?://((www|player)\.)?56\.com/(.+?/)?(v_|(play_album.+-))(?P<textid>.+?)\.(html|swf)' _VALID_URL = r'https?://((www|player)\.)?56\.com/(.+?/)?(v_|(play_album.+-))(?P<textid>.+?)\.(html|swf)'
IE_NAME = u'56.com' IE_NAME = '56.com'
_TEST = {
_TEST ={ 'url': 'http://www.56.com/u39/v_OTM0NDA3MTY.html',
u'url': u'http://www.56.com/u39/v_OTM0NDA3MTY.html', 'file': '93440716.flv',
u'file': u'93440716.flv', 'md5': 'e59995ac63d0457783ea05f93f12a866',
u'md5': u'e59995ac63d0457783ea05f93f12a866', 'info_dict': {
u'info_dict': { 'title': '网事知多少 第32期车怒',
u'title': u'网事知多少 第32期车怒',
}, },
} }
@ -23,14 +23,18 @@ class C56IE(InfoExtractor):
mobj = re.match(self._VALID_URL, url, flags=re.VERBOSE) mobj = re.match(self._VALID_URL, url, flags=re.VERBOSE)
text_id = mobj.group('textid') text_id = mobj.group('textid')
info_page = self._download_webpage('http://vxml.56.com/json/%s/' % text_id, info_page = self._download_webpage('http://vxml.56.com/json/%s/' % text_id,
text_id, u'Downloading video info') text_id, 'Downloading video info')
info = json.loads(info_page)['info'] info = json.loads(info_page)['info']
best_format = sorted(info['rfiles'], key=lambda f: int(f['filesize']))[-1] formats = [{
video_url = best_format['url'] 'format_id': f['type'],
'filesize': int(f['filesize']),
'url': f['url']
} for f in info['rfiles']]
self._sort_formats(formats)
return {'id': info['vid'], return {
'id': info['vid'],
'title': info['Subject'], 'title': info['Subject'],
'url': video_url, 'formats': formats,
'ext': determine_ext(video_url),
'thumbnail': info.get('bimg') or info.get('img'), 'thumbnail': info.get('bimg') or info.get('img'),
} }