class XuiteIE(InfoExtractor):
+ IE_DESC = '隨意窩Xuite影音'
_REGEX_BASE64 = r'(?:[A-Za-z0-9+/]{4})*(?:[A-Za-z0-9+/]{2}==|[A-Za-z0-9+/]{3}=)?'
_VALID_URL = r'https?://vlog\.xuite\.net/(?:play|embed)/(?P<id>%s)' % _REGEX_BASE64
_TESTS = [{
# Audio
'url': 'http://vlog.xuite.net/play/RGkzc1ZULTM4NjA5MTQuZmx2',
- 'md5': '63a42c705772aa53fd4c1a0027f86adf',
+ 'md5': 'e79284c87b371424885448d11f6398c8',
'info_dict': {
'id': '3860914',
'ext': 'mp3',
},
}, {
# Video with only one format
- 'url': 'http://vlog.xuite.net/play/TkRZNjhULTM0NDE2MjkuZmx2',
- 'md5': 'c45737fc8ac5dc8ac2f92ecbcecf505e',
+ 'url': 'http://vlog.xuite.net/play/WUxxR2xCLTI1OTI1MDk5LmZsdg==',
+ 'md5': '21f7b39c009b5a4615b4463df6eb7a46',
'info_dict': {
- 'id': '3441629',
+ 'id': '25925099',
'ext': 'mp4',
- 'title': '孫燕姿 - 眼淚成詩',
+ 'title': 'BigBuckBunny_320x180',
'thumbnail': 're:^https?://.*\.jpg$',
- 'duration': 217.399,
- 'timestamp': 1299383640,
- 'upload_date': '20110306',
- 'uploader': 'Valen',
- 'uploader_id': '10400126',
- 'categories': ['影視娛樂'],
+ 'duration': 596.458,
+ 'timestamp': 1454242500,
+ 'upload_date': '20160131',
+ 'uploader': 'yan12125',
+ 'uploader_id': '12158353',
+ 'categories': ['個人短片'],
+ 'description': 'http://download.blender.org/peach/bigbuckbunny_movies/BigBuckBunny_320x180.mp4',
},
}, {
# Video with two formats
'uploader_id': '242127761',
'categories': ['電玩動漫'],
},
+ 'skip': 'Video removed',
}, {
'url': 'http://vlog.xuite.net/play/S1dDUjdyLTMyOTc3NjcuZmx2/%E5%AD%AB%E7%87%95%E5%A7%BF-%E7%9C%BC%E6%B7%9A%E6%88%90%E8%A9%A9',
'only_matching': True,
}]
+ @staticmethod
+ def base64_decode_utf8(data):
+ return base64.b64decode(data.encode('utf-8')).decode('utf-8')
+
+ @staticmethod
+ def base64_encode_utf8(data):
+ return base64.b64encode(data.encode('utf-8')).decode('utf-8')
+
def _extract_flv_config(self, media_id):
- base64_media_id = base64.b64encode(media_id.encode('utf-8')).decode('utf-8')
+ base64_media_id = self.base64_encode_utf8(media_id)
flv_config = self._download_xml(
'http://vlog.xuite.net/flash/player?media=%s' % base64_media_id,
'flv config')
prop_dict = {}
for prop in flv_config.findall('./property'):
- prop_id = base64.b64decode(prop.attrib['id']).decode('utf-8')
+ prop_id = self.base64_decode_utf8(prop.attrib['id'])
# CDATA may be empty in flv config
if not prop.text:
continue
- encoded_content = base64.b64decode(prop.text).decode('utf-8')
+ encoded_content = self.base64_decode_utf8(prop.text)
prop_dict[prop_id] = compat_urllib_parse_unquote(encoded_content)
return prop_dict