1 from .common
import InfoExtractor
4 compat_urllib_parse_unquote_plus
,
19 class SharedBaseIE(InfoExtractor
):
20 def _real_extract(self
, url
):
21 video_id
= self
._match
_id
(url
)
23 webpage
, urlh
= self
._download
_webpage
_handle
(url
, video_id
)
25 if self
._FILE
_NOT
_FOUND
in webpage
:
27 'Video %s does not exist' % video_id
, expected
=True)
29 video_url
= self
._extract
_video
_url
(webpage
, video_id
, url
)
31 title
= self
._extract
_title
(webpage
)
32 filesize
= int_or_none(self
._extract
_filesize
(webpage
))
42 def _extract_title(self
, webpage
):
43 return compat_b64decode(self
._html
_search
_meta
(
44 'full:title', webpage
, 'title')).decode('utf-8')
46 def _extract_filesize(self
, webpage
):
47 return self
._html
_search
_meta
(
48 'full:size', webpage
, 'file size', fatal
=False)
51 class SharedIE(SharedBaseIE
):
53 _VALID_URL
= r
'https?://shared\.sx/(?P<id>[\da-z]{10})'
54 _FILE_NOT_FOUND
= '>File does not exist<'
57 'url': 'http://shared.sx/0060718775',
58 'md5': '106fefed92a8a2adb8c98e6a0652f49b',
67 def _extract_video_url(self
, webpage
, video_id
, url
):
68 download_form
= self
._hidden
_inputs
(webpage
)
70 video_page
= self
._download
_webpage
(
71 url
, video_id
, 'Downloading video page',
72 data
=urlencode_postdata(download_form
),
74 'Content-Type': 'application/x-www-form-urlencoded',
78 video_url
= self
._html
_search
_regex
(
79 r
'data-url=(["\'])(?P
<url
>(?
:(?
!\
1).)+)\
1',
80 video_page, 'video URL
', group='url
')
85 class VivoIE(SharedBaseIE):
87 _VALID_URL = r'https?
://vivo\
.s
[xt
]/(?P
<id>[\da
-z
]{10}
)'
88 _FILE_NOT_FOUND = '>The
file you have requested does
not exists
or has been removed
'
91 'url
': 'http
://vivo
.sx
/d7ddda0e78
',
92 'md5
': '15b3af41be0b4fe01f4df075c2678b2c
',
100 'url
': 'http
://vivo
.st
/d7ddda0e78
',
101 'only_matching
': True,
104 def _extract_title(self, webpage):
105 title = self._html_search_regex(
106 r'data
-name\s
*=\s
*(["\'])(?P<title>(?:(?!\1).)+)\1', webpage,
107 'title', default=None, group='title')
109 ext = determine_ext(title)
110 if ext.lower() in KNOWN_EXTENSIONS:
111 title = title.rpartition('.' + ext)[0]
113 return self._og_search_title(webpage)
115 def _extract_filesize(self, webpage):
116 return parse_filesize(self._search_regex(
117 r'data-type=["\']video
["\'][^>]*>Watch.*?<strong>\s*\((.+?)\)',
118 webpage, 'filesize', fatal=False))
120 def _extract_video_url(self, webpage, video_id, url):
121 def decode_url_old(encoded_url):
122 return compat_b64decode(encoded_url).decode('utf-8')
124 stream_url = self._search_regex(
125 r'data-stream\s*=\s*(["\'])(?P
<url
>(?
:(?
!\
1).)+)\
1', webpage,
126 'stream url
', default=None, group='url
')
128 stream_url = url_or_none(decode_url_old(stream_url))
132 def decode_url(encoded_url):
133 return rot47(compat_urllib_parse_unquote_plus(encoded_url))
135 return decode_url(self._parse_json(
137 r'(?s
)InitializeStream\s
*\
(\s
*({.+?}
)\s
*\
)\s
*;', webpage,
139 video_id, transform_source=js_to_json)['source
'])