X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;ds=sidebyside;f=youtube_dl%2Fextractor%2Fopenload.py;h=03baf8e32d4ae89b8c54a24ad3a2580e81d1ec01;hb=2fdc7b0e0438e7f72380cfff11b619fe36234ebd;hp=6415b8fdcb7451e6b499d1d4d3515af2775c66c2;hpb=1058f56e96316e778ed74ad008949c634a582e84;p=youtube-dl
diff --git a/youtube_dl/extractor/openload.py b/youtube_dl/extractor/openload.py
index 6415b8fdc..03baf8e32 100644
--- a/youtube_dl/extractor/openload.py
+++ b/youtube_dl/extractor/openload.py
@@ -1,15 +1,14 @@
# coding: utf-8
-from __future__ import unicode_literals
-
-import re
+from __future__ import unicode_literals, division
from .common import InfoExtractor
-from ..compat import compat_chr
+from ..compat import (
+ compat_chr,
+ compat_ord,
+)
from ..utils import (
determine_ext,
- encode_base_n,
ExtractorError,
- mimetype2ext,
)
@@ -41,90 +40,41 @@ class OpenloadIE(InfoExtractor):
'only_matching': True,
}]
- @staticmethod
- def openload_level2_debase(m):
- radix, num = int(m.group(1)) + 27, int(m.group(2))
- return '"' + encode_base_n(num, radix) + '"'
-
- @classmethod
- def openload_level2(cls, txt):
- # The function name is Ç \u01c3
- # Using escaped unicode literals does not work in Python 3.2
- return re.sub(r'Ç\((\d+),(\d+)\)', cls.openload_level2_debase, txt, re.UNICODE).replace('"+"', '')
-
- # Openload uses a variant of aadecode
- # openload_decode and related functions are originally written by
- # vitas@matfyz.cz and released with public domain
- # See https://github.com/rg3/youtube-dl/issues/8489
- @classmethod
- def openload_decode(cls, txt):
- symbol_table = [
- ('_', '(ï¾Ðï¾) [ï¾Îï¾]'),
- ('a', '(ï¾Ðï¾) [ï¾Ïï¾ï¾]'),
- ('b', '(ï¾Ðï¾) [ï¾Îï¾ï¾]'),
- ('c', '(ï¾Ðï¾) [\'c\']'),
- ('d', '(ï¾Ðï¾) [ï¾ï½°ï¾ï¾]'),
- ('e', '(ï¾Ðï¾) [ï¾Ðï¾ï¾]'),
- ('f', '(ï¾Ðï¾) [1]'),
-
- ('o', '(ï¾Ðï¾) [\'o\']'),
- ('u', '(oï¾ï½°ï¾o)'),
- ('c', '(ï¾Ðï¾) [\'c\']'),
-
- ('7', '((ï¾ï½°ï¾) + (o^_^o))'),
- ('6', '((o^_^o) +(o^_^o) +(c^_^o))'),
- ('5', '((ï¾ï½°ï¾) + (ï¾Îï¾))'),
- ('4', '(-~3)'),
- ('3', '(-~-~1)'),
- ('2', '(-~1)'),
- ('1', '(-~0)'),
- ('0', '((c^_^o)-(c^_^o))'),
- ]
- delim = '(ï¾Ðï¾)[ï¾Îµï¾]+'
- ret = ''
- for aachar in txt.split(delim):
- for val, pat in symbol_table:
- aachar = aachar.replace(pat, val)
- aachar = aachar.replace('+ ', '')
- m = re.match(r'^\d+', aachar)
- if m:
- ret += compat_chr(int(m.group(0), 8))
- else:
- m = re.match(r'^u([\da-f]+)', aachar)
- if m:
- ret += compat_chr(int(m.group(1), 16))
- return cls.openload_level2(ret)
-
def _real_extract(self, url):
video_id = self._match_id(url)
- webpage = self._download_webpage(url, video_id)
+ webpage = self._download_webpage('https://openload.co/embed/%s/' % video_id, video_id)
- if 'File not found' in webpage:
+ if 'File not found' in webpage or 'deleted by the owner' in webpage:
raise ExtractorError('File not found', expected=True)
- code = self._search_regex(
- r'\s*\s*\s*',
- webpage, 'JS code')
+ # The following decryption algorithm is written by @yokrysty and
+ # declared to be freely used in youtube-dl
+ # See https://github.com/rg3/youtube-dl/issues/10408
+ enc_data = self._html_search_regex(
+ r']+id="hiddenurl"[^>]*>([^<]+)', webpage, 'encrypted data')
- decoded = self.openload_decode(code)
+ video_url_chars = []
- video_url = self._search_regex(
- r'return\s+"(https?://[^"]+)"', decoded, 'video URL')
+ for idx, c in enumerate(enc_data):
+ j = compat_ord(c)
+ if j >= 33 and j <= 126:
+ j = ((j + 14) % 94) + 33
+ if idx == len(enc_data) - 1:
+ j += 1
+ video_url_chars += compat_chr(j)
+
+ video_url = 'https://openload.co/stream/%s?mime=true' % ''.join(video_url_chars)
title = self._og_search_title(webpage, default=None) or self._search_regex(
r']+class=["\']title["\'][^>]*>([^<]+)', webpage,
'title', default=None) or self._html_search_meta(
'description', webpage, 'title', fatal=True)
- ext = mimetype2ext(self._search_regex(
- r'window\.vt\s*=\s*(["\'])(?P.+?)\1', decoded,
- 'mimetype', default=None, group='mimetype')) or determine_ext(
- video_url, 'mp4')
-
return {
'id': video_id,
'title': title,
- 'ext': ext,
'thumbnail': self._og_search_thumbnail(webpage, default=None),
'url': video_url,
+ # Seems all videos have extensions in their titles
+ 'ext': determine_ext(title),
}