From: Yen Chi Hsuan Date: Sat, 6 Aug 2016 18:44:15 +0000 (+0800) Subject: [openload] Fix extraction (closes #9706) X-Git-Url: http://git.bitcoin.ninja/?a=commitdiff_plain;h=c1decda58c812b3d0a3d4dfa998e7d8bd8f99203;p=youtube-dl [openload] Fix extraction (closes #9706) --- diff --git a/youtube_dl/extractor/openload.py b/youtube_dl/extractor/openload.py index 6415b8fdc..4e80ca9ff 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 +from __future__ import unicode_literals, division -import re +import math from .common import InfoExtractor from ..compat import compat_chr from ..utils import ( + decode_png, determine_ext, - encode_base_n, ExtractorError, - mimetype2ext, ) @@ -41,60 +40,6 @@ 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 ǃ \u01c3 - # Using escaped unicode literals does not work in Python 3.2 - return re.sub(r'ǃ\((\d+),(\d+)\)', 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 = [ - ('_', '(ﾟДﾟ) [ﾟΘﾟ]'), - ('a', '(ﾟДﾟ) [ﾟωﾟﾉ]'), - ('b', '(ﾟДﾟ) [ﾟΘﾟﾉ]'), - ('c', '(ﾟДﾟ) [\'c\']'), - ('d', '(ﾟДﾟ) [ﾟｰﾟﾉ]'), - ('e', '(ﾟДﾟ) [ﾟДﾟﾉ]'), - ('f', '(ﾟДﾟ) [1]'), - - ('o', '(ﾟДﾟ) [\'o\']'), - ('u', '(oﾟｰﾟo)'), - ('c', '(ﾟДﾟ) [\'c\']'), - - ('7', '((ﾟｰﾟ) + (o^_^o))'), - ('6', '((o^_^o) +(o^_^o) +(c^_^o))'), - ('5', '((ﾟｰﾟ) + (ﾟΘﾟ))'), - ('4', '(-~3)'), - ('3', '(-~-~1)'), - ('2', '(-~1)'), - ('1', '(-~0)'), - ('0', '((c^_^o)-(c^_^o))'), - ] - delim = '(ﾟДﾟ)[ﾟεﾟ]+' - 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) @@ -102,29 +47,77 @@ class OpenloadIE(InfoExtractor): if 'File not found' in webpage: raise ExtractorError('File not found', expected=True) - code = self._search_regex( - r'\s*\s*]+>[^>]+\s*]+>([^<]+)', - webpage, 'JS code') - - decoded = self.openload_decode(code) - - video_url = self._search_regex( - r'return\s+"(https?://[^"]+)"', decoded, 'video URL') + # The following extraction logic is proposed by @Belderak and @gdkchan + # and declared to be used freely in youtube-dl + # See https://github.com/rg3/youtube-dl/issues/9706 + + numbers_js = self._download_webpage( + 'https://openload.co/assets/js/obfuscator/n.js', video_id, + note='Downloading signature numbers') + signums = self._search_regex( + r'window\.signatureNumbers\s*=\s*[\'"](?P[a-z]+)[\'"]', + numbers_js, 'signature numbers', group='data') + + linkimg_uri = self._search_regex( + r']+id="linkimg"[^>]+src="([^"]+)"', webpage, 'link image') + linkimg = self._request_webpage( + linkimg_uri, video_id, note=False).read() + + width, height, pixels = decode_png(linkimg) + + output = '' + for y in range(height): + for x in range(width): + r, g, b = pixels[y][3 * x:3 * x + 3] + if r == 0 and g == 0 and b == 0: + break + else: + output += compat_chr(r) + output += compat_chr(g) + output += compat_chr(b) + + img_str_length = len(output) // 200 + img_str = [[0 for x in range(img_str_length)] for y in range(10)] + + sig_str_length = len(signums) // 260 + sig_str = [[0 for x in range(sig_str_length)] for y in range(10)] + + for i in range(10): + for j in range(img_str_length): + begin = i * img_str_length * 20 + j * 20 + img_str[i][j] = output[begin:begin + 20] + for j in range(sig_str_length): + begin = i * sig_str_length * 26 + j * 26 + sig_str[i][j] = signums[begin:begin + 26] + + parts = [] + # TODO: find better names for str_, chr_ and sum_ + str_ = '' + for i in [2, 3, 5, 7]: + str_ = '' + sum_ = float(99) + for j in range(len(sig_str[i])): + for chr_idx in range(len(img_str[i][j])): + if sum_ > float(122): + sum_ = float(98) + chr_ = compat_chr(int(math.floor(sum_))) + if sig_str[i][j][chr_idx] == chr_ and j >= len(str_): + sum_ += float(2.5) + str_ += img_str[i][j][chr_idx] + parts.append(str_.replace(',', '')) + + video_url = 'https://openload.co/stream/%s~%s~%s~%s' % (parts[3], parts[1], parts[2], parts[0]) 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), }