[pornhub] Decode obfuscated video URL (closes #12470)

This commit is contained in:
Throaway 2017-03-20 16:29:39 -07:00 committed by Sergey M․
parent 97952bdb78
commit 21fbf0f955
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -1,7 +1,9 @@
# coding: utf-8 # coding: utf-8
from __future__ import unicode_literals from __future__ import unicode_literals
import functools
import itertools import itertools
import operator
# import os # import os
import re import re
@ -129,9 +131,38 @@ class PornHubIE(InfoExtractor):
tv_webpage = dl_webpage('tv') tv_webpage = dl_webpage('tv')
video_url = self._search_regex( encoded_url = self._search_regex(r'(var.*mediastring.*)</script>',
r'<video[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//.+?)\1', tv_webpage, tv_webpage, 'encoded url')
'video url', group='url') assignments = encoded_url.split(";")
js_vars = {}
def parse_js_value(inp):
inp = re.sub(r'/\*[^*]*\*/', "", inp)
if "+" in inp:
inps = inp.split("+")
return functools.reduce(operator.concat, map(parse_js_value, inps))
inp = inp.strip()
if inp in js_vars:
return js_vars[inp]
# Hope it's a string!
assert inp.startswith('"') and inp.endswith('"')
return inp[1:-1]
for assn in assignments:
assn = assn.strip()
if len(assn) == 0:
continue
assert assn.startswith("var ")
assn = assn[4:]
vname, value = assn.split("=", 1)
js_vars[vname] = parse_js_value(value)
video_url = js_vars["mediastring"]
title = self._search_regex( title = self._search_regex(
r'<h1>([^>]+)</h1>', tv_webpage, 'title', default=None) r'<h1>([^>]+)</h1>', tv_webpage, 'title', default=None)