js_to_json,
orderedSet,
# sanitized_Request,
+ remove_quotes,
str_to_int,
)
# from ..aes import (
_VALID_URL = r'''(?x)
https?://
(?:
- (?:[a-z]+\.)?pornhub\.com/(?:view_video\.php\?viewkey=|embed/)|
+ (?:[a-z]+\.)?pornhub\.com/(?:(?:view_video\.php|video/show)\?viewkey=|embed/)|
(?:www\.)?thumbzilla\.com/video/
)
(?P<id>[\da-z]+)
}, {
'url': 'https://www.thumbzilla.com/video/ph56c6114abd99a/horny-girlfriend-sex',
'only_matching': True,
+ }, {
+ 'url': 'http://www.pornhub.com/video/show?viewkey=648719015',
+ 'only_matching': True,
}]
@staticmethod
tv_webpage = dl_webpage('tv')
- encoded_url = self._search_regex(r'(var.*mediastring.*)</script>',
- tv_webpage, 'encoded url')
- assignments = encoded_url.split(";")
+ assignments = self._search_regex(
+ r'(var.+?mediastring.+?)</script>', tv_webpage,
+ '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 = 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]
+ return remove_quotes(inp)
for assn in assignments:
assn = assn.strip()
- if len(assn) == 0:
+ if not assn:
continue
-
- assert assn.startswith("var ")
- assn = assn[4:]
- vname, value = assn.split("=", 1)
-
+ assn = re.sub(r'var\s+', '', assn)
+ vname, value = assn.split('=', 1)
js_vars[vname] = parse_js_value(value)
- video_url = js_vars["mediastring"]
+ video_url = js_vars['mediastring']
title = self._search_regex(
r'<h1>([^>]+)</h1>', tv_webpage, 'title', default=None)