Fix detection of .42 domains

This commit is contained in:
Valentin Lorentz 2011-01-02 13:22:54 +01:00
parent 3582286658
commit 4649188b96

View File

@ -57,7 +57,7 @@ _octet = r'(?:2(?:[0-4]\d|5[0-5])|1\d\d|\d{1,2})'
_ipAddr = r'%s(?:\.%s){3}' % (_octet, _octet) _ipAddr = r'%s(?:\.%s){3}' % (_octet, _octet)
# Base domain regex off RFC 1034 and 1738 # Base domain regex off RFC 1034 and 1738
_label = r'[0-9a-z][-0-9a-z]*[0-9a-z]?' _label = r'[0-9a-z][-0-9a-z]*[0-9a-z]?'
_domain = r'%s(?:\.%s)*\.[a-z][-0-9a-z]*[a-z]?' % (_label, _label) _domain = r'%s(?:\.%s)*\.[0-9a-z][-0-9a-z]+' % (_label, _label)
_urlRe = r'(\w+://(?:%s|%s)(?::\d+)?(?:/[^\])>\s]*)?)' % (_domain, _ipAddr) _urlRe = r'(\w+://(?:%s|%s)(?::\d+)?(?:/[^\])>\s]*)?)' % (_domain, _ipAddr)
urlRe = re.compile(_urlRe, re.I) urlRe = re.compile(_urlRe, re.I)
_httpUrlRe = r'(https?://(?:%s|%s)(?::\d+)?(?:/[^\])>\s]*)?)' % (_domain, _httpUrlRe = r'(https?://(?:%s|%s)(?::\d+)?(?:/[^\])>\s]*)?)' % (_domain,