def categorized (domain):
""" Provides a discretized categorization on an input domain, and
returns a DCat object dcat, where dcat.cat==
'benign', 'gray', 'suspicious', 'notsure', ...
Note the suspicion here is determined on a per-domain basis,
without considering the clustering effects among a bunch of names.
"""
global Cached
try:
return Cached[domain]
except KeyError:
ds = publicsuffix.DomainStruct(domain)
# 'benign' determination
if (not ds.isFQDN
or whitelist.is_whitelisted_t2ld(ds.eTkLD[1])
or dnsblav.is_dnsblav_service(ds)
or alexa.in_top(ds.eTkLD[1], 50000)
#or good WOT
): cat = 'benign'
# 'suspicious', to be done
elif ds in blacklisted_domains:
cat = 'suspicious'
# 'gray', e.g., non-top-Alexa p2p/p**n sites
elif (
p2p.is_p2p_domain(domain)
or p**n.is_porn_domain(domain)
): cat = 'gray'
# 'notsure'
else:
cat = 'notsure'
dcat = DCat(cat, None) # None: to be implemented
Cached[domain] = dcat
return dcat
def test_is_porn_domain(self):
self.assertFalse(p**n.is_porn_domain("mail.google.com"))
self.assertTrue(p**n.is_porn_domain("boobs.com"))
self.assertTrue(p**n.is_porn_domain("adultvideo.com"))
