mirror of https://github.com/CIRCL/AIL-framework
fix: [vanity] debug
parent
cb4345c871
commit
784579baef
|
@ -664,7 +664,6 @@ def get_vanity_domains(vanity, len_vanity=4, meta=False):
|
||||||
domains = []
|
domains = []
|
||||||
for domain in r_crawler.smembers(f'vanity:4:{vanity[:4]}'):
|
for domain in r_crawler.smembers(f'vanity:4:{vanity[:4]}'):
|
||||||
dom_vanity = get_domain_vanity(domain, len_vanity=len_vanity)
|
dom_vanity = get_domain_vanity(domain, len_vanity=len_vanity)
|
||||||
print(vanity, dom_vanity)
|
|
||||||
if vanity == dom_vanity:
|
if vanity == dom_vanity:
|
||||||
domains.append(domain)
|
domains.append(domain)
|
||||||
if meta:
|
if meta:
|
||||||
|
@ -680,13 +679,14 @@ def get_vanity_cluster(vanity, len_vanity=4, nb_min=4):
|
||||||
return get_vanity_clusters(nb_min=nb_min)
|
return get_vanity_clusters(nb_min=nb_min)
|
||||||
else:
|
else:
|
||||||
clusters = {}
|
clusters = {}
|
||||||
|
print(vanity, len_vanity, get_vanity_domains(vanity[:4], len_vanity=4))
|
||||||
for domain in get_vanity_domains(vanity[:4], len_vanity=4):
|
for domain in get_vanity_domains(vanity[:4], len_vanity=4):
|
||||||
new_vanity = get_domain_vanity(domain, len_vanity=len_vanity)
|
new_vanity = get_domain_vanity(domain, len_vanity=len_vanity)
|
||||||
if vanity not in clusters:
|
if vanity not in clusters:
|
||||||
clusters[new_vanity] = 0
|
clusters[new_vanity] = 0
|
||||||
clusters[new_vanity] += 1
|
clusters[new_vanity] += 1
|
||||||
to_remove = []
|
|
||||||
print(clusters)
|
print(clusters)
|
||||||
|
to_remove = []
|
||||||
for new_vanity in clusters:
|
for new_vanity in clusters:
|
||||||
if clusters[new_vanity] < nb_min:
|
if clusters[new_vanity] < nb_min:
|
||||||
to_remove.append(new_vanity)
|
to_remove.append(new_vanity)
|
||||||
|
|
Loading…
Reference in New Issue