def is_guessable(pattern, segments):
answer = True
if pattern in base_structures:
tags = guesser.unpack(pattern)
for i, segment in enumerate(segments):
if segment.word not in tag_dicts[tags[i]]:
answer = False
break
else:
answer = False
return answer
def main(base_structures, tag_dicts, file):
db = parser.connectToDb()
dictionary = parser.getDictionary(db, parser.dict_sets)
pos_tagger = BackoffTagger()
guessable_count = 0
i = 0
for l in file:
password = l.rstrip()
if not password : continue
segmentations = segment(password, dictionary, db)
guessable = False
for s in segmentations:
tags = pos_tagger.tag([ f.word for f in s if f.dictset_id <= 90])
for j in range(len(s)):
if s[j].dictset_id > 90:
s[j].pos = None
else:
s[j].pos = tags.pop(0)[1]
pattern = grammar.pattern(s)
guessable = is_guessable(pattern, s)
if guessable:
guessable_count += 1
break
print "{}\t{}\t{}".format(password, pattern, guessable, probability(pattern, guesser.unpack(pattern), s))
# print "{}\t{}".format(password, guessable)
i += 1
# if i >= 1000: break
# print # of guessable passwords
print "{} guessable passwords out of {}, ({:%})".format(guessable_count, i, float(guessable_count)/i)
