gistfile1.txt

# Check that nltk and strfry produce same levenshtein distance on
# a bunch of randomly generated strings
import strfry
import nltk.metrics
import random

for i in xrange(0, 10000):
    a = ''.join([chr(random.randint(1, 255)) for x in xrange(0, random.randint(0, 20))])
    b = ''.join([chr(random.randint(1, 255)) for x in xrange(0, random.randint(0, 20))])

    nt = nltk.metrics.edit_distance(a, b)
    st = strfry.levenshtein_distance(a, b)

    if nt != st:
        print "Got different results for '%s' and '%s'" % (a, b)
        print "NLTK: %d" % nt
        print "strfry: %d" % st