Shingle based fuzzy text comparison

shingle-compare.py

def shinglize(s):
    return frozenset(s[i:i + 3] for i in xrange(len(s) - 2))


def shingle_compare(sh1, string):
    '''set(shingle), 'string' -> float 0..1
    '''
    sh2 = shinglize(string)
    if not sh1 or not sh2:
        return 0
    inter = sh1 & sh2
    return float(len(inter)) / max(len(sh1), len(sh2))


def example():
    items = [] # open('file', 'rb') or cursor.fetchall() or else

    s = raw_input()
    needle_sh = shinglize(s)
    matches = []
    for item in items:
        score = shingle_compare(needle_sh, item)
        if score > 0.2:
            matches.append((item, score))
    matches.sort(key=lambda (item, score): (1 - weight, item))
    print(matches[:10])