#!/usr/bin/env python
# -*- coding: utf-8 -*-
from math import sqrt
from text import freq, trigrams, tokenize, depunctuate

def model(filename):
    text = open(filename).read()
    text = unicode(text)
    cleaned = ' '.join(tokenize(depunctuate(text)))
    print cleaned
    return freq(trigrams(cleaned))

def randmodel():
    import os
    from random import choice 
    data= '/home/pat/l/udhr/data/'
    randlang = choice(os.listdir('/home/pat/l/udhr/data'))
    return model(data + os.sep + randlang)

def scalar(vec):
    total = 0 
    for elem in vec:
        total += vec[elem] * vec[elem]
    return sqrt(total)

def sim(v, w):
    total = 0 
    for elem in v:
        if elem in w:
            total += v[elem] * w[elem]
    return float(total) / (scalar(v) * scalar(w))

if __name__ == "__main__":
    import sys
    a, b = sys.argv[1:3]
    #for k,v in model(a).items(): print k, v
    print sim(model(a),model(b))