tach_van.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
import codecs
import string

PATH = '/home/famihug/python/'


f1 = codecs.open(PATH + 'am', 'r', "utf-8")
nguyen_am = f1.readline()

kq = u''
f = codecs.open(PATH + 'in', 'r', 'utf-8')
table = string.maketrans( '', '', )
d = {}
#string = f.readline()
for s in f:
    s = s.lower()
    ls = s.split(" ")
    for word in ls:
        for i in range(0,len(word)):
            if nguyen_am.__contains__(word[i]):
                removed_punc = u"".join(c for c in word[i:] if c not in string.punctuation)
                if d.has_key(removed_punc):
                    d[removed_punc] += 1
                else:
                    d[removed_punc] = 1

                kq += removed_punc
                kq += " "
                break;

f2 = codecs.open(PATH + 'out', 'w', 'utf-8')
f2.write(kq)
for k in d.keys():
    f2.write(str(d[k]) + ":" + k  + "\n")