练习006

#!/usr/bin python
#coding:utf-8
'''
Created on 2016年4月25日
@author: zxc
'''
import os,re
from collections import Counter
FILE_PATH=r'F:\Python\txt'
stop_word = ['the', 'in', 'of', 'and', 'to', 'has', 'that', 's', 'is', 'are', 'a', 'with', 'as', 'an','for']
def getCounter(i):
pattern = r'[A-Za-z]+|\$?\d+%?$'
print i
i = os.path.join(FILE_PATH,i)
print i
f=open(i,'r')
r = re.findall(pattern, f.read())
return Counter(r)

def run(FILE_PATH):
total_counter = Counter()
for i in os.listdir(FILE_PATH):
if os.path.splitext(i)[1] == '.txt':
#print os.path.splitext(i)[1]
total_counter += getCounter(i)
for i in stop_word:
total_counter[i] = 0
print total_counter.most_common()[0][0]

if __name__=="__main__":
run(FILE_PATH)