部分代码1

原文的一部分原文

#!/usr/bin/env python
#-*- coding:utf-8 -*-
#author: Enoch time:2018/10/30 0030

import re
import time
from collections import Counter
t0 = time.clock()
###################################################################################
#Name:count_words
#Inputs:file name,the first n words, stopfile name
#outputs:None
#Author: Enoch
#Date:2018.10.22
###################################################################################
def CountWords(file_name):

    with open(file_name) as f:
        txt = f.read().lower()

    pattern = r"[a-z][a-z0-9]*"
    wordList = re.findall(pattern,txt)
    totalNum = len(wordList)
    tempc = Counter(wordList)

    dicNum = dict(tempc.most_common(10))
    dicNum = sorted(dicNum.items(), key=lambda k:k[0])
    dicNum = sorted(dicNum, key=lambda k:k[1], reverse=True)
    t1 = time.clock()
    for letter, fre in dicNum[:2]:
        print("|\t{:15}|{:<11.2%}|".format(letter, fre))
    print(t1 - t0)

CountWords('../gone_with_the_wind.txt')

猜你喜欢

转载自blog.csdn.net/qq_36097393/article/details/83621095