python 统计 英文 单词

import sys, os, re

def count_words(text):
    # !num !char -> ' '
    text = re.sub("\W+", " ", text)
    
    # number -> ' '
    text = re.sub("[0-9]", " ", text)
    
    # shrink whitespace
    text = re.sub("\s+", " ", text)
    
    return text.count(' ')

def getfiletext(file):
    file = open(file, 'rb')
    textb = file.read()
    file.close()
    
    # ignore gbk
    textb = textb.decode('gbk', 'ignore').encode('utf-8')
    # bytes -> str
    text = bytes.decode(textb, 'utf-8')
    return text

def main():
    for argv in sys.argv[1:]:
        #print(argv)
        print (count_words(getfiletext(argv)))
    
if __name__ == "__main__":
    main()

最近开始SSS阅读法，想看自己看过多少英文单词了，心血来潮用python写了个
主要利用 正则 替换，不是很准确，特别有乱码的时候，精度99.9%吧
反正是大概，无所谓啦