import re
def find_word_freq(file):
file1= open(file,'r')
lines = file1.readlines() #读取全部内容
dic ={}
for line in lines:
re.sub(r'[?.!,""/]",' ',line) #特殊字符替换成空格
re.sub(r" - "," ",line) #替换单独的-,不是同一单词里的连字符
for word in line.split():#分隔默认以空格分
if word[-1]="-":
has_lianjie =True
m=word[:-1]
break
if has_lianjie:
has_lianjie= False
word = m+word
dic.setdefault(word.lower(),0)
dic[word.lower()]+=1
print dic
find_word_freq("D:\pythonex\eng1.txt")
网友评论