综合练习：词频统计

file=open('song.txt','r',encoding='utf-8')
str=file.read()
# 将文章里的标点符号替换为空格
wordList =str.lower().split()
# 将数组变成一个个个单词
sep=''',.'"?!:'''
word = ['it', 'you', 'a', 'and', 'for', 'on', 'the', 'your']
# 产生一个专门存放要替换的标点符号的数组sep
for c in sep:
 str=str.replace(c," ")
# print(str)
# 将大写转化为小写
print(str.lower())
print("-----------------------------------")

# 计算每个单词出现的次数
wordDict={}
wordSet = set(wordList)
for w in wordList:
    wordDict[w]=wordList.count(w)
for w in wordDict:
    print(w,wordDict[w])
print("-----------------------------------")
for z in word:
    if z in wordDict.keys():
        del wordDict[z]
print("-----------------------------------")
count=sorted(wordDict.items(),key=lambda x:x[1],reverse=True)
print(count)
print("-----------------------------------")

for i in range(10):
    print(count[i])
file.close()

 

 

 

 

 

转载于:https://www.cnblogs.com/piglet00/p/8653019.html