JinFile = open('jin.txt','r',encoding='utf-8')
JinText = JinFile.read()
JinFile.close()
print(JinText)
replacelist = ["'",'\n' ]
for c in replacelist:
jinText = JinText.replace(c,' ')
print(JinText)
#列表
print(jinText.split(' '))
jinlist = jinText.split(' ')
#集合中有哪些单词
jinSet = set(jinlist)
print(jinSet)
#统计单词字数
jinDict = {}
for word in jinSet:
jinDict[word]=jinlist.count(word)
print(jinDict)
for d in jinDict:
print(d,jinDict[d])
#排序
wordCountList = list(jinDict.items())
print(wordCountList)
wordCountList.sort(key=lambda x:x[1],reverse=True)
print(wordCountList)
#输出top20
for i in range(20):
print(wordCountList[i])
#写入文件
jinCountFile = open('jinCount.txt',mode='a',encoding='utf-8')
for i in range(len(wordCountList)):
jinCountFile.write(str(wordCountList[i][1])+' '+wordCountList[i][0]+'\n')
jinCountFile.close()
