期末综合大作业:词频统计

geFile = open("ge.txt",mode="r",encoding='utf-8')
geText = geFile.read()
geFile.close()
print(geText)

replaceList = [',','.',"'",'\n']
for c in replaceList:
    geText = geText.replace(c,' ')
print(geText)

print(geText.split(' '))
geList = geText.split(' ')

geSet = set(geList)
print(geSet)

geDict = {}
for word in geSet:
    geDict[word] = geList.count(word)

print(geDict)
for d in geDict:
    print(d,geDict[d])

wordCountList = list(geDict.items())
print(wordCountList)
wordCountList.sort(key=lambda x:x[1],reverse=True)
print(wordCountList)

for i in range(20):
    print(wordCountList)

geCountFile = open('geCount.txt',mode='a',encoding='utf-8')
for i in range(len(wordCountList)):
    geCountFile.write(str(wordCountList[i][1])+' '+wordCountList[i][0]+'\n')
geCountFile.close()

 

posted @ 2018-06-20 20:49  陈奕帆  阅读(102)  评论(0编辑  收藏  举报