期末综合大作业:词频统计

#1.
hryFile = open('hry.txt',mode='r',encoding='utf-8')
hryText = hryFile.read()
hryFile.close()
print(hryText)

#2.
replaceList = [',',"'",'-','\n']
for c in replaceList:
    hryText = hryText.replace(c,' ')
print(hryText)

#3.
print(hryText.split(' '))
hryList = hryText.split(' ')

#4.
hrySet = set(hryList)
print(hrySet)

hryDict = {}
for word in hrySet:
   hryDict[word] = hryList.count(word)

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

#5.
wordCountList = list(hryDict.items())
print(wordCountList)
wordCountList.sort(key=lambda x:x[1],reverse=True)
print(wordCountList)

#6.
for i in range(20):
    print(wordCountList[i])

#7.
hryCountFile = open('ljjCount.txt',mode='a',encoding='utf-8')
for i in range(len(wordCountList)):
    hryCountFile.write(str(wordCountList[i][1])+' '+wordCountList[i][0]+'\n')
hryCountFile.close()

 

posted @ 2018-06-20 20:38  hry丿  阅读(93)  评论(0编辑  收藏  举报