词频统计

#1.
jianFile = open('jian.txt',mode='r',encoding='utf-8')
jianText = jianFile.read()
jianFile.close()
print(jianText)

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

#3.
print(jianText.split(' '))
jianList = jianText.split(' ')

#4.
jianSet = set(jianList)
print(jianSet)

jianDict = {}
for word in jianSet:
    jianDict[word] = jianList.count(word)

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

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

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

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