201709080045 郑绮晴 追光者英文版歌词频统计.py

#1.
zqqFile = open('zqq.txt',mode="r",encoding='utf-8')
zqqText =zqqFile.read()
zqqFile.close()
print(zqqText)

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

#3.
print(zqqText.split(' '))
zqqList = zqqText.split(' ')

#4.
zqqSet = set(zqqList)
print(zqqSet)

zqqDict = {}
for word in zqqSet:
    zqqDict[word] = zqqList.count(word)

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

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

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

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