作业四

 

#读取文件
fo =open(‘bigworld.txt‘,‘r‘)
bigg=fo.read().lower()
fo.close()
print(bigg)
#大小写
str.lower(bigg)
#标点符号
strBig=bigg.replace(‘.‘,‘ ‘)
#特殊符号
sep=‘.,;:?!-_‘
for ch in sep:
    strBig=strBig.replace(ch,‘ ‘)
#字符串分隔
strList=strBig.split()
print(len(strList),strList)
#集合
strSet=set(strList)
exclude={‘the‘,‘i‘}
print(len(strSet),strSet)
#字典
strDict={}
for you in strSet:
    strDict[you]=strList.count(you)

print(len(strDict),strDict)
#列表
wcList=list(strDict.items())
print(wcList)
wcList.sort(key=lambda x:x[1],reverse=True)
print(wcList)
#top 20
for i in range(20):
    print(wcList[i])

classmates=[‘Tracy‘,‘Bod‘,‘Tracy‘,‘李三‘,‘Tracy‘]

print(classmates)
classmates.sort()
print(classmates)

score=[85,90,50,60,95]
cs=dict(zip(classmates,score))
print(cs)
csList=list(cs.items())
print(csList)

def takeSecond(elem):
    x=elem[1]
    return x

csList.sort(key=takeSecond,reverse=True)
print(csList

#英文小说
fo =open(‘blind.txt‘, ‘r‘)
big=fo.read().lower()
fo.close()
print(big)

#三国演义小说
import jieba


fo =open(‘bigbig.txt‘,‘r‘,encoding=‘utf-8‘)
biga=fo.read().lower()
fo.close()
print(biga)
print(jieba.cut(biga))
print(list(jieba.cut(biga)))
print(list(jieba.cut(biga,cut_all=True)))
print(list(jieba.cut_for_search(biga)))