第四次作业

strHello='''...'''.lower()
fo = open('qqq.txt', 'r', encoding='utf-8')
hello = fo.read()
fo.close()
print(hello)
sep = ''',?'''
for ch in sep:
    strHello = strHello.replace(ch, '')

    strList = strHello.split()
    print(len(strList),strList)
    strSet = set(strList)
    exclude = {'i', 'in', 'the''anymore'}
    strSet = strSet-exclude

    print(len(strSet),strSet)

    strDict = {}
    for hello in strSet:
        strDict[hello] = strList.count(hello)

        print(strDict.items())

wcList = list(strDict.items())
wcList.sort()
print(strDict.items())
print(wcList[:20])

import jieba
fo = open('z.txt','r',encoding='utf-8')
ci = fo.read()
fo.close()
print(ci)
ci.replace('，','')
print(ci)
print(list(jieba.cut(ci)))
print(list(jieba.cut(ci,cut_all=True)))
print(list(jieba.cut_for_search(ci)))

cilist = jieba._lcut(ci)
cidict = {}
for word in cilist:
    if len(word) == 1:
        continue
    else:
        cidict[word] = cidict.get(word,0)+1
print(cidict)

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

for i in range(5):
    print(cilist[0])