updateidf.py 文件源码

def get_word_count(filename):
    data_source=open(filename,'r')
    data=data_source.read()
    if(data!=''):
        temp_result = jieba.cut(data,cut_all=True)
        temp_result = '/'.join(temp_result)
        word_result=temp_result.split('/')
        word_view={}#word_view[i]?????????????????i?
        for i in word_result:
            word_view[i]=0
            if(i not in word_doc):
                word_doc[i]=0
        for i in word_result:
            if(word_view[i]==0):
                word_view[i]=1;
                word_doc[i]=word_doc[i]+1