Tfidf_count.py 文件源码

def handel_weibo(filename):
    fp = open("f://emotion/mysite/Label_extract/weibo_corpus/" + filename, 'r')
    contents = []
    for line in fp.readlines():    # ????
        line = line.strip()
        line.decode('utf-8')
        seg_lines = pseg.cut(line)  # ????
        for seg_line in seg_lines:   # ??????????
            if seg_line.flag == 'n' or seg_line.flag == 'nr' or seg_line.flag == 'ns' or seg_line.flag == 'nt' or seg_line.flag == 'nz':
                contents.append(seg_line.word)  # ????
    #print "length:", len(contents)
    fp.close()
    # ??????????
    fp_handel = open('f://emotion/mysite/Label_extract/weibo_corpus_handel/handel_' + filename, 'w+')
    for content in contents:
        fp_handel.write(content)
        fp_handel.write('\n')
    fp_handel.close()


# 2.???????????30????????????????