Tfidf_count.py 文件源码

def read_test_list():
    fp = open("f://emotion/mysite/weibo_crawler/chinese_weibo.txt", '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)  # ????
    fp.close()
    #for w in contents:
    #   print w

    # ??str???????
    str_test = ' '.join(contents)
    return str_test


# 5.??????chinese_weibo.txt??????TF-IDF???????100??