manage.py 文件源码

def parse():
    """parse the comments"""
    import jieba
    import jieba.posseg as pseg

    # Load User's Dictionary
    path_list = os.getcwd().split('/')
    path_list.append("dict.txt")
    dict_path = '/'.join(path_list)
    jieba.load_userdict(dict_path)

    # Disimss These Flags
    dismiss = ['b', 'c', 'r', 'uj', 'u', 'p', 'q', 'uz', 't', 'ul', 'k', 'f',
            'ud', 'ug', 'uv']

    comments = Comment.query.all()
    for comment in comments:
         word_list = []
         pseg_cut = pseg.cut(comment.body)
         for word, flag in pseg_cut:
             if flag not in dismiss:
                 word_list.append(word)
         comment.parsed = '/'.join(word_list)
         db.session.add(comment)
         print "Comment %04d Parsed!" % comment.id

    db.session.commit()
    print "ALL DONE!"