util.py 文件源码

def get_pub_dic_xml(file_name = 'data/proton-beam-all.xml'):
    tree = ET.parse(file_name)
    root = tree.getroot()[0]

    # Create dic of : id -> text features
    pub_dic = {}
    for pub in root:
        rec_number = int (get_text (pub.find('rec-number')))
        abstract   = get_text (pub.find('abstract'))
        title      = get_text (pub.find('titles')[0])
        text = title + abstract
        for kw in pub.find('keywords'):
            text = text + kw.text + ' '
        pub_dic[rec_number] = text

    return pub_dic