去掉标签

"""
author:张鑫
date:2021/6/24 14:30
"""
import re


def remove_label(content):
    if '<' in content:
        pre = re.compile('>(.*?)<')
        content = content.replace('&nbsp', '')
        content = ''.join(pre.findall(content))
        return content
    else:
        content = content.replace('&nbsp', '')
        return content


if __name__ == '__main__':
    # a = remove_label('中国')
    # a=remove_label('<c>中国<c/>')
    # a=remove_label('<>中国</>')
    a = remove_label('&nbsp中国&nbsp钢铁')
    print(a)