data_quality_server/util/info_preprocess/attach_preprocess.py

# coding:utf-8
from util.htmltag import CleanKeepTable
from typing import AnyStr
from loguru import logger


class AttachHelper(object):

    @staticmethod
    def __clear_tag(content: AnyStr) -> AnyStr:
        """
        清理标签
        :param content:
        :return:
        """
        try:
            tag_html = CleanKeepTable(content)  # 保留 table标签
        except Exception as e:
            logger.warning(e)
            return content
        return tag_html

    @staticmethod
    def __add_space(content_html):
        """
        为表格添加空格
        :param content_html:
        :return:
        """
        if "<table" not in content_html:
            return content_html
        if "<thead>" in content_html:
            content_html = content_html.replace('<thead>', '')
        if "<br/" in content_html:
            content_html = content_html.replace(r'<br/', '')
        if "<table " in content_html and "<td " in content_html and "<tr " in content_html:
            return content_html
        content_html = content_html.replace('<tbody', '<tbody ')
        content_html = content_html.replace('<table', '<table ')
        content_html = content_html.replace('<th', '<th ')
        content_html = content_html.replace('<tr', '<tr ')
        content_html = content_html.replace('<td', '<td ')
        return content_html

    def preprocess(self, content: AnyStr) -> AnyStr:
        """
        处理开始
        :param content:
        :return:
        """
        content = self.__add_space(content)
        content = self.__clear_tag(content)
        return content