# coding:utf-8 class TableStruct(object): def __init__(self): """ 定义表格属性 """ self.min_x = None self.max_x = None self.min_y = None self.max_y = None self.contents = "" def parse(self, table): """ 解析表格结构 :param table: :return: """ contents = table.extract() if contents: self.contents = self.__table_format(contents) self.min_x, self.min_y, self.max_x, self.max_y = table.bbox @staticmethod def __table_format(data: list): """ 生成table标签数据,结构化数据 :param data: :return: """ if not data: return "" table = '' for row in data: table += '' for col in row: if col is None: col = '' table += '' % col table += '' table += "
%s
" return table