new_tiku_structure_2021/structure/paper_text_structure.py

from utils.washutil import HtmlWash


class WordParseStructure:
    def __init__(self, html, wordid, is_reparse=0, subject=""):
        self.html = html
        self.is_reparse = is_reparse
        self.wordid = wordid
        self.subject = subject

        # 清洗
        self.row_list, self.subs2src, self.new_html = HtmlWash(self.html,self.wordid,
                                                               self.is_reparse).html_cleal()