new_tiku_structure_v3_art/utils/washutil.py

#!/usr/bin/env/python
# -*- coding:utf-8 -*-

# 本文件包含以下函数
# table_label_cleal：去掉表格中的换行符
# html_cleal ：html文件清洗
# wash_after: 处理最终结果多余的换行符

import re
import shutil
# from operator import itemgetter
# from itertools import groupby
# from PIL import Image
import base64, os, random
import time
import hashlib
from pprint import pprint
# from bs4 import BeautifulSoup
# UPLOAD_FOLDER = config.UPLOAD_FOLDER
import configs
from structure.stems_to_groups import parse_split2group, suojin
from utils.equation_extract import get_equation_instr, get_simpstr2eqn
from utils.field_eq2latex import get_latex
from utils.html_again_parse import css_label_wash
from structure import stems_to_groups


def table_label_cleal(con):
    """
    去掉表格中的【换行符】
    """
    # print(con)
    # print('------------------------------------------')
    con = re.sub(r"<br\s*/?>", "\n", con)
    con = re.sub(r"\n(\n|\t)+", "\n", con)
    count = 1
    while re.search(r"</?[a-z]+>\n(</?[a-z]+>|<td\s+\n*[a-z=\"\d]+>)", con, re.S) and count <= 10:
        con = re.sub("(</?t[dr]>|</?table>|</?tbody>|</?div>)\n(</?t[dr]>|</div>|</?table>|</?tbody>|<p>)",
                     r"\1\2", con, flags=re.S)
        con = re.sub(r'(</?t[rd]>)\n(<td\s.+?>)', r'\1\2', con, flags=re.S)
        count += 1
    # if re.search(r"<table>(.|\n)+?</table>", con, re.S|re.M):
    #     aa = re.search(r"(<table>(.|\n)+?</table>)", con, re.S|re.M)
    #     con = con.replace(aa.group(1),aa.group(1).replace("\n",""))

    # 将空表格的情况去掉
    con = re.sub(r'<table>[\s\n\t]*?<tbody>[\s\n\t]*?(<tr>[\s\n\t]*?<td[^<>]*?>[\s\n\t]*?<p>[\s\n\t]*?</p>'
                 r'[\s\n\t]*?</td>[\s\n\t]*?</tr>[\s\n\t]*?)+</tbody>[\s\n\t]*?</table>[\s\n\t]*?<p>', "", con,
                 flags=re.S)
    con = re.sub(r'(</table><p>)\s*([(（]\s*\d\s*[)）])', r'\1\n\2', con)
    return con


def base642img(html_data, wordid):
    """
    【基于mathjax渲染输出是css-html格式】
    将base64编码的图片保存到本地
    :return:
    """
    # 二进制图片进行转化， 按“word_id”建立文件夹
    # time_str = datetime.datetime.strftime(datetime.datetime.now(), '%Y_%m_%d')
    # file_path = configs.IMG_FOLDER + '/' + str(self.wordid)
    # if not os.path.exists(file_path):
    #     os.makedirs(file_path)
    # else:
    # 思路1：删除图片,重建文件夹，【所有的新图片都是以base64格式传过来的】
    # shutil.rmtree(file_path)
    # os.makedirs(file_path)
    # 思路2：每一次再解析都将base64图片保存到本地再以路径形式返回
    # st = len(os.listdir(file_path))  # 不要以序号索引的形式命名

    # 统计所有base64编码
    all_base64_image = re.findall(r'(<img ([a-z]+="[^"]*?" )?src="(data:image[^>"]+?)"(.*?)\s*/?>)', str(html_data), flags=re.S)
    if all_base64_image:
        file_path = configs.IMG_FOLDER + '/' + str(wordid)
        if not os.path.exists(file_path):
            os.makedirs(file_path)
        # 新图片命名1、时间戳+随机数；2、md5值命名
        # name_list = random.sample(range(100000, 999999), len(all_base64_image))
        for n, img in enumerate(all_base64_image):
            img1 = img[2].split(",", maxsplit=1)
            img_type_info = re.search("data:image/(.+?);base64", img1[0])
            img_type = img_type_info.group(1) if img_type_info else ""  # 图片格式
            # 可能还有alt和style的属性，暂时先不要
            w_info = re.search('( width="\d+")', img[3])
            h_info = re.search('( height="\d+")', img[3])
            img_data = base64.b64decode(str(img1[-1]))
            if img_type:
                # save_path = os.path.join(configs.new_img_ip, get_md5(n)+"."+img_tape)
                pmd5 = hashlib.md5(img_data)
                # img_name = "new_image" + str(int(time.time())) + str(name_list[n]) + "." + img_type  # 时间戳+随机数
                img_name = "new_image" + pmd5.hexdigest() + "." + img_type
                save_path = os.path.join(file_path, img_name)
                with open(save_path, 'wb') as f:
                    f.write(img_data)
                flag_behind = '" />'
                if w_info and h_info:
                    flag_behind = '"' + w_info.group(1) + h_info.group(1) + ' />'
                temp_img = '<img src="' + configs.new_img_ip + '/' + str(wordid) + '/' + img_name + flag_behind
                html_data = html_data.replace(img[0], temp_img)
    return html_data


class HtmlWash:
    """
    html文本清洗
    批量再解析中，新增图片信息替换的文本返回作为ocr保存文本，
    继续往下清洗的文本，则进入结构化解析逻辑中
    """
    def __init__(self, html, wordid, subject, is_reparse=0):
        """
        html文本清洗
        批量再解析中，新增图片信息替换的文本返回作为ocr保存文本，
        继续往下清洗的文本，则进入结构化解析逻辑中
        """
        # super().__init__(html, wordid, is_reparse, must_latex)
        self.html = html
        self.wordid = wordid
        self.is_reparse = is_reparse
        self.subject = subject
        self.sub_list = ["</?div>", "</?b>", "</?caption>", "</?center>", "</?cite>", "</?code>", "</?colgroup>",
                         "</?menu>", "</?dd>", "</?dir>", "</?li>", "</?article>", "</?header>", "</?ruby>",
                         "</?summary>", "</?details>", "</?strike>", "</?small>", "</?select>",
                         "</?section>", "</?script>", "</?[su]>", "</?var>", "</?ul>", "</?tt>", "</?title>",
                         "</?thead>",
                         "</?tfoot>", "<hr />", "<hr>", ""]
        self.sub_dd = {'&times;': '×',
                       '&divide;': '÷',
                       '&deg;': '°',
                       '&middot;': '·',
                       '&plusmn;': '±',
                       '&ordm;': 'º',
                       '&sup1;': '¹',
                       '&sup2;': '²',
                       '&sup3;': '³',
                       '&frac12;': '1/2',
                       '&frac14;': '¼',
                       '&frac34;': '¾',
                       '&yen;': '¥',
                       'm&sup3;': 'm³',
                       # '&lt;': '<',
                       '&pound;': '£',
                       # '∠&lt;': '&lt;',
                       '&gt;': '>',
                       "Ａ": "A",
                       "А": "A",
                       "Α": "A",
                       "Ｂ": "B",
                       "В": "B",
                       "в": "B",
                       "Β": "B",
                       "Ｃ": "C",
                       "С": "C",
                       "ｃ": "c",
                       "с": "c",
                       "Ｄ": "D",
                       "Ε": "E",
                       "Ｅ": "E",
                       "Ｆ": "F",
                       "Ｇ": "G",
                       "ｇ": "g",
                       "ｍ": "m",
                       "Ｎ": "N",
                       "ｓ": "s",
                       "ｔ": "t",
                       "／": "/",
                       "＝": "=",
                       "－": "-",
                       "２": "2", "３": "3", "４": "4", "５": "5", "６": "6",
                       "７": "7", "８": "8", "９": "9", "１": "1", "０": "0",
                       '&nbsp;&nbsp;': ' ',
                       '&nbsp;': ' ',
                       "〖": '【',
                       "〗": '】',
                       "題": '题',
                       "单项选择": '单选',
                       "多项选择": '多选',
                       "不定项选择": '选择',
                       "双项选择": '多选',
                       "实验与探究题": '实验',
                       }

    def new_pic_sub(self):
        """
        针对base64图片先保存到本地，入库时再换成腾讯云线上地址
        # 第一版：再解析中，将二进制图片进行转化,图片怎么保存比较好，先再“天数”建立文件夹
        第一版：再解析中，根据“word_id”建立文件夹
        :return:
        """
        if self.is_reparse:
            # css 标签清洗
            self.html = css_label_wash(self.html)
            # print(self.html)
            # 保存base64编码的图片
            self.html = base642img(self.html, self.wordid)
        self.new_html = self.html
        self.new_html = table_label_cleal(self.new_html)
        self.new_html = self.new_html.replace("\n", "<br/>")

    def html_cleal(self):
        # =======清洗mathjax标签========
        if "MathJax" in self.html:  # 再解析中存在mathjax公式渲染的标签
            all_mathjax = re.findall('(<span class="MathJax_Preview".*?</script>(</span>)*)', self.html)
            for jax in all_mathjax:
                latex = re.findall('<script .+?">(((?!(</)).)*?)</script>(</span>)*', jax[0])
                if latex:
                    latex = "${}$".format(latex[0][0])
                    self.html = self.html.replace(jax[0], latex)
                else:
                    self.html = self.html.replace(jax[0], "")

        # ======再解析中的css清洗及新图片处理=====
        self.new_pic_sub()
        # =====特殊符号处理=====
        html2txt = re.sub(r"|".join(self.sub_list), "", str(self.html))  # ("", " ")   #2020/4/7
        html2txt = re.sub("|".join(self.sub_dd.keys()), lambda x: self.sub_dd[x.group()], html2txt)  # 2020/4/1,4/7,4/20
        html2txt = re.sub("[不非]定[向项]选择", "不定选择", html2txt)
        html2txt = html2txt.replace(r"\\[{\\text{V}}V\]", "Ⓥ").replace(r"\\[{\\text{A}}A\]", "Ⓐ") \
            .replace(r"\\[{\\text{W}}W\]", "Ⓦ").replace(r"\\[{\\text{X}}X\]", "Ⓧ").replace(r"\\[{\\text{G}}G\]", "Ⓖ") \
            .replace("\uf067", "γ").replace('', "γ").replace('\uf020', "").replace("\u3000", " ")\
            .replace("\u2003", " ").replace("\x7f", " ").replace("\xa0", "")
        html2txt = re.sub(r"(<p>\s*)【例题(\d+)】", r"\1\2、", html2txt)
        html2txt = re.sub(r"\\\(|\\\)", "$", html2txt)
        # 把格式标签单独拆分为一行，再在题目切分为小题完了之后再组起来！！！！！
        html2txt = re.sub(r'(<p style="text-(indent|align):.*?">)', r"\1</p><p>", html2txt)

        # 域公式的转化处理；<sub>\<sup>可以在前端显示，不需要用latex渲染
        try:
            html2txt, newhml = get_latex(html2txt, self.is_reparse, self.wordid)
            if newhml:  # 存在域公式转图片时，需要将原文本的域公式也转为图片信息
                self.new_html = newhml
            html2txt = html2txt.replace("【omml-latex】", "")
        except:
            html2txt = html2txt.replace("【omml-latex】", "")

        # 字符串公式的处理：如Fe<sub>2</sub>O<sub>3</sub>, 在结构化之后处理比较好
        # <br/>处理
        html2txt = re.sub(r"<br\s*/?>", "\n", html2txt)

        # =====题型行的统一处理=====
        # ---->>>>>题型行可能放在表格中
        if len(re.findall("</table>", html2txt)) >= 8:  # 这个限制还不太严谨
            for tt in re.finditer('<tr>(((?!(</?tr>)).)*)</tr>', html2txt, re.S):
                tt_list = re.split(r'^\s*<td[^<>]*?>|</p></td>|</td>[\n\s]*?<td[^<>]*?>'
                                   r'|</td>\s*\n|</td>\s*$|\n\s*<td[^<>]*?>|<td[^<>]*?><p>',
                                   tt.group(1).strip())   # </td>\s*[$\n]这样无效
                tt_list = [col for col in tt_list if col.strip()]
                if " ".join(tt_list).replace(" ", "") in ['得分评卷人', '评卷人得分']:
                    html2txt = html2txt.replace(tt.group(0), "")
                else:
                    pass
                    # html2txt = html2txt.replace(tt.group(0), "<p>" + " ".join(tt_list) + "</p>")
                # html2txt = re.sub(r"</?tbody>|</?table>|</?div>", "", html2txt)

        # ---->>>>>end
        html2txt = re.sub(r"(</table>)\s*([一二三四五六七八九十]\s*[、.．､：:]?.{2,6}题)", r"\1</p>\2", html2txt)
        html2txt = re.sub(r'([一二三四五六七八九十])\s*[、.．､，,：:]\s*(论述|填空|探究)题?[与和､、，,\s]*?(计算题|实验题)', r"\1､\3", html2txt)
        html2txt = re.sub(r'<td[^<>]*?><p>(([一二三四五六七八九十])\s*[、.．､，,：:]\s*(.{2,4}题)\s*</p>)</td>[^p]*?<p>', r"\1",
                          str(html2txt), flags=re.S)
        html2txt = re.sub(r'<td[^<>]*?>\s*(([一二三四五六七八九十])\s*[、.．､，,：:]\s*(.{2,4}题)\s*)</td>[\n\s]*?</tr>', r"\1",
                          str(html2txt), flags=re.S)
        html2txt = re.sub(r"<p>\s*([一二三四五六七八九十])\s*[、.．､,，：:]?\s*(计算|[解简]答|实验|作图)题?[与和、､，,\s]*?(计算|[解简]答|实验|作图)",
                          r"<p>\1､\2题", html2txt)
        html2txt = re.sub(r'<p>\s*[(（]\s*[一二三四五六]\s*[)）]\s*必考题\s*(.?|.+?分\s*[.。．]?)\s*</p>', "", html2txt)
        html2txt = re.sub(r'<p>\s*[(（]\s*[一二三四五六]\s*[)）]\s*选考题\s*.?\s*.{,4}(?<!\d)(\d+分)\s*[,，。].{,50}</p>',
                          r"<p>【选做题】:'\1'</p>", html2txt)
        html2txt = re.sub(r'<p>\s*[(（]\s*[一二三四五六]\s*[)）]\s*选考题\s*(.?|.+?分\s*[.。．]?)\s*</p>', "<p>【选做题】</p>", html2txt)
        html2txt = re.sub(r'<p>\s*([一二三四五六七八九十])\s*[、.．､，,：:]?\s*(单项?选择?|非?选择|多项?选择?|不定选择|填空|计算|[解简]答|实验|作图)题?\s*</p>',
                          r"<p>\1､\2题</p>", html2txt)
        html2txt = re.sub(r'([一二三四五六七八九十])\s*[、.．､，,：:]?\s*(单选|单项选择|选择|不定选择|多选|多项选择|填空|计算|[解简]答|实验|作图)\s*(?!题)'
                          r'([(（]\s*本题|.*?\d分)', r"\1" + "､" + r'\2' + "题" + r"\3", html2txt)
        html2txt = re.sub(r'([一二三四五六])\s*[、.．､，,:：]?\s*(单选|单项选择|非?选择|不定选择|多选|多项选择|填空|计算|[解简]答|实验|作图)题',
                          r"\1" + "､" + r'\2' + "题", html2txt)
        html2txt = re.sub(r'([一二三四五六七八九十])\s*[、.．､，,：:]?\s*[(（]\s*本大题(.*?选项中)', r"\1" + "､" + "选择题", html2txt)  # + r"\2"
        html2txt = re.sub(r'<p>\s*([一二三四五六七八九十])\s*[、.．､，,]?\s*[(（本大题]*?(.*?选项中)', r"\1" + "､" + "选择题", html2txt)
        if self.subject != "语文":
            html2txt = re.sub(r'([一二三四五六七八九十])\s*[、.．､，,：:]?\s*([(（]\s*(每小题|本大?题)((?!(选项)).)+?[）)]|综合题)',
                              r"\1" + "､" + "解答题", html2txt)
        else:
            html2txt = re.sub(r'([一二三四五六七八九十])\s*[、.．､，,：:]?\s*([(（]\s*(每小题|本大?题)((?!(选项)).)+?[）)]|综合题)',
                              r"\1" + "､" + "综合题", html2txt)
        html2txt = re.sub(r'(?<!<p>)\s*([一二三四五六七八九十]\s*[、.．､，,：:]?\s*(单项?选择?|选择|不定选择|多项?选择?|填空|计算|[解简]答|实验|作图)题)',
                          r'</p>\n<p>\1', html2txt)
        # html2txt = re.sub(r'<p>\s*第[Ⅰ1I]卷\s*</p>\n?<p>\s*本卷共\d{1,2}个?小?题.*?四个选项.*?\n?<p>\s*(\d\s*[、.．､])'
        #                   r'|<p>第[Ⅰ1I]卷\s*[（(]选择题[）)]</p>\n?<p>本卷共\d{1,2}个?小?题.*?\n?<p>\s*(\d\s*[、.．､])',
        #                   r"<p>一、选择题</p>\n<p>\1\2", html2txt)
        # html2txt = re.sub(r'<p>\s*[^一二三四五六七八九十]{,3}\s*[、.．､]\s*(选择|不定选择|单选|多选|计算|[解简]答|实验|作图)题', r"<p>一､\1题", html2txt)

        # =====答案解析关键字的统一处理=====
        html2txt = re.sub(r'【\s*(<img src=((?!/>).)+?/>\s*)*?([解答])\s*(<img src=((?!/>).)+?/>\s*)*?([析案])\s*'
                          r'(<img src=((?!/>).)+?/>\s*)*?】', r"【\3\6】", str(html2txt))  # 2022/4/28
        html2txt = re.sub(r'<p>\s*(解\s*[：:])', r"<p>【解答】", str(html2txt))
        html2txt = re.sub(r'(<p>|<br\s*/>|\n)\s*(参考译文\s*[：:])', r"\1【参考译文】", str(html2txt))
        html2txt = re.sub(r'【[^【】]*?(答案|[解分][析答]|详解|点[评睛])[^【】]*?】', r"【\1】", str(html2txt))
        html2txt = re.sub(r'(\n\s*|\s{2,})(答案|解析|解答|详解|点评|点睛|考点|专题)\s*[：:]', r"\1【\2】", str(html2txt))
        html2txt = re.sub(r'(\n|^|<p>)\s*(([1-9]|[1-4][0-9])\s*[.．、､])?\s*\[\s*(答案|解析|解答|详解|点评|点睛|考点|专题)\s*\]',
                          r"\1\2【\4】", str(html2txt))
        html2txt = re.sub(r'(\n|^)\s*(分析)\s*[：:]', r"【\2】", str(html2txt))
        if "【解析】" not in html2txt and "【解答】" in html2txt and "【分析】" not in html2txt:
            html2txt = re.sub(r'【解答】', "【解析】", str(html2txt))

        # =====其他关键字的处理=====
        html2txt = re.sub(r'<p>\s*(类型|知识点|考查角度|拔尖角度)[一二三四五六七八九十\d+][^p]*?</p>', "", str(html2txt))
        html2txt = re.sub(r'<p>\s*(选修[\d-]*?[：:].{2,15})\s*</p>', r"<p>【章节】\1</p>", html2txt)
        html2txt = re.sub(r'<p>\s*([一二三四五六]\s*[、.．､]?)?\s*(\[.{2}-*?选修[\d-]*?.*?\])\s*([(（]\d+分[)）])?\s*</p>',
                          r"<p>【章节】\2</p>", html2txt)
        html2txt = re.sub(r'<p>\s*(基础|中档|综合)题[^p题]*?</p>|<p>\s*【(考点|专题)】[^p]*?</p>', "", str(html2txt))
        html2txt = re.sub(r'<p>\s*(基础训练|提升训练|探究培优)</p>', "", str(html2txt))
        html2txt = re.sub(r'<p>注意事项[:：]\s*</p>(\n+\s*<p>\s*\d\s*[、.．､][^/]+?</p>)+', "", html2txt, flags=re.S)
        html2txt = re.sub(r'<p>注意事项[:：]\s*\d\s*[、.．､][^/]+?</p>(\n+\s*<p>\s*\d\s*[、.．､][^/]+?</p>)+', "", html2txt,
                          flags=re.S)
        html2txt = re.sub(r'[(（]\s*([A-Z\dⅠⅡⅢⅣⅤ]+|IV)\s*[)）]', r"(\1)".replace(" ", "").replace("(IV)", "Ⅳ"), html2txt)
        html2txt = re.sub(r'[(（](\s*\d\s*\d?\s*分?\s*)[)）]', "(" + r'\1'.replace(" ", "") + ")", html2txt)
        html2txt = re.sub(r'\[来源:.*?\]', "", html2txt)
        html2txt = re.sub('<p>欢迎访问.*?</p>', '', html2txt)
        html2txt = re.sub('w\s*w\s*w\..*?(\.\s*c\s*o\s*m|\.cn)+|(?<!["“=])http:.*?\.(com|cn|org)', "",
                          html2txt)  # ww w.gkstk.com
        # html2txt = re.sub(r'<(table|td|tr) ([a-z]+="\d+")>', r'<\1>', html2txt)
        html2txt = re.sub(r'<table ([a-z]+="\d+")>', r'<table>', html2txt)
        html2txt = re.sub(r'<p>\s*第\s*[二三四ⅡⅢⅣ]\s*(卷|部分)\s*([(（].*?[)）]|非?选择题.{,8})?\s*</p>', "<p>【非选择题】</p>", html2txt)
        # == == =对可能的题型行的处理 == ==
        html2txt = re.sub("<p>【非选择题】</p>((\s|\n|<p>|</p>)*\d{1,2}\s*[.．、､].+?)", r"<p>二、解答题</p>\1", html2txt)\
                     .replace("【非选择题】", "")

        # =====选项的处理=====
        html2txt = re.sub(r'(<p>\s*([1-9]|[1-9][0-9])\s*[.．、､].+?[(（]\s*[）)])\s*(A\s*[.．、､][^/]*?</p>)',
                          r"\1</p>\n<p>\3", str(html2txt))

        # =====题号的处理=====
        html2txt = re.sub(r'([ED]\s*[、.．､].*?(\s|</su[pb]>\s*))(([1-9]|[1-9][0-9])\s*[、.．､])',
                          r"\1</p>\n<p>\3", html2txt)
        html2txt = re.sub(r'((</?p>|\n)\s*(<img src=.*?"\s*/?>\s*)?([1-9]|[1-9][0-9]))\s*'
                          r'([（(]\s*(\d{1,2}[.\s\d]*?分|.{2,3}题?)\s*[)）]|解析?\s*[:：]|【解析】)', r"</p>\1､\5", html2txt)
        html2txt = re.sub(r"<p>\s*([1-9]|[1-9][0-9])\s*([(（]20\d{2}\s*[\u4e00-\u9fa5、､]{2,9}[)）])", r"<p>\1､\2",
                          html2txt)
        html2txt = re.sub(r"<p>\s*([1-9]|[1-9][0-9])\s*(【(解析?|答案?)】|(解析?|答案?)\s*[:：]|\[(答案|解析)\])", r"<p>\1､\2",
                          html2txt)
        html2txt = re.sub(r"<p>\s*([1-9]|[1-9][0-9])\s*([(（]\s*\d+\s*分?\s*[)）])?(【(解析?|答案?)】|(解析?|答案?)\s*[:：]"
                          r"|\[(答案|解析)\])", r"<p>\1､\2\3", html2txt)
        html2txt = re.sub(r"(</?p>|\n)\s*(<img src=((?!/>).)+?/>)\s*([1-9]|[1-9][0-9])\s*"
                          r"([(（]20\d{2}\s*[\u4e00-\u9fa5、､]{2,9}[)）])", r"<p>\2</p>" + "\n" + r"<p>\4､\5", html2txt)  # 【susp_img】
        html2txt = re.sub(r'(</?p>|\n)((\s*<su[bp]>\s*)?<img src=.*? height="[\d.]+p[tx]"\s*/?>(\s*</su[bp]>)?\s*)'
                          r'(([1-9]|[1-9][0-9])\s*[、.．､])', r"</p>\2</p>" + "\n" + r"\5", html2txt)
        html2txt = re.sub(r"(<p>((?!<p>).)+?(\s|[/\"]>))(([1-9]|[1-9][0-9])\s*[、.．､].{,20}本[大小]?题\d+分)",
                          r"\1</p>" + "\n<p>" + r"\4", html2txt)
        html2txt = re.sub(r"</?p>((\s*<su[bp]>\s*)?<img src=.*?/>(\s*</su[bp]>)?((\s*<su[bp]>\s*)?"
                          r"<img src=((?!/>).)+?/>(\s*</su[bp]>)?)*?\s*)\s*(([1-9]|[1-9][0-9])\s*[、.．､])",
                          r"</p>\1</p>" + "\n<p>" + r"\8", html2txt, flags=re.S)
        html2txt = re.sub(r'(<p>\s*[一二三四五六七八九十].*?题\s*\(.+?分.*?\))\s*(([1-9]|[1-9][0-9])\s*[、.．､].*?)</p>',
                          r"\1</p>\n<p>\2</p>", html2txt)
        html2txt = re.sub(r'(<p>\s*[一二三四五六七八九十].*?题\s*\(.+?分.*?\))\s*(([1-9]|[1-9][0-9])\s*[、.．､].*?)</p>',
                          r"\1</p>\n<p>\2</p>", html2txt)
        html2txt = re.sub(r'(<p>.*?[.．]{6,}\s*\d+分)\s*(([1-9]|[1-9][0-9])\s*[、.．､].*?)</p>', r"\1</p>\n<p>\2</p>", html2txt)
        html2txt = re.sub(r'([1-9]|[1-9][0-9])\s*([（(]\s*\d{1,2}[.\s\d]*?分\s*[)）])\s*[、.．､]', r"\1" + "､" + r"\2", html2txt)

        # =====图片的处理=====
        # 1>>根据图片宽高的异常值判断删除隐藏图片
        def sub1(ss):
            if float(ss.group(1)) <= 3 and float(ss.group(2)) <= 3:
                return ""
            else:
                return ss.group(0)
        html2txt = re.sub(r'<img src=.*? width="([\d.]+)p[xt]" height="([\d.]+)p[xt]"\s*/?>', sub1, html2txt)

        # 2>>将图片中带有的汉字去掉
        html2txt = re.sub(r'(<img src=.*?) alt=".+?"', r"\1", html2txt)
        html2txt = re.sub(r'(<img src=(?!\sstyle=)+?(?<!\\)\")>', r"\1 />", html2txt)  # 将">换为" />

        # 3>>建立图片id字典,对原图片信息第一次替换
        html2txt = re.sub(r'( src=".*?files)\\image', r"\1/image", html2txt)
        all_image = re.findall(r'<img src=".*?image[\da-z]+\..*?[/\"]>', html2txt)
        src2subs = {}
        subs2src = {}
        for src in all_image:
            # 校本题库上传的图片名称是随机数,故设置映射
            # kk = re.search('(<img src=".*?image\d+\.(png|gif|jpg|jpeg))', src)
            # new_src = src.replace(kk.group(1), self.img_url[kk.group(1)]) if type(self.img_url) == dict and kk else src
            # 图片信息简化替换
            new_src = re.sub(r'( data-latex)="\s*\\\[(.*?)\\\]\s*"', r'\1="$\2$"', src)
            new_src = re.sub(r'( data-latex="\$[^"]+?\$")',
                             lambda x: x.group(1).replace("<", " \lt ").replace("  ", " "), new_src)
            latex_info = re.search(r'<img src=".*?/(new_)?image([\da-z]+)\..*?(data-latex=".*?")', src)
            mathpix = " " + latex_info.group(3).replace("\n", "").strip().replace("  ", " ") if latex_info else ""
            if mathpix and len(mathpix) > 20:
                mathpix = ""
            w_h_info = re.search(r'<img src=".*?/(new_)?image([\da-z]+)\..*?width="([\d.]+)[pxt]*?"\s*height="([\d.]+)[pxt]*?"', src)
            w_h = " w_h=" + w_h_info.group(3).split('.')[0] + "*" + w_h_info.group(4).split('.')[0] \
                if w_h_info and not mathpix else ""  # w_h 和 mathpix只存在一个
            image_id = re.search(r'<img src=".*?/(new_)?image([\da-z]+)\.', src).group(2)
            src2subs[src] = '<imgsrc' + image_id + w_h + mathpix + "/>"
            subs2src['<imgsrc' + image_id + w_h + mathpix + "/>"] = new_src
        for k, v in src2subs.items():
            html2txt = html2txt.replace(k, v)
        # ------------------------------------------------------------------------

        # ========html 转 list=========
        html2txt = re.sub(r'(</?div>|</table>|</?body>)(\n\s*)*?<p>', r"\1</p>" + "\n<p>", html2txt, flags=re.S)
        # >>>>>> <table>先替换后再切割
        # 不能简单按 \n 切割，表格里面也可能有换行，应该先替换后再切割
        subs2table = {}
        all_table = re.findall(r'<table>.*?</table>', html2txt, flags=re.S)
        for k, v in enumerate(all_table):
            subs2table["<t{}b>".format(str(k))] = v
            html2txt = html2txt.replace(v, "<t{}b>".format(k))

        # <造成的css标签冲突处理  2021-10-13
        def sub2(ss):
            if re.search(r'^(img|/?h[123456]|/?su[bp]>|t\d+b>|br\s*/?>'
                         r'|/?(p|span|font|article|ul|ol|div|strong|em|table|t?body|html|head|t[drh])(\s*|\s+style=.*?")>'
                         r'|/?[a-z]+ (style|rowspan|colspan|class)=.*?">)', ss.group(1)) is None:
                r1 = "&lt;{}".format(ss.group(1))
                r1 = re.sub(r"&lt;(span (class|style)=)", r"<\1", r1)
                r1 = re.sub(r"&lt;(p (class|style)=)", r"<\1", r1)
                return r1
            else:
                return "<{}".format(ss.group(1)).replace("</body>", "").replace("</html>", "")
        html2txt = re.sub("<([^<]{1,30})", sub2, html2txt)

        # >>>>>> html 切割
        con_list = sum([re.split('<p>|<h[12345]>', i) if len(re.findall("<p>|<h[12345]>", i)) > 1
                        else [i] for i in re.split(r"\n+|</p>(?!</td>)|</h[12345]>", html2txt)], [])  # html2txt)[:-1]
        con_list = [re.sub(r"^\n*\s*(<p>|<h[12345]>)+", "", ii) for ii in con_list]
        # pprint(con_list)
        # >>>>>> <table> 替换回去
        if subs2table:
            con_list = [re.sub(r"|".join(subs2table.keys()), lambda x: subs2table[x.group()], ii) for ii in con_list]
        # 剩余个别标签处理
        con_list = [re.sub(r"^<([a-z]+)>[\s\t\n]*</\1>$", "", i.strip()) for i in con_list]  # 2020/4/7,14
        con_list = [re.sub(r"^(<table>|</td>|<td[^<>]*?>|</?tr>)+?(.|\n)+?([一二三四五六七八九十])\s*[、.．､]\s*(.{2,4}题)(.|\n)+?</table>",
                           r"\3､\4", i.strip())
                    for i in con_list]
        # 把最后可能还存在的</?p>或考号信息去掉
        con_list = [re.sub("</?p>|[…O•.\s]*?密[…O•.\s]*?封[….O•\s]*?装?[…O•.\s]*?订?[….O•\s]*?线?[….O•\s]*?$"
                           "|((学校|班级|姓名|座位号|准考号|[学考]号)[\s：:_]*?){2,}$", "", i.strip()) for i in con_list]
        # =====答案行格式处理====
        temp_list = [re.split(r"^((\s*<imgsrc((?!/>).)*?/>\s*)+)", v.strip(), maxsplit=1)[1::3]
                     if re.match(r'(\s*<imgsrc((?!/>).)*?/>\s*)+?(参考|考试|试[题卷]|物理|理综|数学|化学|生物)'
                                 r'(答案|解析|答案[及与和]评分(标准|意见|细则))\s*$'
                                 r'|(\s*<imgsrc((?!/>).)*?/>\s*)+?评分标准'
                                 r'|(\s*<imgsrc((?!/>).)*?/>\s*)+?(参考|考试|试[题卷])'
                                 r'(答案|解析|答案[及与和]评分(标准|意见|细则))\s*(物理|理综|数学|化学|生物)?\s*$',
                                 re.sub(r"[上下]?学[年期]|[\d—【】.．、､：:(（）)年\s]|[中大]学|模拟|[中高]考|年级|[学期][末中]|[高初][一二三]", "",
                                        v.strip())) else [v] for v in con_list]
        con_list = sum(temp_list, [])

        # =====对可能的题号的处理====  如2、3、4、5、      加了【fei】  # 重新修改！！！！！！！！！！
        con_list = [re.sub(r"^\s*([1-9][0-9]?\s*[.．、､])", r"【fei】\1", i.strip())
                    if (len(re.findall(r"(^|\s*[.．、､])\s*[1-9][0-9]?\s*[.．、､]", i)) >= 3
                        and len(re.sub(r"[\d.．、､\s]", "", i)) < 2) else i for i in con_list]

        # =====头尾清除没用的信息=====
        if con_list and re.search(r"[\u4e00-\u9fa5]|<img | style=\"text-(indent|align)"
                                  r"| style=\"text-decoration| class=\"dots\"|<strong>|<em>", con_list[0]) is None:
            con_list = con_list[1:]
        while con_list and re.search(r"声明[：:].*?著作权属.*?所有|(邮箱|用户|日期|QQ)\s*[：:].+?", con_list[-1]):
            con_list = con_list[:-1]

        return con_list, subs2src, self.new_html


def del_no(item, item_no_type=1):
    """去开头的题号"""
    if item_no_type == 2:
        item = re.sub(r'^\n*\s*[(（]\s*([1-9]|[1-9][0-9])\s*[)）]\s*[.．、､：:]?', "", item)
        return item
    item = re.sub(r'^\n*\s*([1-9]|[1-9][0-9])\s*[.．、､：:]', "", item)
    return item


def html_cleal_test(htmlf):  # 不用
    html2txt = re.sub(r"&nbsp;", "", htmlf.read())  # ("", " ")

    # html2txt.replace("①", "(1).").replace("②", "(2).").replace("③", "(3).")
    con_list = [re.sub(r"^\n+\s+<p>", "", ii) for ii in html2txt.split("</p>")[:-1]]
    # pprint(con_list)
    if re.search(r"[\u4e00-\u9fa5]", con_list[0]) is None:
        con_list = con_list[1:]
    return con_list


def get_md5(image_id):
    """
    由于hash不处理unicode编码的字符串（python3默认字符串是unicode）
        所以这里判断是否字符串，如果是则进行转码
        初始化md5、将image_name进行加密、然后返回加密字串
    """
    image_name = str(image_id) + str(time.time()) + str(random.random())
    image_name = image_name.encode("utf-8")
    md = hashlib.md5()
    md.update(image_name)
    return str(md.hexdigest())


def wash_after(res_dict, item_groups, ans_groups, subject):
    """
    1.处理最终结果多余的换行符;2.对题文中已给答案的选择填空进行替换；3.选择题的细分
    :param res_dict: [{},{},,,]
    :param item_groups: 题组;  ans_groups: 答案分组
    :return:
    """
    pattern1 = re.compile(
        r"([是为点]|等于|=|＝|有|存在)\s*_+((<img src=((?!/>).)*?/>|[^_;；。?!，\n])+?)(?<![=＝_])_+([cdkm上]?m?\s*.?[。.?]?\s*"
        r"($|<br/>|<img src|……))")
    pattern2 = re.compile(r"((有|存在|[是为])[\u4e00-\u9fa5]{0,2})\s*_+(\d+)_+\s*([\u4e00-\u9fa5，,;；。.．])")
    chapter_no = {}
    option_st = 0
    is_optional = False
    option_score = 0
    select_type_id = []

    for num, sr in enumerate(res_dict):
        sr["stem"] = re.sub(r"\n[_\-\s]*密[…O•.\s]*封[….O•\s]*装?[…O•.\s]*订?[….O•\s]*线?"
                               r"|\n\s*((学校|班级|姓名|座位?号|准考号|学号)[\s：:_]*){2,}", "", sr["stem"])
        sr["stem"] = re.sub(r'\n\s*(第\s*[^\s]\s*卷|第[一二三四]部分)\s*([(（].*?[)）]|非?选择题.{,8})?\s*\n', "\n", sr["stem"])
        if num == len(res_dict) - 1:  # 对拆分后的最后一道题进行特殊判断
            end_con = sr["stem"] + sr["parse"]
            if len(re.findall(r"[\u4e00-\u9fa5]", end_con)) > 1000 and (
                    len(re.findall(r"\n\s*([1-9]|1[0-9])\s*[.．、､].+?",
                                   end_con)) > 4 or len(re.findall(r"[(（]\s*[）)]|_{2,}", end_con)) > 6):
                sr['errmsgs'].append("原试卷格式有问题，导致本题可能包含了很多非本题的题文")

        # if not re.sub(r"[(（)）\n\s]", "", sr["stem"]):
        #     sr['errmsgs'].append("本题没有题干，请检查题干格式是否正确")

        if "-" in str(sr["item_id"]) and sr['type'] in ["选择题", "填空题"]:
            if (not sr["key"] or sr["key"] == "见解析") and re.search("[A-H]+", re.sub("[;；、､\n(（)）\s]|\d+分", "", sr["parse"])):
                sr["key"] = re.sub("[;；、､\n(（)）\s]|\d+分", "", sr["parse"])
                sr["parse"] = ""

        # 把首尾的换行都去掉
        # sr["stem"] = table_label_cleal(re.sub(r"\n\s*","<br/>",sr.get("stem", "").lstrip()))
        # 将选择题和填空题中的题干中出现答案的情况 去掉答案
        kuo_con1 = re.search(r'([是为]|等于|[＝=有]|表示)\s*[(（]\s*([A-Zc][A-Zc;；和与、､\s]*?)[）)]\s*(.?($|\n|<br/>|<img))',
                             sr["stem"])
        kuo_con2 = re.search("[(（]\s*([A-Zc][A-Zc;；和与、､\s]*?)[）)]\s*(.?($|\n|<br/>))", sr["stem"])
        if sr['type'].replace("题", "") in ["单选", "多选", "选择", "不定选择"]:
            # sr["type"] = "选择类"
            # 针对选择题在题文中已给出答案的处理
            if kuo_con1:
                sr["stem"] = sr["stem"].replace(kuo_con1.group(0), kuo_con1.group(1) + "( )" + kuo_con1.group(3))
                sr["key"] = kuo_con1.group(2).replace("c", "C") if not sr["key"] else sr["key"]
            elif kuo_con2:
                sr["stem"] = sr["stem"].replace(kuo_con2.group(0), "( )" + kuo_con2.group(2))
                sr["key"] = kuo_con2.group(1).replace("c", "C")  if not sr["key"] else sr["key"]
                # sr['options_text'] = ""
        elif sr['type'] == '填空题':
            # sr["type"] = "解答类"
            ans_list = []
            # 针对填空题在题文中已给出答案的处理
            sub_n = 0
            while re.search(pattern1, sr["stem"]):
                blank_con1 = re.search(pattern1, sr["stem"])
                sr["stem"] = sr["stem"].replace(blank_con1.group(0),
                                                blank_con1.group(1) + "____" + blank_con1.group(5))
                ans_list.append(blank_con1.group(2))
                sub_n += 1
                if sub_n > 5:
                    break
            while re.search(pattern2, sr["stem"]):
                blank_con2 = re.search(pattern2, sr["stem"])
                # 这里的限制条件易出错，可以再判断一下
                sr["stem"] = sr["stem"].replace(blank_con2.group(0),
                                                blank_con2.group(1) + "____" + blank_con2.group(4))
                ans_list.append(blank_con2.group(2))
            if re.findall(r"_{2,}", sr["stem"]):
                sr["blank_num"] = len(re.findall(r"_{2,}", sr["stem"]))
                if not sr["key"] and ans_list:
                    sr["key"] = "; ".join(ans_list)

        # 换行符处理！
        sr["stem"] = sr.get("stem", "").rstrip()
        # sr["stem"] = get_equation_instr(sr["stem"])
        if "options" in sr:  # 对选项部分进行格式处理
            for i in range(len(sr['options'])):
                sr['options'][i] = get_simpstr2eqn(sr['options'][i].strip())
                # sr['options'][i] = get_equation_instr(sr['options'][i].strip()).replace("\n\n", "\n").replace("\n", "<br/>")
        if "slave" in sr and sr["slave"]:
            # 带小题的大题，格式处理，高中数学没有这一功能
            for s in sr["slave"]:
                s["stem"] = s.get("stem", "").rstrip()
                s["parse"] = s.get("parse", "").strip().replace("解答:解：", "解答:").replace("解答:解:", "解答:")
                s["key"] = s.get("key", "").strip()
                # sr["slave"] = sr.get("slave", "").replace("\n", "<br>")
        else:
            # s["parse"] = css_conflict_deal(s["parse"])  # "css 冲突标签处理"
            sr["parse"] = sr.get("parse", "").lstrip()
            sr["parse"] = re.sub("^【解[答析]】\s*", "", sr["parse"])
            # sr["parse"] = get_equation_instr(sr["parse"])
            sr["key"] = sr.get("key", "").lstrip()
            # sr["key"] = get_equation_instr(sr["key"])

            if not sr["parse"] and not sr["key"]:  # 答案和解析都没有
                sr["parse"] = ""
                sr["key"] = ""
                # sr['errmsgs'].append("本题缺少答案和解析")
            elif not sr["key"] and sr["parse"]:
                sr["key"] = "见解析"
            elif re.sub("见解析|略|空|无|没有|答案", "", sr["key"]) and not sr["parse"]:
                sr["parse"] = ""
                # if "本选做题缺少解析" not in sr['errmsgs'] and "本题缺少解析" not in sr['errmsgs']:
                #     sr['errmsgs'].append("本题缺少解析")

        # 辅助标签处理
        # sr["analysis"] = ""
        if "analy" in sr:  # 存在题目分析时，将其放在解析里
            # sr["analy"] = sr.get("analy", "").strip().replace("\n\n", "\n")
            if len(sr["analy"].replace(" ", "")) >= 10:
                sr["parse"] = "【分析】"+sr["analy"] + "\n【详解】" + sr["parse"]
            del sr["analy"]

        if "chapter" in sr:  # 如选修4－5：不等式选讲
            if sr['item_id'] + 1 <= len(res_dict):
                chapter_no[sr['item_id']] = sr["chapter"]
            del sr["chapter"]

        # 是否为选做题"is_optional"，两种形式不会同时出现
        if "option_st" in sr:  # 带有此标签的后面的题目都是选做题option_score
            # option_st = sr['item_id']
            # is_optional = True
            # if "," in sr["option_st"]:
            #     option_score = int(sr["option_st"].split(",")[-1])
            del sr["option_st"]

        # 将选择题改为单选或多选,"is_multiple_choice"
        sr['type'] = re.sub("([单多])项选择题?", r"\1选题", sr['type'])
        sr['type'] = sr['type'].replace("题题", "题")  # .replace("简答", "解答")
        # sr['type'] = re.sub("(计算|简答)题?", "解答题", sr['type'])

        if sr['type'] in ["选择", "选择题"]:  # 有的科目只有选择题，不分单选和多选
            if len(re.findall("[A-Z]", sr["key"])) > 1:
                sr['type'] = '多选题'
            elif len(re.findall("[A-Z]", sr["key"])) == 1:
                sr['type'] = '单选题'
            info_x = re.search("^[（(](多)选题?[）)]", sr["stem"].replace(" ", ""))
            if info_x:
                sr['type'] = '{}选题'.format(info_x.group(1))

        # if "options" in sr:
        #     sr['type'] = '选择'
        # else:
        #     if re.search("_{2,}", str(sr)):
        #         sr['type'] = '填空'
        #     else:
        #         sr['type'] = '解答'
        if sr['type'] == '多选题':
            if len(re.findall("[A-Z]", sr["key"])) == 1:
                sr['errmsgs'].append("本题答案个数与题型(多选题)不符")
            # sr["is_multiple_choice"] = 'true'
        elif sr['type'] == '单选题':
            # sr["is_multiple_choice"] = 'false'
            if "options" in sr and len(sr["options"]) > 4:
                sr['errmsgs'].append("选项个数多于4个，与题型(单选题)不符")
            if len(re.findall("[A-Z]", sr["key"])) > 1:
                sr['errmsgs'].append("本题答案个数与题型(单选题)不符")
        # elif sr['type'] == '不定选择题':
        #     if len(re.findall("[A-Z]", sr["key"])) > 1:
        #         sr['type'] = '多选题'
        #     elif len(re.findall("[A-Z]", sr["key"])) == 1:
        #         sr['type'] = '单选题'
        #     else:
        #         sr['type'] = '选择题'
        #         if "缺少答案" not in "".join(sr['errmsgs']):
        #             sr['errmsgs'].append("本题缺少答案")
        elif sr['type'].replace("题", "") in ["单空", "多空", "填空"]:
            sr['type'] = "填空题"
        elif subject != "语文" and sr['type'] not in ["选择", "选择题"]:
            sr['type'] = "解答题"

        # """按照原先高中数学解析的最后输出格式整理输出"""
        # sr["type"] = sr['type'].replace("非选择", "解答").replace("题题", "题")  #        print(“)
        sr["topic_num"] = sr['item_id']
        sr['errmsgs'] = "；".join(sr['errmsgs'])
        sr["parse"] = re.sub(r"试题【([分解]析)】", r"试题\1:", sr["parse"])  # 解析
        sr["key"] = re.sub("([;；]|<br/>)\s*$", "", sr["key"])

        if 'susp_pic' in sr:
            del sr['susp_pic']
        if 'is_optional' in sr:
            del sr['is_optional']
        if 'spliterr_point' in sr:
            del sr['spliterr_point']
        if 'score' in sr:
            del sr["score"]
        del sr['item_id']

        # ---------------------字符串公式处理--------------------------------
        # sr["stem"] = get_equation_instr(sr["stem"])
        # sr["key"] = get_equation_instr(sr["key"])
        # sr["parse"] = get_equation_instr(sr["parse"])
        # if "options" in sr:
        #     sr["options"] = list(map(get_equation_instr, sr["options"]))
    # -----------------------------------------------------------------------
    # pprint(res_dict)
    if item_groups and item_groups["is_groups"]:
        # print("item_groups:", item_groups)
        # print("ans_groups:", ans_groups.keys())
        res_dict = stems_to_groups.regroup(res_dict, item_groups, ans_groups)
    else:
        for one_res in res_dict:
            one_res = parse_split2group(one_res)
            if "slave" in one_res and one_res["slave"]: # 带小问的试题题文也设置一下字体
                one_res['type'] = '小题多问类'
                one_res["stem"] = re.sub(r'(<p style="text-indent:.*?">\n*|<p style="text-align: center;?">\n*'
                                     r'|<p style="text-align: right;?">\n*)([^\n]+?)',
                                     r'\1<span style="font-family:楷体;">\2',
                                     one_res["stem"])
            elif "options" in one_res and one_res["options"]:
                one_res['type'] = '选择类'
            else:
                one_res['type'] = '解答类'
                if re.search("(阅读|针对|结合).{,4}[资材]料|(\n|^)\s*材料一\s", one_res['stem']):
                    one_res["stem"] = re.sub(r'(<p style="text-indent:.*?">\n*|<p style="text-align: center;?">\n*'
                                                 r'|<p style="text-align: right;?">\n*)([^\n]+?)',
                                                 r'\1<span style="font-family:楷体;">\2',
                                                 one_res["stem"])
    # pprint(res_dict)

    # 换行符替换
    convert_huanhang(res_dict)
    # ------------------------------------------------------------------------
    # if chapter_no:  # 章节标签下移一位
    #     for c, v in chapter_no.items():
    #         res_dict[c]["chapter"] = v

    # 选做题"option_str"处理
    # if select_type_id:
    #     for s in select_type_id:
    #         if len(select_type_id) == 2:
    #             res_dict[s - 1]['option_str'] = "2选1"
    #         elif len(select_type_id) == 4:
    #             res_dict[s - 1]['option_str'] = "4选2"
            # else:
            #     res_dict[s - 1]['errmsgs'] += ";<br/>选做题不是“2选1”和“4选2”类型"

    # if option_st:
    #     print("option_st:", option_st)
    #     for s in range(option_st, len(res_dict)):
    #         if (len(res_dict) - option_st) == 2:
    #             res_dict[s]['option_str'] = "2选1"
    #         elif (len(res_dict) - option_st) == 4:
    #             res_dict[s]['option_str'] = "4选2"
            # else:
            #     res_dict[s]['errmsgs'] += ";<br/>选做题不是“2选1”和“4选2”类型"

    # 再解析中的新图片上传腾讯云
    # 再设置一个入库接口，点击入库，才开始从本地上传图片
    return res_dict


def convert_huanhang(items_list):
    """
    递归 换行符替换：\n -->  <br/>
    :param items_list:
    :return:
    """
    if isinstance(items_list, list):
        for k, one_i in enumerate(items_list):
            items_list[k] = convert_huanhang(one_i)
    elif isinstance(items_list, dict):
        for k, v in items_list.items():
            if k == "answer_type" and type(v) == str:
                items_list[k] = configs.answer_type[v]
            else:
                items_list[k] = convert_huanhang(v)
    elif isinstance(items_list, str):
        items_list = items_list.strip().replace("\n\n", "\n")
        items_list = re.sub('\n*(<p style=".*?">\n*)$', "", items_list)
        items_list = re.sub(r'(<p style=".*?">)\n+', r"\1", items_list)
        items_list = re.sub(r'(<p style=".*?">)(((?!</p>).)*?)\n+(</t[rd]>)', r"\1\2</p>\4", items_list)
        items_list = re.sub(r'(<p style=".*?">.+?)\n[\n\s]*(<img\s)', r"\1</p>\n\2", items_list)
        while re.search(r'(<p style=".*?">[^\n]+?)\n[\n\s]*(<p style=".*?">|<table>)', items_list, flags=re.S):
            items_list = re.sub(r'(<p style=".*?">[^\n]+?)\n[\n\s]*(<p style=".*?">|<table>)',
                                r"\1</p>\2", items_list, flags=re.S)
        items_list = re.sub(r'(<p style=".*?">)(((?!</p>).)*?)\n*$', r"\1\2</p>", items_list)
        items_list = re.sub(r'(<span style="font-family:楷体;?">)(((?!</p>).)*?)</p>', r"\1\2</span></p>", items_list)
        # 将\n 换为 <p></p>
        items_list = linefeed2p(items_list)

        return items_list.replace("\n", "<br/>")
    else:
        return items_list
    return items_list


def linefeed2p(item):
    """
    换行标签替换为<p></p>标签，表格内的还需单独处理
    :return: str
    """
    # def subp_table(ss):  # 表格里换行标签的处理,先看看情况是否需要处理
    if re.sub(r'<table>.*?</table>',"", item).count("\n") > 1:
        subs2table = {}
        all_table = re.findall(r'<table>.*?</table>', item, flags=re.S)
        for k, v in enumerate(all_table):
            subs2table["<t{}b>".format(str(k))] = v
            item = item.replace(v, "<t{}b>".format(k))

        lines = item.split("\n")
        newlines = [line if re.search(r"^<p style=|^<p>|</p>$", line) else "<p>{}</p>".format(line) for line in lines]
        item = "".join(newlines)

        if subs2table:
            item = re.sub(r"|".join(subs2table.keys()), lambda x: subs2table[x.group()], item)
    return item


def css_conflict_deal(item):
    """
    针对<a, <p 符号 在前端显示被过滤掉的问题：对“<”左右加$, 注意条件：“<”前$为双数时加,
    :return: str
    """
    # item = item.replace("<", "&lt;").replace(">", "&gt;")  # 2021-8-24
    # item = re.sub("<(?!img src)", "&lt;", item)  # 还有表格
    item = item.replace("$<$", "【*_*】")
    item = re.sub("<br\s*/?>", "\n", item)
    item = re.sub(r"<(/?su[bp])>", r"【\1】", item)
    item = re.sub(r"<(/?span|p style=|/?p>|/?strong>|/?em>)", r"【\1", item)
    if re.search(r"(?<!\\\()<", item):
        n1=0
        n2=0
        for i in re.finditer("<(?!img)", item):
            if item[:i.start()+2*n1+4*n2].count("$") % 2 == 0:
                item = item[:i.start()+2*n1+4*n2] + "$<$" + item[i.start()+1+2*n1+4*n2:]
                n1 += 1
            else:
                item = item[:i.start() + 2 * n1+4*n2]+" \lt " + item[i.start()+1+2*n1+4*n2:]
                n2 += 1
    item = item.replace("【*_*】", "$<$")
    while re.search(r"\\\(.*?\$<\$.*?\\\)", item):
        item = re.sub(r"(\\\(.*?)\$<\$(.*?\\\))", r"\1 \lt \2", item)
    item = re.sub(r"【(/?su[bp])】", r"<\1>", item)
    item = re.sub(r"【(/?span|p style=|/?p>|/?strong>|/?em>)", r"<\1", item)
    return item


def insert_sort2get_idx(item_list, num):
    """
    :param item_list: 拍好序的列表
    :param num: 插入的数值
    :return: 插入的位置
    """
    add_n = 0
    for i in range(len(item_list)):
        if num > item_list[i]:
            add_n += 1
        else:
            break
    return add_n


# def find_seq_num(num_list):
#     """
#     针对切分题号时切错的序号进行纠正，考虑序号是连续且正常的情况下
#     将连续的数字进行分组
#     :param num_list:输入[3, 4, 8, 9, 12, 13, 14]
#     :return: [[3, 4],[8, 9],[12, 13, 14]]
#     """
#     seq_ranges = []
#     for k, g in groupby(enumerate(num_list), lambda x: x[0] - x[1]):
#         group = (map(itemgetter(1), g))
#         group = list(map(int, group))
#         seq_ranges.append(group)
#     return seq_ranges


# def del_exception_value(item_list):
#     """
#     去列表中的异常值,题目越多，越容易突出异常值
#     :return:
#     """
#     import numpy as np
#     max_v = max(item_list)
#     arr_mean = np.mean(item_list)  # 均值
#     arr_var = np.var(item_list)   # 方差
#     while max_v > len(item_list)+4:
#         item_list.remove(max_v)
#         print(item_list)
#         arr_mean = np.mean(item_list)  # 去最大值后的均值
#         arr_var = np.var(item_list)  # 去最大值后的方差
#         max_v = max(item_list)
#     # print("均值与方差：",arr_mean,arr_var)
#     if abs((item_list[-1] - item_list[0] + 1) - len(item_list)) <= 3:
#         return item_list
#     else:
#         exception_value = []
#         for i in item_list:
#             # print(abs((i - arr_mean) / arr_var), i)
#             if(abs((i - arr_mean)/arr_var)) > 0.3:
#                 exception_value.append(i)
#         right_seq = [i for i in item_list if i not in exception_value]
#         return right_seq


def pic_transfer(con_list):
    aft_opt = []  # 针对选项后是题目图片的情况,进行移位
    if "\n" in con_list[-1]:
        ccon = re.split("\n+", con_list[-1])
        while re.match("<img src=", ccon[-1]) and len(ccon) > 1:
            aft_opt.insert(0, ccon[-1])
            ccon = ccon[:-1]
        if aft_opt:
            con_list[0] += "\n" + "\n".join(aft_opt)
            con_list[-1] = "\n".join(ccon)
    con_list[0] = re.sub(r"\(\d+分\)", "", con_list[0][:9]) + con_list[0][9:]
    return con_list


def judge_split_error(item_list):
    """
    转对试卷切分后的小题判断是否存在切分错误的情况，能纠错就纠错，不能则删除
    :return:
    """
    # for k, v in enumerate(item_list):
    #     if k>0 and v['item_id'] - item_list[k-1]['item_id']>1:
    #         if


if __name__ == '__main__':
    # -------------生成requirements.txt---------------
    # pip freeze > requirements.txt
    # import os, sys
    #
    # project_root = os.path.dirname(os.path.realpath(__file__))  # 找到当前目录
    # print(project_root)
    #
    # # 找到解释器，虚拟环境目录
    # python_root = sys.exec_prefix
    # print(python_root)
    #
    # # 拼接生成requirements命令
    # command = python_root + '\Scripts\pip freeze > ' + project_root + '\\requirements.txt'
    # print(command)
    #
    # # 执行命令。
    # os.system(command)

    # ----------------一键安装 requirements.txt------------
    # pip install -r requirement.txt
    # python_root + '\Scripts\' + pip install -r requirements.txt

    # import os
    # rrr=os.path.basename(r"http:/pstatic.dev.xueping.com/data/word/2020/08/12/5f338d18e2cce.docx")
    # print(rrr)
    # item = "<a 我没发你的接口 $2366<a$   <a 我没发你的接口 $2366<a$  <img 我没发你的接口 $2366<a$ <a 我没发你的接口 $2366<a$   <a 我没发你的接口 $2366<a$"
    # item = r"2.下列选项中,使不等式\( x<\frac{1}{x}< x_{2} \)"
    # ww = css_conflict_deal(item)
    # print(ww)
    # -----------------
    p1 = r"C:\Users\Administrator\Desktop\temp\62314b31a7d375f4518b9afd.html"
    # t1 = open(p1,'r',encoding="utf8").read()
    t1 = """
    <p>7．南洋劝业会是1910年中国在南京举办的第一个世界博览会，历时达半年，中外30多万人参观。如表是南洋劝业会“物产会”展品目录，由此可知</p>
<table border="1">
<tbody>
<tr>
<td>
<p>&nbsp;</p>
</td>
<td>分类</td>
<td>具体内容</td>
</tr>
<tr>
<td rowspan="6">天产品</td>
<td>农业部</td>
<td>五谷、园蔬、树艺、储藏、器具、肥料、水利、益虫害虫标本</td>
</tr>
<tr>
<td>蚕桑部</td>
<td>茧、丝、桑种标本、蚕种标本、器具、场室模型</td>
</tr>
<tr>
<td>水产部</td>
<td>麟醫、介甲、两栖、腔囊、植物、储藏、器具、舟船模型</td>
</tr>
<tr>
<td>药材部</td>
<td>植物、动物、金石、制合器具、储藏器具</td>
</tr>
<tr>
<td>矿采部</td>
<td>五金、石炭、杂矿、采矿器具、炼冶器具、山场模型</td>
</tr>
<tr>
<td>狩猎部</td>
<td>皮革、牙角、毛羽、器械、各种标本</td>
</tr>
<tr>
<td rowspan="8">工艺品</td>
<td>染织部</td>
<td>漂染、机织、机器、场舍模型</td>
</tr>
<tr>
<td>服装部</td>
<td>衣服、冠履、带配、陈设、装饰、各种制造之器具</td>
</tr>
<tr>
<td>陶磁（瓷）部</td>
<td>陶器、磁（瓷）器、土器、制造器具、窑厂模型</td>
</tr>
<tr>
<td>髹漆部</td>
<td>雕填、磨甸、彩绘、八宝、器具、工场模型</td>
</tr>
<tr>
<td>琉璃、珐琅部</td>
<td>琉璃、珐琅、景泰蓝、荡磁（瓷）</td>
</tr>
<tr>
<td>竹木、笺扇部</td>
<td>竹器、木器、藤器、柳条器、竹簧器、笺纸、扇、画纸</td>
</tr>
<tr>
<td>玉石部</td>
<td>玉器、石器、器具、工场模型</td>
</tr>
<tr>
<td>牙角、鞣革部</td>
<td>牙器、角器、骨器、马尾器、牛革器、马革器、羊革器、鹿革器</td>
</tr>
</tbody>
</table>
<p>A．工业体系已经形成B．精耕细作持续繁荣</p>
<p>C．近代工业发展缓慢D．经济推动文化变迁</p>
<p>【答案】C</p>
    """
    res = HtmlWash(t1,'11111111',"高中物理").html_cleal()
    print(res)