PaperText_Segmentation/Utils/html_again_parse.py

# -*- coding: utf-8 -*-
import re
from pprint import pprint

from pyquery import PyQuery as pq


#
# pattern = re.compile(r"\[来源.*?\]|www\..*?com")
#
# filter_words = ["学科网", "高考资源网", "Ziyuanku.com", "WWW.ziyuanku.com",
#                 "高☆考♂资♀源€网", "w.w.w.k.s.5.u.c.o.m", "本资料由《七彩教育网》www.7caiedu.cn 提供!",
#                 "本资料来源于《七彩教育网》http://www.7caiedu.cn", "本资料由《七彩教育网》www.7caiedu.cn 提",
#                 "高考试题来源:"]
#
#
# def filter_word(txt_list):
#     new_txt_list = []
#     for word in txt_list:
#         if not word.strip():
#             continue
#         new_word = re.sub(pattern, "", word)
#         for keys in filter_words:
#             if keys in new_word:
#                 new_word = new_word.replace(keys, "")
#         new_txt_list.append(new_word)
#     return new_txt_list


def filter_data(x):
    if not str(x).replace(" ", "").strip():
        pass
    else:
        return str(x)


def replace_k(con):
    # con = str(con).replace(" ", "+")
    # con = str(con).replace(" ", "+")
    con = re.sub(r'\s(?!(src="http|_src="http|class="tiankong"|data-num=))', "#+#", str(con))
    return pq(con, parser="html")


def css_label_wash(content):
    # todo add 9-4
    """
    清洗文本中带有的css标签
    :param content:
    :return:
    """
    # temp_con = re.sub('</?p(\s*|\s+style=.*?")?>', "", str(content))
    if re.search('</?(span|font|article|ul|ol|div)(\s*|\s+style=.*?")>|text\s*-\s*decoration: underline|border\s*-\s*bottom:', str(content)) is None:
        # content = re.sub('</p\s*>', "\n", content).strip().replace("\n\n", "\n").replace("\n", "<br/>").
        content = re.sub('<p(\s*|\s+style=.*?")>', "<p>", content)
        content = re.sub('</p><p>', "</p>\n<p>", content)
        return content
    else:
        content = re.sub(r'<font\s+style="color: red">', "", str(content))
        content = re.sub(r'<span\s+style="color: red">', "", str(content))
        content = re.sub(r'<span\s+style="color: blue">', "", str(content))
        content = re.sub(r'<font\s+style="color: blue">', "", str(content))
        content = content.replace("</font >", "").replace("</font>", "")
        # content = content.replace("</span >", "").replace("</span>", "")
        content = content.replace('<p style="height: 0;">&nbsp;</p>', "\n").replace('<p><br/></p>', "\n")
        # content = content.replace("</p>", "</p>\n")  # 2022-4-25
        content = re.sub(r"</p>\s*(?!\n)", "</p>\n", content)
        content = re.sub(r"<br\s*/?>", "<p>", content)  # 2022/1/6
        # content = re.sub('<p( style=.*?")?>', "", content)
        # content = re.sub('<br\s*/?>', "\n", content)
        # parm = False
        # if "<article>" not in str(content):
        #     parm = True
        subs2img = {}
        if re.search('<img.*? src="',content) or "</table>" in content:
            all_imgs = re.findall('<img.*? src=.*?[\s/"]>|<table.*?>.*?</table>',
                                  content, flags=re.S)
            for k, img in enumerate(all_imgs):
                content = content.replace(img, "&{}&".format(k))
                subs2img["&{}&".format(k)] = img
        content = re.sub(r"<(su[bp])>(.*?)</(su[bp])>", r"【\1】\2【/\3】", content)
        content = content.replace("&lt;", "【#lt;】")

        html = pq(content, parser="html")
        a = []

        if html.children():
            for line in html.children().items():  # <p>.*?</p>里面的内容可能会被过滤掉

                test = line.text()
                # 保留下划线及着重符标签   <span style="text-decoration: underline;">
                # 波浪线：<span style="text-decoration: underline wavy;">
                # pq会将多个空格换成一个
                if '<span style="text-decoration: underline' in str(line) or '<span class="dots"' in str(line) \
                        or '<p style="text-indent:' in str(line) or '<p style="text-align:' in str(line):
                    line = re.sub(r'<span style="text-decoration: underline(.*?">.+?)</span>', r"【1#\1##】", str(line))
                    line = re.sub(r'<span class="dots">(.+?)</span>', r"【2#\1##】", str(line))
                    line = re.sub(r'<(p style="text-(indent|align):.*?">.+?)</p>', r"【\1##3】", str(line))
                    line = line.replace(" ", "【+】")
                    line = pq(line)
                    new_line = list(map(lambda x: str(x).replace("【1#", '<span style="text-decoration: underline')
                                        .replace("##】", "</span>").replace("【2#", '<span class="dots">')
                                        .replace("【p【+】style=", "<p style=").replace("##3】", "</p>").replace("【+】",
                                                                                                             " "),
                                        line.text().split("\n")))
                    a.extend(new_line)
                elif str(line).startswith("<p") and line.text().strip():
                    if '<img src="http://' in str(line) or "text-decoration: underline" in str(line) or "border-bottom:" in str(line) or "text - decoration: underline" in str(line):
                        # a.append(line.html().replace("<br />", "\n").replace("<br/>", "\n").replace("<br>", "\n") + "\n")
                        a.append(line.html())
                    else:
                        line = replace_k(line)
                        if re.search(r"\n|<br\s*/?>", str(line), flags=re.S):
                            line = re.sub(r"\n|<br\s*/?>", "#*#", str(line), flags=re.S)
                            line = pq(line)
                            new_line = list(map(lambda x: str(x).replace("#+#", " "), line.text().split("#*#")))
                            a.extend(new_line)
                        else:
                            if line.text().strip():
                                a.append(line.text().replace("#+#", " "))

                elif "<article>" in str(line) and "</article>" in str(line):
                    line = re.sub(r'<p.*?>', "", str(line.html()))
                    b = line.replace('</p>', ""). \
                        replace("<br>", "\n"). \
                        replace("<br/>", "\n"). \
                        replace("<br />", "\n"). \
                        replace('<p style="height: 0;">&nbsp;</p>', "\n"). \
                        replace('<p style="height: 0;"> </p>', "\n")
                    b_list = b.split("\n")
                    # b_list = list(filter(lambda x: str(x), b_list))
                    b_list = list(filter(filter_data, b_list))
                    b_list = list(map(lambda x: str(x), b_list))
                    a.extend(b_list)
                elif str(line).startswith("<ul"):
                    a.append(line.text())
                elif line.attr("class") == "slave-datas":
                    for j, ss in enumerate(line.children().items()):
                        if ss(".read-list-title").text():
                            a.append(str(j + 1) + "." + ss(".read-list-title").text())
                        if ss(".read-list-opt").text():
                            a.append(ss(".read-list-opt").text().replace("\n", "\t"))
                        if str(ss).startswith("<ul"):
                            a.append(ss.text())
                elif str(line).startswith("<table"):
                    a.append(str(line))
                elif str(line).startswith("<ol"):
                    for i, ss in enumerate(line.children().items()):
                        a.append(str(i + 1) + "." + ss.text())
                else:
                    # print('test:',line.text())  # 自动去掉了图片
                    if line.text().strip():
                        line = replace_k(line)
                        # a.append(line.text().replace("+", "\xa0") + "\n")
                        a.append(line.text().replace("#+#", " "))

        else:
            # if html.text().strip():  把换行\n 都去掉了
            #     a.append(html.text())
            # a.append(str(html)) # 会自动带上</?p> ；&变为&amp；
            a.append(content.strip())

        new_a = "\n".join(list(map(lambda x: str(x).strip(), a)))
        if subs2img:
            new_a = re.sub("|".join(subs2img.keys()), lambda x: subs2img[x.group()], new_a)
        new_a = "<p>" + new_a.replace("\n\n", "\n").replace("\n", "</p>\n<p>") + "</p>"

        # for sb, img in subs2img.items():  # 2021
        #     new_a = new_a.replace(sb, img)

        # if parm:
        #     new_a[0] = "\xa0" * 4 + new_a[0]
        new_a = re.sub(r"【(/?su[bp])】", r"<\1>", new_a).replace("【#lt;】", "&lt;")
        return new_a


if __name__ == '__main__':
    cons = r'''
    <div>(1)求货物和重物的质量关系;</div>
(2)要使货物运送到B端,若采用在重物下方挖坑的方法,求至少挖多深的坑;<br>(3)要使货物运送到B端,若采用配重落地时传送带立刻顺时针转动的方法(启动时间可<br>忽略),求传送带速度大小的范围和货物从A端传送到B端所用时间的范围(结果保留三位有<br>效数字)。<br><br><br><br>物理参考答案<br>
<div>1.C</div>
<div>【解析】绝大多数α粒子沿直线穿过,偏转角很</div>
<div>小,说明原子核很小;A项错误;少数α粒子穿过金箱</div>
<div>后发生较犬角度的偏转是由于少数α粒子穿过金箱</div>
<div>时距离金原子核较近,受到的库仑斥力较大,B项错</div>
<div>误;极少数α粒子被弹回,说明原子核是一个体积小、</div>
    '''
    cons1 = '''
    9 . 中国古代的政治权力由“传贤”转变为“传子”，“家天下”制度开始形成于<table name=\"optionsTable\" style=\"width:100%;table-layout:fixed;\" cols=\"4\"><tr><td>A．夏朝</td><td>B．商朝</td><td>C．周朝</td><td>D．秦朝</td></tr></table>
    '''
    cons2 = '''
    <p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">1.下列对这首诗的赏析，不正确的一项是（ &nbsp;&nbsp;）</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><strong><span style="mso-spacerun: 'yes'; font-family: 宋体; mso-ansi-font-weight: bold; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">小寒食舟中作</span></strong></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">杜甫</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">佳辰强饮食犹寒，隐几萧条戴鹖冠。春水船如天上坐，老年花似雾中看。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">娟娟戏蝶过闲幔，片片轻鸥下急湍。云白山青万余里，愁看直北是长安。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">[注]这首诗写于大历五年春诗人淹留潭州时，即诗人去世前半年多。鹖（hé）冠：传为楚隐者鹖冠子所戴的帽子。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">A.首联中“强饮”一词是痛快豪饮的意思，表明诗人晚年要纵酒人生。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">B.颔联写诗人在船上所见所感，春来水涨，江流浩瀚，自己老眼昏花。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">C.颈联运用叠词，具有韵律美，写景由近及远，由蝴蝶而鸥鸟，层次分明。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">D.尾联总收全诗，诗人北望长安，思朝廷，忧愁顿生，有沉郁苍茫之美。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【答案】</span><span style="font-family: 宋体;">A</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【解析】</span><span style="font-family: 宋体;">“强饮”理解有误。应是“勉强吃一点饭”的意思。故选A。</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">2.下列对这首诗的赏析，不正确的一项是（ &nbsp;&nbsp;）</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><strong><span style="mso-spacerun: 'yes'; font-family: 宋体; mso-ansi-font-weight: bold; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">送客归江州</span></strong></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">韩翃</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">东归复得采真</span><sup><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt; vertical-align: super;"><span style="font-family: 宋体;">①</span></span></sup><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">游，江水迎君日夜流。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">客舍不离青雀舫，人家旧在白鸥洲</span><sup><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt; vertical-align: super;"><span style="font-family: 宋体;">②</span></span></sup><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">风吹山带遥知雨，露湿荷裳已报秋。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">闻道泉明</span><span style="font-family: 宋体;">③居止近，篮舆相访为淹留。</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【注】</span><span style="font-family: 宋体;">①采真：道教语，指顺乎天性，放任自然。②白鸥洲：指白鸥翔集的沙洲。此处借指客之家乡。③泉明：指晋陶渊明，此称其为泉明，乃避唐高祖李渊之讳。</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">A.这首诗写诗人送客人归江州隐居，但并无送别时的伤感，更多的是一种美好的祝福。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">B.“江水迎君”采用拟人手法，客人归心似箭、归程片刻不能迟的心态跃然纸上。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">C.“青雀舫”“白鸥洲”写出了诗人对客人旅舟华美，家乡景色宜人的赞美与羡慕。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">D.尾联写诗人听说陶渊明居所离客人很近，定会借探访陶渊明居所之机去拜访客人。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【答案】</span><span style="font-family: 宋体;">D</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【解析】</span><span style="font-family: 宋体;">D项，“定会借探访陶渊明居所之机去拜访客人”错误。尾联的意思是听说陶渊明居住的地方就在附近，你可以常常乘着竹轿，前往拜访。表达了诗人对客人隐逸情怀的赞美与羡慕。故选D。</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">3.对下面这首词的赏析，不恰当的一项是（ &nbsp;&nbsp;）</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><strong><span style="mso-spacerun: 'yes'; font-family: 宋体; mso-ansi-font-weight: bold; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">渔家傲</span></strong></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">范仲淹</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">塞下秋来风景异，衡阳雁去无留意。四面边声连角起，千嶂里，长烟落日孤城闭。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">浊酒一杯家万里，燕然未勒归无计。羌管悠悠霜满地，人不寐，将军白发征夫泪。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">A.这首词写出了我国北方秋季的景物特点，从词中的“塞下”“霜”等词语可以看出。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">B.“衡阳雁去”是说“大雁向衡阳飞去”而不是“大雁从衡阳飞走了”。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">C.这首词既表达了将士的爱国之心，又流露出思念亲人和家乡的感情。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">D.这首词感情悲观而消极，表达了鲜明的反战、厌战情绪。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【答案】</span><span style="font-family: 宋体;">D</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【解析】这首《渔家傲》为范仲淹创作，以描写北方秋季景物为背景，表达了作者对家国、亲人的思念以及将士们的英勇豪情。</span><span style="font-family: 宋体;">A项正确，词中的“塞下”“霜”等词语揭示了北方的秋季特点；B项正确，作者借衡阳雁南飞的景象暗示将士们向往家乡的渴望；C项正确，通过浓烈的爱国情感以及思念亲人的情绪表现，展现了作者的家国情怀和将士们的壮志豪情。不过，D项表述错误，词中并未明显表达反战、厌战情绪，其主要表达了将士们为国家和民族拼搏的精神。</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">4.对下面这首唐诗，赏析不恰当的一项是（ &nbsp;&nbsp;）</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><strong><span style="mso-spacerun: 'yes'; font-family: 宋体; mso-ansi-font-weight: bold; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">早梅</span></strong></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">万木冻欲折，孤根暖独回。前村深雪里，昨夜一枝开。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">风递幽香出，禽窥素艳来。明年如应律，先发望春台。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">A.首联把梅花与万木进行对比，万木的干枯摧折既有力地衬托了梅花的迎风斗雪，又好地照应了诗题中的“早”。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">B.颔联用华丽的语言为读者描绘出了一幅浓艳、高贵的雪中梅花图。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">C.颈联写梅花的风韵和姿色，尾联寄寓诗人深深的情思。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">D.这首咏梅诗，语言清丽，笔墨含蓄，有着强烈的艺术感染力。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【答案】</span><span style="font-family: 宋体;">B</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【解析】选项</span><span style="font-family: 宋体;">B不太恰当。颔联并没有用华丽的语言来描绘梅花，只是表达了梅花在寒雪中展现出的独立、高洁的风韵。这里并没有像选项B所说的“浓艳、高贵”。其余选项都能恰当地反映这首诗的特点和内容，因此答案选B。</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">5.对下面这首宋诗理解与赏析，不恰当的一项是（ &nbsp;&nbsp;）</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><strong><span style="mso-spacerun: 'yes'; font-family: 宋体; mso-ansi-font-weight: bold; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">村行</span></strong></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">王禹偁</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">马穿山径菊初黄，信马悠悠野兴长。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">万壑有声含晚籁，数峰无语立斜阳。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">棠梨叶落胭脂色，荞麦花开白雪香。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">何事吟余忽惆怅？村桥原树似吾乡。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">A.首联照应题目，点明地点和时令，写出了诗人信马徐行、观赏山野景色的悠然兴致。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">B.第二联上下句构成对比，生动地表现出山中有时喧响有时静穆的景象。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">C.第三联以“胭脂”和“白雪”为喻，形象地描绘出山村绚丽多彩的秋景。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">D.最后两句设为问答，抒发了诗人由外界景物所触发的浓浓的思乡之情。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【答案】</span><span style="font-family: 宋体;">B</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【解析】本题考查了对宋诗《村行》的理解与赏析。首先，</span><span style="font-family: 宋体;">A选项指出首联照应题目，点明了诗人信马行走在山间小路，看到菊花初黄，意境开阔。B选项提到第二联表现了山中有时喧响有时静穆的景象，但该联实际上并没有对比色彩，而是展示出千山万壑中奔涌着生机勃勃的晚响，无言的数峰沐浴在斜阳中。C选项正确地概括了第三联的内容，诗人通过赞美胭脂色的棠梨叶和白雪般芬芳的荞麦花存在世上，描绘出色彩斑斓的美景。D选项陈述了诗末以问答形式流露出的诗人对故乡情感的深刻思索。因此，答案为B选项，不恰当地解读了第二联。</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">6.下列对这首诗的赏析，不正确的一项是（ &nbsp;&nbsp;）</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><strong><span style="mso-spacerun: 'yes'; font-family: 宋体; mso-ansi-font-weight: bold; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">酬元九侍御赠璧竹鞭长句</span></strong><strong><sup><span style="mso-spacerun: 'yes'; font-family: 宋体; mso-ansi-font-weight: bold; font-size: 10.5000pt; mso-font-kerning: 1.0000pt; vertical-align: super;"><span style="font-family: 宋体;">①</span></span></sup></strong></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">刘禹锡</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">碧玉孤根生在林，美人相赠比双金。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">初开郢客缄封后，想见巴山冰雪深。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">多节本怀端直性，露青犹有岁寒心。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">何时策马同归去，关树扶疏</span><sup><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt; vertical-align: super;"><span style="font-family: 宋体;">②</span></span></sup><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">敲镫吟。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">[注]①此诗写于“永贞革新”失败后，作者被贬为朗州（今湖南）司马之时。元九，即诗人元稹，当时被贬为江陵（今湖北荆州）府士曹参军。②关树：关中之树。扶疏：枝叶繁茂。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">A.首联运用比兴手法，以碧玉般竹鞭的名贵，暗示赠鞭者的高尚，赞扬之情跃然纸上</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">B.颔联写诗人看到朋友赠礼后内心非常欣喜，很想去观赏生长碧竹的巴山冰雪美景。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">C.颈联通过“节”字，将“竹节”与“节操”相关联，把咏鞭与赞人联系在了一起。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">D.尾联由竹鞭引发联想，表达了诗人愿与友人“策马同去”“敲镫吟诗”的美好愿望。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【答案】</span><span style="font-family: 宋体;">B</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【解析】</span><span style="font-family: 宋体;">“很想去观赏生长碧竹的巴山冰雪美景”赏析有误。领联表达的意思是，我一打开郢客的缄封之后，立刻想到冰冻巴山雪深深。目睹竹鞭而展开联想，写出了制鞭之竹在“巴山冰雪”中傲然挺立的景象。这是对元稹不畏权势、宁折不弯的形象写照。是以竹喻人，表达对友人的赞美。译文：绿如碧玉的孤竹生在深林，用它制的璧竹鞭名贵万分；贤稳之人将竹鞭赠送给我，这份厚礼胜过了万两黄金。我一打开郢客的绒封之后，立刻想到冰冻巴山雪深深。鞭上多节，节节怀着端直性，遍体露青犹有岁寒后凋心。我们何时才能策马同归去，在扶疏的关树下敲镫高吟？</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">7.下列对这首诗的赏析，不正确的一项是（ &nbsp;&nbsp;）</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><strong><span style="mso-spacerun: 'yes'; font-family: 宋体; mso-ansi-font-weight: bold; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">见别离者因赠之</span></strong><strong><sup><span style="mso-spacerun: 'yes'; font-family: 宋体; mso-ansi-font-weight: bold; font-size: 10.5000pt; mso-font-kerning: 1.0000pt; vertical-align: super;"><span style="font-family: 宋体;">①</span></span></sup></strong></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">韩偓</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">征人草草尽戎装，征马萧萧立路傍。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">尊酒阑珊将远别，秋山迤逦更斜阳。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">白髭兄弟中年后，瘴海程途万里长。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">曾向天涯怀此恨，见君呜咽更凄凉。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">[注]①诗人生活在唐末战乱之际，当时自北而南，沿路所见，皆发于诗。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">A.标题点明本诗写作的原由，“别离”一词陡生无限伤感情绪，奠定全诗情感基调。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">B.首联紧扣“征人”与“征马”两个形象，真切地描绘了出征时的情景，画面感很强</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">C.颔联描写的是别后想象的虚景，诗人想象征人在离别亲人后沿着秋山远行的景象。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">D.尾联直抒胸臆，眼前的别离勾起了诗人对自身的感叹，抒发了心中的无奈和感慨。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【答案】</span><span style="font-family: 宋体;">C</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【解析】</span><span style="font-family: 宋体;">“颔联描写的是别后想象的虚景”说法错误，“尊酒阑珊将远别”是眼前实景。</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">8.下列对这首诗的赏析，不正确的一项是（ &nbsp;&nbsp;）</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><strong><span style="mso-spacerun: 'yes'; font-family: 宋体; mso-ansi-font-weight: bold; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">大热五首（其一）</span></strong></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">戴复古</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">天地一大窑，阳炭烹六月。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">万物此陶镕，人何怨炎热。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">君看百谷秋，亦自暑中结。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">田水沸如汤，背汗湿如泼。</span></p>
<p class="MsoNormal" style="text-align: center; vertical-align: middle; line-height: 150%;" align="center"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">农夫方夏耘，安坐吾敢食！</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">【注】陶镕：陶铸熔炼，比喻培育、造就。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">A.诗人把六月的天地比作一个大窑，太阳像炭火一样熔炼着其中的一切。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">B.暑天虽极炎热，诗人却认为不应抱怨，因为秋天的谷物均赖此而结实。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">C.田中的水被晒得似乎要沸腾，诗人的背上汗水流得就像刚刚用水泼过。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">D.这首诗描写暑热多用比喻和夸张修辞，语言平易浅近，风格质朴自然。</span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【答案】</span><span style="font-family: 宋体;">C</span></span></p>
<p class="MsoNormal" style="vertical-align: middle; line-height: 150%;"><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;"><span style="font-family: 宋体;">【解析】</span><span style="font-family: 宋体;">“背汗湿如泼”描写的是</span></span><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">农夫在暑热中辛苦劳作的情景</span><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">，</span><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">而不是指</span><span style="mso-spacerun: 'yes'; font-family: 宋体; font-size: 10.5000pt; mso-font-kerning: 1.0000pt;">诗人自己。</span></p>
    '''
    # pprint(cons)
    # print(again_parse(cons))
    # print(again_parse(cons))
    # print(list(map(lambda x: str(x).replace("     ", " "), again_parse(cons))))
    # con1 = r'<p>解：A．研究跨栏动作时，刘翔的大小和形状不能忽略，不能看作质点，故A错误；<br/>B．选取不同的参考系，物体的运动状态是不相同的，故B错误；<br/>C．出租车收费是按路程收费的，故C错误；<br/>D．第<img src="http://192.168.1.145:10811/static/physical_formulas_imgs/16184553930794225.png" data-latex="${4 \rm{s} }$" width="13",height="11" />是指<img src="http://192.168.1.145:10811/static/physical_formulas_imgs/16184553931930702.png" data-latex="${1 \rm{s} }$" width="12",height="11" />的时间，是指从<img src="http://192.168.1.145:10811/static/physical_formulas_imgs/16184553930220437.png" data-latex="${3 \rm{s} }$" width="13",height="11" />末到<img src="http://192.168.1.145:10811/static/physical_formulas_imgs/16184553930794225.png" data-latex="${4 \rm{s} }$" width="13",height="11" />末这一段时间，故D正确；<br/>故选：D．</p>'

    cons = css_label_wash(cons2)

    print(cons)