new_tiku_structure_2021/structure/danti_structure.py

"""
单题再解析、结构化
"""
import re

from structure.option import option_structure
from utils.equation_extract import get_simpstr2eqn, get_equation_instr
from utils.html_again_parse import css_label_wash
from utils.washutil import base642img, css_conflict_deal


def single_parse(one_item, item_type, wordid):
    """
    rtype:题型
    :return:
    """
    # if re.search("选择|不定选择|多选|单选", rtype):
    #     if "【选项】" not in one_item:
    #         return "请不要将本编辑框自带的“【选项】、【答案】、【解析】”字段删除"
    # if "【答案】" not in one_item or "【解析】" not in one_item:
    #     return "请不要将本编辑框自带的“【答案】、【解析】”字段删除"
    print("原始：")
    print(one_item)
    one_item = css_label_wash(one_item)
    one_item = re.sub("</?p>|<h[12345]", "", one_item)
    print("去css：")
    print(one_item)
    one_item = base642img(one_item, wordid)
    one_item = css_conflict_deal(one_item)
    one_item = re.sub(r"\\\(|\\\)", "$", one_item)

    res_list = re.split(r"(\n+【答案】|\n+【解析】)", one_item)
    if len(res_list) > 5:
        return "编辑后的文本出现多个【答案】或【解析】字段，请删除，每个字段只保留唯一且单独成行"
    elif len(res_list) < 5:
        res_list = re.split(r"(【答案】|【解析】)", one_item)
        if len(res_list) > 5:
            return "编辑后的文本出现多个【答案】或【解析】字段，请删除，每个字段只保留唯一且单独成行"
        # elif len(res_list) < 5:
        #     return "编辑后的文本掉了【答案】或【解析】字段，请添加，每个字段保留唯一"

    new_item_struct = {}
    new_item_struct["key"] = ""
    new_item_struct["parse"] = ""
    new_item_struct["stem"] = res_list[0]
    if len(res_list) == 5:
        if res_list[1].strip() == "【答案】" and res_list[3].strip() == "【解析】":
            new_item_struct["key"] = res_list[2]
            new_item_struct["parse"] = res_list[4]
        elif res_list[1].strip() == "【解析】" and res_list[3].strip() == "【答案】":
            new_item_struct["key"] = res_list[4]
            new_item_struct["parse"] = res_list[2]
        else:
            return "编辑后的文本出现多个【答案】或【解析】字段，请修改，每个字段只保留唯一且单独成行"
    elif len(res_list) > 2:
        if res_list[1].strip() == "【答案】":
            new_item_struct["key"] = res_list[2]
        elif res_list[1].strip() == "【解析】":
            new_item_struct["parse"] = res_list[2]

    # 选项再解析
    new_item_struct["errmsgs"] = []
    if re.search("选择|不定选择|多选|单选", item_type):
        new_item_struct = option_structure(new_item_struct, res_list[0], new_item_struct["key"], 1, 1)
        if new_item_struct["errmsgs"]:
            return ";;".join(new_item_struct["errmsgs"])
        if any([True for i in new_item_struct['options'] if not i.replace("：", "").strip()]):  # 空选项中：被当成了内容
            return "存在选项为空，请补充完整"

    item_ids = re.findall("^([1-5][0-9]|[1-9])\s*[.．、､]", new_item_struct["stem"].strip())
    new_item_struct["stem"] = re.sub("^([1-5][0-9]|[1-9])\s*[.．、､]", "", new_item_struct["stem"].strip())
    if len(new_item_struct["stem"].strip())<3:
        return "题干为空，请补充完整"

    new_item_struct["topic_num"] = item_ids[0] if item_ids else 0
    # 换行符替换
    new_item_struct["stem"] = new_item_struct["stem"].strip().replace("\n\n", "\n").replace("\n", "<br/>")  # 2020/4/10 gai
    new_item_struct["key"] = new_item_struct["key"].strip().replace("\n\n", "\n").replace("\n", "<br/>")
    new_item_struct["parse"] = new_item_struct["parse"].strip().replace("\n\n", "\n").replace("\n", "<br/>")
    if "options" in new_item_struct:  # 对选项部分进行格式处理  get_equation_instr
        for i in range(len(new_item_struct['options'])):
            new_item_struct['options'][i] = new_item_struct['options'][i].strip().replace("\n\n", "\n")\
                                                            .replace("\n", "<br/>")
    print(new_item_struct)
    return new_item_struct


if __name__ == '__main__':
    html = '''
    17.为测定干电池的电动势和内阻，提供的实验器材如下所示：
A．干电池2节，每节干电池的电动势为1.5 V<img src="http://zxhx-1302712961.cos.ap-shanghai.myqcloud.com/zyk/uploadfiles/wording/6167c1360d2da861d11c950a/image34.png" width="26px" height="26px">左右，内阻较小
B．电流表A(量程为0～0.6 A，内阻约0.5 Ω)
C．滑动变阻器R1(0～10 Ω，10 A)D．滑动变阻器R2(0～100 Ω，1 A)
E．电流表G(0～3.0 mA，Rg=10 Ω)F．定值电阻R1=990 Ω
G．定值电阻R2=90 ΩH．开关S和导线若干
(1)由于两节干电池的内阻较小，现将定值电阻R0=3.0 Ω与两节干电池串联后作为一个整体进行测量。在进行实验时，滑动变阻器应选用________，定值电阻应选用________。(填写实验器材前的编号)

    '''

    aa = single_parse(html, "解答题", "6167c1360d2da861d11c950a")
    print(aa)