platform-spiders/hb_hbzwfww_bacx_njpc/湖北政务服务网-备案查询-详情页.py

# -*- coding: utf-8 -*-
"""
Created on 2024-03-15
---------
@summary: 湖北政务服务网-备案查询
---------
@author: lzz
"""
import feapder
from items.njpc_item import DataNjpcItem

from hbzwfww_area_dict import area_dict
from utils import token as get_token


class Spider(feapder.PlanToBuildDetailSpider):

    def start_requests(self):
        data_list = self.get_tasks_by_rabbitmq(limit=50)
        for item in data_list:
            # log.debug(item)
            request_params = item.get("request_params")
            timeout = request_params.pop('timeout', 10)
            yield feapder.Request(url=item.get("parser_url"),
                                  item=item,
                                  deal_detail=item.get("deal_detail"),
                                  callback=item.get("parser"),
                                  timeout=timeout,
                                  **request_params)

    def download_midware(self, request):
        token = get_token()
        request.headers = {
            "Accept": "application/json, text/plain, */*",
            "Accept-Language": "zh-CN,zh;q=0.9",
            "Connection": "keep-alive",
            "Content-Type": "application/json",
            "Origin": "http://zwfw.hubei.gov.cn",
            "Referer": "http://zwfw.hubei.gov.cn/",
            "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/122.0.0.0 Safari/537.36",
            "authtoken": "null",
            "compkey": "hb",
            "extoken": f"{token}",
        }

    def detail_get(self, request, response):
        items = request.item
        data_item = DataNjpcItem(**items)

        detail_info = response.json.get('data')
        baz_dict = {"BA01": "尚未审核", "BA02": "有效", "BA03": "已注销", "BA04": "无此信息"}
        ba_examine_status = baz_dict.get(detail_info.get('hb_Record_card'))

        unit_nature_dict = {"A00006": "集体企业", "A00001": "政府机关", "A00007": "私营企业", "A00012": "其他",
                            "A00004": "国有及国有控股企业", "A00011": "村民委员会", "A00005": "股份制企业",
                            "A00008": "港澳台及外资企业"}
        unit_nature = unit_nature_dict.get(detail_info.get('unit_nature')) or "暂无内容"
        ba_apply_status_dict = {"A00001": "备案注销", "A00002": "变更备案证", "A00003": "生成备案证", }
        ba_apply_status = ba_apply_status_dict.get(detail_info.get('ba_apply_status')) or "暂无内容"
        construct_nature_dict = {"4": "技改及其他", "0": "新建", "1": "扩建", "2": "迁建", "3": "改建", }
        construct_nature = construct_nature_dict.get(detail_info.get('construct_nature')) or "暂无内容"

        project_address = ''
        if detail_info.get('project_address'):
            area_code = detail_info.get('project_address')[0]
            darea = ""
            dcity = ""
            ddistrict = ""
            for tf in area_dict:
                if area_code and len(area_code) > 0:
                    if area_code[0] == tf.get('_keye'):
                        darea = tf.get('name')
                        if len(area_code) > 1:
                            for ts in tf.get('children'):
                                if area_code[1] == ts.get('_keye'):
                                    dcity = ts.get('name')
                                    if len(area_code) > 2:
                                        for tt in ts.get('children'):
                                            if area_code[2] == tt.get('_keye'):
                                                ddistrict = tt.get('name')
            project_address = f"{darea}/{dcity}/{ddistrict}"

        html = f'''
        <form>
            <div>
                <div>
                    <div><span>项目名称</span></div>
                    <div><span>{detail_info.get('project_name')}</span></div>
                </div>

                <div>
                    <div><span>项目代码</span></div>
                    <div><span>{detail_info.get('project_code')}</span></div>
                </div>

                <div>
                    <div><span>建设内容及规</span></div>
                    <div><span>{detail_info.get('description')}</span></div>
                </div>

                <div>
                    <div><span>项目单位</span></div>
                    <div><span>{detail_info.get('unit_name')}</span></div>
                </div>

                <div>
                    <div><span>法人代表姓名</span></div>
                    <div><span>{detail_info.get('legal_name')}</span></div>
                </div>

                <div>
                    <div><span>项目单位性质</span></div>
                    <div><span>{unit_nature}</span></div>
                </div>

                <div>
                    <div><span>所属行政区划</span></div>
                    <div><span>{project_address}</span></div>
                </div>

                <div>
                    <div><span>项目总投资</span></div>
                    <div><span>{detail_info.get('amount')}</span></div>
                </div>

                <div>
                    <div><span>建设性质</span></div>
                    <div><span>{construct_nature}</span></div>
                </div>

                <div>
                    <div><span>拟开工时间</span></div>
                    <div><span>{detail_info.get('nstart_date')}</span></div>
                </div>

                <div>
                    <div><span>备案审核状态</span></div>
                    <div><span>{ba_examine_status}</span></div>
                </div>

                <div>
                    <div><span>备案证状态</span></div>
                    <div><span>{ba_apply_status}</span></div>
                </div>

                <div>
                    <div><span>申报日期</span></div>
                    <div><span>{detail_info.get('declaration_date')}</span></div>
                </div>

            </div>
        </form>
        '''

        data_item.contenthtml = html.replace('None', '暂无内容').replace('//', '')
        total_investment = detail_info.get('amount') or ''
        if total_investment:
            data_item.total_investment = str(total_investment) + "万元"

        yield data_item


if __name__ == '__main__':
    Spider(redis_key="lzz:hbzwfww_bacx").start()