data_quality_server/tables/fields/area.py

import os
import re
import pandas as pd

class AreaChecker(object):
    def __init__(self):
        self.errors_tables = {
            "0101": {
                "name": "全国类数据",
                "parent_name": "全国类型",
                "parent_code": "01",
                "checkFn": self.check0101
            },
            "0201": {
                "name": "没有市县正文中有",
                "parent_name": "市县类型",
                "parent_code": "01",
                "checkFn": self.check0201
            }
        }

    def check0101(self, area: str) -> bool:
        return area == "全国"

    def check0201(self, area, city, district, detail) -> bool:
        def find_city_info(query):
            for index, row in df.iterrows():
                if query in str(row["地市"]) or query in str(row["地区代码"]):
                    return {
                        "省份": row["省份"],
                        "地市": row["地市"],
                        "区县": row["区县"],
                        "父级地区代码": row["父级地区代码"],
                        "地区代码": row["地区代码"]
                    }
            return None

        # 获取当前脚本所在目录的上一级目录
        current_dir = os.path.dirname(__file__)
        parent_dir = os.path.abspath(os.path.join(current_dir, os.pardir))

        # 构建 Excel 文件路径
        xls_file = os.path.join(parent_dir, 'docs', 'area.xlsx')
        # 加载地区代码的XLS文件
        # xls_file = "C:\\Users\\25503\\PycharmProjects\\data_quality\\docs\\table_head_doc\\aera.xls"
        df = pd.read_excel(xls_file)

        detail = re.sub(r"<.*?>", "", detail)
        # 提取采购单位名称
        unit_name_pattern = r"采购单位名称:([\s\S]+?)\n"
        unit_name_match = re.search(unit_name_pattern, detail)
        purchase_unit_name = unit_name_match.group(1) if unit_name_match else None

        # 提取采购单位地址
        purchase_unit_address = r"采购单位地址:([\s\S]+?)\n"
        address_match = re.search(purchase_unit_address, detail)
        city_from_unit_address = address_match.group(1) if address_match else None

        # 提取项目所在行政区划编码
        project_district_code = r"项目所在行政区划编码:(\d+)"
        district_code_match = re.search(project_district_code, detail)
        district = district_code_match.group(1) if district_code_match else None

        print(f"提取到的区县: {district}")
        print(f"提取到的地市: {city}")
        print(f"提取到的省份: {area}")

        if not district:
            if purchase_unit_name:
                city_info = find_city_info(purchase_unit_name)
                print(f"根据单位名称找到的城市信息: {city_info}")
            if city_info is None and city_from_unit_address:
                city_info = find_city_info(city_from_unit_address)

            if city_info:
                if city_info["省份"] == area and city_info["地市"] == city:
                    print("省份与区域一致，地市与城市一致，不做处理")
                else:
                    print("省份或地市与区域不一致，标记为true")
                    return True
            else:
                print("未找到匹配的城市信息，返回true")
                return True

        else:
            print("district不为空，保持原有标记")
            return False   # 保持原有的标记