craw_spider/area_info/city_name_to_id.py

import json
from pathlib import Path

# 直接导出一个全局映射，供其他脚本直接 import
_DEFAULT_PATH = Path(__file__).with_name("city.json")


def build_city_name_to_id():
    """
    递归读取 city.json，生成 {name: id} 映射。
    会遍历每个节点的 `sons` 字段（如存在）。
    """
    city_json_path = Path(_DEFAULT_PATH)
    raw = city_json_path.read_text(encoding="utf-8")
    data = json.loads(raw)

    name_to_id = {}

    def walk(nodes):
        for node in nodes:
            name = node.get("name")
            node_id = node.get("id")
            if name is not None and node_id is not None:
                # 若出现重名，后出现的会覆盖前面的值
                name_to_id[str(name)] = int(node_id)

            sons = node.get("sons") or []
            if sons:
                walk(sons)

    walk(data)
    return name_to_id


def build_city_name_to_id_2():
    city_json_path = Path(_DEFAULT_PATH)
    raw = city_json_path.read_text(encoding="utf-8")
    data_list = json.loads(raw)
    new_dict = {}
    for data in data_list:
        province = data["name"]
        province_id = data["id"]
        sons = data.get("sons") or []
        for son in sons:
            city = son.get("name", "")
            city_id = son.get("id", "")
            new_dict[city] = {"province": province, "province_id": province_id,
                              "city": city, "city_id": city_id}
    return new_dict


def get_city(city_str):
    city_dict = build_city_name_to_id()
    province = ""
    city = ""
    if "北京" in city_str:
        city = "北京市"
        province = "北京"
    if "重庆" in city_str:
        city = "重庆市"
        province = "重庆"
    if "上海" in city_str:
        city = "上海市"
        province = "上海"
    if "天津" in city_str:
        city = "天津市"
        province = "天津"
    if "新疆" in city_str:
        province = "新疆维吾尔自治区"
    if "宁夏" in city_str:
        province = "宁夏回族自治区"
    if "内蒙古" in city_str:
        province = "内蒙古自治区"
    if "西藏" in city_str:
        province = "西藏自治区"
    if "广西" in city_str:
        province = "广西壮族自治区"

    for _city, _city_id in city_dict.items():
        city_name = _city.replace("省", "").replace("市", "")
        if city_name in city_str and "省" in _city:
            if not province:
                province = _city
        if city_name in city_str and "市" in _city:
            if not city:
                city = _city
    city_id = province_id = 0
    if city:
        city_id = city_dict[city]
    if province:
        province_id = city_dict[province]
    return city_id, province_id, city, province


def get_city_2(city_str):
    """可以通过市获取省份"""
    city_dict = build_city_name_to_id_2()
    province = ""
    city = ""
    if "北京" in city_str:
        city = "北京市"
        province = "北京"
    if "重庆" in city_str:
        city = "重庆市"
        province = "重庆"
    if "上海" in city_str:
        city = "上海市"
        province = "上海"
    if "天津" in city_str:
        city = "天津市"
        province = "天津"
    if "新疆" in city_str:
        province = "新疆维吾尔自治区"
    if "宁夏" in city_str:
        province = "宁夏回族自治区"
    if "内蒙古" in city_str:
        province = "内蒙古自治区"
    if "西藏" in city_str:
        province = "西藏自治区"
    if "广西" in city_str:
        province = "广西壮族自治区"
    city_id = province_id = 0
    for _city, _city_value in city_dict.items():
        city_name = _city.replace("市", "")
        if city_name in city_str:
            # if not province:
            province = _city_value["province"]
            province_id = _city_value["province_id"]
            city = _city
            city_id = _city_value["city_id"]

    print(city_id, province_id, city, province)

    return city_id, province_id, city, province


if __name__ == "__main__":
    city_id, province_id, city, province = get_city("香港")
    print(city_id, province_id, city, province)
    get_city_2("哈尔滨")