GK/3.31GK/frame_code/insert_flights_gk_route.py

from flights_mongodb import mongo_con_parse
from settings import FLIGHTS_WEBSITES_ROUTE_CONF_TAB, FLIGHTS_CITY_AIRPORT_CODE_TAB

db = mongo_con_parse()
website = "gk"


def get_from_to_data(input_str):
    input_str_list = input_str.split("\t")
    from_airport_code = input_str_list[0]
    to_airport_code = input_str_list[1]
    return {
        "from_airport_code": from_airport_code,
        "to_airport_code": to_airport_code
    }


def get_city_data(airport_code):
    city_code = None
    find_data = db.get_collection(FLIGHTS_CITY_AIRPORT_CODE_TAB).find_one(  # 这里拿出每个机场码所对应的城市码
        {
            "airport_code": airport_code,
            "website": website
        }
    )
    if find_data:
        city_code = find_data.get("city_code")
    return city_code


def get_airport_data(city_code):
    airport_code_list = []
    find_data = db.get_collection(FLIGHTS_CITY_AIRPORT_CODE_TAB).find(  # 这里要拿出每个城市所有对应的机场码
        {
            "city_code": city_code,
            "website": website
        }
    )
    for item in find_data:
        airport_code = item.get("airport_code")
        airport_code_list.append(airport_code)
    return airport_code_list


# 根据城市码获取机场对
def find_route_airport_pair(from_city_code, to_city_code):
    airport_code_pair_list = []
    find_data = db.get_collection(FLIGHTS_WEBSITES_ROUTE_CONF_TAB).find(  # 这里拿出在航线表里定义的城市码对关联的机场码对
        {
            "from_city_code": from_city_code,
            "to_city_code": to_city_code,
            "source_website": website
        }
    )
    for item in find_data:
        from_airport_code = item.get("from_airport_code")
        to_airport_code = item.get("to_airport_code")
        airport_code_pair_list.append((from_airport_code, to_airport_code))  # 存元组
    return airport_code_pair_list


def insert_routes():
    # 这里是城市代码, 网站没有的城市/机场码: OSA SPK TYO SHI, 删去它们的航段
    routes = """
    KCZ	FUK
    KCZ	KOJ
    KCZ	KMJ
    KCZ	KMI
    KCZ	OIT
    KCZ	OKA
    KCZ	AKJ
    KCZ	NGS
    FUK	KCZ
    FUK	MYJ
    FUK	NGO
    FUK	OKA
    FUK	TAK
    FUK	AKJ
    KOJ	KCZ
    KOJ	KMJ
    KOJ	MYJ
    KOJ	NGO
    KOJ	OKA
    KOJ	TAK
    KOJ	AKJ
    KMJ	KCZ
    KMJ	KOJ
    KMJ	MYJ
    KMJ	NGO
    KMJ	OKA
    KMJ	TAK
    KMJ	AKJ
    MYJ	FUK
    MYJ	KOJ
    MYJ	KMJ
    MYJ	KMI
    MYJ	OIT
    MYJ	OKA
    MYJ	AKJ
    MYJ	NGS
    KMI	KCZ
    KMI	MYJ
    KMI	OKA
    KMI	TAK
    KMI	AKJ
    NGO	FUK
    NGO	KOJ
    NGO	KMJ
    NGO	OKA
    OIT	KCZ
    OIT	MYJ
    OIT	OKA
    OIT	TAK
    OIT	AKJ
    OKA	KCZ
    OKA	FUK
    OKA	KOJ
    OKA	KMJ
    OKA	MYJ
    OKA	KMI
    OKA	NGO
    OKA	OIT
    OKA	TAK
    OKA	AKJ
    OKA	NGS
    TAK	FUK
    TAK	KOJ
    TAK	KMJ
    TAK	KMI
    TAK	OIT
    TAK	OKA
    TAK	AKJ
    TAK	NGS
    AKJ	KCZ
    AKJ	FUK
    AKJ	KOJ
    AKJ	KMJ
    AKJ	MYJ
    AKJ	KMI
    AKJ	OIT
    AKJ	OKA
    AKJ	TAK
    AKJ	NGS
    NGS	KCZ
    NGS	MYJ
    NGS	OKA
    NGS	TAK
    NGS	AKJ

    KCZ	FUK
    KCZ	KOJ
    KCZ	KMJ
    KCZ	KMI
    KCZ	OIT
    KCZ	OKA
    KCZ	KIX
    KCZ	CTS
    KCZ	NRT
    KCZ	HKO
    KCZ	AKJ
    KCZ	NGS
    FUK	KCZ
    FUK	MYJ
    FUK	NGO
    FUK	OKA
    FUK	KIX
    FUK	CTS
    FUK	TAK
    FUK	NRT
    FUK	HKO
    FUK	EKK
    FUK	AKJ
    KOJ	KCZ
    KOJ	KMJ
    KOJ	MYJ
    KOJ	NGO
    KOJ	OKA
    KOJ	CTS
    KOJ	TAK
    KOJ	NRT
    KOJ	HKO
    KOJ	EKK
    KOJ	AKJ
    KMJ	KCZ
    KMJ	KOJ
    KMJ	MYJ
    KMJ	NGO
    KMJ	OKA
    KMJ	KIX
    KMJ	CTS
    KMJ	TAK
    KMJ	NRT
    KMJ	HKO
    KMJ	EKK
    KMJ	AKJ
    MYJ	FUK
    MYJ	KOJ
    MYJ	KMJ
    MYJ	KMI
    MYJ	OIT
    MYJ	OKA
    MYJ	CTS
    MYJ	NRT
    MYJ	HKO
    MYJ	AKJ
    MYJ	NGS
    KMI	KCZ
    KMI	MYJ
    KMI	OKA
    KMI	KIX
    KMI	CTS
    KMI	TAK
    KMI	NRT
    KMI	HKO
    KMI	EKK
    KMI	AKJ
    NGO	FUK
    NGO	KOJ
    NGO	KMJ
    NGO	OKA
    NGO	CTS
    NGO	HKO
    OIT	KCZ
    OIT	MYJ
    OIT	OKA
    OIT	KIX
    OIT	CTS
    OIT	TAK
    OIT	NRT
    OIT	HKO
    OIT	EKK
    OIT	AKJ
    OKA	KCZ
    OKA	FUK
    OKA	KOJ
    OKA	KMJ
    OKA	MYJ
    OKA	KMI
    OKA	NGO
    OKA	OIT
    OKA	KIX
    OKA	CTS
    OKA	TAK
    OKA	NRT
    OKA	HKO
    OKA	EKK
    OKA	AKJ
    OKA	NGS
    KIX	KCZ
    KIX	FUK
    KIX	KMJ
    KIX	KMI
    KIX	OIT
    KIX	OKA
    KIX	CTS
    KIX	NRT
    KIX	HKO
    KIX	EKK
    KIX	AKJ
    KIX	NGS
    CTS	KCZ
    CTS	FUK
    CTS	KOJ
    CTS	KMJ
    CTS	MYJ
    CTS	KMI
    CTS	NGO
    CTS	OIT
    CTS	OKA
    CTS	KIX
    CTS	TAK
    CTS	NRT
    CTS	EKK
    CTS	NGS
    TAK	FUK
    TAK	KOJ
    TAK	KMJ
    TAK	KMI
    TAK	OIT
    TAK	OKA
    TAK	CTS
    TAK	NRT
    TAK	HKO
    TAK	AKJ
    TAK	NGS
    NRT	KCZ
    NRT	FUK
    NRT	KOJ
    NRT	KMJ
    NRT	MYJ
    NRT	KMI
    NRT	OIT
    NRT	OKA
    NRT	KIX
    NRT	CTS
    NRT	TAK
    NRT	HKO
    NRT	EKK
    NRT	AKJ
    NRT	NGS
    HKO	KCZ
    HKO	FUK
    HKO	KOJ
    HKO	KMJ
    HKO	MYJ
    HKO	KMI
    HKO	NGO
    HKO	OIT
    HKO	OKA
    HKO	KIX
    HKO	TAK
    HKO	NRT
    HKO	EKK
    HKO	NGS
    EKK	FUK
    EKK	KOJ
    EKK	KMJ
    EKK	KMI
    EKK	OIT
    EKK	OKA
    EKK	KIX
    EKK	CTS
    EKK	NRT
    EKK	HKO
    EKK	AKJ
    EKK	NGS
    AKJ	KCZ
    AKJ	FUK
    AKJ	KOJ
    AKJ	KMJ
    AKJ	MYJ
    AKJ	KMI
    AKJ	OIT
    AKJ	OKA
    AKJ	KIX
    AKJ	TAK
    AKJ	NRT
    AKJ	EKK
    AKJ	NGS
    NGS	KCZ
    NGS	MYJ
    NGS	OKA
    NGS	KIX
    NGS	CTS
    NGS	TAK
    NGS	NRT
    NGS	HKO
    NGS	EKK
    NGS	AKJ
       """

    db.get_collection(FLIGHTS_WEBSITES_ROUTE_CONF_TAB).delete_many({"source_website": website})  # 清空之前数据
    temp_list = [i.strip() for i in routes.split("\n") if i.strip()]
    routes = list(set(temp_list))  # 去重
    routes.sort(key=temp_list.index)  # 保持原有顺序
    # print(len(routes))
    # total_route = {x for route in routes for x in route.split('\t')}
    # print(f'一共需要 {len(total_route)} 条航线的城市机场对应信息')

    insert_datas = []  # 列表里准备存字典

    count = 0
    for route in routes:
        name_code_datas = get_from_to_data(route)
        # print(route, name_code_datas)
        from_city_code = name_code_datas.get("from_airport_code")
        to_city_code = name_code_datas.get("to_airport_code")

        from_airport_code = get_airport_data(from_city_code)
        to_airport_code = get_airport_data(to_city_code)
        dict_mode = {
            "source_website": website,  # 数据来源网站
            "website_status": 1,  # 网站是否采集状态
            "flight_route_status": 1,  # 航线是否采集状态
            "from_city_code": from_city_code,
            "from_city_name": "",
            "from_airport_code": from_airport_code,
            "to_city_code": to_city_code,
            "to_city_name": "",
            "to_airport_code": to_airport_code,
        }
        insert_datas.append(dict_mode)
        count += 1
        if count % 50 == 0:
            print("add {} ...".format(count))
    db.get_collection(FLIGHTS_WEBSITES_ROUTE_CONF_TAB).insert_many(insert_datas)  # 一次录入多条数据
    print("finish...")


if __name__ == "__main__":
    insert_routes()