station_data.py

"""
全国车站坐标数据库
- 首次运行从开源数据库 (Seo-4d696b75/station_database) 下载全日本车站数据，
  解析为紧凑的本地缓存 stations.json (name -> [(lat, lng, prefecture), ...])。
- 之后离线可用。
- 提供按名称查坐标（含同名站消歧）、都道府县 -> 地区 映射。
- 下载失败时回退到内置的主要车站表，保证 App 仍可用。
"""
from __future__ import annotations
import json
import os
import threading
import urllib.request

_HERE = os.path.dirname(os.path.abspath(__file__))
# v2: 记录改为 [lat, lng, pref, lines]，新增线路字段(用于真实线路轨迹)
_CACHE = os.path.join(_HERE, "stations_v2.json")

# 全量数据源（含 name / lat / lng / prefecture 字段）
_SOURCE_URL = "https://raw.githubusercontent.com/Seo-4d696b75/station_database/main/out/main/station.json"

# ── 都道府县(1-47) → 地区 ─────────────────────────────────────────────────────
# (地区名, 概览图上的近似中心 lat/lon, [都道府县编号...])
REGIONS = [
    ("北海道", 43.1, 142.4, [1]),
    ("東北",   39.5, 140.6, [2, 3, 4, 5, 6, 7]),
    ("関東",   35.9, 139.8, [8, 9, 10, 11, 12, 13, 14]),
    ("中部",   36.3, 137.9, [15, 16, 17, 18, 19, 20, 21, 22, 23]),
    ("近畿",   34.7, 135.6, [24, 25, 26, 27, 28, 29, 30]),
    ("中国",   34.7, 132.7, [31, 32, 33, 34, 35]),
    ("四国",   33.8, 133.5, [36, 37, 38, 39]),
    ("九州",   32.2, 130.8, [40, 41, 42, 43, 44, 45, 46]),
    ("沖縄",   26.3, 127.8, [47]),
]

_PREF_TO_REGION = {}
_REGION_CENTER = {}
for _name, _lat, _lon, _prefs in REGIONS:
    _REGION_CENTER[_name] = (_lat, _lon)
    for _p in _prefs:
        _PREF_TO_REGION[_p] = _name


def region_of_pref(pref: int) -> str:
    return _PREF_TO_REGION.get(pref, "その他")


def region_center(name: str):
    return _REGION_CENTER.get(name, (37.0, 138.0))


# ── 内置回退表（下载失败时用，覆盖主要车站）──────────────────────────────────
# name -> (lat, lng, prefecture)
_FALLBACK = {
    "東京": (35.6812, 139.7671, 13), "新宿": (35.6896, 139.7006, 13),
    "渋谷": (35.6580, 139.7016, 13), "池袋": (35.7295, 139.7109, 13),
    "品川": (35.6284, 139.7387, 13), "上野": (35.7141, 139.7774, 13),
    "秋葉原": (35.6984, 139.7731, 13), "新橋": (35.6661, 139.7579, 13),
    "高田馬場": (35.7122, 139.7034, 13), "表参道": (35.6653, 139.7124, 13),
    "代々木上原": (35.6690, 139.6800, 13), "田原町": (35.7106, 139.7910, 13),
    "本駒込": (35.7330, 139.7470, 13), "飯田橋": (35.7021, 139.7449, 13),
    "早稲田": (35.7058, 139.7197, 13), "豊島園": (35.7430, 139.6470, 13),
    "横浜": (35.4658, 139.6223, 14), "川崎": (35.5308, 139.7025, 14),
    "大宮": (35.9062, 139.6237, 11), "千葉": (35.6133, 140.1130, 12),
    "大阪": (34.7024, 135.4959, 27), "梅田": (34.7025, 135.4983, 27),
    "難波": (34.6659, 135.5012, 27), "京都": (34.9858, 135.7588, 26),
    "三宮": (34.6947, 135.1980, 28), "名古屋": (35.1709, 136.8815, 23),
    "札幌": (43.0686, 141.3508, 1), "仙台": (38.2601, 140.8821, 4),
    "広島": (34.3975, 132.4753, 34), "博多": (33.5897, 130.4207, 40),
    "天神": (33.5912, 130.3990, 40), "那覇": (26.2124, 127.6792, 47),
}


# ── 站库 ──────────────────────────────────────────────────────────────────────
class StationDB:
    def __init__(self, table: dict):
        # table: name -> list[[lat, lng, pref]]
        self._t = {}
        self._normalized = {}
        for name, lst in table.items():
            if isinstance(lst, (list, tuple)) and lst and isinstance(lst[0], (int, float)):
                lst = [list(lst)]              # 单条 (lat,lng,pref)
            values = [tuple(x) for x in lst]
            self._t[name] = values
            self._normalized.setdefault(self._norm(name), []).extend(values)

    @staticmethod
    def _norm(name: str) -> str:
        if not name:
            return ""
        return (name.strip()
                    .replace("ヶ", "ケ").replace("ヵ", "カ")
                    .replace("　", ""))

    def candidates(self, name: str):
        """返回该站名所有候选 (lat,lng,pref)。"""
        if not name:
            return []
        if name in self._t:
            return self._t[name]
        return self._normalized.get(self._norm(name), [])

    def resolve(self, name: str, anchor=None):
        """选最合适的一个坐标。anchor=(lat,lng) 时选离 anchor 最近的同名站。"""
        cands = self.candidates(name)
        if not cands:
            return None
        if anchor is None or len(cands) == 1:
            return cands[0]
        ay, ax = anchor
        return min(cands, key=lambda c: (c[0]-ay)**2 + (c[1]-ax)**2)

    def lines_of(self, entry):
        """从一个 resolve() 返回的记录里取线路 line_cd 列表。"""
        if entry and len(entry) > 3 and isinstance(entry[3], (list, tuple)):
            return list(entry[3])
        return []

    def __len__(self):
        return len(self._t)


# ── 下载 / 加载 ────────────────────────────────────────────────────────────────
def _build_cache_from_source(timeout=90) -> dict:
    """下载全量数据，解析为 name -> [[lat,lng,pref],...]，写入缓存。"""
    req = urllib.request.Request(_SOURCE_URL, headers={"User-Agent": "SFCardViewerPro/1.0"})
    with urllib.request.urlopen(req, timeout=timeout) as r:
        raw = r.read()
    arr = json.loads(raw.decode("utf-8"))
    table: dict = {}
    for s in arr:
        try:
            if s.get("closed"):
                continue
            name = s["name"]
            lat = round(float(s["lat"]), 6)
            lng = round(float(s["lng"]), 6)
            pref = int(s.get("prefecture", 0))
        except (KeyError, TypeError, ValueError):
            continue
        try:
            lines = [int(x) for x in (s.get("lines") or [])]
        except (TypeError, ValueError):
            lines = []
        table.setdefault(name, []).append([lat, lng, pref, lines])
    if table:
        try:
            with open(_CACHE, "w", encoding="utf-8") as f:
                json.dump(table, f, ensure_ascii=False, separators=(",", ":"))
        except OSError:
            pass
    return table


def load(force_download=False) -> StationDB:
    """
    返回 StationDB。优先用本地缓存；没有则下载；下载失败回退内置表。
    这是一个可能联网的阻塞调用，建议放在后台线程里执行。
    """
    if not force_download and os.path.exists(_CACHE):
        try:
            with open(_CACHE, "r", encoding="utf-8") as f:
                return StationDB(json.load(f))
        except (OSError, json.JSONDecodeError):
            pass
    try:
        table = _build_cache_from_source()
        if table:
            return StationDB(table)
    except Exception:
        pass
    return StationDB(_FALLBACK)


def has_cache() -> bool:
    return os.path.exists(_CACHE)


# 异步加载封装
class AsyncLoader:
    def __init__(self):
        self.db: StationDB | None = None
        self._thread: threading.Thread | None = None

    def start(self, on_ready):
        def run():
            db = load()
            self.db = db
            on_ready(db)
        self._thread = threading.Thread(target=run, daemon=True)
        self._thread.start()


if __name__ == "__main__":
    db = load()
    print(f"车站数: {len(db)}  来源: {'缓存/下载' if len(db) > 100 else '回退表'}")
    for n in ("表参道", "代々木上原", "豊島園", "梅田", "札幌"):
        print(f"  {n}: {db.resolve(n)}")