Minjin1050/mn-geocoding-api

1import os2import faiss3import joblib4import csv5import io6import json7import uvicorn8import numpy as np9from fastapi import FastAPI, UploadFile, File, HTTPException10from fastapi.responses import FileResponse11from fastapi.middleware.cors import CORSMiddleware12from sentence_transformers import SentenceTransformer13from pydantic import BaseModel14from typing import Optional, List15from contextlib import asynccontextmanager16from math import radians, cos, sin, asin, sqrt17def haversine_distance(lat1, lon1, lat2, lon2):18    try:19        # Утгуудыг заавал float төрөлд шилжүүлэх (Алдаанаас сэргийлнэ)20        lat1, lon1, lat2, lon2 = map(float, [lat1, lon1, lat2, lon2])21        22        # Хэрэв координатууд ижил бол зай 0 байна23        if lat1 == lat2 and lon1 == lon2:24            return 0.025 26        R = 6371000.0  # Дэлхийн радиус метрээр27        28        # Радиан руу шилжүүлэх29        phi1 = radians(lat1)30        phi2 = radians(lat2)31        d_phi = radians(lat2 - lat1)32        d_lambda = radians(lon2 - lon1)33 34        # Haversine томьёо35        a = sin(d_phi / 2)**2 + cos(phi1) * cos(phi2) * sin(d_lambda / 2)**236        c = 2 * asin(sqrt(a))37        38        distance = R * c39        return distance40    except Exception as e:41        print(f"Зай тооцоолоход алдаа гарлаа: {e}")42        return 0.043DISTRICT_MAPPING = {44    'baynzurkh duureg': 'bayanzurkh duureg', 'baynzurh duureg': 'bayanzurkh duureg',45    'bayanzurh': 'bayanzurkh duureg', 'baynzurh': 'bayanzurkh duureg', 'bzd': 'bayanzurkh duureg',46    'sukhbaatar duureg': 'sukhbaatar duureg', 'sbd': 'sukhbaatar duureg', 'sukhbaatr': 'sukhbaatar duureg',47    'bayangol duureg': 'bayangol duureg', 'bayngol duureg': 'bayangol duureg', 'bgd': 'bayangol duureg',48    'khan uul duureg': 'khan-uul duureg', 'khanuul duureg': 'khan-uul duureg', 'hud': 'khan-uul duureg',49    'han uul duureg': 'khan-uul duureg', 'hanuul duureg': 'khan-uul duureg', 'khud': 'khan-uul duureg',50    'chingeltei duureg': 'chingeltei duureg', 'chingeltii duureg': 'chingeltei duureg', 'chd': 'chingeltei duureg',51    'songinokhairkhan duureg': 'songinokhairkhan duureg', 'songinohairhan duureg': 'songinokhairkhan duureg',52    'shd': 'songinokhairkhan duureg', 'skhd': 'songinokhairkhan duureg'53}54 55CYRILLIC_TO_LATIN_MAP = {56    'А':'a','а':'a','Б':'b','б':'b','В':'v','в':'v','Г':'g','г':'g','Д':'d','д':'d','Е':'e','е':'e','Ё':'yo','ё':'yo',57    'Ж':'j','ж':'j','З':'z','з':'z','И':'i','и':'i','Й':'i','й':'i','К':'k','к':'k','Л':'l','л':'l','М':'m','м':'m',58    'Н':'n','н':'n','О':'o','о':'o','Ө':'o','ө':'o','П':'p','п':'p','Р':'r','р':'r','С':'s','с':'s','Т':'t','т':'t',59    'У':'u','у':'u','Ү':'u','ү':'u','Ф':'f','ф':'f','Х':'kh','х':'kh','Ц':'ts','ц':'ts','Ч':'ch','ч':'ch','Ш':'sh','ш':'sh',60    'Щ':'shch','щ':'shch','Ъ':'','ъ':'','Ы':'ii','ы':'ii','Ь':'','ь':'','Э':'e','э':'e','Ю':'yu','ю':'yu','Я':'ya','я':'ya'61}62 63def preprocess_address(address: str) -> str:64    latin_address = "".join(CYRILLIC_TO_LATIN_MAP.get(c, c) for c in address)65    address_lower = latin_address.lower()66    words = address_lower.split()67    processed_words = [DISTRICT_MAPPING.get(w, w) for w in words]68    final_address = " ".join(processed_words)69    for key, value in DISTRICT_MAPPING.items():70        if key in final_address:71            final_address = final_address.replace(key, value)72    return " ".join(final_address.split())73 74# =======================================================75# 2. ТОХИРГОО БА LIFESPAN76# =======================================================77BASE_DIR = os.path.dirname(os.path.abspath(__file__))78MODEL_PATH = os.path.join(BASE_DIR, "models", "address.pkl")79INDEX_PATH = os.path.join(BASE_DIR, "models", "faiss_final.bin")80FEEDBACK_FILE = "incorrect_predictions.csv"81 82geocoding_model = None83faiss_index = None84embed_model = None85 86@asynccontextmanager87async def lifespan(app: FastAPI):88    global geocoding_model, faiss_index, embed_model89    print("⏳ Hugging Face дээр моделуудыг ачаалж байна...")90    try:91        if os.path.exists(MODEL_PATH) and os.path.exists(INDEX_PATH):92            geocoding_model = joblib.load(MODEL_PATH)93            faiss_index = faiss.read_index(INDEX_PATH)94            embed_model = SentenceTransformer('paraphrase-multilingual-mpnet-base-v2')95            print("✅ Бүх модел амжилттай ачаалагдлаа!")96        else:97            print("❌ Файл олдсонгүй!")98    except Exception as e:99        print(f"❌ Ачаалахад алдаа гарлаа: {e}")100    yield101 102app = FastAPI(lifespan=lifespan)103 104app.add_middleware(105    CORSMiddleware,106    allow_origins=["*"],107    allow_credentials=True,108    allow_methods=["*"],109    allow_headers=["*"],110)111 112# =======================================================113# 3. MODELS & CORE SEARCH LOGIC114# =======================================================115class AddressRequest(BaseModel):116    address: str117 118class ValidationRequest(BaseModel):119    address: str120    user_lat: float121    user_lon: float122 123class AddressResponse(BaseModel):124    input_address: Optional[str] = None125    address: str126    latitude: float127    longitude: float128    match_quality: float129 130def get_coordinates(address: str) -> AddressResponse:131    if embed_model is None:132        return AddressResponse(address="Model not loaded", latitude=0, longitude=0, match_quality=0)133    134    query_vector = embed_model.encode(address, convert_to_numpy=True).astype('float32').reshape(1, -1)135    D, I = faiss_index.search(query_vector, 1)136    idx, dist = I[0][0], D[0][0]137    138    if idx == -1 or dist > 0.8:139        return AddressResponse(address="Олдсонгүй", latitude=0, longitude=0, match_quality=0)140    141    matched_data = geocoding_model.iloc[idx]142    return AddressResponse(143        address=str(matched_data['transliterated_address']),144        latitude=float(matched_data['latitude']),145        longitude=float(matched_data['longitude']),146        match_quality=float(1.0 - dist)147    )148 149# =======================================================150# 4. ENDPOINTS151# =======================================================152 153@app.get("/")154async def serve_frontend():155    return FileResponse("index.html")156 157@app.post("/geocode/single", response_model=AddressResponse)158async def geocode_single(request: AddressRequest):159    processed = preprocess_address(request.address)160    res = get_coordinates(processed)161    res.input_address = request.address162    return res163 164@app.post("/geocode/validate")165async def validate_address(request: ValidationRequest):166    """2-р табад зориулсан: Хаягаар хайж, гараас өгсөн координаттай тулгах"""167    processed = preprocess_address(request.address)168    res = get_coordinates(processed)169    170    if res.address == "Олдсонгүй":171        return {"status": "error", "message": "Модел хаягийг олсонгүй"}172 173    # Зайг тооцоолох174    distance = haversine_distance(request.user_lat, request.user_lon, res.latitude, res.longitude)175    176    return {177        "input_address": request.address,178        "user_coords": {"lat": request.user_lat, "lon": request.user_lon},179        "predicted_coords": {"lat": res.latitude, "lon": res.longitude},180        "predicted_address": res.address,181        "distance_meters": round(distance, 2),182        "is_correct": distance <= 50183    }184 185@app.post("/geocode/batch-file")186async def geocode_batch_file(file: UploadFile = File(...)):187    contents = await file.read()188    addresses = []189    if file.filename.endswith('.csv'):190        reader = csv.DictReader(io.StringIO(contents.decode('utf-8')))191        addresses = [row['address'] for row in reader if 'address' in row]192    elif file.filename.endswith('.json'):193        addresses = [item['address'] for item in json.loads(contents) if 'address' in item]194    195    results = []196    for addr in addresses:197        res = get_coordinates(preprocess_address(addr))198        res.input_address = addr199        results.append(res.dict())200    return results201 202@app.post("/feedback/log-incorrect")203async def log_incorrect(data: dict):204    file_exists = os.path.isfile(FEEDBACK_FILE)205    with open(FEEDBACK_FILE, mode='a', newline='', encoding='utf-8') as f:206        writer = csv.DictWriter(f, fieldnames=data.keys())207        if not file_exists: writer.writeheader()208        writer.writerow(data)209    return {"status": "logged"}210 211@app.get("/feedback/download-csv")212async def download_incorrect_csv():213    if not os.path.exists(FEEDBACK_FILE):214        raise HTTPException(status_code=404, detail="File not found")215    return FileResponse(FEEDBACK_FILE, filename="errors.csv")216 217if __name__ == "__main__":218    uvicorn.run(app, host="0.0.0.0", port=7860)