"""
Fish Quality Check — FastAPI Backend
=====================================
Endpoint:
    POST /predict   → kirim foto mata + insang, terima grade A/B/C
    GET  /health    → cek status server & model
    GET  /docs      → dokumentasi API otomatis (Swagger UI)

Jalankan:
    uvicorn api:app --host 0.0.0.0 --port 8000 --reload
"""

import io
import os
import sys
import time
import logging
from datetime import datetime
from typing import Optional

from fastapi import FastAPI, File, UploadFile, HTTPException, Request
from fastapi.middleware.cors import CORSMiddleware
from fastapi.responses import JSONResponse
from pydantic import BaseModel
from PIL import Image

# Pastikan utils bisa diimport dari mana pun uvicorn dijalankan
BASE_DIR = os.path.dirname(os.path.abspath(__file__))
sys.path.insert(0, BASE_DIR)
from utils.inference import classify_fish, is_model_ready, load_model

# ── Logging ──────────────────────────────────────────────────────────────────
logging.basicConfig(
    level=logging.INFO,
    format="%(asctime)s  %(levelname)s  %(message)s",
)
logger = logging.getLogger(__name__)

# ── App ───────────────────────────────────────────────────────────────────────
app = FastAPI(
    title="Fish Quality Check API",
    description="""
## API untuk grading kualitas ikan ekspor

Upload foto **mata** dan **insang** ikan, dapatkan **Grade A / B / C** secara otomatis.

### Logika Grading
| Mata     | Insang    | Grade | Keterangan       |
|----------|-----------|-------|-----------------|
| Segar    | Segar     | **A** | Sangat Segar    |
| Segar    | Tdk Segar | **B** | Cukup Segar     |
| Tdk Segar| Segar     | **B** | Cukup Segar     |
| Tdk Segar| Tdk Segar | **C** | Tidak Segar     |
    """,
    version="1.0.0",
)

# ── CORS ─────────────────────────────────────────────────────────────────────
# Tambahkan URL frontend kamu di sini
ALLOWED_ORIGINS = [
    "http://localhost:3000",      # React dev server
    "http://localhost:5173",      # Vite dev server
    "http://localhost:8080",      # Vue dev server
    "https://your-frontend.com",  # Ganti dengan domain produksi
]

app.add_middleware(
    CORSMiddleware,
    allow_origins=ALLOWED_ORIGINS,
    allow_credentials=True,
    allow_methods=["GET", "POST"],
    allow_headers=["*"],
)

# ── Response schemas (Pydantic) ───────────────────────────────────────────────
class PartResult(BaseModel):
    kelas: str
    status: str
    confidence: float
    probabilitas: dict

class PredictResponse(BaseModel):
    grade: str
    label: str
    warna: str
    deskripsi: str
    layak_ekspor: bool
    mata: PartResult
    insang: PartResult
    waktu_proses_ms: float
    timestamp: str

class HealthResponse(BaseModel):
    status: str
    model_ready: bool
    device: str
    timestamp: str

# ── Startup: pre-load model ───────────────────────────────────────────────────
@app.on_event("startup")
async def startup_event():
    logger.info("Server starting up...")
    if is_model_ready():
        try:
            load_model()
            logger.info("Model loaded successfully on startup.")
        except Exception as e:
            logger.warning(f"Model load failed on startup: {e}")
    else:
        logger.warning("Model file not found. Run train.py first.")

# ── Middleware: log setiap request ────────────────────────────────────────────
@app.middleware("http")
async def log_requests(request: Request, call_next):
    start = time.time()
    response = await call_next(request)
    duration = (time.time() - start) * 1000
    logger.info(f"{request.method} {request.url.path} → {response.status_code} ({duration:.1f}ms)")
    return response

# ── Helper: validasi file upload ──────────────────────────────────────────────
ALLOWED_TYPES = {"image/jpeg", "image/png", "image/jpg", "image/webp"}
MAX_SIZE_MB   = 10

def validate_image(file: UploadFile, field_name: str) -> None:
    if file.content_type not in ALLOWED_TYPES:
        raise HTTPException(
            status_code=400,
            detail=f"Format file '{field_name}' tidak didukung. Gunakan JPG atau PNG."
        )

async def read_image(file: UploadFile, field_name: str) -> Image.Image:
    validate_image(file, field_name)
    raw = await file.read()

    if len(raw) > MAX_SIZE_MB * 1024 * 1024:
        raise HTTPException(
            status_code=400,
            detail=f"Ukuran file '{field_name}' terlalu besar. Maksimal {MAX_SIZE_MB}MB."
        )

    try:
        img = Image.open(io.BytesIO(raw)).convert("RGB")
    except Exception:
        raise HTTPException(
            status_code=400,
            detail=f"File '{field_name}' tidak dapat dibaca sebagai gambar."
        )
    return img

# ── Endpoints ─────────────────────────────────────────────────────────────────

@app.get("/health", response_model=HealthResponse, tags=["System"])
def health():
    """Cek status server dan apakah model sudah siap."""
    import torch
    device = "cuda" if torch.cuda.is_available() else "cpu"
    return HealthResponse(
        status="ok",
        model_ready=is_model_ready(),
        device=device,
        timestamp=datetime.now().isoformat(),
    )


@app.post(
    "/predict",
    response_model=PredictResponse,
    tags=["Prediction"],
    summary="Prediksi grade kualitas ikan",
    responses={
        200: {"description": "Prediksi berhasil"},
        400: {"description": "File tidak valid"},
        503: {"description": "Model belum siap"},
    },
)
async def predict(
    eye:  UploadFile = File(..., description="Foto mata ikan (JPG/PNG)"),
    gill: UploadFile = File(..., description="Foto insang ikan (JPG/PNG)"),
):
    """
    **Upload dua gambar:**
    - `eye`  → foto close-up mata ikan
    - `gill` → foto close-up insang ikan

    **Response:** grade A / B / C beserta detail confidence per bagian.
    """
    # Cek model
    if not is_model_ready():
        raise HTTPException(
            status_code=503,
            detail="Model belum tersedia. Jalankan train.py terlebih dahulu."
        )

    # Baca & validasi gambar
    eye_img  = await read_image(eye,  "eye")
    gill_img = await read_image(gill, "gill")

    # Jalankan inferensi
    try:
        start_t = time.time()
        result  = classify_fish(eye_img, gill_img)
        elapsed_ms = (time.time() - start_t) * 1000
    except Exception as e:
        logger.error(f"Inference error: {e}")
        raise HTTPException(
            status_code=500,
            detail=f"Terjadi error saat analisis: {str(e)}"
        )

    # Susun response
    return PredictResponse(
        grade        = result["grade"],
        label        = result["label"],
        warna        = result["color"],
        deskripsi    = result["description"],
        layak_ekspor = result["grade"] in ("A", "B"),
        mata=PartResult(
            kelas        = result["eye"]["class"],
            status       = result["eye"]["status"],
            confidence   = round(result["eye"]["confidence"], 4),
            probabilitas = {k: round(v, 4) for k, v in result["eye"]["probs"].items()},
        ),
        insang=PartResult(
            kelas        = result["gill"]["class"],
            status       = result["gill"]["status"],
            confidence   = round(result["gill"]["confidence"], 4),
            probabilitas = {k: round(v, 4) for k, v in result["gill"]["probs"].items()},
        ),
        waktu_proses_ms = round(elapsed_ms, 2),
        timestamp       = datetime.now().isoformat(),
    )


# ── Error handlers ────────────────────────────────────────────────────────────
@app.exception_handler(404)
async def not_found(request: Request, exc):
    return JSONResponse(
        status_code=404,
        content={"detail": f"Endpoint '{request.url.path}' tidak ditemukan."}
    )

@app.exception_handler(500)
async def server_error(request: Request, exc):
    return JSONResponse(
        status_code=500,
        content={"detail": "Internal server error. Cek log server."}
    )