from fastapi import FastAPI, UploadFile, File
from fastapi.responses import JSONResponse
from fastapi.middleware.cors import CORSMiddleware
from huggingface_hub import hf_hub_download
from pydantic import BaseModel

import numpy as np
import tensorflow as tf
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing.image import img_to_array
from tensorflow.keras.applications.efficientnet import preprocess_input

from PIL import Image
import json
import io

# ==== CONFIG ====
REPO_ID = "MAS-AI-0000/GameNet-1"
MODEL_FILENAME = "GameNetModel.h5"
#MODEL_FILENAME = "GameNetModel.keras"
LABELS_FILENAME = "label_to_index.json"
GENRE_FILENAME = "game_genre_map.json"
IMG_SIZE = (300, 300)

# ==== Load assets ====
model_path = hf_hub_download(repo_id=REPO_ID, filename=MODEL_FILENAME)
labels_path = hf_hub_download(repo_id=REPO_ID, filename=LABELS_FILENAME)
genre_path = hf_hub_download(repo_id=REPO_ID, filename=GENRE_FILENAME)

model = load_model(model_path)

with open(labels_path, "r") as f:
    label_to_index = json.load(f)
index_to_label = {v: k for k, v in label_to_index.items()}

with open(genre_path, "r") as f:
    genre_map = json.load(f)

# ==== FastAPI Setup ====
app = FastAPI()

# Optional: CORS if frontend is on different domain
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],  # change this in production
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# Response schema
class Prediction(BaseModel):
    game: str
    genre: str
    confidence: float

# Inference route
@app.post("/predict", response_model=Prediction)
async def predict(file: UploadFile = File(...)):
    try:
        # Step 1: Load image
        image_bytes = await file.read()
        img = Image.open(io.BytesIO(image_bytes)).convert("RGB")

        # Step 2: Resize for EfficientNetB3 (300x300)
        img = img.resize(IMG_SIZE, Image.Resampling.BICUBIC)

        # Step 3: Convert to array and preprocess
        arr = img_to_array(img)
        arr = preprocess_input(arr)  # normalize like in Colab
        arr = np.expand_dims(arr, axis=0)

        # Step 4: Inference
        preds = model.predict(arr)
        class_idx = int(np.argmax(preds))
        confidence = float(np.max(preds))

        # Step 5: Get label and genre
        game = index_to_label.get(class_idx, "Unknown")
        genre = genre_map.get(game, "Unknown")

        return Prediction(game=game, genre=genre, confidence=confidence)

    except Exception as e:
        return JSONResponse(content={"error": str(e)}, status_code=500)