app.py

import os
import numpy as np
import tensorflow as tf
from fastapi import FastAPI, UploadFile, File
from fastapi.staticfiles import StaticFiles
from scipy import signal
import soundfile as sf
from datetime import datetime
import threading, time

from team_code import base_model  # ton architecture

# ----------------------------
# CONFIG
# ----------------------------
SIG_LEN = 32256
N_FEATURES = 1
MODEL_PATH = "pretrained_model.h5"
OUTPUT_DIR = "/tmp/audio_results"
os.makedirs(OUTPUT_DIR, exist_ok=True)

BASE_URL = "https://stroke-ia-murmur-api.hf.space"  # adapte selon ton déploiement

# ----------------------------
# CHARGEMENT DU MODÈLE
# ----------------------------
print("[INFO] Chargement du modèle TensorFlow...")
model = base_model(SIG_LEN, N_FEATURES)
model.load_weights(MODEL_PATH)
model.compile()  # juste pour être sûr
print("[INFO] Modèle chargé ✅")

# ----------------------------
# FASTAPI
# ----------------------------
app = FastAPI(title="Heart Murmur Detection API")
app.mount("/files", StaticFiles(directory=OUTPUT_DIR), name="files")

# ----------------------------
# FONCTION DE PRÉTRAITEMENT
# ----------------------------
def preprocess_audio(file_path):
    data, sr = sf.read(file_path)
    if data.ndim > 1:
        data = np.mean(data, axis=1)
    resampled = signal.resample(data, SIG_LEN)
    return np.expand_dims(resampled, axis=(0, 2))  # (1, sig_len, 1)

# ----------------------------
# ENDPOINT PRINCIPAL
# ----------------------------
@app.post("/predict/")
async def predict_murmur(audio_file: UploadFile = File(...)):
    """
    Upload un fichier audio (.wav, .mp3) → renvoie diagnostic + probabilité
    """
    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
    tmp_path = os.path.join(OUTPUT_DIR, f"{timestamp}_{audio_file.filename}")
    with open(tmp_path, "wb") as f:
        f.write(await audio_file.read())

    try:
        x = preprocess_audio(tmp_path)
        pred = float(model.predict(x)[0][0])
        label = "Abnormal" if pred > 0.5 else "Normal"
        prob = round(pred, 3)

        report_path = os.path.join(OUTPUT_DIR, f"report_{timestamp}.txt")
        with open(report_path, "w") as f:
            f.write(f"Result: {label}\nProbability: {prob}\n")

        return {
            "diagnosis": label,
            "probability": prob,
            "rapport_url": f"{BASE_URL}/files/{os.path.basename(report_path)}",
            "message": "✅ Analyse audio terminée."
        }

    except Exception as e:
        return {"error": str(e)}

    finally:
        if os.path.exists(tmp_path):
            os.remove(tmp_path)

# ----------------------------
# AUTO-CLEANUP
# ----------------------------
def auto_cleanup(interval_minutes=10):
    while True:
        time.sleep(interval_minutes * 60)
        for file in os.listdir(OUTPUT_DIR):
            try:
                os.remove(os.path.join(OUTPUT_DIR, file))
                print(f"[CLEANUP] Fichier supprimé : {file}")
            except Exception as e:
                print(f"[CLEANUP] Erreur : {e}")

threading.Thread(target=auto_cleanup, daemon=True).start()