app.py

#!/usr/bin/env python3
"""
Serveur HTTP minimal pour DeepSeek-Coder
"""
import http.server
import socketserver
import json
import subprocess
import threading
import time

# Démarrer Ollama
def start_ollama():
    subprocess.run(["ollama", "serve"])

# Lancer Ollama en thread
threading.Thread(target=start_ollama, daemon=True).start()
time.sleep(30)

# Télécharger le modèle
print("Téléchargement du modèle...")
subprocess.run(["ollama", "pull", "deepseek-coder:1.3b"], capture_output=True)

# HTML simple
HTML = open("index.html").read() if os.path.exists("index.html") else "<h1>DeepSeek-Coder 1.3B</h1>"

class OllamaHandler(http.server.SimpleHTTPRequestHandler):
    def do_GET(self):
        if self.path == "/":
            self.send_response(200)
            self.send_header("Content-type", "text/html")
            self.end_headers()
            self.wfile.write(HTML.encode())
        else:
            super().do_GET()
    
    def do_POST(self):
        if self.path == "/generate":
            content_length = int(self.headers['Content-Length'])
            post_data = self.rfile.read(content_length)
            data = json.loads(post_data)
            
            # Appeler Ollama
            result = subprocess.run(
                ["ollama", "run", "deepseek-coder:1.3b", data.get("prompt", "")],
                capture_output=True,
                text=True
            )
            
            self.send_response(200)
            self.send_header("Content-type", "application/json")
            self.end_headers()
            self.wfile.write(json.dumps({"response": result.stdout}).encode())

# Démarrer le serveur
if __name__ == "__main__":
    PORT = 7860
    with socketserver.TCPServer(("", PORT), OllamaHandler) as httpd:
        print(f"Serveur démarré sur le port {PORT}")
        httpd.serve_forever()