jean-router: streaming de messages d'attente pendant le WoL

- Quand le PC est eteint/en veille, stream immediat de messages
- 'PC de jeu se reveille...' → 'Magic packet envoye...' → 'Demarrage X%...'
- Une fois pret, forward la vraie reponse en stream
- Plus de blocage silencieux, l'utilisateur voit la progression

jean-router/app.py

@@ -13,7 +13,7 @@ import socket
 from typing import Optional
 
 import requests
-from flask import Flask, request, Response, jsonify
+from flask import Flask, request, Response, jsonify, stream_with_context
 
 JEAN_WEB_HOST = os.getenv("JEAN_WEB_HOST", "192.168.1.11")
 JEAN_WEB_PORT = int(os.getenv("JEAN_WEB_PORT", "8090"))
@@ -34,7 +34,7 @@ OLLAMA_UNLOAD_TIMEOUT = int(os.getenv("OLLAMA_UNLOAD_TIMEOUT", "15"))
 ENABLE_WOL = os.getenv("ENABLE_WOL", "false").lower() == "true"
 TARGET_MAC = os.getenv("TARGET_MAC", os.getenv("PC_GAMING_MAC", ""))
 BROADCAST_IP = os.getenv("BROADCAST_IP", "192.168.1.255")
-WAKE_TIMEOUT = int(os.getenv("WAKE_TIMEOUT", "120"))
+WAKE_TIMEOUT = int(os.getenv("WAKE_TIMEOUT", "180"))
 wol_in_progress = threading.Lock()
 jean_alive_cache = {"alive": False, "timestamp": 0}
 
@@ -152,6 +152,83 @@ def wait_for_jean(trigger_wol=True):
         log.error(f"[WoL] Jean ne repond pas apres {WAKE_TIMEOUT}s")
 
 
+def stream_wake_then_forward(upstream_url, data, headers):
+    """Stream des messages d'attente pendant le reveil, puis forward la vraie reponse"""
+    import uuid
+    chat_id = f"chatcmpl-{uuid.uuid4().hex[:29]}"
+
+    def wake_chunk(content):
+        return (
+            f'data: {{"id":"{chat_id}","object":"chat.completion.chunk",'
+            f'"created":{int(time.time())},"model":"wake","choices":['
+            f'{{"delta":{{"content":"{content} "}},"index":0}}]}}\n\n'
+        ).encode()
+
+    def generate():
+        yield wake_chunk("🖥️ Le PC de jeu se réveille...")
+
+        if not jean_alive() and ENABLE_WOL:
+            with wol_in_progress:
+                if not jean_alive():
+                    log.info("[WoL-Stream] Envoi WoL + attente...")
+                    send_wol()
+                    yield wake_chunk("⚡ Magic packet envoyé, démarrage en cours...")
+
+                    deadline = time.time() + WAKE_TIMEOUT
+                    last_pct = 0
+                    while time.time() < deadline:
+                        if jean_alive():
+                            jean_alive_cache["timestamp"] = 0
+                            yield wake_chunk("✅ PC allumé, chargement du modèle...")
+                            break
+                        pct = int((1 - (deadline - time.time()) / WAKE_TIMEOUT) * 100)
+                        if pct >= last_pct + 10:
+                            yield wake_chunk(f"⏳ Démarrage... {min(pct, 95)}%")
+                            last_pct = pct
+                        time.sleep(CHECK_INTERVAL)
+                    else:
+                        yield wake_chunk("❌ Le PC n'a pas répondu, réessayez.")
+                        yield b"data: [DONE]\n\n"
+                        return
+
+        # Attendre que Jean soit vraiment prêt (proxy + modele)
+        deadline = time.time() + SWITCH_TIMEOUT
+        while time.time() < deadline:
+            try:
+                r = requests.get(f"http://{JEAN_PROXY_HOST}:{JEAN_PROXY_PORT}/health", timeout=2)
+                if r.ok and r.json().get("status") == "ok":
+                    break
+            except Exception:
+                pass
+            yield wake_chunk(".")
+            time.sleep(2)
+
+        yield wake_chunk("🚀 Modèle prêt, voici la réponse :\n\n")
+
+        # Forward la vraie requête et streamer
+        try:
+            resp = requests.request(
+                method="POST", url=upstream_url, headers=headers,
+                json=data, stream=True, timeout=None,
+            )
+            for line in resp.iter_lines():
+                if line:
+                    yield line + b"\n"
+            yield b"data: [DONE]\n\n"
+        except Exception as e:
+            yield wake_chunk(f"❌ Erreur: {str(e)[:100]}")
+            yield b"data: [DONE]\n\n"
+
+    return Response(
+        stream_with_context(generate()),
+        content_type="text/event-stream",
+        headers={
+            "Cache-Control": "no-cache",
+            "X-Accel-Buffering": "no",
+        },
+    )
+
+
 def get_ollama_models(force_refresh=False):
     """Récupère la liste des modèles depuis Ollama"""
     if not ENABLE_OLLAMA:
@@ -761,7 +838,14 @@ def chat_completions():
     if model_id in jean_preset_ids or model_id in jean_names:
         internal_id = jean_name_to_id.get(model_id, model_id)
 
-        wait_for_jean(trigger_wol=True)
+        if not jean_alive() and ENABLE_WOL:
+            log.info("[WoL-Stream] PC down, demarrage stream d'attente...")
+            fwd_headers = dict(request.headers)
+            fwd_headers.pop("Host", None)
+            return stream_wake_then_forward(
+                f"http://{JEAN_PROXY_HOST}:{JEAN_PROXY_PORT}/v1/chat/completions",
+                data, fwd_headers
+            )
 
         if not ensure_vram_for_backend("jean"):
             return jsonify({"error": "Failed to prepare VRAM for Jean"}), 503