MAJ main.py

main.py

@@ -9,64 +9,92 @@ from dotenv import load_dotenv
 
 load_dotenv()
 
-# --- 1. GESTION DE L'HISTORIQUE (EPINGLAGE, RENOMMAGE, SUPPRESSION) ---
+# --- 0. INITIALISATION DES DOSSIERS (BDD & LOGS) ---
+os.makedirs("sessions", exist_ok=True)
+os.makedirs("logs", exist_ok=True)
+
+def log_event(session_id, action, level="INFO"):
+    """Enregistre chaque action dans un fichier texte dédié à la session."""
+    sid = session_id if session_id else "system"
+    log_file = f"logs/session_{sid}.log"
+    timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+    with open(log_file, "a", encoding="utf-8") as f:
+        f.write(f"[{timestamp}] [{level}] {action}\n")
+
+# --- 1. GESTION DE L'HISTORIQUE (INDEX & SESSIONS) ---
 HISTORY_FILE = "history.json"
 
-def load_history():
+def load_index():
     if not os.path.exists(HISTORY_FILE): return {}
     with open(HISTORY_FILE, "r", encoding="utf-8") as f:
-        try: return json.load(f)
+        try: 
+            idx = json.load(f)
+            # Migration automatique si l'ancien format (avec "messages" dedans) est détecté
+            for sid, data in list(idx.items()):
+                if "messages" in data:
+                    save_session_data(sid, data["messages"])
+                    del data["messages"]
+            return idx
         except: return {}
 
-def save_history(history_data):
+def save_index(idx):
     with open(HISTORY_FILE, "w", encoding="utf-8") as f:
-        json.dump(history_data, f, indent=4, ensure_ascii=False)
+        json.dump(idx, f, indent=4, ensure_ascii=False)
+
+def load_session_data(session_id):
+    path = f"sessions/{session_id}.json"
+    if not os.path.exists(path): return []
+    with open(path, "r", encoding="utf-8") as f:
+        try: return json.load(f)
+        except: return []
+
+def save_session_data(session_id, messages):
+    with open(f"sessions/{session_id}.json", "w", encoding="utf-8") as f:
+        json.dump(messages, f, indent=4, ensure_ascii=False)
 
 def create_new_session():
     session_id = str(uuid.uuid4())
-    history = load_history()
-    history[session_id] = {"title": f"Discussion du {datetime.now().strftime('%d/%m %H:%M')}", "pinned": False, "messages": []}
-    save_history(history)
+    idx = load_index()
+    idx[session_id] = {"title": f"Discussion du {datetime.now().strftime('%d/%m %H:%M')}", "pinned": False}
+    save_index(idx)
+    save_session_data(session_id, [])
+    log_event(session_id, "Nouvelle session initialisée.")
     return session_id
 
 def render_history_list():
-    """Génère la liste HTML de l'historique avec les boutons d'action."""
-    history = load_history()
-    sorted_sessions = sorted(history.items(), key=lambda x: (not x[1].get('pinned', False), x[1]['title']), reverse=False)
+    idx = load_index()
+    sorted_sessions = sorted(idx.items(), key=lambda x: (not x[1].get('pinned', False), x[1]['title']), reverse=False)
+    
+    items = [
+        Button("➕ Nouvelle Discussion", cls="history-btn w-full mb-3 text-center", style="background:#00e5ff; color:#0f172a; font-weight:bold;", hx_get="/session/new", hx_target="#chat-history")
+    ]
     
-    items = []
     for sid, data in sorted_sessions:
         pin_icon = "📍" if data.get('pinned') else "📌"
         pin_color = "#10b981" if data.get('pinned') else "#94a3b8"
         
         item = Div(
-            Div(data['title'], cls="truncate flex-grow cursor-pointer"),
+            Div(data['title'], cls="truncate flex-grow cursor-pointer hover:text-cyan-400", hx_get=f"/session/load/{sid}", hx_target="#chat-history"),
             Div(
-                # Bouton Renommer
-                Button("✏️", cls="text-xs mx-1 hover:text-white", 
-                       onclick=f"let name = prompt('Nouveau nom:'); if(name) {{ htmx.ajax('POST', '/history/rename/{sid}', {{values: {{title: name}}, target: '#history-list'}}); }}"),
-                # Bouton Epingler
-                Button(pin_icon, style=f"color:{pin_color};", cls="text-xs mx-1 hover:text-white", 
-                       hx_post=f"/history/pin/{sid}", hx_target="#history-list"),
-                # Bouton Supprimer (Nouveau)
-                Button("🗑️", cls="text-xs hover:text-red-500", 
-                       onclick=f"if(confirm('Supprimer définitivement cette discussion ?')) {{ htmx.ajax('POST', '/history/delete/{sid}', {{target: '#history-list'}}); }}"),
+                Button("✏️", cls="text-xs mx-1 hover:text-white", onclick=f"let name = prompt('Nouveau nom:'); if(name) {{ htmx.ajax('POST', '/history/rename/{sid}', {{values: {{title: name}}, target: '#history-list-container'}}); }}"),
+                Button(pin_icon, style=f"color:{pin_color};", cls="text-xs mx-1 hover:text-white", hx_post=f"/history/pin/{sid}", hx_target="#history-list-container"),
+                Button("🗑️", cls="text-xs hover:text-red-500", onclick=f"if(confirm('Supprimer définitivement cette discussion ?')) {{ htmx.ajax('POST', '/history/delete/{sid}', {{target: '#history-list-container'}}); }}"),
                 cls="flex-shrink-0"
             ),
             cls="flex justify-between items-center history-btn w-full mb-1"
         )
         items.append(item)
-    return Div(*items, id="history-list", style="flex-grow: 1; overflow-y: auto;")
+    return Div(*items, id="history-list-container", hx_swap_oob="true")
 
 
-# --- 2. GESTION DU CATALOGUE (GENERIQUE) ---
+# --- 2. GESTION DU CATALOGUE (GENERIQUE & OPENROUTER) ---
 MODELS_DATA = []
 def init_models():
     global MODELS_DATA
     MODELS_DATA = [
-        {"id": "groq|llama3-8b-8192", "name": "Groq - Llama 3 (8B)", "is_free": True},
-        {"id": "gemini|gemini-1.5-flash", "name": "Google - Gemini 1.5 Flash", "is_free": True},
-        {"id": "deepseek|deepseek-coder", "name": "DeepSeek - Coder", "is_free": False}
+        {"id": "groq|llama3-8b-8192", "name": "Groq - Llama 3 (8B)", "is_free": True, "description": "Modèle ultra-rapide hébergé par Groq. Idéal pour les tâches générales, la structuration de données et le code basique.", "architecture": {"modality": "text"}},
+        {"id": "gemini|gemini-1.5-flash", "name": "Google - Gemini 1.5 Flash", "is_free": True, "description": "Modèle multimodal très rapide et léger de Google. Parfait pour le raisonnement rapide, le code et l'analyse de longs contextes.", "architecture": {"modality": "text/vision"}},
+        {"id": "deepseek|deepseek-coder", "name": "DeepSeek - Coder", "is_free": False, "description": "Modèle d'élite spécialisé dans l'ingénierie logicielle, la création d'algorithmes et la génération de code complexe.", "architecture": {"modality": "text/code"}}
     ]
     try:
         res = requests.get("https://openrouter.ai/api/v1/models", timeout=5)
@@ -75,7 +103,8 @@ def init_models():
                 m["id"] = f"openrouter|{m['id']}"
                 m["name"] = f"OR - {m['name']}"
                 MODELS_DATA.append(m)
-    except: pass
+    except Exception as e:
+        log_event("system", f"Erreur chargement catalogue OpenRouter : {e}", "ERROR")
 init_models()
 
 def get_model_options(filter_type="all"):
@@ -98,7 +127,8 @@ def get_budget():
         return f"{data.get('total_credits', 0) - data.get('total_usage', 0):.4f} $"
     except: return "Erreur lecture"
 
-def handle_api_error(provider, raw_error):
+def handle_api_error(session_id, provider, raw_error):
+    log_event(session_id, f"Erreur API {provider} : {raw_error}", "ERROR")
     sys_prompt = "Tu es un assistant technique. Une API distante a renvoyé l'erreur suivante. Traduis-la en français simplement et propose une explication claire pour aider l'utilisateur."
     try:
         qwen_res = requests.post("http://localhost:11434/api/generate", json={"model": "qwen2.5-coder:7b", "prompt": f"{sys_prompt}\n\nErreur ({provider}): {raw_error}", "stream": False}, timeout=15).json()["response"]
@@ -106,54 +136,57 @@ def handle_api_error(provider, raw_error):
     except:
         return f"⚠️ Erreur brute ({provider}) : {raw_error}"
 
-async def async_ask_llm(provider, actual_model, msg):
-    return await asyncio.to_thread(ask_llm, provider, actual_model, msg)
+async def async_ask_llm(session_id, provider, actual_model, msg):
+    return await asyncio.to_thread(ask_llm, session_id, provider, actual_model, msg)
 
-def ask_llm(provider, actual_model, msg):
+def ask_llm(session_id, provider, actual_model, msg):
+    log_event(session_id, f"Interrogation API distantes : {provider} | Modèle : {actual_model}")
     try:
         if provider == "openrouter":
             res = requests.post("https://openrouter.ai/api/v1/chat/completions", headers={"Authorization": f"Bearer {os.getenv('OPENROUTER_API_KEY')}"}, json={"model": actual_model, "messages": [{"role": "user", "content": msg}]})
-            if res.status_code != 200: return handle_api_error(provider, res.text)
+            if res.status_code != 200: return handle_api_error(session_id, provider, res.text)
             return res.json()["choices"][0]["message"]["content"]
             
         elif provider == "groq":
             res = requests.post("https://api.groq.com/openai/v1/chat/completions", headers={"Authorization": f"Bearer {os.getenv('GROQ_API_KEY')}"}, json={"model": actual_model, "messages": [{"role": "user", "content": msg}]})
-            if res.status_code != 200: return handle_api_error(provider, res.text)
+            if res.status_code != 200: return handle_api_error(session_id, provider, res.text)
             return res.json()["choices"][0]["message"]["content"]
             
         elif provider == "gemini":
             res = requests.post(f"https://generativelanguage.googleapis.com/v1beta/models/{actual_model}:generateContent?key={os.getenv('GEMINI_API_KEY')}", json={"contents": [{"parts": [{"text": msg}]}]})
-            if res.status_code != 200: return handle_api_error(provider, res.text)
+            if res.status_code != 200: return handle_api_error(session_id, provider, res.text)
             return res.json()["candidates"][0]["content"]["parts"][0]["text"]
             
-        return handle_api_error(provider, "Fournisseur non implémenté.")
+        return handle_api_error(session_id, provider, "Fournisseur non implémenté.")
     except Exception as e:
-        return handle_api_error(provider, str(e))
+        return handle_api_error(session_id, provider, str(e))
 
 
 # --- 4. PIPELINES DE TRAITEMENT ---
-def pipeline_minecraft(raw_msg):
+def pipeline_minecraft(session_id, raw_msg):
+    log_event(session_id, "Lancement Pipeline 100% Local Minecraft (Qwen -> Phuzzy -> Qwen)")
     try:
         en_prompt = requests.post("http://localhost:11434/api/generate", json={"model": "qwen2.5-coder:7b", "prompt": f"Translate this Minecraft modding request into technical English. Return ONLY English text.\n\n{raw_msg}", "stream": False}).json()["response"]
         phuzzy_res = requests.post("http://localhost:11434/api/generate", json={"model": "phuzzy:latest", "prompt": en_prompt, "stream": False}).json()["response"]
         final_fr = requests.post("http://localhost:11434/api/generate", json={"model": "qwen2.5-coder:7b", "prompt": f"Traduis les explications techniques en français. NE TRADUIS PAS le code.\n\n{phuzzy_res}", "stream": False}).json()["response"]
         return final_fr, "Qwen ➔ Phuzzy ➔ Qwen"
     except Exception as e:
-        return handle_api_error("Ollama Local", str(e)), "Erreur Locale"
+        return handle_api_error(session_id, "Ollama Local", str(e)), "Erreur Locale"
 
-async def pipeline_consolidated(raw_msg):
+async def pipeline_consolidated(session_id, raw_msg):
+    log_event(session_id, "Lancement Pipeline Consolidation Multi-IA")
     try:
         opt_prompt = requests.post("http://localhost:11434/api/generate", json={"model": "qwen2.5-coder:7b", "prompt": f"Optimise cette demande pour des LLM codeurs. Sois ultra précis.\n\n{raw_msg}", "stream": False}).json()["response"]
         rep_groq, rep_gemini = await asyncio.gather(
-            async_ask_llm("groq", "llama3-8b-8192", opt_prompt),
-            async_ask_llm("gemini", "gemini-1.5-flash", opt_prompt)
+            async_ask_llm(session_id, "groq", "llama3-8b-8192", opt_prompt),
+            async_ask_llm(session_id, "gemini", "gemini-1.5-flash", opt_prompt)
         )
         sys_synth = "Tu es un Architecte Logiciel Senior. Voici la même demande traitée par deux IA différentes. Lis leurs propositions, corrige les erreurs potentielles, garde le meilleur des deux, et génère le code final absolu et parfait en français."
         final_prompt = f"{sys_synth}\n\n--- IA 1 (Groq) ---\n{rep_groq}\n\n--- IA 2 (Gemini) ---\n{rep_gemini}"
         final_res = requests.post("http://localhost:11434/api/generate", json={"model": "qwen2.5-coder:7b", "prompt": final_prompt, "stream": False}).json()["response"]
         return final_res, "Qwen ➔ [Groq + Gemini] ➔ Synthèse Qwen"
     except Exception as e:
-        return handle_api_error("Pipeline Multi-IA", str(e)), "Erreur Consolidation"
+        return handle_api_error(session_id, "Pipeline Multi-IA", str(e)), "Erreur Consolidation"
 
 
 # --- 5. INTERFACE UTILISATEUR & ROUTES ---
@@ -180,7 +213,10 @@ theme_hdrs = [
         .context-radio-group { display: flex; gap: 15px; background: #1e293b; padding: 10px 15px; border-radius: 8px; border: 1px solid #333; margin-bottom: 10px; width: fit-content; }
         .htmx-indicator { display: none; position: absolute; bottom: 120px; left: 50%; transform: translateX(-50%); background: rgba(15, 23, 42, 0.9); border: 1px solid #00e5ff; padding: 15px 30px; border-radius: 50px; color: #00e5ff; font-weight: bold;}
         .htmx-request .htmx-indicator { display: block; }
-        .history-btn { background: #2a2a35; color: #e2e8f0; border: 1px solid #444; padding: 8px; border-radius: 5px; }
+        .history-btn { background: #2a2a35; color: #e2e8f0; border: 1px solid #444; padding: 8px; border-radius: 5px; cursor: pointer; text-align: left; }
+        .history-btn:hover { background: #38bdf8; color: #0f172a; }
+        .export-btn { display: block; background: #0f172a; border: 1px solid #00e5ff; color: #00e5ff; font-weight: bold; padding: 10px; border-radius: 5px; text-align: center; text-decoration: none; transition: 0.2s;}
+        .export-btn:hover { background: #00e5ff; color: #0f172a; }
     """),
     Script("""
         function updateUI() {
@@ -192,8 +228,13 @@ theme_hdrs = [
         
         function updateFileCount(input) {
             let label = document.getElementById('file-label-text');
-            if(input.files.length > 0) { label.innerText = `📎 ${input.files.length} fichier(s)`; label.style.color = '#00e5ff'; }
-            else { label.innerText = '📎 Fichiers'; label.style.color = '#94a3b8'; }
+            if(input && input.files && input.files.length > 0) { 
+                label.innerText = `📎 ${input.files.length} fichier(s)`; 
+                label.style.color = '#00e5ff'; 
+            } else { 
+                label.innerText = '📎 Fichiers'; 
+                label.style.color = '#94a3b8'; 
+            }
         }
 
         htmx.onLoad(function(content) {
@@ -209,70 +250,106 @@ theme_hdrs = [
 
 app, rt = fast_app(hdrs=theme_hdrs)
 
+@rt('/session/new')
+def get():
+    sid = create_new_session()
+    new_hidden = Input(type="hidden", name="session_id", value=sid, id="current-session-id", hx_swap_oob="true")
+    return Div(), new_hidden, render_history_list()
+
+@rt('/session/load/{sid}')
+def get(sid: str):
+    log_event(sid, "Chargement de la session.")
+    messages = load_session_data(sid)
+    bubbles = []
+    for msg in messages:
+        bubbles.append(Div(
+            P("Vous", cls="msg-user"), Div(NotStr(f"{msg['user']}{msg.get('info_fichiers', '')}{msg.get('info_local', '')}"), cls="bubble-user"),
+            P(f"AETHAS38 ({msg.get('nom_affichage', 'IA')})", cls="msg-ia"), Div(msg['ia'], cls="bubble-ia")
+        ))
+    new_hidden = Input(type="hidden", name="session_id", value=sid, id="current-session-id", hx_swap_oob="true")
+    return tuple(bubbles) + (new_hidden,)
+
 @rt('/history/pin/{sid}')
 def post_pin(sid: str):
-    history = load_history()
-    if sid in history:
-        history[sid]['pinned'] = not history[sid].get('pinned', False)
-        save_history(history)
+    idx = load_index()
+    if sid in idx:
+        idx[sid]['pinned'] = not idx[sid].get('pinned', False)
+        save_index(idx)
+        log_event(sid, "Épinglage modifié.")
     return render_history_list()
 
 @rt('/history/rename/{sid}')
 def post_rename(sid: str, title: str):
-    history = load_history()
-    if sid in history:
-        history[sid]['title'] = title
-        save_history(history)
+    idx = load_index()
+    if sid in idx:
+        idx[sid]['title'] = title
+        save_index(idx)
+        log_event(sid, f"Renommé en: {title}")
     return render_history_list()
 
 @rt('/history/delete/{sid}')
 def post_delete(sid: str):
-    """Nouveau : Route pour la suppression d'une conversation."""
-    history = load_history()
-    if sid in history:
-        del history[sid]
-        save_history(history)
+    idx = load_index()
+    if sid in idx:
+        del idx[sid]
+        save_index(idx)
+        if os.path.exists(f"sessions/{sid}.json"): os.remove(f"sessions/{sid}.json")
+        log_event(sid, "Session supprimée définitivement.", "WARNING")
     return render_history_list()
 
 @rt('/export_models')
-def get_export_models():
-    """Nouveau : Route pour extraire tous les modèles, trier par type et télécharger en TXT."""
-    txt_content = "=== CATALOGUE DES MODÈLES (OPENROUTER & LOCAUX) ===\n\n"
+def get():
+    log_event("system", "Demande d'exportation du catalogue de modèles.")
+    now = datetime.now()
+    date_str = now.strftime('%d/%m/%Y à %H:%M:%S')
+    file_date = now.strftime('%Y%m%d_%H%M%S')
     
-    try:
-        res = requests.get("https://openrouter.ai/api/v1/models", timeout=10)
-        if res.status_code == 200:
-            for m in res.json().get("data", []):
-                name = m.get('name', 'Inconnu')
-                m_id = m.get('id', 'Inconnu')
-                desc = str(m.get('description', '')).replace('\n', ' ')
-                desc_courte = desc[:150] + "..." if len(desc) > 150 else desc
+    txt_content = f"=== CATALOGUE COMPLET DES MODÈLES (Généré le {date_str}) ===\n\n"
+    
+    for m in MODELS_DATA:
+        name = m.get('name', 'Inconnu')
+        m_id = m.get('id', 'Inconnu')
+        
+        usage = "Général / Texte"
+        arch_mod = str(m.get('architecture', {}).get('modality', '')).lower()
+        m_id_lower = m_id.lower()
+        
+        if "vision" in arch_mod or "vision" in m_id_lower or "vl" in m_id_lower or "image" in arch_mod:
+            usage = "Vision (Analyse d'images & Multimodal)"
+        elif "coder" in m_id_lower or "code" in m_id_lower or "math" in m_id_lower:
+            usage = "Codage & Mathématiques"
+        elif "video" in m_id_lower:
+            usage = "Vidéo"
+        elif "rp" in m_id_lower or "uncensored" in m_id_lower or "roleplay" in m_id_lower:
+            usage = "Non-censuré / Roleplay"
+            
+        is_free = m.get("is_free")
+        if is_free is None:
+            pricing = m.get('pricing', {})
+            p_prompt = float(pricing.get('prompt', -1))
+            p_comp = float(pricing.get('completion', -1))
+            cout = "Gratuit" if p_prompt == 0.0 and p_comp == 0.0 else f"Payant (In: {p_prompt}, Out: {p_comp})"
+        else:
+            cout = "Gratuit" if is_free else "Payant"
+            
+        desc = str(m.get('description', '')).replace('\n', ' ')
+        desc_courte = desc[:150] + "..." if len(desc) > 150 else desc
+        desc_fr = desc_courte
+        
+        if "openrouter" in m_id and desc_courte:
+            try:
+                # Appel direct Google Translate
+                url = "https://translate.googleapis.com/translate_a/single?client=gtx&sl=en&tl=fr&dt=t&q=" + requests.utils.quote(desc_courte)
+                trans_res = requests.get(url, timeout=2)
+                if trans_res.status_code == 200:
+                    desc_fr = "".join([x[0] for x in trans_res.json()[0]])
+            except: pass
                 
-                # Détection intelligente du domaine de spécialité
-                usage = "Général / Texte"
-                arch_mod = str(m.get('architecture', {}).get('modality', '')).lower()
-                m_id_lower = m_id.lower()
-                
-                if "vision" in arch_mod or "vision" in m_id_lower or "vl" in m_id_lower or "image" in arch_mod:
-                    usage = "Vision (Analyse d'images & Multimodal)"
-                elif "coder" in m_id_lower or "code" in m_id_lower or "math" in m_id_lower:
-                    usage = "Codage & Mathématiques"
-                elif "video" in m_id_lower:
-                    usage = "Vidéo"
-                elif "rp" in m_id_lower or "uncensored" in m_id_lower or "roleplay" in m_id_lower:
-                    usage = "Non-censuré / Roleplay"
-                    
-                # Détection du prix
-                pricing = m.get('pricing', {})
-                p_prompt = float(pricing.get('prompt', -1))
-                p_comp = float(pricing.get('completion', -1))
-                cout = "Gratuit" if p_prompt == 0.0 and p_comp == 0.0 else f"Payant (In: {p_prompt}, Out: {p_comp})"
-                
-                txt_content += f"Nom : {name}\nID : {m_id}\nTarif : {cout}\nSpécialité : {usage}\nDescription : {desc_courte}\n{'-'*60}\n"
-    except Exception as e:
-        txt_content += f"Erreur de récupération : {e}\n"
+        if not desc_fr: desc_fr = "Aucune description fournie."
+        
+        txt_content += f"Nom : {name}\nID : {m_id}\nTarif : {cout}\nSpécialité : {usage}\nDescription : {desc_fr}\n{'-'*60}\n"
 
-    return Response(txt_content, media_type="text/plain", headers={"Content-Disposition": "attachment; filename=modeles_ia.txt"})
+    return Response(txt_content, media_type="text/plain", headers={"Content-Disposition": f"attachment; filename=modeles_ia_{file_date}.txt"})
 
 
 @rt('/')
@@ -290,8 +367,8 @@ def get():
                     style="display:flex; flex-direction:column; flex-grow:1; margin-bottom: 10px; overflow:hidden;"
                 ),
                 
-                # Nouveau bouton d'extraction des modèles
-                A(Button("📥 Exporter Modèles (.txt)", cls="history-btn w-full mt-2 text-center", style="background:#0f172a; border-color:#00e5ff; color:#00e5ff; font-weight:bold;"), href="/export_models", download="modeles.txt"),
+                # hx_disable=True autorise le navigateur à télécharger le fichier généré
+                A("📥 Exporter Modèles (.txt)", cls="export-btn w-full mt-2", href="/export_models", hx_disable=True),
                 
                 Div(
                     P("BUDGET OPENROUTER", style="color:#94a3b8; font-size:10px; font-weight:bold; margin-bottom:5px;"),
@@ -305,7 +382,7 @@ def get():
                 Div("⚙️ Traitement de l'Architecture IA en cours...", id="loading-tracker", cls="htmx-indicator"),
                 
                 Form(
-                    Input(type="hidden", name="session_id", value=session_id),
+                    Input(type="hidden", name="session_id", value=session_id, id="current-session-id"),
                     Div(
                         Label(Input(type="radio", name="context_type", value="generic", checked=True, onchange="updateUI()"), " 💻 Code Générique (Modèle Unique)"),
                         Label(Input(type="radio", name="context_type", value="consolidated", onchange="updateUI()"), " 🧠 Consolidation (Multi-IA ➔ Qwen)"),
@@ -318,14 +395,14 @@ def get():
                         cls="flex gap-2 mb-2 w-full", id="model-selection-area"
                     ),
                     Div(
-                        # Le tag multiple permet nativement de sélectionner 10, 50, ou 100+ fichiers simultanément
                         Input(type="file", name="fichiers", id="file-upload", multiple=True, style="display:none;", onchange="updateFileCount(this)"),
                         Label(Span("📎 Fichiers", id="file-label-text"), _for="file-upload", cls="file-upload-btn"),
                         Input(type="text", name="msg", placeholder="Insérez votre requête...", cls="input-box", required=True),
                         Button("Envoyer", type="submit", cls="send-btn"),
                         cls="input-row"
                     ),
-                    hx_post="/chat", hx_target="#chat-history", hx_swap="beforeend", hx_indicator="#loading-tracker", enctype="multipart/form-data"
+                    hx_post="/chat", hx_target="#chat-history", hx_swap="beforeend", hx_indicator="#loading-tracker", enctype="multipart/form-data",
+                    **{"hx-on:htmx:afterRequest": "if(event.detail.successful) { this.reset(); updateFileCount(document.getElementById('file-upload')); }"}
                 ),
                 cls="main-content"
             ),
@@ -338,6 +415,8 @@ def get(filter_type: str): return get_model_options(filter_type)
 
 @rt('/chat')
 async def post(msg: str, model_id: str, session_id: str, context_type: str = "generic", fichiers: list[UploadFile] = None):
+    log_event(session_id, f"Envoi d'un message (Contexte: {context_type})")
+    
     files_context = ""
     noms_fichiers = []
     if fichiers:
@@ -345,34 +424,43 @@ async def post(msg: str, model_id: str, session_id: str, context_type: str = "ge
             if f.filename:
                 noms_fichiers.append(f.filename)
                 try: files_context += f"\n--- {f.filename} ---\n{(await f.read()).decode('utf-8')}\n"
-                except: files_context += f"\n--- {f.filename} (Binaire ignoré) ---\n"
+                except: 
+                    files_context += f"\n--- {f.filename} (Binaire ignoré) ---\n"
+                    log_event(session_id, f"Fichier non lisible : {f.filename}", "WARNING")
     
     full_req = f"Fichiers fournis:\n{files_context}\nDemande: {msg}" if files_context else msg
 
     if context_type == "minecraft":
-        ia_reponse, pipeline_info = pipeline_minecraft(full_req)
+        ia_reponse, pipeline_info = pipeline_minecraft(session_id, full_req)
         nom_affichage = "Local - Phuzzy/Minecraft"
     
     elif context_type == "consolidated":
-        ia_reponse, pipeline_info = await pipeline_consolidated(full_req)
+        ia_reponse, pipeline_info = await pipeline_consolidated(session_id, full_req)
         nom_affichage = "Qwen Synthèse (via Groq/Gemini)"
         
     else:
         opt_prompt = requests.post("http://localhost:11434/api/generate", json={"model": "qwen2.5-coder:7b", "prompt": f"Optimise cette demande pour un LLM codeur. Retourne UNIQUEMENT le prompt.\n\n{full_req}", "stream": False}).json()["response"]
         parts = model_id.split("|", 1)
         provider, actual_model = parts[0], parts[1] if len(parts) > 1 else model_id
-        ia_reponse = ask_llm(provider, actual_model, opt_prompt)
+        ia_reponse = ask_llm(session_id, provider, actual_model, opt_prompt)
         nom_affichage = f"{provider.capitalize()} - {actual_model.split('/')[-1]}"
         pipeline_info = "Qwen ➔ API Distante Unique"
 
-    history = load_history()
-    if session_id in history:
-        history[session_id]["messages"].append({"user": msg, "ia": ia_reponse})
-        save_history(history)
-
     info_fichiers = f"<br><span style='color:#a855f7; font-size:10px;'>📎 {len(noms_fichiers)} fichier(s) joint(s)</span>" if noms_fichiers else ""
     info_local = f"<br><span style='color:#38bdf8; font-size:10px;'>⚙️ Pipeline : {pipeline_info}</span>"
     
+    # Séparation et sauvegarde des données complètes pour la session
+    session_data = load_session_data(session_id)
+    session_data.append({
+        "user": msg, 
+        "ia": ia_reponse, 
+        "nom_affichage": nom_affichage, 
+        "info_fichiers": info_fichiers, 
+        "info_local": info_local
+    })
+    save_session_data(session_id, session_data)
+    log_event(session_id, "Réponse sauvegardée avec succès.")
+
     chat_bubble = Div(
         P("Vous", cls="msg-user"), Div(NotStr(f"{msg}{info_fichiers}{info_local}"), cls="bubble-user"),
         P(f"AETHAS38 ({nom_affichage})", cls="msg-ia"), Div(ia_reponse, cls="bubble-ia")