fix(pos): retry con backoff para rate limit 429 de OpenRouter

3 reintentos con espera 5s/10s/15s. Si agota reintentos,
muestra mensaje amigable en vez de error.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

pos/services/ai_chat.py

@@ -60,42 +60,53 @@ def chat(user_message, conversation_history=None):
         messages.extend(conversation_history)
     messages.append({"role": "user", "content": user_message})
 
-    try:
+    import time
-        resp = requests.post(
+    max_retries = 3
-            OPENROUTER_URL,
-            headers={
-                "Authorization": f"Bearer {OPENROUTER_API_KEY}",
-                "Content-Type": "application/json",
-            },
-            json={
-                "model": MODEL,
-                "messages": messages,
-                "max_tokens": 500,
-                "temperature": 0.3,
-            },
-            timeout=15,
-        )
-        resp.raise_for_status()
-        data = resp.json()
-        content = data["choices"][0]["message"]["content"]
 
-        # Try to parse JSON response
+    for attempt in range(max_retries):
         try:
-            # Handle markdown-wrapped JSON (```json ... ```)
+            resp = requests.post(
-            stripped = content.strip()
+                OPENROUTER_URL,
-            if stripped.startswith("```"):
+                headers={
-                lines = stripped.split("\n")
+                    "Authorization": f"Bearer {OPENROUTER_API_KEY}",
-                # Remove first and last lines (``` markers)
+                    "Content-Type": "application/json",
-                json_str = "\n".join(lines[1:-1])
+                },
-                parsed = json.loads(json_str)
+                json={
-            else:
+                    "model": MODEL,
-                parsed = json.loads(stripped)
+                    "messages": messages,
-            return parsed
+                    "max_tokens": 500,
-        except (json.JSONDecodeError, IndexError):
+                    "temperature": 0.3,
-            return {"message": content, "search_query": None, "vehicle": None}
+                },
-    except Exception as e:
+                timeout=20,
-        return {
+            )
-            "message": f"Error de conexion: {str(e)}",
+            if resp.status_code == 429:
-            "search_query": None,
+                # Rate limited — wait and retry
-            "vehicle": None,
+                wait = (attempt + 1) * 5  # 5s, 10s, 15s
-        }
+                if attempt < max_retries - 1:
+                    time.sleep(wait)
+                    continue
+                return {"message": "El asistente está ocupado. Intenta de nuevo en unos segundos.", "search_query": None, "vehicle": None}
+            resp.raise_for_status()
+            data = resp.json()
+            content = data["choices"][0]["message"]["content"]
+
+            # Try to parse JSON response
+            try:
+                stripped = content.strip()
+                if stripped.startswith("```"):
+                    lines = stripped.split("\n")
+                    json_str = "\n".join(lines[1:-1])
+                    parsed = json.loads(json_str)
+                else:
+                    parsed = json.loads(stripped)
+                return parsed
+            except (json.JSONDecodeError, IndexError):
+                return {"message": content, "search_query": None, "vehicle": None}
+        except Exception as e:
+            if attempt < max_retries - 1:
+                continue
+            return {
+                "message": f"Error de conexion: {str(e)}",
+                "search_query": None,
+                "vehicle": None,
+            }