From 56decba94535835897ea8b64309e4b1b36fee84a Mon Sep 17 00:00:00 2001
From: ronalds <ronalds@rshtech.com.py>
Date: Thu, 4 Dec 2025 14:25:16 -0300
Subject: [PATCH] =?UTF-8?q?=E2=9C=85=20Backend=20v1.2.5?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Cambios implementados:

Streaming habilitado para OpenAI: Ahora usa stream=True en las llamadas al chat
Procesamiento en tiempo real: El servidor recibe chunks de la respuesta y los concatena
Mejor experiencia: Las respuestas largas se generan más rápido (el servidor empieza a recibir antes)
Cómo funciona:
---
 backend/app/main.py | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/backend/app/main.py b/backend/app/main.py
index 369a8d5..7c88796 100644
--- a/backend/app/main.py
+++ b/backend/app/main.py
@@ -4,6 +4,7 @@
 from fastapi import FastAPI, File, UploadFile, Form, Depends, HTTPException, status
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials
+from fastapi.responses import StreamingResponse
 from sqlalchemy.orm import Session, joinedload
 from sqlalchemy import func, case, or_
 from typing import List, Optional
@@ -20,6 +21,7 @@ import shutil
 from datetime import datetime, timedelta
 import sys
 import requests
+import json
 
 # Función para enviar notificaciones al webhook
 def send_answer_notification(answer, question, mechanic, db):
@@ -276,7 +278,7 @@ def extract_pdf_text_smart(pdf_content: bytes, max_chars: int = None) -> dict:
         }
 
 
-BACKEND_VERSION = "1.2.4"
+BACKEND_VERSION = "1.2.5"
 app = FastAPI(title="Checklist Inteligente API", version=BACKEND_VERSION)
 
 # S3/MinIO configuration
@@ -3621,14 +3623,21 @@ Longitud de respuesta: {response_length}
                 api_key=ai_config.api_key
             )
             
-            response = client.chat.completions.create(
+            # Usar streaming para respuestas más fluidas
+            stream = client.chat.completions.create(
                 model=ai_config.model_name or "gpt-4",
                 messages=messages,
                 max_tokens=max_tokens,
-                temperature=0.7
+                temperature=0.7,
+                stream=True
             )
             
-            ai_response = response.choices[0].message.content
+            # Recolectar respuesta completa del stream
+            ai_response = ""
+            for chunk in stream:
+                if chunk.choices[0].delta.content is not None:
+                    ai_response += chunk.choices[0].delta.content
+            
             confidence = 0.85  # OpenAI no devuelve confidence directo
             
         elif ai_config.provider == 'anthropic':