Update app.py

app.py

@@ -1,6 +1,3 @@
-TITLE = """<h1 align="center">Gemini Playground ✨</h1>"""
-SUBTITLE = """<h2 align="center">Play with Gemini Pro and Gemini Pro Vision</h2>"""
-
 import os
 import time
 import uuid
@@ -11,22 +8,21 @@ import gradio as gr
 from PIL import Image
 from dotenv import load_dotenv
 
-# Cargar las variables de entorno desde el archivo .env
+# Load environment variables
 load_dotenv()
 
-print("google-generativeai:", genai.__version__)
-
-# Obtener la clave de la API de las variables de entorno
+# Constants
+TITLE = """<h1 align="center">Gemini Playground ✨</h1>"""
+SUBTITLE = """<h2 align="center">Play with Gemini Pro and Gemini Pro Vision</h2>"""
 GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY")
-
-# Verificar que la clave de la API esté configurada
-if not GOOGLE_API_KEY:
-    raise ValueError("GOOGLE_API_KEY is not set in environment variables.")
-
 IMAGE_CACHE_DIRECTORY = "/tmp"
 IMAGE_WIDTH = 512
 CHAT_HISTORY = List[Tuple[Optional[Union[Tuple[str], str]], Optional[str]]]
 
+# Check API key
+if not GOOGLE_API_KEY:
+    raise ValueError("GOOGLE_API_KEY is not set in environment variables.")
+
 def preprocess_image(image: Image.Image) -> Optional[Image.Image]:
     if image:
         image_height = int(image.height * IMAGE_WIDTH / image.width)
@@ -39,6 +35,16 @@ def cache_pil_image(image: Image.Image) -> str:
     image.save(image_path, "JPEG")
     return image_path
 
+def transform_history(history: CHAT_HISTORY):
+    """Transform Gradio history into Gemini-compatible format."""
+    new_history = []
+    for chat in history:
+        if chat[0]:
+            new_history.append({"parts": [{"text": chat[0]}], "role": "user"})
+        if chat[1]:
+            new_history.append({"parts": [{"text": chat[1]}], "role": "model"})
+    return new_history
+
 def upload(files: Optional[List[str]], chatbot: CHAT_HISTORY) -> CHAT_HISTORY:
     for file in files:
         image = Image.open(file).convert('RGB')
@@ -57,7 +63,7 @@ def user(text_prompt: str, chatbot: CHAT_HISTORY):
 def bot(
     files: Optional[List[str]],
     model_choice: str,
-    system_instruction: Optional[str],  # Sistema de instrucciones opcional
+    system_instruction: Optional[str],
     chatbot: CHAT_HISTORY
 ):
     if not GOOGLE_API_KEY:
@@ -71,37 +77,31 @@ def bot(
         top_p=0.9
     )
 
-    # Usar el valor por defecto para system_instruction si está vacío
     if not system_instruction:
-        system_instruction = "1"  # O puedes poner un valor predeterminado como "No system instruction provided."
-
-    text_prompt = [chatbot[-1][0]] if chatbot and chatbot[-1][0] and isinstance(chatbot[-1][0], str) else []
-    image_prompt = [preprocess_image(Image.open(file).convert('RGB')) for file in files] if files else []
-    
-    model = genai.GenerativeModel(
-        model_name=model_choice,
-        generation_config=generation_config,
-        system_instruction=system_instruction  # Usar el valor por defecto si está vacío
-    )
+        system_instruction = "Default system instruction."
+
+    text_prompt = [chat[0] for chat in chatbot if isinstance(chat[0], str)]
+    chat = genai.Chat(model=model_choice, messages=transform_history(chatbot))
 
-    response = model.generate_content(text_prompt + image_prompt, stream=True, generation_config=generation_config)
+    response = chat.send_message(
+        text_prompt[-1] if text_prompt else "",  # Use the last user message
+        system_instruction=system_instruction
+    )
 
-    chatbot[-1][1] = ""
+    chatbot[-1] = (chatbot[-1][0], "")
     for chunk in response:
         for i in range(0, len(chunk.text), 10):
             section = chunk.text[i:i + 10]
-            chatbot[-1][1] += section
+            chatbot[-1] = (chatbot[-1][0], chatbot[-1][1] + section)
             time.sleep(0.01)
             yield chatbot
 
-# Componente para el acordeón que contiene el cuadro de texto para la instrucción del sistema
 system_instruction_component = gr.Textbox(
     placeholder="Enter system instruction...",
     show_label=True,
     scale=8
 )
 
-# Definir los componentes de entrada y salida
 chatbot_component = gr.Chatbot(label='Gemini', bubble_full_width=False, scale=2, height=300)
 text_prompt_component = gr.Textbox(placeholder="Message...", show_label=False, autofocus=True, scale=8)
 upload_button_component = gr.UploadButton(label="Upload Images", file_count="multiple", file_types=["image"], scale=1)
@@ -116,21 +116,17 @@ model_choice_component = gr.Dropdown(
 user_inputs = [text_prompt_component, chatbot_component]
 bot_inputs = [upload_button_component, model_choice_component, system_instruction_component, chatbot_component]
 
-# Definir la interfaz de usuario
 with gr.Blocks() as demo:
     gr.HTML(TITLE)
     gr.HTML(SUBTITLE)
     with gr.Column():
-        # Campo de selección de modelo arriba
         model_choice_component.render()
         chatbot_component.render()
         with gr.Row():
             text_prompt_component.render()
             upload_button_component.render()
             run_button_component.render()
-
-        # Crear el acordeón para la instrucción del sistema al final
-        with gr.Accordion("System Instruction", open=False):  # Acordeón cerrado por defecto
+        with gr.Accordion("System Instruction", open=False):
             system_instruction_component.render()
 
     run_button_component.click(
@@ -158,5 +154,4 @@ with gr.Blocks() as demo:
         queue=False
     )
 
-# Lanzar la aplicación
 demo.queue(max_size=99).launch(debug=False, show_error=True)