Update app.py

app.py

@@ -1,73 +1,162 @@
-import gradio as gr
-import google.generativeai as genai
+TITLE = """<h1 align="center">Gemini Playground ✨</h1>"""
+SUBTITLE = """<h2 align="center">Play with Gemini Pro and Gemini Pro Vision</h2>"""
+
 import os
-from pypdf import PdfReader
+import time
+import uuid
+from typing import List, Tuple, Optional, Union
+
+import google.generativeai as genai
+import gradio as gr
+from PIL import Image
+from dotenv import load_dotenv
+
+# Cargar las variables de entorno desde el archivo .env
+load_dotenv()
+
+print("google-generativeai:", genai.__version__)
 
-# Cargar la clave API desde el entorno
-GOOGLE_API_KEY = os.environ.get("GOOGLE_API_KEY")
+# Obtener la clave de la API de las variables de entorno
+GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY")
+
+# Verificar que la clave de la API esté configurada
 if not GOOGLE_API_KEY:
-    raise ValueError("La variable de entorno GOOGLE_API_KEY no está definida.")
-
-genai.configure(api_key=GOOGLE_API_KEY)
-
-def read_text_file(filepath):
-    with open(filepath, 'r', encoding='utf-8') as f:
-        return f.read()
-
-def read_pdf_file(filepath):
-    text = ""
-    with open(filepath, 'rb') as f:
-        reader = PdfReader(f)
-        for page in reader.pages:
-            text += page.extract_text() + "\n"
-    return text
-
-def process_input(instruction, text_file, pdf_file, image_file):
-    prompt_parts = []
-
-    if instruction:
-        prompt_parts.append(instruction)
-
-    if text_file:
-        file_content = read_text_file(text_file.name)
-        prompt_parts.append(f"Contenido del archivo de texto:\n{file_content}")
-
-    if pdf_file:
-        file_content = read_pdf_file(pdf_file.name)
-        prompt_parts.append(f"Contenido del archivo PDF:\n{file_content}")
-
-    if image_file:
-        # Para el modelo vision, necesitas pasar la imagen como datos
-        prompt_parts.append(f"Imagen: {image_file.name}")
-        response = genai.ChatCompletion.create(
-            model="gemini-1.5-flash",
-            messages=[{"role": "user", "content": "Contenido de la imagen: " + image_file.name}],
-            max_tokens=8192,
-        )
-        return response['choices'][0]['message']['content']
-
-    if prompt_parts:
-        prompt = "\n\n".join(prompt_parts)
-        response = genai.ChatCompletion.create(
-            model="gemini-1.5-flash",
-            messages=[{"role": "user", "content": prompt}],
-            max_tokens=8192,
-        )
-        return response['choices'][0]['message']['content']
-    else:
-        return "Por favor, proporciona alguna instrucción o archivo."
-
-iface = gr.Interface(
-    fn=process_input,
-    inputs=[
-        gr.Textbox(label="Instrucciones"),
-        gr.File(label="Archivo de Texto (.txt)"),
-        gr.File(label="Archivo PDF (.pdf)"),
-        gr.Image(label="Imagen", type="filepath"),  # Cambié el tipo a 'filepath'
-    ],
-    outputs=gr.Textbox(label="Respuesta de Gemini"),
-    title="Interactúa con Gemini",
-    description="Sube archivos de texto, PDF o imágenes y proporciona instrucciones para que Gemini los procese.",
+    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]]]
+
+def preprocess_image(image: Image.Image) -> Optional[Image.Image]:
+    if image:
+        image_height = int(image.height * IMAGE_WIDTH / image.width)
+        return image.resize((IMAGE_WIDTH, image_height))
+
+def cache_pil_image(image: Image.Image) -> str:
+    image_filename = f"{uuid.uuid4()}.jpeg"
+    os.makedirs(IMAGE_CACHE_DIRECTORY, exist_ok=True)
+    image_path = os.path.join(IMAGE_CACHE_DIRECTORY, image_filename)
+    image.save(image_path, "JPEG")
+    return image_path
+
+def upload(files: Optional[List[str]], chatbot: CHAT_HISTORY) -> CHAT_HISTORY:
+    for file in files:
+        image = Image.open(file).convert('RGB')
+        image_preview = preprocess_image(image)
+        if image_preview:
+            gr.Image(image_preview).render()
+        image_path = cache_pil_image(image)
+        chatbot.append(((image_path,), None))
+    return chatbot
+
+def user(text_prompt: str, chatbot: CHAT_HISTORY):
+    if text_prompt:
+        chatbot.append((text_prompt, None))
+    return "", chatbot
+
+def bot(
+    files: Optional[List[str]],
+    model_choice: str,
+    system_instruction: Optional[str],  # Sistema de instrucciones opcional
+    chatbot: CHAT_HISTORY
+):
+    if not GOOGLE_API_KEY:
+        raise ValueError("GOOGLE_API_KEY is not set.")
+
+    genai.configure(api_key=GOOGLE_API_KEY)
+    generation_config = genai.types.GenerationConfig(
+        temperature=0.7,
+        max_output_tokens=8192,
+        top_k=10,
+        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
+    )
+
+    response = model.generate_content(text_prompt + image_prompt, stream=True, generation_config=generation_config)
+
+    chatbot[-1][1] = ""
+    for chunk in response:
+        for i in range(0, len(chunk.text), 10):
+            section = chunk.text[i:i + 10]
+            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)
+run_button_component = gr.Button(value="Run", variant="primary", scale=1)
+model_choice_component = gr.Dropdown(
+    choices=["gemini-1.5-flash", "gemini-2.0-flash-exp", "gemini-1.5-pro"],
+    value="gemini-1.5-flash",
+    label="Select Model",
+    scale=2
 )
 
-iface.launch()
+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
+            system_instruction_component.render()
+
+    run_button_component.click(
+        fn=user,
+        inputs=user_inputs,
+        outputs=[text_prompt_component, chatbot_component],
+        queue=False
+    ).then(
+        fn=bot, inputs=bot_inputs, outputs=[chatbot_component],
+    )
+
+    text_prompt_component.submit(
+        fn=user,
+        inputs=user_inputs,
+        outputs=[text_prompt_component, chatbot_component],
+        queue=False
+    ).then(
+        fn=bot, inputs=bot_inputs, outputs=[chatbot_component],
+    )
+
+    upload_button_component.upload(
+        fn=upload,
+        inputs=[upload_button_component, chatbot_component],
+        outputs=[chatbot_component],
+        queue=False
+    )
+
+# Lanzar la aplicación
+demo.queue(max_size=99).launch(debug=False, show_error=True)