TITLE = """
Gemini Playground ✨
"""
SUBTITLE = """Play with Gemini Pro and Gemini Pro Vision
"""
import os
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__)
# 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("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:
# Display a preview of the uploaded image
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,
chatbot: CHAT_HISTORY
):
if not GOOGLE_API_KEY:
raise ValueError("GOOGLE_API_KEY is not set.")
# Configurar la API con la clave
genai.configure(api_key=GOOGLE_API_KEY)
generation_config = genai.types.GenerationConfig(
temperature=0.7, # Valor predeterminado
max_output_tokens=8192, # Fijar el límite de tokens a 8,192
top_k=10, # Valor predeterminado
top_p=0.9 # Valor predeterminado
)
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_choice)
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
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
)
user_inputs = [
text_prompt_component,
chatbot_component
]
bot_inputs = [
upload_button_component,
model_choice_component,
chatbot_component
]
with gr.Blocks() as demo:
gr.HTML(TITLE)
gr.HTML(SUBTITLE)
with gr.Column():
chatbot_component.render()
with gr.Row():
text_prompt_component.render()
upload_button_component.render()
run_button_component.render()
model_choice_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
)
demo.queue(max_size=99).launch(debug=False, show_error=True)