app.py

import gradio as gr
import os
import google.generativeai as genai
import logging
import time
import backoff

# Configure Logging
logging.basicConfig(level=logging.ERROR, format='%(asctime)s - %(levelname)s - %(message)s')

# Load environment variables
try:
    genai.configure(api_key=os.environ["geminiapikey"])
except KeyError:
    logging.error("Error: 'geminiapikey' environment variable not found.")
    exit(1)

read_key = os.environ.get('HF_TOKEN', None)

custom_css = """
#md {
    height: 400px;  
    font-size: 30px;
    background: #202020;
    padding: 20px;
    color: white;
    border: 1px solid white;
}
"""

@backoff.on_exception(backoff.expo,
                      (genai.APIError),
                      max_tries=3) # retry up to 3 times
def predict(prompt):
    # Create the model
    generation_config = {
        "temperature": 0.7,
        "top_p": 0.95,
        "top_k": 40,
        "max_output_tokens": 2048,
        "response_mime_type": "text/plain",
    }

    model = genai.GenerativeModel(
        model_name="gemini-1.5-pro",
        generation_config=generation_config,
    )

    try:
        contents_to_send = [genai.Content(parts=[prompt])]

        response = model.generate_content(contents=contents_to_send, tools='google_search_retrieval')

        if response and response.text:
          return response.text
        else:
          logging.error(f"Unexpected response: {response}")
          return "Error: Could not extract text from the response."
    except genai.APIError as e:
       logging.error(f"API error occurred: {e}")
       raise
    except Exception as e:
        logging.error(f"An error occurred: {e}")
        return f"An error occurred: {e}"


# Create the Gradio interface
with gr.Blocks(css=custom_css) as demo:
    with gr.Row():
        details_output = gr.Markdown(label="answer", elem_id="md")
    with gr.Row():
        ort_input = gr.Textbox(label="prompt", placeholder="ask anything...")
    with gr.Row():
        button = gr.Button("Senden")

    # Connect the button to the function
    button.click(fn=predict, inputs=ort_input, outputs=details_output)

# Launch the Gradio application
demo.launch()