app.py

import gradio as gr
from transformers import pipeline

# Load the JinaAI ReaderLM-v2 model
model_name = "jinaai/ReaderLM-v2"
html_converter = pipeline("text-generation", model=model_name)

# Function to convert HTML to Markdown or JSON
def convert_html(html_input, output_format):
    # Prepare the prompt for the model
    prompt = f"Convert the following HTML into {output_format}:\n\n{html_input}"
    
    # Generate the output using the model
    response = html_converter(prompt, max_length=500, num_return_sequences=1)
    converted_output = response[0]['generated_text']
    
    # Extract the relevant part of the output (remove the prompt)
    converted_output = converted_output.replace(prompt, "").strip()
    return converted_output

# Gradio Interface with NoCrypt/miku theme
interface = gr.Interface(
    fn=convert_html,
    inputs=[
        gr.Textbox(lines=10, placeholder="Paste your raw HTML here...", label="Raw HTML Input"),
        gr.Radio(["Markdown", "JSON"], label="Output Format", value="Markdown")
    ],
    outputs=gr.Textbox(lines=10, label="Converted Output"),
    title="HTML to Markdown/JSON Converter",
    description="Convert raw HTML into beautifully formatted Markdown or JSON using JinaAI ReaderLM-v2.",
    theme="NoCrypt/miku",  # Apply the NoCrypt/miku theme
    examples=[
        ["<h1>Hello World</h1><p>This is a <strong>test</strong>.</p>", "Markdown"],
        ["<ul><li>Item 1</li><li>Item 2</li></ul>", "JSON"]
    ]
)

# Launch the interface
interface.launch()