app.py

import gradio as gr
import markdown
from markdown.extensions.tables import TableExtension
from markdown.extensions.fenced_code import FencedCodeExtension
from markdown.extensions.toc import TocExtension
from markdown.extensions.attr_list import AttrListExtension
from markdown.extensions.codehilite import CodeHiliteExtension
from transformers import pipeline

# Function to render markdown to HTML with extensions
def render_markdown(md_text):
    print("[DEBUG] render_markdown called with input:", md_text)  # Debug log for input
    # Convert the input markdown text to HTML using various extensions for additional functionality
    rendered_html = markdown.markdown(
        md_text,
        extensions=[
            TableExtension(),  # Enables rendering of tables in markdown
            FencedCodeExtension(),  # Supports fenced code blocks
            TocExtension(baselevel=2),  # Generates a table of contents starting at level 2
            AttrListExtension(),  # Allows adding attributes to markdown elements
            CodeHiliteExtension(linenums=False, css_class="highlight"),  # Syntax highlighting for code blocks
        ],
    )
    print("[DEBUG] Rendered HTML output:", rendered_html)  # Debug log for output
    return rendered_html

# Load the JinaAI ReaderLM-v2 model
model_name = "jinaai/ReaderLM-v2"
print("[DEBUG] Loading model:", model_name)  # Debug log for model loading
html_converter = pipeline("text-generation", model=model_name)  # Initialize the text-generation pipeline with the specified model

# Function to convert HTML to Markdown or JSON
def convert_html(html_input, output_format):
    print("[DEBUG] convert_html called with inputs:", html_input, output_format)  # Debug log for inputs
    # Prepare the prompt for the model, specifying the desired output format (Markdown or JSON)
    prompt = f"Convert the following HTML into {output_format}:\n\n{html_input}"
    print("[DEBUG] Generated prompt:", prompt)  # Debug log for the prompt
    
    # Use the model to generate the conversion output
    response = html_converter(prompt, max_length=99999, num_return_sequences=1)
    print("[DEBUG] Model response:", response)  # Debug log for model response
    converted_output = response[0]['generated_text']  # Extract the generated text from the model response
    
    # Remove the prompt text from the generated output and clean up the result
    converted_output = converted_output.replace(prompt, "").strip()
    print("[DEBUG] Converted output:", converted_output)  # Debug log for the final output
    return converted_output

# Creating the Gradio Interface
with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:

    # Tab for the Markdown live preview feature
    with gr.Tab("Live Preview"):
        gr.Markdown("# Markdown Suite")  # Header for the tab

        with gr.Row():
            with gr.Column():
                # Input textbox for entering Markdown text
                md_input = gr.Textbox(
                    lines=20, 
                    placeholder="Write your markdown here...", 
                    label="Markdown Input",
                    elem_classes=["gr-textbox"]
                )
            with gr.Column():
                # Output area to display the rendered HTML from the Markdown input
                md_output = gr.HTML(label="Rendered Output", elem_classes=["gr-html"])

        # Define the interaction: Update the HTML preview whenever the Markdown input changes
        md_input.change(render_markdown, inputs=md_input, outputs=md_output)

    # Tab for HTML to Markdown/JSON conversion feature
    with gr.Tab("HTML to Markdown/JSON"):
        gr.Markdown("# HTML to Markdown/JSON Converter")  # Header for the tab

        with gr.Row():
            # Input textbox for raw HTML input
            html_input = gr.Textbox(
                lines=10, 
                placeholder="Paste your raw HTML here...", 
                label="Raw HTML Input"
            )

            # Radio buttons to select the output format (Markdown or JSON)
            output_format = gr.Radio([
                "Markdown", 
                "JSON"
            ], label="Output Format", value="Markdown")

        # Output textbox to display the converted Markdown or JSON
        converted_output = gr.Textbox(
            lines=10, 
            label="Converted Output"
        )

        # Define the interaction: Convert HTML when the "Convert" button is clicked
        convert_button = gr.Button("Convert")
        convert_button.click(
            convert_html,  # Function to handle conversion
            inputs=[html_input, output_format],  # Inputs: Raw HTML and desired output format
            outputs=converted_output  # Output: Converted text
        )

# Launch the app
print("[DEBUG] Launching the app")  # Debug log for app launch
demo.launch()