app.py

import subprocess

import gradio as gr


def run_script():
    try:
        result = subprocess.run(
            ["python", "hub_datasets_by_language.py"],
            capture_output=True,
            text=True,
            check=True,
        )
        return "Script executed successfully! Plots have been updated."
    except subprocess.CalledProcessError as e:
        return f"Failed to execute script: {str(e.stderr)}"


def create_app():
    with gr.Blocks() as app:
        gr.Markdown(
            """
            # Visualizing The Language Gap In The Hugging Face Hub
                    
            The open-source community is creating more and more resources in languages other than English but there is still a huge gap. This Space showcases plots that can help visualize this gap in the case of Spanish and can easily be adapted to other languages.
        """
        )

        gr.Markdown(
            """
            ## English vs Spanish Monolingual Datasets
                    
            Note: We consider only **monolingual** datasets in these plots, i.e. datasets that only contain data in one language. This is because *most* of the multilingual datasets are usually machine-translated and we want to focus on original data.        
        """
        )

        with gr.Row():
            with gr.Column():
                gr.Image(
                    value="plots/bar_plot_horizontal.png",
                    label="Distribution by Year (Horizontal)",
                    show_label=True,
                    show_download_button=True,
                    show_share_button=True,
                )
                gr.Image(
                    value="plots/stack_area_en_es.png",
                    label="Stacked Area Plot",
                    show_label=True,
                    show_download_button=True,
                    show_share_button=True,
                )
            with gr.Column():
                gr.Image(
                    value="plots/bar_plot_vertical.png",
                    label="Distribution by Year (Vertical)",
                    show_label=True,
                    show_download_button=True,
                    show_share_button=True,
                )
                gr.Image(
                    value="plots/time_series.png",
                    label="Cumulative Growth Over Time",
                    show_label=True,
                    show_download_button=True,
                    show_share_button=True,
                )

        with gr.Row():
            update_button = gr.Button("Update Plots with Latest Data")
            output_label = gr.Label()

        gr.Markdown(
            """
            ## Adapt to other languages

            This Space is WIP and more languages and visuals will be included shortly. Meanwhile, you can clone the Space, adapt the code in the script and run it to generate plots for other languages.
            """
        )

        gr.Markdown("## Citation")
        with gr.Accordion("Citation information", open=False):
            gr.Markdown(
                r"""
                If you use these plots or the code please cite:

                ```
                @misc{grandury2024gaphf,
                    author = {María Grandury},
                    title = {Visualizing The Language Gap In The Hugging Face Hub},
                    year = {2024},
                    publisher = {Hugging Face},
                    howpublished = {\url{https://huggingface.co/spaces/mariagrandury/language-gap-in-hf-hub}},
                }
                ```
                """
            )

        update_button.click(
            fn=run_script,
            outputs=output_label,
        )

    return app


app = create_app()
app.launch()