src/view/metric_view_tab.py

from datetime import datetime
import tempfile
from typing import Callable
import gradio as gr
from functools import partial
import re
import json

from src.logic.data_fetching import fetch_datasets, fetch_graph_data, fetch_groups, fetch_metrics, update_datasets_with_regex, update_datasets_with_regex
from src.logic.data_processing import export_data
from src.logic.graph_settings import update_graph_options
from src.logic.plotting import plot_data

def create_metric_view_tab(METRICS_LOCATION_DEFAULT: str, available_datasets: gr.State):
    metric_data = gr.State([])

    with gr.Row():
        with gr.Column(scale=2):
            with gr.Row():
                with gr.Column(scale=1):
                    base_folder = gr.Textbox(
                        label="Metrics Location",
                        value=METRICS_LOCATION_DEFAULT,
                    )
                    datasets_fetch = gr.Button("Fetch Datasets")

                with gr.Column(scale=1):
                    regex_select = gr.Text(label="Regex filter", value=".*")
                    regex_button = gr.Button("Search")
            with gr.Row():
                selected_datasets_dropdown = gr.Dropdown(
                    choices=[],
                    label="Datasets",
                    multiselect=True,
                    interactive=True,
                )

        with gr.Column(scale=1):
            grouping_dropdown = gr.Dropdown(
                choices=[],
                label="Grouping",
                multiselect=False,
            )
            metric_name_dropdown = gr.Dropdown(
                choices=[],
                label="Metric name",
                multiselect=False,
            )

            render_button = gr.Button("Render Metric", variant="primary")

    with gr.Tabs():
        with gr.TabItem("Graph Settings"):
            log_scale_x_checkbox = gr.Checkbox(
                label="Log scale x",
                value=False,
            )
            log_scale_y_checkbox = gr.Checkbox(
                label="Log scale y",
                value=False,
            )
            rounding = gr.Number(
                label="Rounding",
                value=2,
            )

        with gr.TabItem("Grouping Settings") as group_settings:
            with gr.Row() as group_choices:
                with gr.Column(scale=2):
                    group_regex = gr.Text(
                        label="Group Regex",
                        value=None,
                    )
                    with gr.Row():
                        top_select = gr.Number(
                            label="N Groups",
                            value=100,
                            interactive=True,
                        )

                        direction_checkbox = gr.Radio(
                            label="Partition",
                            choices=[
                                "Top",
                                "Bottom",
                                "Most frequent (n_docs)",
                            ],
                            value="Most frequent (n_docs)",
                        )

        with gr.TabItem("Histogram Settings") as histogram_settings:
            normalization_checkbox = gr.Checkbox(
                label="Normalize",
                value=True,
                visible=False
            )
            cdf_checkbox = gr.Checkbox(
                label="CDF",
                value=False,
            )
            perc_checkbox = gr.Checkbox(
                label="%",
                value=False,
            )

        with gr.TabItem("Summary Settings") as summary_settings:
            show_stds_checkbox = gr.Checkbox(
                label="Show standard deviations",
                value=False,
            )

    with gr.Row():
        graph_output = gr.Plot(label="Graph")
    with gr.Row(visible=False) as min_max_hist:
        with gr.Column(scale=3):
            min_max_hist_data = gr.Markdown()
        with gr.Column(scale=1):
            export_data_button = gr.Button("Export Data")
            export_data_json = gr.File(visible=False)
            


    def update_selected_datasets_dropdown(available_datasets, selected_datasets_dropdown):
        selected_datasets = selected_datasets_dropdown or []
        selected_datasets = set(selected_datasets) & set(available_datasets)
        return gr.Dropdown(choices=available_datasets, value=sorted(list(selected_datasets)))


    datasets_fetch.click(
        fn=fetch_datasets,
        inputs=[base_folder],
        outputs=[available_datasets, selected_datasets_dropdown],
    )

    available_datasets.change(
        fn=update_selected_datasets_dropdown,
        inputs=[available_datasets, selected_datasets_dropdown],
        outputs=selected_datasets_dropdown,
    )

    regex_button.click(
        fn=update_datasets_with_regex,
        inputs=[regex_select, selected_datasets_dropdown, available_datasets],
        outputs=selected_datasets_dropdown,
    )
    
    
    selected_datasets_dropdown.change(
        fn=fetch_groups,
        inputs=[base_folder, selected_datasets_dropdown, grouping_dropdown],
        outputs=grouping_dropdown,
    )

    grouping_dropdown.change(
        fn=fetch_metrics,
        inputs=[base_folder, selected_datasets_dropdown, grouping_dropdown, metric_name_dropdown],
        outputs=metric_name_dropdown,
    )

    render_button.click(
        fn=fetch_graph_data,
        inputs=[
            base_folder,
            selected_datasets_dropdown,
            metric_name_dropdown,
            grouping_dropdown,
        ],
        # We also output the graph_output = None to show the progress
        outputs=[metric_data, graph_output],
    )


    grouping_dropdown.change(
        fn=update_graph_options,
        inputs=[grouping_dropdown],
        outputs=[group_settings, histogram_settings, summary_settings],
    )


    gr.on(
        triggers=[normalization_checkbox.input, rounding.input, group_regex.input, direction_checkbox.input,
                  top_select.input, log_scale_x_checkbox.input,
                  log_scale_y_checkbox.input, cdf_checkbox.input, perc_checkbox.input, show_stds_checkbox.input, metric_data.change],
        fn=plot_data,
        inputs=[
            metric_data,
            metric_name_dropdown,
            normalization_checkbox,
            rounding,
            grouping_dropdown,
            top_select,
            direction_checkbox,
            group_regex,
            log_scale_x_checkbox,
            log_scale_y_checkbox,
            cdf_checkbox,
            perc_checkbox,
            show_stds_checkbox
        ],
        outputs=[graph_output, min_max_hist, min_max_hist_data],
    )

    export_data_button.click(
        fn=export_data,
        inputs=[metric_data, metric_name_dropdown, grouping_dropdown],
        outputs=[export_data_json],
    )
    
    return base_folder, selected_datasets_dropdown