app.py

import gradio as gr
from gradio_leaderboard import Leaderboard, ColumnFilter, SelectColumns
import pandas as pd
from apscheduler.schedulers.background import BackgroundScheduler
from src.populate import get_model_info_df, get_merged_df

from src.about import (
    CITATION_BUTTON_LABEL,
    CITATION_BUTTON_TEXT,
    EVALUATION_QUEUE_TEXT,
    INTRODUCTION_TEXT,
    LLM_BENCHMARKS_TEXT,
    TITLE,
)
from src.display.css_html_js import custom_css
from src.display.utils import (
    BENCHMARK_COLS,
    COLS,
    EVAL_COLS,
    EVAL_TYPES,
    AutoEvalColumn,
    ModelType,
    fields,
    WeightType,
    Precision,
)
from src.envs import API, EVAL_REQUESTS_PATH, EVAL_RESULTS_PATH, QUEUE_REPO, REPO_ID, RESULTS_REPO, TOKEN
from src.populate import get_evaluation_queue_df, get_leaderboard_df
from src.submission.submit import add_new_eval


def restart_space():
    API.restart_space(repo_id=REPO_ID, token=TOKEN)


### Space initialisation
# try:
#     print(EVAL_REQUESTS_PATH)
#     snapshot_download(
#         repo_id=QUEUE_REPO,
#         local_dir=EVAL_REQUESTS_PATH,
#         repo_type="dataset",
#         tqdm_class=None,
#         etag_timeout=30,
#         token=TOKEN,
#     )
# except Exception:
#     restart_space()
# try:
#     print(EVAL_RESULTS_PATH)
#     snapshot_download(
#         repo_id=RESULTS_REPO,
#         local_dir=EVAL_RESULTS_PATH,
#         repo_type="dataset",
#         tqdm_class=None,
#         etag_timeout=30,
#         token=TOKEN,
#     )
# except Exception:
#     restart_space()


LEADERBOARD_DF = get_leaderboard_df(
    EVAL_RESULTS_PATH + "/leaderboards/BOOM_leaderboard.csv", EVAL_REQUESTS_PATH, COLS, BENCHMARK_COLS
)
LEADERBOARD_DF_DOMAIN = get_leaderboard_df(
    EVAL_RESULTS_PATH + "/leaderboards/BOOM_domain_leaderboard.csv", EVAL_REQUESTS_PATH, COLS, BENCHMARK_COLS
)
LEADERBOARD_DF_METRIC_TYPE = get_leaderboard_df(
    EVAL_RESULTS_PATH + "/leaderboards/BOOM_metric_type_leaderboard.csv", EVAL_REQUESTS_PATH, COLS, BENCHMARK_COLS
)
LEADERBOARD_DF_TERM = get_leaderboard_df(
    EVAL_RESULTS_PATH + "/leaderboards/BOOM_term_leaderboard.csv", EVAL_REQUESTS_PATH, COLS, BENCHMARK_COLS
)
LEADERBOARD_DF_BOOMLET = get_leaderboard_df(
    EVAL_RESULTS_PATH + "/leaderboards/BOOMLET_leaderboard.csv", EVAL_REQUESTS_PATH, COLS, BENCHMARK_COLS
)
model_info_df = get_model_info_df(EVAL_RESULTS_PATH)

# (
#     finished_eval_queue_df,
#     running_eval_queue_df,
#     pending_eval_queue_df,
# ) = get_evaluation_queue_df(EVAL_REQUESTS_PATH, EVAL_COLS)


def init_leaderboard(dataframe, model_info_df):
    if dataframe is None or dataframe.empty:
        raise ValueError("Leaderboard DataFrame is empty or None.")

    merged_df = get_merged_df(dataframe, model_info_df)

    if "Rank" in merged_df.columns:
        merged_df = merged_df.sort_values(by=["Rank"], ascending=True)
    else:
        # Sort by the first CRPS column if the Rank column is not present
        crps_cols = [col for col in merged_df.columns if "CRPS" in col]
        if crps_cols:
            merged_df = merged_df.sort_values(by=crps_cols[0], ascending=True)

    # Move the model_type_symbol column to the beginning
    cols = [AutoEvalColumn.model_type_symbol.name, AutoEvalColumn.model.name] + sorted(
        [
            col
            for col in merged_df.columns
            if col not in [AutoEvalColumn.model_type_symbol.name, AutoEvalColumn.model.name]
        ]
    )
    merged_df = merged_df[cols]
    col2type_dict = {c.name: c.type for c in fields(AutoEvalColumn)}
    datatype_list = [col2type_dict[col] if col in col2type_dict else "number" for col in merged_df.columns]
    model_info_col_list = [c.name for c in fields(AutoEvalColumn) if c.displayed_by_default]
    default_selection_list = list(dataframe.columns) + model_info_col_list
    return Leaderboard(
        value=merged_df,
        datatype=datatype_list,
        select_columns=SelectColumns(
            default_selection=default_selection_list,
            cant_deselect=[c.name for c in fields(AutoEvalColumn) if c.never_hidden],
            label="Select Columns to Display:",
        ),
        search_columns=[AutoEvalColumn.model.name],
        hide_columns=[c.name for c in fields(AutoEvalColumn) if c.hidden],
        filter_columns=[
            ColumnFilter(AutoEvalColumn.model_type.name, type="checkboxgroup", label="Model types"),
        ],
        bool_checkboxgroup_label="Hide models",
        column_widths=[40, 180] + [160 for _ in range(len(merged_df.columns) - 2)],
        wrap=True,
        interactive=False,
    )


demo = gr.Blocks(css=custom_css)
with demo:
    gr.HTML(TITLE)
    gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")

    with gr.Tabs(elem_classes="tab-buttons") as tabs:
        with gr.TabItem("🏅 Overall", elem_id="boom-benchmark-tab-table", id=0):
            leaderboard = init_leaderboard(LEADERBOARD_DF, model_info_df)

        with gr.TabItem("🏅 By Domain", elem_id="boom-benchmark-tab-table", id=1):
            leaderboard = init_leaderboard(LEADERBOARD_DF_DOMAIN, model_info_df)

        with gr.TabItem("🏅 By Metric Type", elem_id="boom-benchmark-tab-table", id=2):
            leaderboard = init_leaderboard(LEADERBOARD_DF_METRIC_TYPE, model_info_df)

        with gr.TabItem("🏅 By Forecast Horizon", elem_id="boom-benchmark-tab-table", id=3):
            leaderboard = init_leaderboard(LEADERBOARD_DF_TERM, model_info_df)

        with gr.TabItem("🏅 BOOMLET", elem_id="boom-benchmark-tab-table", id=4):
            leaderboard = init_leaderboard(LEADERBOARD_DF_BOOMLET, model_info_df)

        with gr.TabItem("📝 About", elem_id="boom-benchmark-tab-table", id=5):
            gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")

    with gr.Row():
        with gr.Accordion("📙 Citation", open=False):
            citation_button = gr.Textbox(
                value=CITATION_BUTTON_TEXT,
                label=CITATION_BUTTON_LABEL,
                lines=20,
                elem_id="citation-button",
                show_copy_button=True,
            )

scheduler = BackgroundScheduler()
scheduler.add_job(restart_space, "interval", seconds=1800)
scheduler.start()
demo.queue(default_concurrency_limit=40).launch()