import csv
from collections import defaultdict

import gradio as gr
import pandas as pd


def strip_colname(x):
    if x.startswith("score_"):
        return x[6:]
    return x


INTRO = """The current leaderboard displays performance across all filtered directions based on the dev subset of BOUQuET.

A smarter leaderboard and the code for reproducing the evaluations will be published soon!
"""

LANGS_EXPLANATION = """## Languages
Below, we give a brief description of each language variety participating in the leaderboard.
Each language variety is identified by
an [ISO 639-3 code](https://en.wikipedia.org/wiki/List_of_ISO_639-3_codes) (the first 3 letters) for the language,
an [ISO 15924](https://en.wikipedia.org/wiki/ISO_15924) code (the next 4 letters) for the writing system,
and optionally, a [Glottolog code](https://glottolog.org/) for the dialect.

The varieties with a secondary language code (Egyptian Arabic, Colloquial Malay) use code-switching,
i.e. the speakers switch between the two languages (a colloquial and a standardized variety)
depending on the context (e.g. the formality level).

For a fuller description of the languages and the codes used to represent them, please refer
to https://huggingface.co/datasets/facebook/bouquet#languages and the [BOUQuET paper](https://arxiv.org/abs/2502.04314).
"""


def languages_tab():

    langs_df = pd.read_csv("data/language_metadata.tsv", sep="\t")

    with gr.Tab("Languages"):
        gr.Markdown("# BOUQuET languages list")
        gr.Markdown(LANGS_EXPLANATION)
        gr.Dataframe(langs_df.style.format(na_rep=""))