src/presenter/description_language_presenter.py

from datetime import date

from gradio import BarPlot
from langcodes import Language
from pandas import DataFrame, concat

from dataset_data import DatasetData
from df_utils import DfUtils

_TOP_N: int = 5
_OTHER_LABEL: str = "Other"


class DescriptionLanguagePresenter:
    def __init__(self, data: DatasetData):
        self.__data = data

    @staticmethod
    def __language_name(code: str) -> str:
        return Language.get(code).display_name()

    def present(self) -> BarPlot:
        all_dfs: dict[date, DataFrame] = self.__data.get_parquet_dict()
        non_null_dfs: dict[date, DataFrame] = {}
        for day, df in all_dfs.items():
            language_df: DataFrame = DfUtils.extract_description_language(df[["id", "anki_web"]])
            non_null_dfs[day] = language_df[language_df["description_language"].notna()]

        all_languages_df: DataFrame = concat(non_null_dfs.values(), ignore_index=True)
        top_languages: list[str] = all_languages_df["description_language"].value_counts().head(_TOP_N).index.tolist()

        rows: list[dict] = []
        for day, df in non_null_dfs.items():
            total: int = len(df)
            if total == 0:
                continue
            counted_top: int = 0
            for language in top_languages:
                count: int = int((df["description_language"] == language).sum())
                counted_top += count
                rows.append(
                    {
                        "Date": str(day),
                        "Language": self.__language_name(language),
                        "Percentage": round(100 * count / total, 1),
                    }
                )
            other_count: int = total - counted_top
            rows.append(
                {"Date": str(day), "Language": _OTHER_LABEL, "Percentage": round(100 * other_count / total, 1)}
            )

        plot_df: DataFrame = DataFrame(rows, columns=["Date", "Language", "Percentage"])
        return BarPlot(
            plot_df,
            x="Date",
            y="Percentage",
            color="Language",
            y_title="Percentage of addons (%)",
            x_title="Snapshot date",
            y_lim=[0, 100],
            y_axis_format=".1f",
            sort="x",
        )