src/leaderboard/data.py

import logging
import datasets
from src.populate import get_leaderboard_df
from src.envs import AGGREGATED_REPO, HF_HOME
from src.display.utils import COLS, BENCHMARK_COLS

class LeaderboardData:
    def __init__(self):
        self.__data = None
        self.aggregated_repo = AGGREGATED_REPO  # Replace with your actual repo
        self.hf_home = HF_HOME  # Replace with your actual HF_HOME
        self.cols = COLS  # Replace with your actual COLS
        self.benchmark_cols = BENCHMARK_COLS  # Replace with your actual BENCHMARK_COLS

    def __update(self):
        """Internal method to download and process leaderboard data."""
        try:
            leaderboard_dataset = datasets.load_dataset(
                self.aggregated_repo,
                "default",
                split="train",
                cache_dir=self.hf_home,
                download_mode=datasets.DownloadMode.FORCE_REDOWNLOAD,
                verification_mode="no_checks"
            )
            
            self.__data = get_leaderboard_df(
                leaderboard_dataset=leaderboard_dataset,
                cols=self.cols,
                benchmark_cols=self.benchmark_cols,
            )
            
            logging.info("Leaderboard dataset successfully downloaded.")
            return self.__data
            
        except Exception as e:
            logging.error(f"Failed to download leaderboard dataset: {e}")
            return None

    def update(self):
        """Public method to trigger leaderboard data update."""
        logging.info("Leaderboard updated at reload!")
        return self.__update()

    def get_data(self):
        """Returns the current leaderboard data."""
        return self.__data


# def get_latest_data_leaderboard(leaderboard_initial_df=None):
#     global NEW_DATA_ON_LEADERBOARD
#     global LEADERBOARD_DF
#     if NEW_DATA_ON_LEADERBOARD:
#         logging.info("Leaderboard updated at reload!")
#         try:
#             leaderboard_dataset = datasets.load_dataset(
#                 AGGREGATED_REPO,
#                 "default",
#                 split="train",
#                 cache_dir=HF_HOME,
#                 download_mode=datasets.DownloadMode.FORCE_REDOWNLOAD,  # Always download fresh data
#                 verification_mode="no_checks"
#             )
#             LEADERBOARD_DF = get_leaderboard_df(
#                 leaderboard_dataset=leaderboard_dataset,
#                 cols=COLS,
#                 benchmark_cols=BENCHMARK_COLS,
#             )
#             logging.info("Leaderboard dataset successfully downloaded.")
#         except Exception as e:
#             logging.error(f"Failed to download leaderboard dataset: {e}")
#             return

#         # Reset the flag after successful download
#         NEW_DATA_ON_LEADERBOARD = False
#     else:
#         LEADERBOARD_DF = leaderboard_initial_df
#         logging.info("Using cached leaderboard dataset.")
#     return LEADERBOARD_DF