src/populate.py

# src/populate.py

import os
import pandas as pd
import json

from src.display.utils import COLUMNS, EVAL_COLS

def get_leaderboard_df(eval_results_path, eval_requests_path, cols, benchmark_cols):
    # Initialize an empty DataFrame
    df = pd.DataFrame(columns=cols)

    # Load evaluation results from JSON files
    if os.path.exists(eval_results_path):
        result_files = [os.path.join(eval_results_path, f) for f in os.listdir(eval_results_path) if f.endswith('.json')]
        data_list = []
        for file in result_files:
            with open(file, 'r') as f:
                data = json.load(f)
                # Flatten the JSON structure if needed
                flattened_data = {}
                flattened_data.update(data.get('config', {}))
                flattened_data.update(data.get('results', {}))
                data_list.append(flattened_data)
        if data_list:
            df = pd.DataFrame(data_list)

    # Ensure DataFrame has all columns
    for col in cols:
        if col not in df.columns:
            df[col] = None

    # Sort by 'average' column if it exists
    if 'average' in df.columns:
        df = df.sort_values(by=['average'], ascending=False)

    return df

def get_evaluation_queue_df(eval_requests_path, eval_cols):
    # Initialize empty DataFrames
    finished_df = pd.DataFrame(columns=eval_cols)
    running_df = pd.DataFrame(columns=eval_cols)
    pending_df = pd.DataFrame(columns=eval_cols)

    # Load evaluation requests from JSON files
    if os.path.exists(eval_requests_path):
        request_files = [os.path.join(eval_requests_path, f) for f in os.listdir(eval_requests_path) if f.endswith('.json')]
        data_list = []
        for file in request_files:
            with open(file, 'r') as f:
                data = json.load(f)
                data_list.append(data)
        if data_list:
            df = pd.DataFrame(data_list)
            # Split DataFrame based on status
            finished_df = df[df['status'] == 'finished']
            running_df = df[df['status'] == 'running']
            pending_df = df[df['status'] == 'pending']

    return finished_df, running_df, pending_df