import pandas as pd import os # --- Setup Paths --- # Get the directory where this script is located (resources/) script_dir = os.path.dirname(os.path.abspath(__file__)) # Define the datasets directory (resources/datasets/) datasets_dir = os.path.join(script_dir, "datasets") # [NEW] Create the directory if it doesn't exist (Safety check) os.makedirs(datasets_dir, exist_ok=True) input_file = os.path.join(datasets_dir, "routerbench_0shot.pkl") full_csv_output = os.path.join(datasets_dir, "routerbench_0shot.csv") train_output = os.path.join(datasets_dir, "routerbench_0shot_train.csv") test_output = os.path.join(datasets_dir, "routerbench_0shot_test.csv") test_sample_output = os.path.join(datasets_dir, "routerbench_0shot_test_500.csv") try: print(f"Loading dataset from: {input_file}") # Load the pickle dataset df = pd.read_pickle(input_file) # --- 0. Convert Original to CSV --- df.to_csv(full_csv_output, index=False) print(f"Converted pickle to CSV: {full_csv_output}") # --- 1. Randomly sample 1% for Train (No Replacement) --- train_df = df.sample(frac=0.01, random_state=42) train_df.to_csv(train_output, index=False) print(f"Created 'train' split with {len(train_df)} rows.") # --- 2. Remaining 99% for Test (Keep Original Ordering) --- test_df = df.drop(train_df.index) test_df.to_csv(test_output, index=False) print(f"Created 'test' split with {len(test_df)} rows.") # --- 3. Randomly sample 500 rows from the Test set --- sample_size = 500 if len(test_df) < sample_size: print(f"Warning: Test data only has {len(test_df)} rows. Sampling all of them.") sample_size = len(test_df) test_sample_df = test_df.sample(n=sample_size, random_state=42) test_sample_df.to_csv(test_sample_output, index=False) print(f"Created 'test_500' split with {len(test_sample_df)} rows.") except FileNotFoundError: print(f"Error: The file '{input_file}' was not found. Please run the download script first.") except Exception as e: print(f"An unexpected error occurred: {e}")