prepare/est.py

import os
from transformers import AutoTokenizer

def count_tokens_in_file(file_path, tokenizer):
    """
    Count tokens in a text file using the specified tokenizer.
    
    Args:
        file_path (str): Path to the text file
        tokenizer: HuggingFace tokenizer
        
    Returns:
        int: Number of tokens in the file
    """
    try:
        with open(file_path, 'r', encoding='utf-8') as file:
            text = file.read()
        
        # Tokenize the text
        tokens = tokenizer.encode(text)
        
        return len(tokens)
    except Exception as e:
        print(f"Error processing {file_path}: {e}")
        return 0

def main():
    # Load tokenizer
    print("Loading tokenizer...")
    tokenizer = AutoTokenizer.from_pretrained("intfloat/multilingual-e5-large")
    
    # List of files to process
    files = ["fb.txt", "threads.txt", "tbrain.txt", "ptt.txt", "dcard.txt", "discord.txt"]
    
    total_tokens = 0
    # 1B tokens
    goal = 1000000000
    
    
    # Process each file
    for file_path in files:
        if os.path.exists(file_path):
            tokens = count_tokens_in_file(file_path, tokenizer)
            total_tokens += tokens
            print(f"{file_path}: {tokens} tokens")
        else:
            print(f"File not found: {file_path}")
    
    print(f"\nTotal tokens across all files: {total_tokens}")
    print(f"Goal: {goal}")
    print(f"Remaining(percentage): {(goal - total_tokens) / goal * 100:.2f}%")

if __name__ == "__main__":
    main()