genfile.py

#!/usr/bin/env python3

import os
import argparse
from pathlib import Path

def duplicate_file(input_file: str, output_file: str, target_size: str) -> None:
    # Convert size string (like '1GB', '100MB') to bytes
    size_map = {'B': 1, 'KB': 1024, 'MB': 1024**2, 'GB': 1024**3}
    size_value = float(target_size[:-2])
    size_unit = target_size[-2:].upper()
    target_bytes = int(size_value * size_map[size_unit])

    # Read input file in chunks of 1MB
    chunk_size = 1024 * 1024

    with open(input_file, 'rb') as source, open(output_file, 'wb') as dest:
        # Read the entire source file
        source_data = source.read()
        source_size = len(source_data)

        # Calculate how many times we need to write the source
        bytes_written = 0

        while bytes_written < target_bytes:
            # Calculate remaining bytes
            remaining = target_bytes - bytes_written
            # Write either the full source or just the remaining bytes needed
            write_size = min(source_size, remaining)
            dest.write(source_data[:write_size])
            bytes_written += write_size

if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='Generate large file by duplicating content')
    parser.add_argument('input_file', help='Source file to duplicate')
    parser.add_argument('output_file', help='Output file path')
    parser.add_argument('size', help='Target file size (e.g., 1GB, 100MB)')

    args = parser.parse_args()

    try:
        duplicate_file(args.input_file, args.output_file, args.size)
        print(f"Successfully generated {args.output_file} with size {args.size}")
    except Exception as e:
        print(f"Error: {str(e)}")