Python script to clean trailing whitespace, preserving individual line endings

cleantrailspace.py

import os
import argparse
import chardet

def is_text_file(filename, chunk_size=1024):
    """Check if the file is likely a text file by reading a small chunk."""
    try:
        with open(filename, 'rb') as f:
            chunk = f.read(chunk_size)
            # Use chardet to detect encoding and ensure it's text-like
            result = chardet.detect(chunk)
            encoding = result.get('encoding')
            return encoding is not None
    except Exception as e:
        print(f"Error reading file {filename}: {e}")
        return False

def remove_trailing_whitespace(file_path):
    with open(file_path, 'rb') as f:
        lines = f.readlines()

    new_lines = []
    for line in lines:
        # Detect the end-of-line (EOL) sequence (\r\n, \n, or no EOL).
        if line.endswith(b'\r\n'):
            eol = b'\r\n'
            stripped_line = line.rstrip(b' \t\r\n')
        elif line.endswith(b'\n'):
            eol = b'\n'
            stripped_line = line.rstrip(b' \t\n')
        else:
            # No EOL, meaning this could be the last line in the file
            eol = b''
            stripped_line = line.rstrip(b' \t')

        new_lines.append(stripped_line + eol)

    # Write the cleaned lines back to the file only if there were changes
    with open(file_path, 'wb') as f:
        f.writelines(new_lines)

def main():
    parser = argparse.ArgumentParser(description="Remove trailing whitespace from text files.")
    parser.add_argument('files', metavar='file', type=str, nargs='+', help='Text files to clean')
    args = parser.parse_args()

    for file_path in args.files:
        if not os.path.isfile(file_path):
            print(f"Skipping {file_path}: not a file.")
            continue

        if not is_text_file(file_path):
            print(f"Skipping {file_path}: not a text file.")
            continue

        print(f"Cleaning {file_path}...")
        try:
            remove_trailing_whitespace(file_path)
            print(f"Finished cleaning {file_path}.")
        except Exception as e:
            print(f"Error cleaning {file_path}: {e}")

if __name__ == "__main__":
    main()