clean-ascii

#!/usr/bin/env python3
# vim: ft=python

import unittest
import re
import string
import sys

# Function being tested
non_printable_pattern = f"[^{re.escape(string.printable)}]"

def remove_non_printable(s):
    """Removes all non-printable characters from a string."""
    return re.sub(non_printable_pattern, "", s)

# Test Suite
class TestRemoveNonPrintable(unittest.TestCase):

    def test_basic_text(self):
        """Test that a normal string is unchanged."""
        self.assertEqual(remove_non_printable("Hello, World!"), "Hello, World!")

    def test_non_printable_chars(self):
        """Test removal of non-printable ASCII characters."""
        self.assertEqual(remove_non_printable("Hello\x00 World!\x1F\x7F"), "Hello World!")

    def test_mixed_text(self):
        """Test removal from a mix of printable and non-printable characters."""
        self.assertEqual(remove_non_printable("ABC\x01\x02DEF\x03"), "ABCDEF")

    def test_only_non_printable(self):
        """Test a string with only non-printable characters."""
        self.assertEqual(remove_non_printable("\x00\x01\x02\x03\x04"), "")

    def test_whitespace_retained(self):
        """Ensure that standard whitespace characters are not removed."""
        self.assertEqual(remove_non_printable("Hello\tWorld\n"), "Hello\tWorld\n")

    def test_extended_ascii(self):
        """Ensure that extended ASCII characters are not preserved."""
        self.assertEqual(remove_non_printable("Café — Déjà vu!"), "Caf  Dj vu!")

    def test_unicode_handling(self):
        """Ensure that Unicode characters (beyond ASCII) are not preserved."""
        self.assertEqual(remove_non_printable("你好, мир, 안녕하세요!"), ", , !")

if __name__ == "__main__":
    if len(sys.argv) > 1 and sys.argv[1] == "--self-test":
        unittest.main(argv=[sys.argv[0]])  # Ensures unittest doesn't parse other args
    else:
        # Loop over lines from stdin
        for line in sys.stdin:
            cleaned_line = remove_non_printable(line)
            sys.stdout.write(cleaned_line)  # Use sys.stdout.write to avoid extra newlines