First off, let’s break down what’s happening here. Your code uses re.sub("[^[:print:]]", '?', line) to swap non-printable characters with a question mark, but it’s unexpectedly replacing the '5' in your string. Here are the most likely reasons and fixes:

1. The '5' Isn’t a Standard ASCII Digit #

Chances are, the character you see as '5' in your line variable isn’t the regular ASCII digit (ordinal value 53). It might be a Unicode character that looks identical (like a full-width digit U+FF15, ordinal 65293) but isn’t part of the ASCII printable set.

To confirm this, run this quick check to inspect each character’s ordinal value:

for char in line:
    print(f"Character: '{char}' | Ordinal: {ord(char)}")

If the '5' shows an ordinal other than 53, that’s your culprit.

2. Python’s [:print:] Uses ASCII by Default #

By default, Python’s re module treats POSIX character classes like [:print:] as ASCII-only. That means any non-ASCII printable character (like the full-width 5) will be matched by [^[:print:]] and replaced.

Fix 1: Use the re.UNICODE Flag #

Add the re.UNICODE (or re.U) flag to make [:print:] include all Unicode printable characters:

import re
line = 'Testing offset [4,-5]'
fixed_line = re.sub("[^[:print:]]", '?', line, flags=re.UNICODE)
print(fixed_line)  # Should output your expected string

Fix 2: Target Only Actual Control Characters #

If you want to replace only true non-printable control characters (not Unicode printable ones), use a regex that explicitly matches ASCII control characters:

fixed_line = re.sub(r'[\x00-\x1F\x7F]', '?', line)

This swaps only control characters (0-31 and 127) while leaving all printable characters—including Unicode ones—intact.

3. Don’t Forget to Assign the Result! #

A quick sanity check: re.sub returns a new string (it doesn’t modify the original line in place). Make sure you’re assigning the result to a variable, like:

line = re.sub("[^[:print:]]", '?', line)  # Overwrite the original line

Or store it in a new variable like fixed_line as shown above.

内容的提问来源于stack exchange，提问作者Deepanshu Arora