strip_short_diff.py

#!/usr/bin/env python3
"""Strip \\DIFadd{...} and \\DIFaddFL{...} markup when the content has fewer than N words."""
import re
import sys


def count_words(text: str) -> int:
    # Remove LaTeX commands
    plain = re.sub(r'\\[a-zA-Z]+\*?(?:\[[^\]]*\])?(?:\{[^{}]*\})*', '', text)
    # Remove braces, tildes, dollars
    plain = re.sub(r'[{}~$]', '', plain)
    return len(plain.split())


def strip_short(content: str, min_words: int) -> str:
    # Match \DIFadd{...} or \DIFaddFL{...} with balanced braces
    pattern = re.compile(r'\\DIFadd(?:FL)?\{')

    result = []
    i = 0
    while i < len(content):
        m = pattern.search(content, i)
        if not m:
            result.append(content[i:])
            break

        result.append(content[i:m.start()])

        # Find matching closing brace
        depth = 0
        j = m.end() - 1  # position of opening {
        for j in range(m.end() - 1, len(content)):
            if content[j] == '{':
                depth += 1
            elif content[j] == '}':
                depth -= 1
                if depth == 0:
                    break

        body = content[m.end():j]  # content inside { }

        if count_words(body) < min_words:
            result.append(body)  # strip markup, keep content
        else:
            result.append(content[m.start():j + 1])  # keep markup

        i = j + 1

    return ''.join(result)


if __name__ == '__main__':
    min_words = int(sys.argv[1]) if len(sys.argv) > 1 else 3
    content = sys.stdin.read()
    sys.stdout.write(strip_short(content, min_words))