check_refactor.py

#!/usr/bin/env python
# SPDX-License-Identifier: Apache-2.0
# Copyright 2025 Google LLC.
# Author: Marco Elver

import sys
from collections import Counter

def main():
    """
    Reads a diff from stdin and checks if lines were simply moved around.
    Ignores blank lines and leading/trailing whitespace.
    """
    removed = Counter()
    added = Counter()

    for line in sys.stdin:
        line = line.rstrip('\n')

        # Skip git diff header lines
        if line.startswith('diff --git'): continue
        if line.startswith('index '): continue
        if line.startswith('old mode '): continue
        if line.startswith('new mode '): continue
        if line.startswith('new file mode'): continue
        if line.startswith('deleted file mode'): continue
        if line.startswith('similarity index'): continue
        if line.startswith('copy from'): continue
        if line.startswith('copy to'): continue
        if line.startswith('rename from'): continue
        if line.startswith('rename to'): continue
        if line.startswith('--- '): continue
        if line.startswith('+++ '): continue
        if line.startswith('@@ '): continue
        if line.startswith(r'\ No newline'): continue

        # Check for added/removed lines
        if line.startswith('-'):
            content = line[1:].strip()
            if content: # Ignore blank lines
                removed[content] += 1
        elif line.startswith('+'):
            content = line[1:].strip()
            if content: # Ignore blank lines
                added[content] += 1

    # Calculate differences
    lost_lines = removed - added
    extra_lines = added - removed

    if not lost_lines and not extra_lines:
        print("OK: No lines lost or added (ignoring whitespace).")
        sys.exit(0)

    found_issues = False
    if lost_lines:
        found_issues = True
        print("--- LOST LINES (Removed but not re-added) ---")
        for line, count in lost_lines.items():
            prefix = f"{count} copies" if count > 1 else "1 copy"
            print(f"[{prefix}] {line}")
        print()

    if extra_lines:
        found_issues = True
        print("--- ADDED LINES (Added but not present originally) ---")
        for line, count in extra_lines.items():
            prefix = f"{count} copies" if count > 1 else "1 copy"
            print(f"[{prefix}] {line}")

    if found_issues:
        sys.exit(1)

if __name__ == "__main__":
    main()