Add check-dco hook: verify commit message Signed-off-by

.pre-commit-hooks.yaml

@@ -210,3 +210,14 @@
     types: [text]
     stages: [pre-commit, pre-push, manual]
     minimum_pre_commit_version: 3.2.0
+
+-   id: check-dco
+    name: Check Developer Certificate of Origin (DCO)
+    description: >
+        Verifies that the commit message contains a valid
+        ``Signed-off-by: Name <email>`` line per the
+        `Developer Certificate of Origin <https://developercertificate.org/>`__.
+    entry: check-dco
+    language: python
+    stages: [commit-msg]
+    minimum_pre_commit_version: 2.9.2

pre_commit_hooks/check_dco.py

@@ -0,0 +1,85 @@
+"""check-dco: verify commit messages contain a Signed-off-by line.
+
+This hook reads the commit message file (passed as argv[1] by pre-commit
+in commit-msg stage) and checks that it contains a valid
+``Signed-off-by: Name <email>`` line per the Developer Certificate of Origin.
+
+This is a pure Python implementation with no external dependencies —
+only the standard library is used (``re``, ``sys``).
+"""
+
+from __future__ import annotations
+
+import argparse
+import re
+from collections.abc import Sequence
+from typing import Optional
+
+DCO_PATTERN = re.compile(
+    r'^Signed-off-by:\s+'
+    r'(?P<name>[^<]+)'
+    r'\s+'
+    r'<(?P<email>[^>]+)>'
+    r'\s*$',
+)
+
+EXIT_PASS = 0
+EXIT_FAIL = 1
+
+
+def check_dco(commit_msg_path: str) -> tuple[int, list[str]]:
+    """Check a commit message file for a valid DCO sign-off.
+
+    Returns (exit_code, diagnostic_messages).
+    """
+    with open(commit_msg_path, 'r', encoding='utf-8') as f:
+        lines = f.read().splitlines()
+
+    errors: list[str] = []
+    found_valid = False
+
+    for i, line in enumerate(lines, start=1):
+        # Check if any line matches the DCO pattern
+        if DCO_PATTERN.match(line):
+            found_valid = True
+        # Detect malformed sign-off attempts
+        lower = line.lower().strip()
+        if lower.startswith('signed-off-by'):
+            if not DCO_PATTERN.match(line):
+                errors.append(
+                    f'{commit_msg_path}:{i}: malformed Signed-off-by line — '
+                    f'expected "Signed-off-by: Name <email>"',
+                )
+
+    if not found_valid and not errors:
+        # No sign-off found at all
+        errors.append(
+            f'{commit_msg_path}: missing Signed-off-by line. '
+            f'Add "Signed-off-by: Your Name <your@email>" to the commit message.',
+        )
+
+    for err in errors:
+        print(err)
+
+    if not found_valid:
+        return EXIT_FAIL, errors
+
+    return EXIT_PASS, []
+
+
+def main(argv: Sequence[str] | None = None) -> int:
+    parser = argparse.ArgumentParser(
+        description='Check that the commit message has a DCO sign-off.',
+    )
+    parser.add_argument(
+        'commit_msg_file',
+        help='Path to the commit message file (provided by pre-commit).',
+    )
+    args = parser.parse_args(argv)
+
+    exit_code, _ = check_dco(args.commit_msg_file)
+    return exit_code
+
+
+if __name__ == '__main__':
+    raise SystemExit(main())

tests/check_dco_test.py

@@ -0,0 +1,223 @@
+from __future__ import annotations
+
+import os
+import pytest
+import tempfile
+
+from pre_commit_hooks.check_dco import check_dco, main
+
+
+def _write_commit_msg(content: str) -> str:
+    """Write a temporary commit message file and return its path."""
+    tmp = tempfile.NamedTemporaryFile(mode='w', delete=False, encoding='utf-8')
+    tmp.write(content)
+    tmp.close()
+    return tmp.name
+
+
+# ── Success cases ──
+
+
+def test_standard_signoff():
+    """A standard Signed-off-by line should pass."""
+    msg = _write_commit_msg(
+        'feat: add new feature\n'
+        '\n'
+        'Signed-off-by: Alice Smith <alice@example.com>\n',
+    )
+    try:
+        code, _ = check_dco(msg)
+        assert code == 0
+    finally:
+        os.unlink(msg)
+
+
+def test_multiple_signoffs():
+    """Multiple co-author sign-offs should pass."""
+    msg = _write_commit_msg(
+        'fix: resolve encoding issue\n'
+        '\n'
+        'Co-authored-by: Bob <bob@example.com>\n'
+        'Signed-off-by: Alice <alice@example.com>\n'
+        'Signed-off-by: Bob <bob@example.com>\n',
+    )
+    try:
+        code, _ = check_dco(msg)
+        assert code == 0
+    finally:
+        os.unlink(msg)
+
+
+def test_signoff_in_body_not_trailer():
+    """Sign-off in the middle of the body should still pass."""
+    msg = _write_commit_msg(
+        'docs: update readme\n'
+        '\n'
+        'Signed-off-by: Charlie <charlie@example.com>\n'
+        '\n'
+        'More content after sign-off.\n',
+    )
+    try:
+        code, _ = check_dco(msg)
+        assert code == 0
+    finally:
+        os.unlink(msg)
+
+
+def test_signoff_with_full_name():
+    """A sign-off with full name and email should pass."""
+    msg = _write_commit_msg(
+        'chore: bump version\n'
+        '\n'
+        'Signed-off-by: John Michael Doe <john.m.doe@company.com>\n',
+    )
+    try:
+        code, _ = check_dco(msg)
+        assert code == 0
+    finally:
+        os.unlink(msg)
+
+
+def test_signoff_with_plus_email():
+    """Email with + addressing should pass."""
+    msg = _write_commit_msg(
+        'refactor: extract method\n'
+        '\n'
+        'Signed-off-by: Dev <dev+feature@example.com>\n',
+    )
+    try:
+        code, _ = check_dco(msg)
+        assert code == 0
+    finally:
+        os.unlink(msg)
+
+
+def test_multiline_commit_with_signoff():
+    """A multi-paragraph commit message with a trailing sign-off."""
+    msg = _write_commit_msg(
+        'feat: implement BM25 search\n'
+        '\n'
+        'This adds BM25 scoring to the search module for\n'
+        'better relevance ranking in RAG pipelines.\n'
+        '\n'
+        'Includes unit tests and benchmark script.\n'
+        '\n'
+        'Signed-off-by: Ikalus <ikalus1988@example.com>\n',
+    )
+    try:
+        code, _ = check_dco(msg)
+        assert code == 0
+    finally:
+        os.unlink(msg)
+
+
+# ── Failure cases ──
+
+
+def test_missing_signoff():
+    """A commit message without any Signed-off-by line should fail."""
+    msg = _write_commit_msg('fix: resolve timeout issue\n\nJust a quick fix.\n')
+    try:
+        code, errors = check_dco(msg)
+        assert code == 1
+        assert any('missing Signed-off-by' in e for e in errors)
+    finally:
+        os.unlink(msg)
+
+
+def test_empty_message():
+    """An empty commit message should fail."""
+    msg = _write_commit_msg('')
+    try:
+        code, errors = check_dco(msg)
+        assert code == 1
+        assert any('missing Signed-off-by' in e for e in errors)
+    finally:
+        os.unlink(msg)
+
+
+def test_signoff_without_email():
+    """Signed-off-by: without email should fail."""
+    msg = _write_commit_msg(
+        'fix: typo\n'
+        '\n'
+        'Signed-off-by: JustName\n',
+    )
+    try:
+        code, errors = check_dco(msg)
+        assert code == 1
+        assert any('malformed' in e for e in errors)
+    finally:
+        os.unlink(msg)
+
+
+def test_signoff_without_name():
+    """Signed-off-by: <email> without name should fail."""
+    msg = _write_commit_msg(
+        'fix: typo\n'
+        '\n'
+        'Signed-off-by: <anon@example.com>\n',
+    )
+    try:
+        code, errors = check_dco(msg)
+        assert code == 1
+        assert any('malformed' in e for e in errors)
+    finally:
+        os.unlink(msg)
+
+
+def test_lowercase_signoff_only():
+    """lowercase 'signed-off-by:'  without proper format should fail."""
+    msg = _write_commit_msg(
+        'fix: typo\n'
+        '\n'
+        'signed-off-by: alice <alice@example.com>\n',
+    )
+    try:
+        code, _ = check_dco(msg)
+        assert code == 1
+    finally:
+        os.unlink(msg)
+
+
+def test_signoff_only_no_colon():
+    """Signed-off-by without colon should fail."""
+    msg = _write_commit_msg(
+        'fix: typo\n'
+        '\n'
+        'Signed-off-by alice <alice@example.com>\n',
+    )
+    try:
+        code, errors = check_dco(msg)
+        assert code == 1
+        assert any('malformed' in e for e in errors)
+    finally:
+        os.unlink(msg)
+
+
+# ── main() integration ──
+
+
+def test_main_passes_with_valid_signoff():
+    msg = _write_commit_msg(
+        'feat: add search\n'
+        '\n'
+        'Signed-off-by: Test <test@example.com>\n',
+    )
+    try:
+        assert main([msg]) == 0
+    finally:
+        os.unlink(msg)
+
+
+def test_main_fails_without_signoff():
+    msg = _write_commit_msg('fix: quick patch\n')
+    try:
+        assert main([msg]) == 1
+    finally:
+        os.unlink(msg)
+
+
+def test_main_help():
+    with pytest.raises(SystemExit):
+        main(['--help'])