SERVER-45949 Update validate commit message client to work with new patch description standard

(cherry picked from commit 09b9e9316877084501db453d08510345465a6770)

2 files changed, 109 insertions, 27 deletions

diff --git a/buildscripts/tests/test_validate_commit_message.py b/buildscripts/tests/test_validate_commit_message.py
index 1cfd37a1d0c..0539776fd71 100644
--- a/buildscripts/tests/test_validate_commit_message.py
+++ b/buildscripts/tests/test_validate_commit_message.py
@@ -1,5 +1,5 @@
 """Unit tests for the evergreen_task_timeout script."""
-
+import itertools
 import unittest
 from mock import MagicMock, patch
 
@@ -8,11 +8,11 @@ from buildscripts.validate_commit_message import main, STATUS_OK, STATUS_ERROR,
 # pylint: disable=missing-docstring,no-self-use
 
 INVALID_MESSAGES = [
-    [""],  # You must provide a message
-    ["RevertEVG-1"],  # revert and ticket must be formatted
-    ["revert EVG-1"],  # revert must be capitalized
-    ["This is not a valid message"],  # message must be valid
-    ["Fix lint plus extras is not a valid message"],  # Fix lint is strict
+    "",  # You must provide a message
+    "RevertEVG-1",  # revert and ticket must be formatted
+    "revert EVG-1",  # revert must be capitalized
+    "This is not a valid message",  # message must be valid
+    "Fix Lint",  # Fix lint is strict in terms of caps
 ]
 
 NS = "buildscripts.validate_commit_message"
@@ -23,33 +23,49 @@ def ns(relative_name):  # pylint: disable=invalid-name
     return NS + "." + relative_name
 
 
+def interleave_new_format(older):
+    """Create a new list containing a new and old format copy of each string."""
+    newer = [
+        f"Commit Queue Merge: '{old}' into 'mongodb/mongo:SERVER-45949-validate-message-format'"
+        for old in older
+    ]
+    return list(itertools.chain(*zip(older, newer)))
+
+
 class ValidateCommitMessageTest(unittest.TestCase):
     def test_valid(self):
         messages = [
-            ["Fix lint"],
-            ["EVG-1"],  # Test valid projects with various number lengths
-            ["SERVER-20"],
-            ["WT-300"],
-            ["SERVER-44338"],
-            ["Revert EVG-5"],
-            ["Revert SERVER-60"],
-            ["Revert WT-700"],
-            ["Revert 'SERVER-8000"],
-            ['Revert "SERVER-90000'],
-            ["Import wiredtiger: 58115abb6fbb3c1cc7bfd087d41a47347bce9a69 from branch mongodb-4.4"],
-            ["Import tools: 58115abb6fbb3c1cc7bfd087d41a47347bce9a69 from branch mongodb-4.4"]
+            "Fix lint",
+            "EVG-1",  # Test valid projects with various number lengths
+            "SERVER-20",
+            "WT-300",
+            "SERVER-44338",
+            "Revert EVG-5",
+            "Revert SERVER-60",
+            "Revert WT-700",
+            "Revert 'SERVER-8000",
+            'Revert "SERVER-90000',
+            "Import wiredtiger: 58115abb6fbb3c1cc7bfd087d41a47347bce9a69 from branch mongodb-4.4",
+            "Import tools: 58115abb6fbb3c1cc7bfd087d41a47347bce9a69 from branch mongodb-4.4"
         ]
 
-        self.assertTrue(all(main(message) == STATUS_OK for message in messages))
+        self.assertTrue(
+            all(main([message]) == STATUS_OK for message in interleave_new_format(messages)))
 
     def test_private(self):
         self.assertEqual(main(["XYZ-1"]), STATUS_ERROR)
 
     def test_catch_all(self):
-        self.assertTrue(all(main(message) == STATUS_ERROR for message in INVALID_MESSAGES))
+        self.assertTrue(
+            all(
+                main([message]) == STATUS_ERROR
+                for message in interleave_new_format(INVALID_MESSAGES)))
 
     def test_ignore_warnings(self):
-        self.assertTrue(all(main(["-i"] + message) == STATUS_OK for message in INVALID_MESSAGES))
+        self.assertTrue(
+            all(
+                main(["-i"] + [message]) == STATUS_OK
+                for message in interleave_new_format(INVALID_MESSAGES)))
 
     def test_last_git_commit_success(self):
         with patch(
diff --git a/buildscripts/validate_commit_message.py b/buildscripts/validate_commit_message.py
index 03283b9d940..e9e186dd25c 100755
--- a/buildscripts/validate_commit_message.py
+++ b/buildscripts/validate_commit_message.py
@@ -33,12 +33,26 @@ import re
 import subprocess
 import sys
 
-VALID_PATTERNS = [
-    re.compile(r"^Fix lint$"),  # Allow "Fix lint" as the sole commit summary
-    re.compile(r'^(Revert ["\']?)?(EVG|SERVER|WT)-[0-9]+'),  # These are public tickets
-    re.compile(r'^Import (wiredtiger|tools):'),  # These are public tickets
-]
-PRIVATE_PATTERNS = [re.compile(r"^[A-Z]+-[0-9]+")]
+COMMON_PUBLIC_PATTERN = r'''
+    ((?P<revert>Revert)\s+[\"\']?)?                         # Revert (optional)
+    ((?P<ticket>(?:EVG|SERVER|WT)-[0-9]+)[\"\']?\s*)               # ticket identifier
+    (?P<body>(?:(?!\(cherry\spicked\sfrom).)*)?             # To also capture the body
+    (?P<backport>\(cherry\spicked\sfrom.*)?                 # back port (optional)
+    '''
+"""Common Public pattern format."""
+
+COMMON_LINT_PATTERN = r'(?P<lint>Fix\slint)'
+"""Common Lint pattern format."""
+
+COMMON_IMPORT_PATTERN = r'(?P<imported>Import\s(wiredtiger|tools):\s.*)'
+"""Common Import pattern format."""
+
+COMMON_PRIVATE_PATTERN = r'''
+    ((?P<revert>Revert)\s+[\"\']?)?                                     # Revert (optional)
+    ((?P<ticket>[A-Z]+-[0-9]+)[\"\']?\s*)                               # ticket identifier
+    (?P<body>(?:(?!('\s(into\s'(([^/]+))/(([^:]+)):(([^']+))'))).)*)?   # To also capture the body
+'''
+"""Common Private pattern format."""
 
 STATUS_OK = 0
 STATUS_ERROR = 1
@@ -46,6 +60,58 @@ STATUS_ERROR = 1
 GIT_SHOW_COMMAND = ["git", "show", "-1", "-s", "--format=%s"]
 
 
+def new_patch_description(pattern: str) -> str:
+    """
+    Wrap the pattern to conform to the new commit queue patch description format.
+
+    Add the commit queue prefix and suffix to the pattern. The format looks like:
+
+    Commit Queue Merge: '<commit message>' into '<owner>/<repo>:<branch>'
+
+    :param pattern: The pattern to wrap.
+    :return: A pattern to match the new format for the patch description.
+    """
+    return (r"""^((?P<commitqueue>Commit\sQueue\sMerge:)\s')"""
+            f'{pattern}'
+            # r"""('\s(?P<into>into\s'((?P<owner>[^/]+))/((?P<repo>[^:]+)):((?P<branch>[^']+))'))"""
+            )
+
+
+def old_patch_description(pattern: str) -> str:
+    """
+    Wrap the pattern to conform to the new commit queue patch description format.
+
+    Just add a start anchor. The format looks like:
+
+    <commit message>
+
+    :param pattern: The pattern to wrap.
+    :return: A pattern to match the old format for the patch description.
+    """
+    return r'^' f'{pattern}'
+
+
+# NOTE: re.VERBOSE is for visibility / debugging. As such significant white space must be
+# escaped (e.g ' ' to \s).
+VALID_PATTERNS = [
+    re.compile(new_patch_description(COMMON_PUBLIC_PATTERN), re.MULTILINE | re.DOTALL | re.VERBOSE),
+    re.compile(old_patch_description(COMMON_PUBLIC_PATTERN), re.MULTILINE | re.DOTALL | re.VERBOSE),
+    re.compile(new_patch_description(COMMON_LINT_PATTERN), re.MULTILINE | re.DOTALL | re.VERBOSE),
+    re.compile(old_patch_description(COMMON_LINT_PATTERN), re.MULTILINE | re.DOTALL | re.VERBOSE),
+    re.compile(new_patch_description(COMMON_IMPORT_PATTERN), re.MULTILINE | re.DOTALL | re.VERBOSE),
+    re.compile(old_patch_description(COMMON_IMPORT_PATTERN), re.MULTILINE | re.DOTALL | re.VERBOSE),
+]
+"""valid public patterns."""
+
+PRIVATE_PATTERNS = [
+    re.compile(
+        new_patch_description(COMMON_PRIVATE_PATTERN), re.MULTILINE | re.DOTALL | re.VERBOSE),
+    re.compile(
+        old_patch_description(COMMON_PRIVATE_PATTERN), re.MULTILINE | re.DOTALL | re.VERBOSE),
+]
+"""private patterns."""
+
+
 def main(argv=None):
     """Execute Main function to validate commit messages."""
     parser = argparse.ArgumentParser(