From 0eb892fd74e3e129ca308393545e9f82b87dbaf3 Mon Sep 17 00:00:00 2001
From: Nikolay Martyanov <nikolay@zededa.com>
Date: Tue, 27 Aug 2024 02:30:46 +0200
Subject: [PATCH] github: Add commit message linting workflow.

Introduced a GitHub Actions workflow (`commit-messages.yml`) to lint
commit messages on `master`, version branches, and stable branches.

Added `check_commit_messages.py` script to validate commit messages
against recommended format (subject, empty line, body).

Signed-off-by: Nikolay Martyanov <nikolay@zededa.com>
---
 .github/workflows/commit-messages.yml         | 44 +++++++++++
 tools/check-commit-messages/.gitignore        |  1 +
 .../check_commit_messages.py                  | 78 +++++++++++++++++++
 tools/check-commit-messages/requirements.txt  |  3 +
 4 files changed, 126 insertions(+)
 create mode 100644 .github/workflows/commit-messages.yml
 create mode 100644 tools/check-commit-messages/.gitignore
 create mode 100644 tools/check-commit-messages/check_commit_messages.py
 create mode 100644 tools/check-commit-messages/requirements.txt

diff --git a/.github/workflows/commit-messages.yml b/.github/workflows/commit-messages.yml
new file mode 100644
index 0000000000..f70cdbea7d
--- /dev/null
+++ b/.github/workflows/commit-messages.yml
@@ -0,0 +1,44 @@
+# Copyright (c) 2024, Zededa, Inc.
+# SPDX-License-Identifier: Apache-2.0
+---
+name: Lint Commit Messages
+
+on:  # yamllint disable-line rule:truthy
+  push:
+    branches:
+      - "master"
+      - "[0-9]+.[0-9]+"
+      - "[0-9]+.[0-9]+-stable"
+  pull_request:
+    branches:
+      - "master"
+      - "[0-9]+.[0-9]+"
+      - "[0-9]+.[0-9]+-stable"
+
+jobs:
+  commitlint:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ github.base_ref }}
+          fetch-depth: 0
+
+      - name: Fetch the PR's head ref
+        run: |
+          git fetch origin ${{ github.event.pull_request.head.sha }}:${{ github.event.pull_request.head.ref }}
+          git checkout ${{ github.event.pull_request.head.ref }}
+
+      - name: Create virtual environment
+        run: |
+          cd tools/check-commit-messages
+          python3 -m venv .venv
+          source .venv/bin/activate
+          pip install -r requirements.txt
+
+      - name: Lint commit messages
+        run: |
+          cd tools/check-commit-messages
+          source .venv/bin/activate
+          python check_commit_messages.py ${{ github.event.pull_request.base.sha }}
diff --git a/tools/check-commit-messages/.gitignore b/tools/check-commit-messages/.gitignore
new file mode 100644
index 0000000000..5ceb3864c2
--- /dev/null
+++ b/tools/check-commit-messages/.gitignore
@@ -0,0 +1 @@
+venv
diff --git a/tools/check-commit-messages/check_commit_messages.py b/tools/check-commit-messages/check_commit_messages.py
new file mode 100644
index 0000000000..ff7ebad26f
--- /dev/null
+++ b/tools/check-commit-messages/check_commit_messages.py
@@ -0,0 +1,78 @@
+"""
+Copyright (c) 2024, Zededa Inc.
+SPDX-License-Identifier: Apache-2.0
+
+This script checks the commit messages in the current branch to ensure they follow the
+recommended format. The recommended format is as follows:
+* Commit message should have a subject and body
+* An empty line should separate the subject and body
+* Body should start with a capital letter
+* The body should not be empty (it should contain more than just "Signed-off-by")
+"""
+
+import sys
+
+import git
+
+def check_commit_message(commit):
+    """
+    Check if a commit message follows the recommended format.
+    """
+    message_lines = commit.message.strip().splitlines()
+
+    if len(message_lines) < 2:
+        return False, f"Commit {commit.hexsha} has no body."
+
+    subject = message_lines[0].strip()
+    second_line = message_lines[1].strip()
+
+    if not subject:
+        return False, f"Commit {commit.hexsha} has no subject."
+
+    # Check if the second line is empty (indicating a blank line between subject and body)
+    if second_line:
+        return False, f"Commit {commit.hexsha} does not have an empty line after the subject."
+
+    # Remove the subject and empty line to get the body
+    body_lines = [line.strip() for line in message_lines[2:] if line.strip()]
+
+    # Remove lines that are just "Signed-off-by"
+    body_lines = [line for line in body_lines if not line.lower().startswith("signed-off-by")]
+
+    if not body_lines:
+        return False, f"Commit {commit.hexsha} has a body but only contains Signed-off-by."
+
+    # Check if the body starts with a capital letter
+    if not body_lines[0][0].isupper():
+        return False, f"Body of commit {commit.hexsha} does not start with a capital letter."
+
+    return True, None
+
+
+def main():
+    """
+    Main function to check the commit messages in the current branch.
+    """
+    repo = git.Repo(search_parent_directories=True)
+
+    # Base hash should be provided as an argument
+    base_hash = sys.argv[1]
+
+    commits = list(repo.iter_commits(f'{base_hash}..HEAD'))
+
+    if not commits:
+        print(f"No commits between {base_hash} and HEAD.")
+        sys.exit(1)
+
+    print(f"Checking {len(commits)} commits between {base_hash} and HEAD...")
+
+    for commit in commits:
+        valid, error_message = check_commit_message(commit)
+        if not valid:
+            print(error_message)
+            sys.exit(1)
+
+    print("All commits are valid.")
+
+if __name__ == "__main__":
+    main()
diff --git a/tools/check-commit-messages/requirements.txt b/tools/check-commit-messages/requirements.txt
new file mode 100644
index 0000000000..dba1edf678
--- /dev/null
+++ b/tools/check-commit-messages/requirements.txt
@@ -0,0 +1,3 @@
+gitdb==4.0.11
+GitPython==3.1.43
+smmap==5.0.1