CI: Check commit message compliance

This commit adds these checks.
- Title consists of module name(s) and subject or just subject.
- If there are two or more module names, the module names has to be
  separated by a comma followed by an optional space.
- First word of the subject is starting from an upper case letter and
  remaining characters in the first word are lower case letters.
  "Don't" and a word with a hyphen at the middle are also acceptable.
- The title is 72 characters at max including module prefix.
- If the subject exceed 50 characters excluding the module prefix,
  display a warning message.
- Titles for commits of revert and merge are ignored.
- Full description lines are 72 columns max.

.github/workflows/commit-msg.yml

@@ -0,0 +1,24 @@
+name: Commit Message Check
+
+on: [pull_request]
+
+jobs:
+  ubuntu64:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+        with:
+          fetch-depth: 0
+
+      - name: Setup environment
+        run: |
+          pip3 install -U gitpython
+
+      - name: Fetch OBS Studio master
+        run: |
+          git fetch origin
+
+      - name: Check the log messages
+        run: |
+          ./CI/check-log-msg.py origin/master..

CI/check-log-msg.py

@@ -0,0 +1,132 @@
+#! /bin/env python3
+
+import git
+import re
+
+has_error = False
+
+LOG_ERROR = 100
+LOG_WARNING = 200
+LOG_INFO = 300
+current_commit = None
+
+def blog(level, txt):
+    if level == LOG_ERROR:
+        global has_error
+        has_error = True
+        s_level = "Error: "
+    elif level == LOG_WARNING:
+        s_level = "Warning: "
+    elif level == LOG_INFO:
+        s_level = "Info: "
+    print('{level}commit {hexsha}: {txt}'.format(level = s_level, hexsha = current_commit.hexsha[:9], txt = txt))
+
+def find_directory_name(name, tree, max_depth):
+    for t in tree.trees:
+        if name == t.name:
+            return True
+    if max_depth > 1:
+        for t in tree.trees:
+            if find_directory_name(name, t, max_depth - 1):
+                return True
+        for b in tree.blobs:
+            if name == b.name:
+                return True
+    return False
+
+def check_path(path, tree):
+    paths = path.split('/', 1)
+    name = paths[0]
+    if len(paths) == 1:
+        return find_directory_name(name, tree, 1)
+    else:
+        for t in tree.trees:
+            if name == t.name:
+                return check_path(paths[1], t)
+    return False
+
+def find_submodule_name(name, submodules):
+    for s in submodules:
+        sname = s.name.split('/')[-1]
+        if name == sname:
+            return True
+
+def check_module_names(names):
+    for name in names:
+        if name.find('/') >= 0:
+            if check_path(name, current_commit.tree):
+                return True
+        else:
+            if find_directory_name(name, current_commit.tree, 3):
+                return True
+            # TODO: Cannot handle removed submodules. Implement to parse .gitmodules file.
+            if find_submodule_name(name, current_commit.repo.submodules):
+                return True
+    blog(LOG_ERROR, "unknown module name '%s'" % name)
+
+def check_message_title(title):
+    global has_error
+    title_split = title.split(': ', 1)
+    if len(title_split) == 2:
+        check_module_names(re.split(r', *', title_split[0]))
+        title_text = title_split[1]
+    else:
+        title_text = title_split[0]
+
+    if len(title) > 72:
+        blog(LOG_ERROR, 'Too long title: %s' % title)
+
+    if len(title_text) > 50:
+        blog(LOG_WARNING, 'Too long title excluding module name: %s' % title_text)
+
+    if not re.match(r"([A-Z][a-z]*(-[a-z]+)*|Don't) ", title_text):
+        blog(LOG_ERROR, 'Invalid first word: %s' % title_text.split(' ', 1)[0])
+        has_error = True
+
+def check_message_body(body):
+    for line in body.split('\n'):
+        if len(line) > 72 and line.find(' ') > 0:
+            blog(LOG_ERROR, 'Too long description in a line: %s' % line)
+    pass
+
+def check_message(c):
+    msg = c.message.split('\n', 2)
+    if len(msg) == 0:
+        blog(LOG_ERROR, 'Commit message is empty.')
+        return True
+
+    if re.match(r'Revert ', msg[0]):
+        return False
+    if re.match(r'Merge [0-9a-f]{40} into [0-9a-f]{40}$', msg[0]):
+        return False
+    if re.match(r'Merge pull request', msg[0]):
+        return False
+
+    if len(msg) > 0:
+        check_message_title(msg[0])
+    if len(msg) > 1:
+        if len(msg[1]):
+            blog(LOG_ERROR, '2nd line is not empty.')
+    if len(msg) > 2:
+        check_message_body(msg[2])
+    return has_error
+
+def main():
+    import sys
+    import argparse
+    parser = argparse.ArgumentParser(prog='check-log-msg.py', description='Log message compliance checker')
+    parser.add_argument('commits')
+    args = parser.parse_args()
+
+    repo = git.Repo('.')
+    global has_error
+    for c in repo.iter_commits(args.commits): # '27.1.0-rc2..27.1.0'):
+        global current_commit
+        current_commit = c
+        blog(LOG_INFO, "Checking commit '%s'" % c.message.split('\n', 1)[0])
+        check_message(c)
+    if has_error:
+        sys.exit(1)
+
+if __name__ == '__main__':
+    ret = main()