Fix updating of repos other than matplotlib.github.com

On those repos, the `gh-pages` branch is force pushed to prevent the
repository from ballooning in size. This means we have to do a hard
reset when pulling it.

Author: QuLogic

test_webhook.py

@@ -35,18 +35,11 @@ async def test_signature(monkeypatch):
         'unused')
 
 
-async def test_update_repo_empty(tmp_path_factory):
-    """Test that updating an empty repository works as expected."""
-    repo1 = tmp_path_factory.mktemp('repo1')
-    run_check(['git', 'init', '-b', 'main', repo1])
-    await update_repo(repo1, 'unused', 'matplotlib/repo1')
-
-
 async def test_update_repo(tmp_path_factory):
     """Test that updating a repository works as expected."""
     # Set up a source repository.
     src = tmp_path_factory.mktemp('src')
-    run_check(['git', 'init', '-b', 'main', src])
+    run_check(['git', 'init', '-b', 'gh-pages', src])
     (src / 'readme.txt').write_text('Test repo information')
     run_check(['git', 'add', 'readme.txt'], cwd=src)
     run_check(['git', 'commit', '-m', 'Initial commit'], cwd=src)
@@ -65,7 +58,7 @@ async def test_update_repo(tmp_path_factory):
         (line for line in src_stdout if line.split()[-1] == 'HEAD'), '')
 
     # Now this should correctly update the first repository.
-    await update_repo(dest, 'unused', 'matplotlib/dest')
+    assert await update_repo(dest, 'unused', 'matplotlib/dest')
     dest_stdout = run_check(['git', 'show-ref', '--head', 'HEAD'], cwd=dest,
                             capture_output=True).stdout.splitlines()
     dest_commit = next(

webhook.py

@@ -32,11 +32,25 @@
 async def update_repo(repo: Path, delivery: str, name: str):
     """Update a git repo at the given path."""
     log.info('%s: Running git pull for %s at %s', delivery, name, repo)
-    proc = await asyncio.create_subprocess_exec('git', 'pull', cwd=repo)
+
+    proc = await asyncio.create_subprocess_exec('git', 'fetch', cwd=repo)
+    await proc.wait()
+    if proc.returncode != 0:
+        log.error('%s: Running git fetch for %s at %s failed: %d',
+                  delivery, name, repo, proc.returncode)
+        return False
+
+    # Updates must be done this way because we force push away previous
+    # gh-pages commits to keep the repository from ballooning in size.
+    proc = await asyncio.create_subprocess_exec(
+        'git', 'reset', '--hard', '@{upstream}', cwd=repo)
     await proc.wait()
     if proc.returncode != 0:
-        log.error('%s: Running git pull for %s at %s failed: %d',
+        log.error('%s: Running git reset for %s at %s failed: %d',
                   delivery, name, repo, proc.returncode)
+        return False
+
+    return True
 
 
 def handle_update_repo_result(task: asyncio.Task):
@@ -134,10 +148,12 @@ async def github_webhook(request: web.Request):
         log.info('%s: Ignoring webhook for unused event %s', delivery, event)
         return web.Response(status=200)
 
+    expected_branch = ('main' if repository == 'matplotlib.github.com'
+                       else 'gh-pages')
     ref = data.get('ref', '')
-    if ref != 'refs/heads/gh-pages':
-        log.info('%s: Ignoring push event on branch %s other than gh-pages',
-                 delivery, ref)
+    if ref != f'refs/heads/{expected_branch}':
+        log.info('%s: Ignoring push event on branch %s other than %s',
+                 delivery, ref, expected_branch)
         return web.Response(status=200)
 
     checkout = Path(os.environ.get('SITE_DIR', 'sites'), repository)