Bug-2038705: Migrate the get_commit to pygithub

tests/changelog/test_collector.py

@@ -36,18 +36,12 @@ def _commit():
             },
         }
 
-    def commit(request):
-        return 200, {}, json.dumps(_commit())
-
     def commits(request):
         return 200, {}, json.dumps([_commit()])
 
     responses.add_callback(
         responses.GET, COMMITS, callback=commits, content_type="application/json"
     )
-    responses.add_callback(
-        responses.GET, COMMIT_INFO, callback=commit, content_type="application/json"
-    )
 
 
 @responses.activate
@@ -68,8 +62,22 @@ def test_collect(mock_pygithub_get_repo):
     mock_release.html_url = "mock_release_url"
     mock_release.author = mock_author
 
+    # Mock the file and commit object
+    mock_file1 = mock.Mock()
+    mock_file1.filename = "file1"
+    mock_file2 = mock.Mock()
+    mock_file2.filename = "file2"
+
+    mock_commit = mock.Mock()
+    mock_commit.files = [mock_file1, mock_file2]
+    mock_commit.commit.committer.date = now.isoformat()
+    mock_parent = mock.Mock()
+    mock_parent.sha = "mock_parent_sha"
+    mock_commit.parents = [mock_parent]
+
     mock_repo = mock.Mock()
     mock_repo.get_releases.return_value = [mock_release]
+    mock_repo.get_commit.return_value = mock_commit
     mock_pygithub_get_repo.return_value = mock_repo
 
     prepare_responses()

tests/changelog/test_tasks.py

@@ -25,8 +25,21 @@ def test_update_changelog(mock_pygithub_get_repo):
     mock_release.html_url = "mock_release_url"
     mock_release.author = mock_author
 
+    mock_file1 = mock.Mock()
+    mock_file1.filename = "file1"
+    mock_file2 = mock.Mock()
+    mock_file2.filename = "file2"
+
+    mock_commit = mock.Mock()
+    mock_commit.files = [mock_file1, mock_file2]
+    mock_commit.commit.committer.date = now.isoformat()
+    mock_parent = mock.Mock()
+    mock_parent.sha = "mock_parent_sha"
+    mock_commit.parents = [mock_parent]
+
     mock_repo = mock.Mock()
     mock_repo.get_releases.return_value = [mock_release]
+    mock_repo.get_commit.return_value = mock_commit
     mock_pygithub_get_repo.return_value = mock_repo
 
     prepare_responses()

tests/utils/test_github.py

@@ -1,10 +1,61 @@
 from datetime import UTC, datetime
 from unittest.mock import patch
 
+import pytest
+
 # Import the function to be tested
 from treeherder.utils.github import get_releases
 
 
+# Mock GitCommit and it's related classes
+class MockCommitParent:
+    def __init__(self, sha):
+        self.sha = sha
+
+
+class MockCommitFile:
+    def __init__(self, filename):
+        self.filename = filename
+
+
+class MockCommitter:
+    def __init__(self, date):
+        self.date = date
+
+
+class MockInnerCommit:
+    def __init__(self, committer_date):
+        self.committer = MockCommitter(committer_date)
+
+
+class MockCommit:
+    def __init__(self, sha, committer_date, parents=None, files=None):
+        self.sha = sha
+        self.commit = MockInnerCommit(committer_date)
+        self.parents = [MockCommitParent(p_sha) for p_sha in parents] if parents else []
+        self.files = [MockCommitFile(f_name) for f_name in files] if files else []
+
+
+@pytest.fixture
+def github_commit_mock():
+    """
+    A factory fixture that patches the github object, sets up a MockRepository,
+    and returns a helper function to easily register commits.
+    """
+    with patch("treeherder.utils.github.github") as mock_github:
+        mock_repo = MockRepository()
+        mock_github.get_repo.return_value = mock_repo
+
+        def _register(sha, committer_date, parents=None, files=None):
+            commit_obj = MockCommit(
+                sha=sha, committer_date=committer_date, parents=parents, files=files
+            )
+            mock_repo._commits[sha] = commit_obj
+            return mock_github, mock_repo, commit_obj
+
+        yield _register
+
+
 # Helper for MockGitRelease
 class MockAuthor:
     def __init__(self, login):
@@ -63,15 +114,19 @@ def __repr__(self):
 
 # Mock Repository class to simulate PyGithub's Repository objects
 class MockRepository:
-    def __init__(self, releases):
-        self._releases = releases
+    def __init__(self, releases=None, commits=None):
+        self._releases = releases or []
+        self._commits = commits or {}
 
     def get_releases(self):
         # PyGithub's get_releases returns an iterable (PaginatedList),
         # with releases in reverse chronological order.
         # Returning a list directly simulates this behavior for the mock.
         return self._releases
 
+    def get_commit(self, sha):
+        return self._commits[sha]
+
 
 @patch("treeherder.utils.github.github")
 def test_get_releases_no_params(mock_github):
@@ -293,3 +348,86 @@ def test_get_releases_with_number_and_since_params(mock_github):
     ]
     assert len(result_s3) == 3
     assert result_s3 == expected_s3
+
+
+def test_get_commit_standard(github_commit_mock):
+    """
+    Test get_commit returns a dictionary representing a standard commit with files, parents, and committer date.
+    """
+    owner = "test-owner"
+    repo = "test-repo"
+    sha = "abc123commitsha"
+    date_str = "2023-01-01T12:00:00Z"
+
+    mock_github, _, _ = github_commit_mock(
+        sha=sha,
+        committer_date=date_str,
+        parents=["parentsha1", "parentsha2"],
+        files=["file1.py", "file2.py"],
+    )
+
+    from treeherder.utils.github import get_commit
+
+    result = get_commit(owner, repo, sha)
+
+    # Assertions
+    mock_github.get_repo.assert_called_once_with(f"{owner}/{repo}")
+    assert result == {
+        "files": [{"filename": "file1.py"}, {"filename": "file2.py"}],
+        "commit": {"committer": {"date": date_str}},
+        "parents": [{"sha": "parentsha1"}, {"sha": "parentsha2"}],
+    }
+
+
+def test_get_commit_initial_commit(github_commit_mock):
+    """
+    Test get_commit handles an initial/root commit with no parents.
+    """
+    owner = "test-owner"
+    repo = "test-repo"
+    sha = "initialcommitsha"
+    date_str = "2023-01-01T00:00:00Z"
+
+    github_commit_mock(
+        sha=sha,
+        committer_date=date_str,
+        parents=[],
+        files=["README.md"],
+    )
+
+    from treeherder.utils.github import get_commit
+
+    result = get_commit(owner, repo, sha)
+
+    assert result == {
+        "files": [{"filename": "README.md"}],
+        "commit": {"committer": {"date": date_str}},
+        "parents": [],
+    }
+
+
+def test_get_commit_no_files(github_commit_mock):
+    """
+    Test get_commit handles a commit with no files changed.
+    """
+    owner = "test-owner"
+    repo = "test-repo"
+    sha = "nofilescommitsha"
+    date_str = "2023-01-02T10:00:00Z"
+
+    github_commit_mock(
+        sha=sha,
+        committer_date=date_str,
+        parents=["parentsha"],
+        files=[],
+    )
+
+    from treeherder.utils.github import get_commit
+
+    result = get_commit(owner, repo, sha)
+
+    assert result == {
+        "files": [],
+        "commit": {"committer": {"date": date_str}},
+        "parents": [{"sha": "parentsha"}],
+    }

treeherder/utils/github.py

@@ -89,7 +89,30 @@ def get_all_commits(owner, repo, params=None):
 
 
 def get_commit(owner, repo, sha, params=None):
-    return fetch_api(f"repos/{owner}/{repo}/commits/{sha}", params)
+    """
+    Retrieve GitHub commit for a given sha.
+    Returns a standardized dictionary representing a commit.
+    """
+    repo_object = pygithub_get_repo(owner, repo)
+    commit = repo_object.get_commit(sha)
+    # Create a commit dict to be returned
+    commit_dict = {}
+
+    # Append file objects required by collector.py
+    commit_dict["files"] = []
+    for file in commit.files:
+        f = {}
+        f["filename"] = file.filename
+        commit_dict["files"].append(f)
+
+    # Append object required by ingest.py
+    ## Add committer date
+    commit_dict["commit"] = {"committer": {"date": commit.commit.committer.date}}
+    ## Add parent sha's
+    commit_dict["parents"] = []
+    for parent in commit.parents:
+        commit_dict["parents"].append({"sha": parent.sha})
+    return commit_dict
 
 
 def get_pull_request(owner, repo, pr_id):