[OJ-55106] Handle NotFoundException for deleted repos in BBS ingestion

jf_agent/git/bitbucket_server.py

@@ -221,7 +221,11 @@ def get_repos(client, api_projects, include_repos, exclude_repos, redact_names_a
         for repo in project.repos.list():
             if all(filt(repo) for filt in filters):
                 api_repo = project.repos[repo['name']]
-                yield api_repo, _standardize_repo(api_project, api_repo, redact_names_and_urls)
+                try:
+                    yield api_repo, _standardize_repo(api_project, api_repo, redact_names_and_urls)
+                except stashy.errors.NotFoundException as e:
+                    logger.warning(f'WARN: Got NotFoundException fetching repo, skipping: {e}')
+                    continue
 
     logger.info('✓')
 
@@ -257,7 +261,11 @@ def get_commits_for_included_branches(
 ):
     for i, api_repo in enumerate(api_repos, start=1):
         with logging_helper.log_loop_iters('repo for branch commits', i, 1):
-            repo = api_repo.get()
+            try:
+                repo = api_repo.get()
+            except stashy.errors.NotFoundException as e:
+                logger.warning(f'WARN: Got NotFoundException fetching repo, skipping: {e}')
+                continue
             if verbose:
                 logger.info(f"Beginning download of commits for repo {repo}")
             api_project = client.projects[repo['project']['key']]
@@ -270,7 +278,7 @@ def get_commits_for_included_branches(
             # We are working with the BBS api object rather than a StandardizedRepository here,
             # so we can not use get_branches_for_standardized_repo  as we do in bitbucket_cloud_adapter and gitlab_adapter.
             branches_to_process = [_get_default_branch_name(api_repo)]
-            additional_branch_patterns = included_branches.get(api_repo.get()['name'])
+            additional_branch_patterns = included_branches.get(repo['name'])
 
             if additional_branch_patterns:
                 repo_branches = [b['displayId'] for b in api_repo.branches()]
@@ -333,7 +341,11 @@ def get_pull_requests(
 ):
     for i, api_repo in enumerate(api_repos, start=1):
         with logging_helper.log_loop_iters('repo for pull requests', i, 1):
-            repo = api_repo.get()
+            try:
+                repo = api_repo.get()
+            except stashy.errors.NotFoundException as e:
+                logger.warning(f'WARN: Got NotFoundException fetching repo, skipping: {e}')
+                continue
             if verbose:
                 logger.info(f"Beginning download of PRs for repo {repo}")
             api_project = client.projects[repo['project']['key']]
@@ -369,7 +381,7 @@ def get_pull_requests(
                     additions, deletions, changed_files = None, None, None
                 except RetryError:
                     logger.warning(
-                        f"Could not retrieve diff data for PR {pr['id']} in repo {api_repo.get()['name']}"
+                        f"Could not retrieve diff data for PR {pr['id']} in repo {repo['name']}"
                     )
                     additions, deletions, changed_files = None, None, None
                 except ChunkedEncodingError as e:

tests/test_bitbucket_server.py

@@ -1,9 +1,10 @@
 import json
 import unittest
-
 from unittest import TestCase
 from unittest.mock import MagicMock
 
+import stashy.errors
+
 from jf_agent.git import bitbucket_server
 
 TEST_INPUT_FILE_PATH = f'tests/test_data/bitbucket_server/'
@@ -324,6 +325,109 @@
             "resulting PR body does not match input",
         )
 
+    def test_get_repos_skips_deleted_repo(self):
+        # Arrange
+        test_projects = _get_test_data('test_projects.json')
+        test_repos = _get_test_data('test_repos.json')
+        test_branches = _get_test_data('test_branches.json')
+
+        mock_client = MagicMock()
+        mock_project = MagicMock()
+
+        mock_deleted_repo = MagicMock()
+        mock_deleted_repo.get.side_effect = stashy.errors.NotFoundException(MagicMock())
+
+        mock_live_repo = MagicMock()
+        mock_live_repo.get.return_value = test_repos[0]
+        mock_live_repo.branches.return_value = test_branches
+        mock_live_repo.default_branch = MagicMock()
+
+        mock_client.projects = {'test_project_key': mock_project}
+        mock_project.repos.list.return_value = [test_repos[0], test_repos[0]]
+        mock_project.repos.__getitem__.side_effect = [mock_deleted_repo, mock_live_repo]
+
+        # Act — should not raise despite first repo being deleted
+        result_repos = list(bitbucket_server.get_repos(mock_client, test_projects, {}, {}, False))
+
+        # Assert — deleted repo skipped, live repo still returned
+        self.assertEqual(len(result_repos), 1)
+        self.assertEqual(result_repos[0][0], mock_live_repo)
+
+    def test_get_branch_commits_skips_deleted_repo(self):
+        # Arrange
+        test_repos = _get_test_data('test_repos.json')
+        test_branches = _get_test_data('test_branches.json')
+        test_commits = _get_test_data('test_commits.json')
+
+        mock_client = MagicMock()
+        mock_project = MagicMock()
+
+        mock_deleted_repo = MagicMock()
+        mock_deleted_repo.get.side_effect = stashy.errors.NotFoundException(MagicMock())
+
+        mock_live_repo = MagicMock()
+        mock_live_repo.get.return_value = test_repos[0]
+        mock_live_repo.default_branch = test_branches[0]
+
+        mock_client.projects = {'test_project_key': mock_project}
+        mock_project.repos = {'test_repo_name': mock_live_repo}
+        mock_live_repo.commits.return_value = test_commits
+
+        test_git_instance_info = {'pull_from': '1900-07-23', 'repos_dict_v2': {}}
+
+        # Act — should not raise despite the first repo being deleted
+        result_commits = list(
+            bitbucket_server.get_commits_for_included_branches(
+                mock_client,
+                [mock_deleted_repo, mock_live_repo],
+                {},
+                False,
+                test_git_instance_info,
+                False,
+                False,
+            )
+        )
+
+        # Assert — deleted repo produced nothing, live repo's commits still returned
+        self.assertEqual(len(result_commits), len(test_commits))
+
+    def test_get_pull_requests_skips_deleted_repo(self):
+        # Arrange
+        test_prs = _get_test_data('test_prs.json')
+        test_repos = _get_test_data('test_repos.json')
+        test_commits = _get_test_data('test_commits.json')
+
+        mock_client = MagicMock()
+        mock_project = MagicMock()
+
+        mock_deleted_repo = MagicMock()
+        mock_deleted_repo.get.side_effect = stashy.errors.NotFoundException(MagicMock())
+
+        mock_live_repo = MagicMock()
+        mock_live_repo.get.return_value = test_repos[0]
+        mock_live_repo.pull_requests.all.return_value = test_prs
+
+        mock_client.projects = {'test_project_key': mock_project}
+        mock_project.repos = {'test_repo_name': mock_live_repo}
+        mock_live_repo.commits.return_value = test_commits
+
+        test_git_instance_info = {'pull_from': '1900-07-23', 'repos_dict_v2': {}}
+
+        # Act — should not raise despite the first repo being deleted
+        result_prs = list(
+            bitbucket_server.get_pull_requests(
+                mock_client,
+                [mock_deleted_repo, mock_live_repo],
+                False,
+                test_git_instance_info,
+                False,
+                False,
+            )
+        )
+
+        # Assert — deleted repo produced nothing, live repo's PRs still returned
+        self.assertEqual(len(result_prs), len(test_prs))
+
 
 def _get_test_data(file_name):
     with open(f'{TEST_INPUT_FILE_PATH}{file_name}', 'r') as f: