SIENTIAPDE-1350: Add unit tests for the Cleanup activity, ensuring 100% code coverage, and update todo-list.

This commit is contained in:
Bruno Domingues
2025-11-25 16:01:10 -03:00
parent f3c88885ea
commit d2c4803a75
2 changed files with 555 additions and 0 deletions

View File

@@ -0,0 +1,554 @@
"""Unit tests for the Cleanup activity, ensuring 100% code coverage."""
import asyncio
import os
import shutil
import tempfile
from datetime import UTC, datetime, timedelta
from importlib import reload
from unittest.mock import AsyncMock, MagicMock, patch
import pytest
# Define mocks at the top level to be accessible by all tests
@pytest.fixture
def mock_logger():
"""Fixture for a mock logger."""
return MagicMock()
@pytest.fixture
def mock_notification_handler():
"""Fixture for a mock notification handler."""
return MagicMock()
@pytest.fixture
def mock_metrics_controller():
"""Fixture for a mock metrics controller with async methods."""
controller = MagicMock()
controller.shutdown = AsyncMock()
controller.emit = AsyncMock()
return controller
@pytest.fixture
def mock_storage_repository():
"""Fixture for a mock storage repository."""
repo = MagicMock()
repo.delete_file = MagicMock()
repo.list_bucket_objects = MagicMock()
return repo
@pytest.fixture
def temp_dir():
"""Fixture to create and clean up a temporary directory."""
path = tempfile.mkdtemp()
yield path
shutil.rmtree(path)
# --- Initialization Tests ---
def test_cleanup_init_default_values(
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test Cleanup initialization uses default environment values."""
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
assert cleanup.retention_hours == 24
assert cleanup.dry_run is False
assert cleanup.max_keys_cleanup == 1000
def test_cleanup_init_custom_env_values(
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test Cleanup initialization with custom environment values."""
with patch.dict(
os.environ,
{
'CLEANUP_RETENTION_HOURS': '48',
'CLEANUP_DRY_RUN': 'true',
'MAX_KEYS_CLEANUP': '500',
},
):
import model_manager.activities.cleanup
reload(model_manager.activities.cleanup)
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
assert cleanup.retention_hours == 48
assert cleanup.dry_run is True
assert cleanup.max_keys_cleanup == 500
@patch.dict(os.environ, {'CLEANUP_RETENTION_HOURS': 'invalid'})
def test_cleanup_init_invalid_env_value_raises_error():
"""Test Cleanup module raises ValueError for invalid environment variables on import."""
import model_manager.activities.cleanup
with pytest.raises(ValueError):
reload(model_manager.activities.cleanup)
# --- MinIO Cleanup Tests ---
def test_cleanup_minio_files_missing_bucket_name(
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test cleanup_minio_files raises ValueError if bucket_name is missing."""
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
with pytest.raises(ValueError, match='bucket_name must be provided'):
asyncio.run(cleanup.cleanup_minio_files({'metadata': {}}))
def test_cleanup_minio_files_success_with_deletions(
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test successful deletion of old files from MinIO."""
import model_manager.activities.cleanup
reload(model_manager.activities.cleanup)
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
old_ts = int((datetime.now(UTC) - timedelta(hours=48)).timestamp() * 1000)
recent_ts = int((datetime.now(UTC) - timedelta(hours=1)).timestamp() * 1000)
mock_storage_repository.list_bucket_objects.return_value = [
f'{old_ts}-old-file.txt',
f'{recent_ts}-recent-file.txt',
'no-timestamp-file.txt',
]
asyncio.run(cleanup.cleanup_minio_files({'bucket_name': 'test-bucket', 'metadata': {}}))
mock_storage_repository.delete_file.assert_called_once_with(
'test-bucket', f'{old_ts}-old-file.txt'
)
cleanup._emit_metrics.assert_called_once()
def test_cleanup_minio_files_dry_run(
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test MinIO cleanup in dry_run mode does not delete files."""
with patch.dict(os.environ, {'CLEANUP_DRY_RUN': 'true'}):
import model_manager.activities.cleanup
reload(model_manager.activities.cleanup)
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
old_ts = int((datetime.now(UTC) - timedelta(hours=48)).timestamp() * 1000)
mock_storage_repository.list_bucket_objects.return_value = [f'{old_ts}-old-file.txt']
asyncio.run(cleanup.cleanup_minio_files({'bucket_name': 'test-bucket', 'metadata': {}}))
mock_storage_repository.delete_file.assert_not_called()
cleanup._emit_metrics.assert_called_once()
def test_cleanup_minio_files_delete_error(
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test error during MinIO file deletion is handled gracefully."""
import model_manager.activities.cleanup
reload(model_manager.activities.cleanup)
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
cleanup.error = AsyncMock()
old_ts = int((datetime.now(UTC) - timedelta(hours=48)).timestamp() * 1000)
mock_storage_repository.list_bucket_objects.return_value = [f'{old_ts}-old-file.txt']
mock_storage_repository.delete_file.side_effect = OSError('Permission Denied')
asyncio.run(cleanup.cleanup_minio_files({'bucket_name': 'test-bucket', 'metadata': {}}))
cleanup.error.assert_called_once()
cleanup._emit_metrics.assert_called_once()
def test_cleanup_minio_files_exception_handling(
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test exception during MinIO cleanup triggers notification and metrics."""
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup.send_notification = AsyncMock()
cleanup._emit_metrics = AsyncMock()
mock_storage_repository.list_bucket_objects.side_effect = Exception('Connection Error')
with pytest.raises(Exception, match='Connection Error'):
asyncio.run(cleanup.cleanup_minio_files({'bucket_name': 'test-bucket', 'metadata': {}}))
cleanup.send_notification.assert_called_once()
cleanup._emit_metrics.assert_called_once()
# --- Temp Directory Cleanup Tests ---
def test_cleanup_temp_directories_nonexistent_path(
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test temp directory cleanup with a non-existent path."""
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
cleanup.warning = AsyncMock()
asyncio.run(
cleanup.cleanup_temp_directories({'temp_path': '/nonexistent/path', 'metadata': {}})
)
cleanup.warning.assert_called_once()
cleanup._emit_metrics.assert_called_once()
def test_cleanup_temp_directories_success_with_deletions(
temp_dir,
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test successful deletion of old temporary directories."""
import model_manager.activities.cleanup
reload(model_manager.activities.cleanup)
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
old_time = (datetime.now() - timedelta(hours=48)).strftime('%Y%m%d_%H%M%S_000000')
old_dir = os.path.join(temp_dir, f'old_dir_{old_time}')
os.makedirs(old_dir)
recent_time = (datetime.now() - timedelta(hours=1)).strftime('%Y%m%d_%H%M%S_000000')
recent_dir = os.path.join(temp_dir, f'recent_dir_{recent_time}')
os.makedirs(recent_dir)
asyncio.run(cleanup.cleanup_temp_directories({'temp_path': temp_dir, 'metadata': {}}))
assert not os.path.exists(old_dir)
assert os.path.exists(recent_dir)
cleanup._emit_metrics.assert_called_once()
@patch.dict(os.environ, {'CLEANUP_DRY_RUN': 'true'})
def test_cleanup_temp_directories_dry_run(
temp_dir,
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test temp directory cleanup in dry_run mode does not delete."""
import model_manager.activities.cleanup
reload(model_manager.activities.cleanup)
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
old_time = (datetime.now() - timedelta(hours=48)).strftime('%Y%m%d_%H%M%S_000000')
old_dir = os.path.join(temp_dir, f'old_dir_{old_time}')
os.makedirs(old_dir)
asyncio.run(cleanup.cleanup_temp_directories({'temp_path': temp_dir, 'metadata': {}}))
assert os.path.exists(old_dir)
cleanup._emit_metrics.assert_called_once()
def test_cleanup_temp_directories_delete_error(
temp_dir,
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test graceful handling of errors during directory deletion."""
import model_manager.activities.cleanup
reload(model_manager.activities.cleanup)
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
cleanup.error = AsyncMock()
old_time = (datetime.now() - timedelta(hours=48)).strftime('%Y%m%d_%H%M%S_000000')
old_dir = os.path.join(temp_dir, f'old_dir_{old_time}')
os.makedirs(old_dir)
with patch('shutil.rmtree', side_effect=OSError('Permission Denied')):
asyncio.run(cleanup.cleanup_temp_directories({'temp_path': temp_dir, 'metadata': {}}))
cleanup.error.assert_called_once()
cleanup._emit_metrics.assert_called_once()
# --- Metrics and Utility Tests ---
def test_emit_metrics(
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test that _emit_metrics calls the public emit_metric method."""
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup.emit_metric = AsyncMock()
asyncio.run(
cleanup._emit_metrics(
metadata={'pod_id': 'p1', 'workflow_name': 'wf1'},
metrics_status='success',
activity_name='test_activity',
emit_workflow_metric=True,
)
)
assert cleanup.emit_metric.call_count == 2
def test_cleanup_temp_directories_with_files_and_unmatched_dirs(
temp_dir,
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test that files and directories with non-matching names are skipped."""
import model_manager.activities.cleanup
reload(model_manager.activities.cleanup)
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
cleanup.debug = AsyncMock()
# Create a file and a directory with a non-matching name
with open(os.path.join(temp_dir, 'a_file.txt'), 'w') as f:
f.write('hello')
os.makedirs(os.path.join(temp_dir, 'a_directory_with_no_timestamp'))
asyncio.run(cleanup.cleanup_temp_directories({'temp_path': temp_dir, 'metadata': {}}))
# Ensure the debug message for skipping was called for the unmatched directory
cleanup.debug.assert_called_with(
'Skipping directory without timestamp pattern: a_directory_with_no_timestamp', {}
)
cleanup._emit_metrics.assert_called_once()
def test_cleanup_temp_directories_invalid_timestamp_format(
temp_dir,
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test that a directory with an invalid timestamp format is handled correctly."""
import model_manager.activities.cleanup
reload(model_manager.activities.cleanup)
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
cleanup.error = AsyncMock()
# Create a directory with a malformed timestamp that matches the regex but fails parsing
malformed_dir_name = 'dir_20239999_999999_999999'
os.makedirs(os.path.join(temp_dir, malformed_dir_name))
asyncio.run(cleanup.cleanup_temp_directories({'temp_path': temp_dir, 'metadata': {}}))
cleanup.error.assert_called_once()
cleanup._emit_metrics.assert_called_once()
def test_cleanup_temp_directories_generic_exception(
temp_dir,
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test that a generic exception during directory cleanup is handled."""
import model_manager.activities.cleanup
reload(model_manager.activities.cleanup)
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup._emit_metrics = AsyncMock()
cleanup.send_notification = AsyncMock()
with patch('os.listdir', side_effect=Exception('Unexpected OS Error')):
with pytest.raises(Exception, match='Unexpected OS Error'):
asyncio.run(cleanup.cleanup_temp_directories({'temp_path': temp_dir, 'metadata': {}}))
cleanup.send_notification.assert_called_once()
cleanup._emit_metrics.assert_called_once()
def test_emit_metrics_activity_only(
mock_storage_repository,
mock_logger,
mock_notification_handler,
mock_metrics_controller,
):
"""Test that _emit_metrics can emit only the activity metric."""
from model_manager.activities.cleanup import Cleanup
cleanup = Cleanup(
storage_repository=mock_storage_repository,
logger=mock_logger,
notification_handler=mock_notification_handler,
metrics_controller=mock_metrics_controller,
)
cleanup.emit_metric = AsyncMock()
asyncio.run(
cleanup._emit_metrics(
metadata={'pod_id': 'p1', 'workflow_name': 'wf1'},
metrics_status='success',
activity_name='test_activity',
emit_workflow_metric=False,
)
)
cleanup.emit_metric.assert_called_once()

View File

@@ -1,3 +1,4 @@
- Adcionar configuração do cron job ao helm chart
- Refatorar o arquivo .dockerignore para só deixar copiar os arquivos que forem necessários para a execução do container, pois ele está copiando muitos arquivos desnecessários.
- atualizar as variáveis de ambiente no helm chart
- Criar um gráfico no grafana para cada nova atividade.