Added JobRepository and JobServices

2025-09-21 19:11:57 +02:00
parent 9564cfadd5
commit 98c43feadf
8 changed files with 1616 additions and 1 deletions
--- a/tests/repositories/test_job_repository.py
+++ b/tests/repositories/test_job_repository.py
@@ -0,0 +1,523 @@
+"""
+Test suite for JobRepository with async/await support.
+
+This module contains comprehensive tests for all JobRepository methods
+using mongomock-motor for in-memory MongoDB testing.
+"""
+
+from datetime import datetime
+
+import pytest
+import pytest_asyncio
+from bson import ObjectId
+from mongomock_motor import AsyncMongoMockClient
+from pymongo.errors import PyMongoError
+
+from app.database.repositories.job_repository import JobRepository
+from app.exceptions.job_exceptions import JobRepositoryError
+from app.models.job import ProcessingJob, ProcessingStatus
+from app.models.types import PyObjectId
+
+
+@pytest_asyncio.fixture
+async def in_memory_repository():
+  """Create an in-memory JobRepository for testing."""
+  client = AsyncMongoMockClient()
+  db = client.test_database
+  repo = JobRepository(db)
+  await repo.initialize()
+  return repo
+
+
+@pytest.fixture
+def sample_document_id():
+  """Sample document ObjectId for testing."""
+  return PyObjectId()
+
+
+@pytest.fixture
+def sample_task_id():
+  """Sample Celery task ID for testing."""
+  return "celery-task-12345-abcde"
+
+
+@pytest.fixture
+def multiple_sample_jobs():
+  """Multiple ProcessingJob objects for testing."""
+  doc_id_1 = ObjectId()
+  doc_id_2 = ObjectId()
+  base_time = datetime.utcnow()
+  
+  return [
+      ProcessingJob(
+        document_id=doc_id_1,
+        status=ProcessingStatus.PENDING,
+        task_id="task-1",
+        created_at=base_time,
+        started_at=None,
+        completed_at=None,
+        error_message=None
+      ),
+      ProcessingJob(
+        document_id=doc_id_2,
+        status=ProcessingStatus.PROCESSING,
+        task_id="task-2",
+        created_at=base_time,
+        started_at=base_time,
+        completed_at=None,
+        error_message=None
+      ),
+      ProcessingJob(
+        document_id=doc_id_1,
+        status=ProcessingStatus.COMPLETED,
+        task_id="task-3",
+        created_at=base_time,
+        started_at=base_time,
+        completed_at=base_time,
+        error_message=None
+      )
+  ]
+
+
+class TestJobRepositoryInitialization:
+  """Tests for repository initialization."""
+  
+  @pytest.mark.asyncio
+  async def test_i_can_initialize_repository(self):
+    """Test repository initialization."""
+    # Arrange
+    client = AsyncMongoMockClient()
+    db = client.test_database
+    repo = JobRepository(db)
+    
+    # Act
+    initialized_repo = await repo.initialize()
+    
+    # Assert
+    assert initialized_repo is repo
+    assert repo.db is not None
+    assert repo.collection is not None
+
+
+class TestJobRepositoryCreation:
+  """Tests for job creation functionality."""
+  
+  @pytest.mark.asyncio
+  async def test_i_can_create_job_with_task_id(self, in_memory_repository, sample_document_id, sample_task_id):
+    """Test successful job creation with task ID."""
+    # Act
+    created_job = await in_memory_repository.create_job(sample_document_id, sample_task_id)
+    
+    # Assert
+    assert created_job is not None
+    assert created_job.document_id == sample_document_id
+    assert created_job.task_id == sample_task_id
+    assert created_job.status == ProcessingStatus.PENDING
+    assert created_job.created_at is not None
+    assert created_job.started_at is None
+    assert created_job.completed_at is None
+    assert created_job.error_message is None
+    assert created_job.id is not None
+    assert isinstance(created_job.id, ObjectId)
+  
+  @pytest.mark.asyncio
+  async def test_i_can_create_job_without_task_id(self, in_memory_repository, sample_document_id):
+    """Test successful job creation without task ID."""
+    # Act
+    created_job = await in_memory_repository.create_job(sample_document_id)
+    
+    # Assert
+    assert created_job is not None
+    assert created_job.document_id == sample_document_id
+    assert created_job.task_id is None
+    assert created_job.status == ProcessingStatus.PENDING
+    assert created_job.created_at is not None
+    assert created_job.started_at is None
+    assert created_job.completed_at is None
+    assert created_job.error_message is None
+    assert created_job.id is not None
+    assert isinstance(created_job.id, ObjectId)
+  
+  @pytest.mark.asyncio
+  async def test_i_cannot_create_duplicate_job_for_document(self, in_memory_repository, sample_document_id,
+                                                            sample_task_id):
+    """Test that creating job with duplicate document_id raises DuplicateKeyError."""
+    # Arrange
+    await in_memory_repository.create_job(sample_document_id, sample_task_id)
+    
+    # Act & Assert
+    with pytest.raises(JobRepositoryError) as exc_info:
+      await in_memory_repository.create_job(sample_document_id, "different-task-id")
+    
+    assert "create_job" in str(exc_info.value)
+  
+  @pytest.mark.asyncio
+  async def test_i_cannot_create_job_with_pymongo_error(self, in_memory_repository, sample_document_id, mocker):
+    """Test handling of PyMongo errors during job creation."""
+    # Arrange
+    mocker.patch.object(in_memory_repository.collection, 'insert_one', side_effect=PyMongoError("Database error"))
+    
+    # Act & Assert
+    with pytest.raises(JobRepositoryError) as exc_info:
+      await in_memory_repository.create_job(sample_document_id)
+    
+    assert "create_job" in str(exc_info.value)
+
+
+class TestJobRepositoryFinding:
+  """Tests for job finding functionality."""
+  
+  @pytest.mark.asyncio
+  async def test_i_can_find_job_by_valid_id(self, in_memory_repository, sample_document_id, sample_task_id):
+    """Test finding job by valid ObjectId."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id, sample_task_id)
+    
+    # Act
+    found_job = await in_memory_repository.find_job_by_id(created_job.id)
+    
+    # Assert
+    assert found_job is not None
+    assert found_job.id == created_job.id
+    assert found_job.document_id == created_job.document_id
+    assert found_job.task_id == created_job.task_id
+    assert found_job.status == created_job.status
+  
+  @pytest.mark.asyncio
+  async def test_i_cannot_find_job_by_nonexistent_id(self, in_memory_repository):
+    """Test that nonexistent ObjectId returns None."""
+    # Arrange
+    nonexistent_id = PyObjectId()
+    
+    # Act
+    found_job = await in_memory_repository.find_job_by_id(nonexistent_id)
+    
+    # Assert
+    assert found_job is None
+  
+  @pytest.mark.asyncio
+  async def test_i_cannot_find_job_with_pymongo_error(self, in_memory_repository, mocker):
+    """Test handling of PyMongo errors during job finding."""
+    # Arrange
+    mocker.patch.object(in_memory_repository.collection, 'find_one', side_effect=PyMongoError("Database error"))
+    
+    # Act & Assert
+    with pytest.raises(JobRepositoryError) as exc_info:
+      await in_memory_repository.find_job_by_id(PyObjectId())
+    
+    assert "get_job_by_id" in str(exc_info.value)
+  
+  @pytest.mark.asyncio
+  async def test_i_can_find_jobs_by_document_id(self, in_memory_repository, sample_document_id, sample_task_id):
+    """Test finding jobs by document ID."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id, sample_task_id)
+    
+    # Act
+    found_jobs = await in_memory_repository.find_jobs_by_document_id(sample_document_id)
+    
+    # Assert
+    assert len(found_jobs) == 1
+    assert found_jobs[0].id == created_job.id
+    assert found_jobs[0].document_id == sample_document_id
+  
+  @pytest.mark.asyncio
+  async def test_i_can_find_empty_jobs_list_for_nonexistent_document(self, in_memory_repository):
+    """Test that nonexistent document ID returns empty list."""
+    # Arrange
+    nonexistent_id = ObjectId()
+    
+    # Act
+    found_jobs = await in_memory_repository.find_jobs_by_document_id(nonexistent_id)
+    
+    # Assert
+    assert found_jobs == []
+  
+  @pytest.mark.asyncio
+  async def test_i_cannot_find_jobs_by_document_with_pymongo_error(self, in_memory_repository, mocker):
+    """Test handling of PyMongo errors during finding jobs by document ID."""
+    # Arrange
+    mocker.patch.object(in_memory_repository.collection, 'find', side_effect=PyMongoError("Database error"))
+    
+    # Act & Assert
+    with pytest.raises(JobRepositoryError) as exc_info:
+      await in_memory_repository.find_jobs_by_document_id(PyObjectId())
+    
+    assert "get_jobs_by_file_id" in str(exc_info.value)
+  
+  @pytest.mark.asyncio
+  @pytest.mark.parametrize("status", [
+      ProcessingStatus.PENDING,
+      ProcessingStatus.PROCESSING,
+      ProcessingStatus.COMPLETED
+  ])
+  async def test_i_can_find_jobs_by_pending_status(self, in_memory_repository, sample_document_id, status):
+    """Test finding jobs by PENDING status."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id)
+    await in_memory_repository.update_job_status(created_job.id, status)
+    
+    # Act
+    found_jobs = await in_memory_repository.get_jobs_by_status(status)
+    
+    # Assert
+    assert len(found_jobs) == 1
+    assert found_jobs[0].id == created_job.id
+    assert found_jobs[0].status == status
+  
+  @pytest.mark.asyncio
+  async def test_i_can_find_jobs_by_failed_status(self, in_memory_repository, sample_document_id):
+    """Test finding jobs by FAILED status."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id)
+    await in_memory_repository.update_job_status(created_job.id, ProcessingStatus.FAILED, "Test error")
+    
+    # Act
+    found_jobs = await in_memory_repository.get_jobs_by_status(ProcessingStatus.FAILED)
+    
+    # Assert
+    assert len(found_jobs) == 1
+    assert found_jobs[0].id == created_job.id
+    assert found_jobs[0].status == ProcessingStatus.FAILED
+    assert found_jobs[0].error_message == "Test error"
+  
+  @pytest.mark.asyncio
+  async def test_i_can_find_empty_jobs_list_for_unused_status(self, in_memory_repository):
+    """Test that unused status returns empty list."""
+    # Act
+    found_jobs = await in_memory_repository.get_jobs_by_status(ProcessingStatus.COMPLETED)
+    
+    # Assert
+    assert found_jobs == []
+  
+  @pytest.mark.asyncio
+  async def test_i_cannot_find_jobs_by_status_with_pymongo_error(self, in_memory_repository, mocker):
+    """Test handling of PyMongo errors during finding jobs by status."""
+    # Arrange
+    mocker.patch.object(in_memory_repository.collection, 'find', side_effect=PyMongoError("Database error"))
+    
+    # Act & Assert
+    with pytest.raises(JobRepositoryError) as exc_info:
+      await in_memory_repository.get_jobs_by_status(ProcessingStatus.PENDING)
+    
+    assert "get_jobs_by_status" in str(exc_info.value)
+
+
+class TestJobRepositoryStatusUpdate:
+  """Tests for job status update functionality."""
+  
+  @pytest.mark.asyncio
+  async def test_i_can_update_job_status_to_processing(self, in_memory_repository, sample_document_id):
+    """Test updating job status to PROCESSING with started_at timestamp."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id)
+    
+    # Act
+    updated_job = await in_memory_repository.update_job_status(created_job.id, ProcessingStatus.PROCESSING)
+    
+    # Assert
+    assert updated_job is not None
+    assert updated_job.id == created_job.id
+    assert updated_job.status == ProcessingStatus.PROCESSING
+    assert updated_job.started_at is not None
+    assert updated_job.completed_at is None
+    assert updated_job.error_message is None
+  
+  @pytest.mark.asyncio
+  async def test_i_can_update_job_status_to_completed(self, in_memory_repository, sample_document_id):
+    """Test updating job status to COMPLETED with completed_at timestamp."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id)
+    await in_memory_repository.update_job_status(created_job.id, ProcessingStatus.PROCESSING)
+    
+    # Act
+    updated_job = await in_memory_repository.update_job_status(created_job.id, ProcessingStatus.COMPLETED)
+    
+    # Assert
+    assert updated_job is not None
+    assert updated_job.id == created_job.id
+    assert updated_job.status == ProcessingStatus.COMPLETED
+    assert updated_job.started_at is not None
+    assert updated_job.completed_at is not None
+    assert updated_job.error_message is None
+  
+  @pytest.mark.asyncio
+  async def test_i_can_update_job_status_to_failed_with_error(self, in_memory_repository, sample_document_id):
+    """Test updating job status to FAILED with error message and completed_at timestamp."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id)
+    error_message = "Processing failed due to invalid format"
+    
+    # Act
+    updated_job = await in_memory_repository.update_job_status(
+      created_job.id, ProcessingStatus.FAILED, error_message
+    )
+    
+    # Assert
+    assert updated_job is not None
+    assert updated_job.id == created_job.id
+    assert updated_job.status == ProcessingStatus.FAILED
+    assert updated_job.completed_at is not None
+    assert updated_job.error_message == error_message
+  
+  @pytest.mark.asyncio
+  async def test_i_can_update_job_status_to_failed_without_error(self, in_memory_repository, sample_document_id):
+    """Test updating job status to FAILED without error message."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id)
+    
+    # Act
+    updated_job = await in_memory_repository.update_job_status(created_job.id, ProcessingStatus.FAILED)
+    
+    # Assert
+    assert updated_job is not None
+    assert updated_job.id == created_job.id
+    assert updated_job.status == ProcessingStatus.FAILED
+    assert updated_job.completed_at is not None
+    assert updated_job.error_message is None
+  
+  @pytest.mark.asyncio
+  async def test_i_cannot_update_nonexistent_job_status(self, in_memory_repository):
+    """Test that updating nonexistent job returns None."""
+    # Arrange
+    nonexistent_id = ObjectId()
+    
+    # Act
+    result = await in_memory_repository.update_job_status(nonexistent_id, ProcessingStatus.COMPLETED)
+    
+    # Assert
+    assert result is None
+  
+  @pytest.mark.asyncio
+  async def test_i_cannot_update_job_status_with_pymongo_error(self, in_memory_repository, sample_document_id, mocker):
+    """Test handling of PyMongo errors during job status update."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id)
+    mocker.patch.object(in_memory_repository.collection, 'find_one_and_update',
+                        side_effect=PyMongoError("Database error"))
+    
+    # Act & Assert
+    with pytest.raises(JobRepositoryError) as exc_info:
+      await in_memory_repository.update_job_status(created_job.id, ProcessingStatus.COMPLETED)
+    
+    assert "update_job_status" in str(exc_info.value)
+
+
+class TestJobRepositoryDeletion:
+  """Tests for job deletion functionality."""
+  
+  @pytest.mark.asyncio
+  async def test_i_can_delete_existing_job(self, in_memory_repository, sample_document_id):
+    """Test successful job deletion."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id)
+    
+    # Act
+    deletion_result = await in_memory_repository.delete_job(created_job.id)
+    
+    # Assert
+    assert deletion_result is True
+    
+    # Verify job is actually deleted
+    found_job = await in_memory_repository.find_job_by_id(created_job.id)
+    assert found_job is None
+  
+  @pytest.mark.asyncio
+  async def test_i_cannot_delete_nonexistent_job(self, in_memory_repository):
+    """Test that deleting nonexistent job returns False."""
+    # Arrange
+    nonexistent_id = ObjectId()
+    
+    # Act
+    result = await in_memory_repository.delete_job(nonexistent_id)
+    
+    # Assert
+    assert result is False
+  
+  @pytest.mark.asyncio
+  async def test_i_cannot_delete_job_with_pymongo_error(self, in_memory_repository, sample_document_id, mocker):
+    """Test handling of PyMongo errors during job deletion."""
+    # Arrange
+    created_job = await in_memory_repository.create_job(sample_document_id)
+    mocker.patch.object(in_memory_repository.collection, 'delete_one', side_effect=PyMongoError("Database error"))
+    
+    # Act & Assert
+    with pytest.raises(JobRepositoryError) as exc_info:
+      await in_memory_repository.delete_job(created_job.id)
+    
+    assert "delete_job" in str(exc_info.value)
+
+
+class TestJobRepositoryComplexScenarios:
+  """Tests for complex job repository scenarios."""
+  
+  @pytest.mark.asyncio
+  async def test_i_can_handle_complete_job_lifecycle(self, in_memory_repository, sample_document_id, sample_task_id):
+    """Test complete job lifecycle from creation to completion."""
+    # Create job
+    job = await in_memory_repository.create_job(sample_document_id, sample_task_id)
+    assert job.status == ProcessingStatus.PENDING
+    assert job.started_at is None
+    assert job.completed_at is None
+    
+    # Start processing
+    job = await in_memory_repository.update_job_status(job.id, ProcessingStatus.PROCESSING)
+    assert job.status == ProcessingStatus.PROCESSING
+    assert job.started_at is not None
+    assert job.completed_at is None
+    
+    # Complete job
+    job = await in_memory_repository.update_job_status(job.id, ProcessingStatus.COMPLETED)
+    assert job.status == ProcessingStatus.COMPLETED
+    assert job.started_at is not None
+    assert job.completed_at is not None
+    assert job.error_message is None
+  
+  @pytest.mark.asyncio
+  async def test_i_can_handle_job_failure_scenario(self, in_memory_repository, sample_document_id, sample_task_id):
+    """Test job failure scenario with error message."""
+    # Create and start job
+    job = await in_memory_repository.create_job(sample_document_id, sample_task_id)
+    job = await in_memory_repository.update_job_status(job.id, ProcessingStatus.PROCESSING)
+    
+    # Fail job with error
+    error_msg = "File format not supported"
+    job = await in_memory_repository.update_job_status(job.id, ProcessingStatus.FAILED, error_msg)
+    
+    # Assert failure state
+    assert job.status == ProcessingStatus.FAILED
+    assert job.started_at is not None
+    assert job.completed_at is not None
+    assert job.error_message == error_msg
+  
+  @pytest.mark.asyncio
+  async def test_i_can_handle_multiple_documents_with_different_statuses(self, in_memory_repository):
+    """Test managing multiple jobs for different documents with various statuses."""
+    # Create jobs for different documents
+    doc1 = PyObjectId()
+    doc2 = PyObjectId()
+    doc3 = PyObjectId()
+    
+    job1 = await in_memory_repository.create_job(doc1, "task-1")
+    job2 = await in_memory_repository.create_job(doc2, "task-2")
+    job3 = await in_memory_repository.create_job(doc3, "task-3")
+    
+    # Update to different statuses
+    await in_memory_repository.update_job_status(job1.id, ProcessingStatus.PROCESSING)
+    await in_memory_repository.update_job_status(job2.id, ProcessingStatus.COMPLETED)
+    await in_memory_repository.update_job_status(job3.id, ProcessingStatus.FAILED, "Error occurred")
+    
+    # Verify status queries
+    pending_jobs = await in_memory_repository.get_jobs_by_status(ProcessingStatus.PENDING)
+    processing_jobs = await in_memory_repository.get_jobs_by_status(ProcessingStatus.PROCESSING)
+    completed_jobs = await in_memory_repository.get_jobs_by_status(ProcessingStatus.COMPLETED)
+    failed_jobs = await in_memory_repository.get_jobs_by_status(ProcessingStatus.FAILED)
+    
+    assert len(pending_jobs) == 0
+    assert len(processing_jobs) == 1
+    assert len(completed_jobs) == 1
+    assert len(failed_jobs) == 1
+    
+    assert processing_jobs[0].id == job1.id
+    assert completed_jobs[0].id == job2.id
+    assert failed_jobs[0].id == job3.id