Implemented default pipeline

2025-09-26 22:08:39 +02:00
parent f1b551d243
commit 4de732b0ae
56 changed files with 4534 additions and 2837 deletions
--- a/tests/services/test_job_service.py
+++ b/tests/services/test_job_service.py
@@ -0,0 +1,518 @@
+"""
+Unit tests for JobService using in-memory MongoDB.
+
+Tests the business logic operations with real MongoDB operations
+using mongomock for better integration testing.
+"""
+
+import pytest
+from bson import ObjectId
+from mongomock.mongo_client import MongoClient
+
+from app.exceptions.job_exceptions import InvalidStatusTransitionError
+from app.models.job import ProcessingStatus
+from app.models.types import PyObjectId
+from app.services.job_service import JobService
+
+
+@pytest.fixture
+def in_memory_database():
+  """Create an in-memory database for testing."""
+  client = MongoClient()
+  return client.test_database
+
+
+@pytest.fixture
+def job_service(in_memory_database):
+  """Create JobService with in-memory repositories."""
+  service = JobService(in_memory_database).initialize()
+  return service
+
+
+@pytest.fixture
+def sample_document_id():
+  """Sample file ObjectId."""
+  return PyObjectId()
+
+
+@pytest.fixture
+def sample_task_id():
+  """Sample Celery task UUID."""
+  return "550e8400-e29b-41d4-a716-446655440000"
+
+
+class TestCreateJob:
+  """Tests for create_job method."""
+  
+  def test_i_can_create_job_with_task_id(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test creating job with task ID."""
+    # Execute
+    result = job_service.create_job(sample_document_id, sample_task_id)
+    
+    # Verify job creation
+    assert result is not None
+    assert result.document_id == sample_document_id
+    assert result.task_id == sample_task_id
+    assert result.status == ProcessingStatus.PENDING
+    assert result.created_at is not None
+    assert result.started_at is None
+    assert result.error_message is None
+    
+    # Verify job exists in database
+    job_in_db = job_service.get_job_by_id(result.id)
+    assert job_in_db is not None
+    assert job_in_db.id == result.id
+    assert job_in_db.document_id == sample_document_id
+    assert job_in_db.task_id == sample_task_id
+    assert job_in_db.status == ProcessingStatus.PENDING
+  
+  def test_i_can_create_job_without_task_id(
+      self,
+      job_service,
+      sample_document_id
+  ):
+    """Test creating job without task ID."""
+    # Execute
+    result = job_service.create_job(sample_document_id)
+    
+    # Verify job creation
+    assert result is not None
+    assert result.document_id == sample_document_id
+    assert result.task_id is None
+    assert result.status == ProcessingStatus.PENDING
+    assert result.created_at is not None
+    assert result.started_at is None
+    assert result.error_message is None
+
+
+class TestGetJobMethods:
+  """Tests for job retrieval methods."""
+  
+  def test_i_can_get_job_by_id(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test retrieving job by ID."""
+    # Create a job first
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    
+    # Execute
+    result = job_service.get_job_by_id(created_job.id)
+    
+    # Verify
+    assert result is not None
+    assert result.id == created_job.id
+    assert result.document_id == created_job.document_id
+    assert result.task_id == created_job.task_id
+    assert result.status == created_job.status
+  
+  def test_i_can_get_jobs_by_status(
+      self,
+      job_service,
+      sample_document_id
+  ):
+    """Test retrieving jobs by status."""
+    # Create jobs with different statuses
+    pending_job = job_service.create_job(sample_document_id, "pending-task")
+    
+    processing_job = job_service.create_job(ObjectId(), "processing-task")
+    job_service.mark_job_as_started(processing_job.id)
+    
+    completed_job = job_service.create_job(ObjectId(), "completed-task")
+    job_service.mark_job_as_started(completed_job.id)
+    job_service.mark_job_as_completed(completed_job.id)
+    
+    # Execute - get pending jobs
+    pending_results = job_service.get_jobs_by_status(ProcessingStatus.PENDING)
+    
+    # Verify
+    assert len(pending_results) == 1
+    assert pending_results[0].id == pending_job.id
+    assert pending_results[0].status == ProcessingStatus.PENDING
+    
+    # Execute - get processing jobs
+    processing_results = job_service.get_jobs_by_status(ProcessingStatus.PROCESSING)
+    assert len(processing_results) == 1
+    assert processing_results[0].status == ProcessingStatus.PROCESSING
+    
+    # Execute - get completed jobs
+    completed_results = job_service.get_jobs_by_status(ProcessingStatus.COMPLETED)
+    assert len(completed_results) == 1
+    assert completed_results[0].status == ProcessingStatus.COMPLETED
+
+
+class TestUpdateStatus:
+  """Tests for mark_job_as_started method."""
+  
+  def test_i_can_mark_pending_job_as_started(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test marking pending job as started (PENDING → PROCESSING)."""
+    # Create a pending job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    assert created_job.status == ProcessingStatus.PENDING
+    
+    # Execute
+    result = job_service.mark_job_as_started(created_job.id)
+    
+    # Verify status transition
+    assert result is not None
+    assert result.id == created_job.id
+    assert result.status == ProcessingStatus.PROCESSING
+    
+    # Verify in database
+    updated_job = job_service.get_job_by_id(created_job.id)
+    assert updated_job.status == ProcessingStatus.PROCESSING
+  
+  def test_i_cannot_mark_processing_job_as_started(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test that processing job cannot be marked as started."""
+    # Create and start a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    job_service.mark_job_as_started(created_job.id)
+    
+    # Try to start it again
+    with pytest.raises(InvalidStatusTransitionError) as exc_info:
+      job_service.mark_job_as_started(created_job.id)
+    
+    # Verify exception details
+    assert exc_info.value.current_status == ProcessingStatus.PROCESSING
+    assert exc_info.value.target_status == ProcessingStatus.PROCESSING
+  
+  def test_i_cannot_mark_completed_job_as_started(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test that completed job cannot be marked as started."""
+    # Create, start, and complete a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    job_service.mark_job_as_started(created_job.id)
+    job_service.mark_job_as_completed(created_job.id)
+    
+    # Try to start it again
+    with pytest.raises(InvalidStatusTransitionError) as exc_info:
+      job_service.mark_job_as_started(created_job.id)
+    
+    # Verify exception details
+    assert exc_info.value.current_status == ProcessingStatus.COMPLETED
+    assert exc_info.value.target_status == ProcessingStatus.PROCESSING
+  
+  def test_i_cannot_mark_failed_job_as_started(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test that failed job cannot be marked as started."""
+    # Create, start, and fail a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    job_service.mark_job_as_started(created_job.id)
+    job_service.mark_job_as_failed(created_job.id, "Test error")
+    
+    # Try to start it again
+    with pytest.raises(InvalidStatusTransitionError) as exc_info:
+      job_service.mark_job_as_started(created_job.id)
+    
+    # Verify exception details
+    assert exc_info.value.current_status == ProcessingStatus.FAILED
+    assert exc_info.value.target_status == ProcessingStatus.PROCESSING
+  
+  def test_i_can_mark_processing_job_as_completed(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test marking processing job as completed (PROCESSING → COMPLETED)."""
+    # Create and start a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    started_job = job_service.mark_job_as_started(created_job.id)
+    
+    # Execute
+    result = job_service.mark_job_as_completed(created_job.id)
+    
+    # Verify status transition
+    assert result is not None
+    assert result.id == created_job.id
+    assert result.status == ProcessingStatus.COMPLETED
+    
+    # Verify in database
+    updated_job = job_service.get_job_by_id(created_job.id)
+    assert updated_job.status == ProcessingStatus.COMPLETED
+  
+  def test_i_cannot_mark_pending_job_as_completed(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test that pending job cannot be marked as completed."""
+    # Create a pending job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    
+    # Try to complete it directly
+    with pytest.raises(InvalidStatusTransitionError) as exc_info:
+      job_service.mark_job_as_completed(created_job.id)
+    
+    # Verify exception details
+    assert exc_info.value.current_status == ProcessingStatus.PENDING
+    assert exc_info.value.target_status == ProcessingStatus.COMPLETED
+  
+  def test_i_cannot_mark_completed_job_as_completed(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test that completed job cannot be marked as completed again."""
+    # Create, start, and complete a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    job_service.mark_job_as_started(created_job.id)
+    job_service.mark_job_as_completed(created_job.id)
+    
+    # Try to complete it again
+    with pytest.raises(InvalidStatusTransitionError) as exc_info:
+      job_service.mark_job_as_completed(created_job.id)
+    
+    # Verify exception details
+    assert exc_info.value.current_status == ProcessingStatus.COMPLETED
+    assert exc_info.value.target_status == ProcessingStatus.COMPLETED
+  
+  def test_i_cannot_mark_failed_job_as_completed(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test that failed job cannot be marked as completed."""
+    # Create, start, and fail a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    job_service.mark_job_as_started(created_job.id)
+    job_service.mark_job_as_failed(created_job.id, "Test error")
+    
+    # Try to complete it
+    with pytest.raises(InvalidStatusTransitionError) as exc_info:
+      job_service.mark_job_as_completed(created_job.id)
+    
+    # Verify exception details
+    assert exc_info.value.current_status == ProcessingStatus.FAILED
+    assert exc_info.value.target_status == ProcessingStatus.COMPLETED
+  
+  def test_i_can_mark_processing_job_as_failed_with_error_message(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test marking processing job as failed with error message."""
+    # Create and start a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    started_job = job_service.mark_job_as_started(created_job.id)
+    
+    error_message = "Processing failed due to invalid file format"
+    
+    # Execute
+    result = job_service.mark_job_as_failed(created_job.id, error_message)
+    
+    # Verify status transition
+    assert result is not None
+    assert result.id == created_job.id
+    assert result.status == ProcessingStatus.FAILED
+    assert result.error_message == error_message
+    
+    # Verify in database
+    updated_job = job_service.get_job_by_id(created_job.id)
+    assert updated_job.status == ProcessingStatus.FAILED
+    assert updated_job.error_message == error_message
+  
+  def test_i_can_mark_processing_job_as_failed_without_error_message(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test marking processing job as failed without error message."""
+    # Create and start a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    job_service.mark_job_as_started(created_job.id)
+    
+    # Execute without error message
+    result = job_service.mark_job_as_failed(created_job.id)
+    
+    # Verify status transition
+    assert result is not None
+    assert result.status == ProcessingStatus.FAILED
+    assert result.error_message is None
+  
+  def test_i_cannot_mark_pending_job_as_failed(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test that pending job cannot be marked as failed."""
+    # Create a pending job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    
+    # Try to fail it directly
+    with pytest.raises(InvalidStatusTransitionError) as exc_info:
+      job_service.mark_job_as_failed(created_job.id, "Test error")
+    
+    # Verify exception details
+    assert exc_info.value.current_status == ProcessingStatus.PENDING
+    assert exc_info.value.target_status == ProcessingStatus.FAILED
+  
+  def test_i_cannot_mark_completed_job_as_failed(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test that completed job cannot be marked as failed."""
+    # Create, start, and complete a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    job_service.mark_job_as_started(created_job.id)
+    job_service.mark_job_as_completed(created_job.id)
+    
+    # Try to fail it
+    with pytest.raises(InvalidStatusTransitionError) as exc_info:
+      job_service.mark_job_as_failed(created_job.id, "Test error")
+    
+    # Verify exception details
+    assert exc_info.value.current_status == ProcessingStatus.COMPLETED
+    assert exc_info.value.target_status == ProcessingStatus.FAILED
+  
+  def test_i_cannot_mark_failed_job_as_failed(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test that failed job cannot be marked as failed again."""
+    # Create, start, and fail a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    job_service.mark_job_as_started(created_job.id)
+    job_service.mark_job_as_failed(created_job.id, "First error")
+    
+    # Try to fail it again
+    with pytest.raises(InvalidStatusTransitionError) as exc_info:
+      job_service.mark_job_as_failed(created_job.id, "Second error")
+    
+    # Verify exception details
+    assert exc_info.value.current_status == ProcessingStatus.FAILED
+    assert exc_info.value.target_status == ProcessingStatus.FAILED
+
+
+class TestDeleteJob:
+  """Tests for delete_job method."""
+  
+  def test_i_can_delete_existing_job(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test deleting an existing job."""
+    # Create a job
+    created_job = job_service.create_job(sample_document_id, sample_task_id)
+    
+    # Verify job exists
+    job_before_delete = job_service.get_job_by_id(created_job.id)
+    assert job_before_delete is not None
+    
+    # Execute deletion
+    result = job_service.delete_job(created_job.id)
+    
+    # Verify deletion
+    assert result is True
+    
+    # Verify job no longer exists
+    deleted_job = job_service.get_job_by_id(created_job.id)
+    assert deleted_job is None
+  
+  def test_i_cannot_delete_nonexistent_job(
+      self,
+      job_service
+  ):
+    """Test deleting a nonexistent job returns False."""
+    # Execute deletion with random ObjectId
+    result = job_service.delete_job(ObjectId())
+    
+    # Verify
+    assert result is False
+
+
+class TestStatusTransitionValidation:
+  """Tests for status transition validation across different scenarios."""
+  
+  def test_valid_job_lifecycle_flow(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test complete valid job lifecycle: PENDING → PROCESSING → COMPLETED."""
+    # Create job (PENDING)
+    job = job_service.create_job(sample_document_id, sample_task_id)
+    assert job.status == ProcessingStatus.PENDING
+    
+    # Start job (PENDING → PROCESSING)
+    started_job = job_service.mark_job_as_started(job.id)
+    assert started_job.status == ProcessingStatus.PROCESSING
+    
+    # Complete job (PROCESSING → COMPLETED)
+    completed_job = job_service.mark_job_as_completed(job.id)
+    assert completed_job.status == ProcessingStatus.COMPLETED
+  
+  def test_valid_job_failure_flow(
+      self,
+      job_service,
+      sample_document_id,
+      sample_task_id
+  ):
+    """Test valid job failure: PENDING → PROCESSING → FAILED."""
+    # Create job (PENDING)
+    job = job_service.create_job(sample_document_id, sample_task_id)
+    assert job.status == ProcessingStatus.PENDING
+    
+    # Start job (PENDING → PROCESSING)
+    started_job = job_service.mark_job_as_started(job.id)
+    assert started_job.status == ProcessingStatus.PROCESSING
+    
+    # Fail job (PROCESSING → FAILED)
+    failed_job = job_service.mark_job_as_failed(job.id, "Test failure")
+    assert failed_job.status == ProcessingStatus.FAILED
+    assert failed_job.error_message == "Test failure"
+  
+  def test_job_operations_with_empty_database(
+      self,
+      job_service
+  ):
+    """Test job operations when database is empty."""
+    # Try to get nonexistent job
+    result = job_service.get_job_by_id(ObjectId())
+    assert result is None
+    
+    # Try to get jobs by status when none exist
+    pending_jobs = job_service.get_jobs_by_status(ProcessingStatus.PENDING)
+    assert pending_jobs == []
+    
+    # Try to delete nonexistent job
+    delete_result = job_service.delete_job(ObjectId())
+    assert delete_result is False