Initial DocsMCP stack

tests/conftest.py

@@ -0,0 +1,191 @@
+"""
+Pytest configuration and fixtures for local-context7 tests.
+
+This module provides:
+- Mocks for external dependencies (Qdrant, FastEmbed)
+- Database fixtures for SQLite operations
+- Common test utilities
+"""
+from unittest.mock import MagicMock, patch
+import pytest
+import os
+import json
+from pathlib import Path
+
+from backend.app.db import init_db, upsert_library, insert_document_chunk, get_chunks_for_library, list_libraries, clear_library_documents, get_connection
+
+
+# =============================================================================
+# FIXTURES
+# =============================================================================
+
+@pytest.fixture(scope="function")
+def test_database():
+    """
+    Create a fresh SQLite database for testing.
+    
+    Yields:
+        Database connection with tables initialized
+    """
+    # Use an in-memory or temporary file database
+    db_path = Path(__file__).parent.parent / "backend" / "data" / "test_db.sqlite"
+    
+    # Ensure data directory exists
+    db_path.parent.mkdir(parents=True, exist_ok=True)
+    
+    # Remove existing test DB if present
+    if db_path.exists():
+        db_path.unlink()
+    
+    # Initialize database with tables
+    result = init_db()
+    assert result["success"], f"Failed to initialize test DB: {result.get('error')}"
+    
+    yield
+    
+    # Cleanup: remove test database after tests
+    if db_path.exists():
+        db_path.unlink()
+
+
+@pytest.fixture(scope="function")
+def sample_text():
+    """Sample text for chunking tests."""
+    return """# Introduction
+
+This is the introduction section.
+
+## Background
+
+Background information goes here to make this longer and test chunking.
+
+This paragraph has more content about the background topic.
+
+### Details
+
+Specific details about the background are provided in this subsection.
+
+More details follow here to ensure we have enough text to properly test heading preservation.
+
+## Conclusion
+
+The conclusion wraps up everything nicely."""
+
+
+# =============================================================================
+# MOCKS
+# =============================================================================
+
+@pytest.fixture
+def mock_embedding_model():
+    """
+    Mock FastEmbed model that returns dummy vectors.
+    
+    This avoids needing to download and load the actual embedding model.
+    Returns 384-dimensional zero vectors for any input.
+    """
+    mock_model = MagicMock()
+    
+    # Mock embed method - returns list of lists with float values
+    def mock_embed(texts):
+        return [
+            [0.0] * 384  # Zero vector placeholder
+            for _ in texts
+        ]
+    
+    mock_model.embed = mock_embed
+    
+    return mock_model
+
+
+@pytest.fixture
+def mock_qdrant_client():
+    """
+    Mock Qdrant client that returns empty or test results.
+    
+    Allows testing search logic without needing a running Qdrant server.
+    """
+    mock_client = MagicMock()
+    
+    # Mock search method
+    def mock_search(collection_name, query_vector, limit=10, search_filter=None):
+        # Return empty list (simulating no results)
+        return []
+    
+    mock_client.search = mock_search
+    
+    # Mock delete_collection for cleanup
+    mock_client.delete_collection = MagicMock(return_value=True)
+    
+    return mock_client
+
+
+@pytest.fixture
+def mock_embedding_model_batch():
+    """
+    Batch embedding model mock that returns deterministic fake vectors.
+    
+    Returns slightly different vectors for different input lengths/first chars,
+    allowing tests to verify vector retrieval if needed.
+    """
+    def hash_text(text):
+        # Simple hash-based pseudo-random vector generation
+        text_hash = hash(text) % 1000000
+        return [(hash_text(text) / 1000000 + (i * 0.001)) for i in range(384)]
+    
+    mock_model = MagicMock()
+    mock_model.embed = lambda texts: [hash_text(t) for t in texts]
+    
+    return mock_model
+
+
+# =============================================================================
+# SETUP TEARDOWN FIXTURES
+# =============================================================================
+
+@pytest.fixture(autouse=True)
+def clear_test_database(test_database):
+    """
+    Clear test database before and after each test function.
+    
+    Note: This fixture runs the teardown (cleanup) AFTER the test,
+    so we manually clear at the end of the yield context.
+    The db_path is cleaned up by the test_database fixture's yield block.
+    """
+    pass  # Cleanup handled in test_database fixture
+
+
+@pytest.fixture
+def empty_vector():
+    """Empty/dummy embedding vector for tests."""
+    return [0.0] * 384
+
+
+@pytest.fixture
+def fake_embeddings(sample_text):
+    """Fake embedding vectors for sample text."""
+    def hash_text(text):
+        return [(hash(text) + len(text)) % 1000 / 10000 for _ in range(384)]
+    
+    return [hash_text(s) for s in sample_text.split("\n\n") if s.strip()]
+
+
+# =============================================================================
+# UTILITY FUNCTIONS
+# =============================================================================
+
+@pytest.fixture
+def temp_file(tmp_path):
+    """Create a temporary file and yield its path."""
+    test_file = tmp_path / "test.txt"
+    return test_file
+
+
+# Register custom marker for slow tests (if needed)
+def pytest_configure(config):
+    config.addinivalue_line("markers", "slow: marks tests as slow (deselect with '-m \"not slow\"')")
+
+
+def pytest_runtest_setup(item):
+    """Add custom markers if needed."""
+    pass