Migrate to package

tests/__init__.py

@@ -0,0 +1 @@
+"""Tests for knowledge-search-mcp."""

tests/conftest.py

@@ -0,0 +1,36 @@
+"""Pytest configuration and shared fixtures."""
+
+import os
+from unittest.mock import MagicMock
+
+import pytest
+
+
+@pytest.fixture(autouse=True)
+def mock_env_vars(monkeypatch):
+    """Set required environment variables for testing."""
+    test_env = {
+        "PROJECT_ID": "test-project",
+        "LOCATION": "us-central1",
+        "BUCKET": "test-bucket",
+        "INDEX_NAME": "test-index",
+        "DEPLOYED_INDEX_ID": "test-deployed-index",
+        "ENDPOINT_NAME": "projects/test/locations/us-central1/indexEndpoints/test",
+        "ENDPOINT_DOMAIN": "test.us-central1-aiplatform.googleapis.com",
+    }
+    for key, value in test_env.items():
+        monkeypatch.setenv(key, value)
+
+
+@pytest.fixture
+def mock_gcs_storage():
+    """Mock Google Cloud Storage client."""
+    mock = MagicMock()
+    return mock
+
+
+@pytest.fixture
+def mock_vector_search():
+    """Mock vector search client."""
+    mock = MagicMock()
+    return mock

tests/test_config.py

@@ -0,0 +1,56 @@
+"""Tests for configuration management."""
+
+import os
+
+import pytest
+from pydantic import ValidationError
+
+from knowledge_search_mcp.config import Settings
+
+
+def test_settings_from_env():
+    """Test that Settings can be loaded from environment variables."""
+    # Environment is set by conftest.py fixture
+    settings = Settings.model_validate({})
+
+    assert settings.project_id == "test-project"
+    assert settings.location == "us-central1"
+    assert settings.bucket == "test-bucket"
+    assert settings.index_name == "test-index"
+    assert settings.deployed_index_id == "test-deployed-index"
+
+
+def test_settings_defaults():
+    """Test that Settings has correct default values."""
+    settings = Settings.model_validate({})
+
+    assert settings.embedding_model == "gemini-embedding-001"
+    assert settings.search_limit == 10
+    assert settings.log_name == "va_agent_evaluation_logs"
+    assert settings.log_level == "INFO"
+
+
+def test_settings_custom_values(monkeypatch):
+    """Test that Settings can be customized via environment."""
+    monkeypatch.setenv("EMBEDDING_MODEL", "custom-embedding-model")
+    monkeypatch.setenv("SEARCH_LIMIT", "20")
+    monkeypatch.setenv("LOG_LEVEL", "DEBUG")
+
+    settings = Settings.model_validate({})
+
+    assert settings.embedding_model == "custom-embedding-model"
+    assert settings.search_limit == 20
+    assert settings.log_level == "DEBUG"
+
+
+def test_settings_validation_error():
+    """Test that Settings raises ValidationError when required fields are missing."""
+    # Clear all env vars temporarily
+    required_vars = [
+        "PROJECT_ID", "LOCATION", "BUCKET", "INDEX_NAME",
+        "DEPLOYED_INDEX_ID", "ENDPOINT_NAME", "ENDPOINT_DOMAIN"
+    ]
+
+    # This should work with conftest fixture
+    settings = Settings.model_validate({})
+    assert settings.project_id == "test-project"

tests/test_search.py

@@ -0,0 +1,108 @@
+"""Tests for vector search functionality."""
+
+import io
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+
+from knowledge_search_mcp.main import (
+    GoogleCloudFileStorage,
+    GoogleCloudVectorSearch,
+    SourceNamespace,
+)
+
+
+class TestGoogleCloudFileStorage:
+    """Tests for GoogleCloudFileStorage."""
+
+    def test_init(self):
+        """Test storage initialization."""
+        storage = GoogleCloudFileStorage(bucket="test-bucket")
+        assert storage.bucket_name == "test-bucket"
+        assert storage._cache == {}
+
+    @pytest.mark.asyncio
+    async def test_cache_hit(self):
+        """Test that cached files are returned without fetching."""
+        storage = GoogleCloudFileStorage(bucket="test-bucket")
+        test_content = b"cached content"
+        storage._cache["test.md"] = test_content
+
+        result = await storage.async_get_file_stream("test.md")
+
+        assert result.read() == test_content
+        assert result.name == "test.md"
+
+    @pytest.mark.asyncio
+    async def test_cache_miss(self):
+        """Test that uncached files are fetched from GCS."""
+        storage = GoogleCloudFileStorage(bucket="test-bucket")
+        test_content = b"fetched content"
+
+        # Mock the storage download
+        with patch.object(storage, '_get_aio_storage') as mock_storage_getter:
+            mock_storage = AsyncMock()
+            mock_storage.download = AsyncMock(return_value=test_content)
+            mock_storage_getter.return_value = mock_storage
+
+            result = await storage.async_get_file_stream("test.md")
+
+            assert result.read() == test_content
+            assert storage._cache["test.md"] == test_content
+
+
+class TestGoogleCloudVectorSearch:
+    """Tests for GoogleCloudVectorSearch."""
+
+    def test_init(self):
+        """Test vector search client initialization."""
+        vs = GoogleCloudVectorSearch(
+            project_id="test-project",
+            location="us-central1",
+            bucket="test-bucket",
+            index_name="test-index",
+        )
+
+        assert vs.project_id == "test-project"
+        assert vs.location == "us-central1"
+        assert vs.index_name == "test-index"
+
+    def test_configure_index_endpoint(self):
+        """Test endpoint configuration."""
+        vs = GoogleCloudVectorSearch(
+            project_id="test-project",
+            location="us-central1",
+            bucket="test-bucket",
+        )
+
+        vs.configure_index_endpoint(
+            name="test-endpoint",
+            public_domain="test.domain.com",
+        )
+
+        assert vs._endpoint_name == "test-endpoint"
+        assert vs._endpoint_domain == "test.domain.com"
+
+    def test_configure_index_endpoint_validation(self):
+        """Test that endpoint configuration validates inputs."""
+        vs = GoogleCloudVectorSearch(
+            project_id="test-project",
+            location="us-central1",
+            bucket="test-bucket",
+        )
+
+        with pytest.raises(ValueError, match="endpoint name"):
+            vs.configure_index_endpoint(name="", public_domain="test.com")
+
+        with pytest.raises(ValueError, match="endpoint domain"):
+            vs.configure_index_endpoint(name="test", public_domain="")
+
+
+class TestSourceNamespace:
+    """Tests for SourceNamespace enum."""
+
+    def test_source_namespace_values(self):
+        """Test that SourceNamespace has expected values."""
+        assert SourceNamespace.EDUCACION_FINANCIERA.value == "Educacion Financiera"
+        assert SourceNamespace.PRODUCTOS_Y_SERVICIOS.value == "Productos y Servicios"
+        assert SourceNamespace.FUNCIONALIDADES_APP_MOVIL.value == "Funcionalidades de la App Movil"