Claude Skills MCP Server

"""Tests for search engine functionality.""" import pytest from src.claude_skills_mcp.search_engine import SkillSearchEngine def test_search_engine_initialization(): """Test search engine initialization.""" engine = SkillSearchEngine("all-MiniLM-L6-v2") assert engine.model is not None assert engine.skills == [] assert engine.embeddings is None def test_index_skills(mock_skills): """Test indexing skills.""" engine = SkillSearchEngine("all-MiniLM-L6-v2") engine.index_skills(mock_skills) assert len(engine.skills) == 3 assert engine.embeddings is not None assert engine.embeddings.shape[0] == 3 def test_index_empty_skills(): """Test indexing with empty skill list.""" engine = SkillSearchEngine("all-MiniLM-L6-v2") engine.index_skills([]) assert engine.skills == [] assert engine.embeddings is None def test_search_basic(mock_skills): """Test basic search functionality.""" engine = SkillSearchEngine("all-MiniLM-L6-v2") engine.index_skills(mock_skills) results = engine.search("analyze gene expression", top_k=2) assert len(results) == 2 assert results[0]["name"] in ["RNA Analysis", "Protein Folding", "Drug Discovery"] assert "relevance_score" in results[0] assert 0 <= results[0]["relevance_score"] <= 1 def test_search_top_k_limit(mock_skills): """Test that top_k limits results correctly.""" engine = SkillSearchEngine("all-MiniLM-L6-v2") engine.index_skills(mock_skills) results_1 = engine.search("protein analysis", top_k=1) results_2 = engine.search("protein analysis", top_k=2) results_all = engine.search("protein analysis", top_k=10) assert len(results_1) == 1 assert len(results_2) == 2 assert len(results_all) == 3 # Max available def test_search_relevance_ordering(mock_skills): """Test that results are ordered by relevance.""" engine = SkillSearchEngine("all-MiniLM-L6-v2") engine.index_skills(mock_skills) results = engine.search("RNA sequencing gene expression", top_k=3) # Results should be sorted by relevance score (highest first) for i in range(len(results) - 1): assert results[i]["relevance_score"] >= results[i + 1]["relevance_score"] def test_search_empty_index(): """Test searching with no indexed skills.""" engine = SkillSearchEngine("all-MiniLM-L6-v2") results = engine.search("test query", top_k=3) assert results == [] def test_search_result_structure(mock_skills): """Test that search results have the expected structure.""" engine = SkillSearchEngine("all-MiniLM-L6-v2") engine.index_skills(mock_skills) results = engine.search("test query", top_k=1) assert len(results) > 0 result = results[0] assert "name" in result assert "description" in result assert "content" in result assert "source" in result assert "relevance_score" in result @pytest.mark.parametrize( "query,expected_top_skill", [ ("RNA sequencing data", "RNA Analysis"), ("protein structure prediction", "Protein Folding"), ("drug compound screening", "Drug Discovery"), ], ) def test_search_finds_relevant_skills(mock_skills, query, expected_top_skill): """Test that search finds the most relevant skill for specific queries.""" engine = SkillSearchEngine("all-MiniLM-L6-v2") engine.index_skills(mock_skills) results = engine.search(query, top_k=1) assert len(results) > 0 # The top result should be the expected skill (with high probability) # Note: This is not guaranteed 100% but should work with good embeddings assert results[0]["name"] == expected_top_skill def test_cosine_similarity(): """Test cosine similarity computation.""" import numpy as np from src.claude_skills_mcp.search_engine import SkillSearchEngine vec1 = np.array([1.0, 0.0, 0.0]) vec2 = np.array([1.0, 0.0, 0.0]) vec3 = np.array([0.0, 1.0, 0.0]) matrix = np.array([vec2, vec3]) similarities = SkillSearchEngine._cosine_similarity(vec1, matrix) assert len(similarities) == 2 assert abs(similarities[0] - 1.0) < 0.001 # Same vector assert abs(similarities[1] - 0.0) < 0.001 # Orthogonal