PitchLense MCP

""" Tests for MCP tools and analyzers using mocks (Gemini, SerpAPI, Perplexity). """ import os from unittest.mock import Mock, patch import pytest from pitchlense_mcp import ( CustomerRiskMCPTool, FinancialRiskMCPTool, MarketRiskMCPTool, TeamRiskMCPTool, OperationalRiskMCPTool, CompetitiveRiskMCPTool, LegalRiskMCPTool, ExitRiskMCPTool, ProductRiskMCPTool, PeerBenchmarkMCPTool, SerpNewsMCPTool, PerplexityMCPTool, ) # ----------------------------- # SerpAPI (Google News) tests # ----------------------------- @patch.dict(os.environ, {"SERPAPI_API_KEY": "test_key"}, clear=True) @patch("pitchlense_mcp.tools.serp_news.GoogleSearch") def test_serp_news_success(mock_search): mock_instance = Mock() mock_instance.get_dict.return_value = { "news_results": [ { "title": "Headline", "link": "https://news.google.com/article", "source": "Source", "date": "1 hour ago", "snippet": "Snippet", "thumbnail": "https://image/url.jpg", } ] } mock_search.return_value = mock_instance tool = SerpNewsMCPTool() res = tool.fetch_google_news("openai funding", num_results=5) assert res["query"] == "openai funding" assert isinstance(res.get("results"), list) assert res["results"][0]["link"].startswith("https://") @patch.dict(os.environ, {}, clear=True) def test_serp_news_missing_key(): tool = SerpNewsMCPTool() res = tool.fetch_google_news("openai") assert "error" in res assert "SERPAPI_API_KEY" in res["error"] # ----------------------------- # Perplexity tests # ----------------------------- @patch.dict(os.environ, {"PERPLEXITY_API_KEY": "ppx_key"}, clear=True) @patch("httpx.Client.post") def test_perplexity_success(mock_post): class Resp: def raise_for_status(self): return None def json(self): return { "choices": [ { "message": { "content": "Answer text", "citations": [ "https://example.com/a", "https://example.com/b", ], } } ] } mock_post.return_value = Resp() tool = PerplexityMCPTool() out = tool.search_perplexity("What is RAG?") assert out["query"] == "What is RAG?" assert out["answer"] == "Answer text" assert {"url": "https://example.com/a", "title": None} in out["sources"] @patch.dict(os.environ, {}, clear=True) def test_perplexity_missing_key(): tool = PerplexityMCPTool() res = tool.search_perplexity("Anything") assert "error" in res assert "PERPLEXITY_API_KEY" in res["error"] # ----------------------------- # Analyzer MCP tools tests # ----------------------------- @pytest.mark.parametrize( "tool_cls, method_name", [ (CustomerRiskMCPTool, "analyze_customer_risks"), (FinancialRiskMCPTool, "analyze_financial_risks"), (MarketRiskMCPTool, "analyze_market_risks"), (TeamRiskMCPTool, "analyze_team_risks"), (OperationalRiskMCPTool, "analyze_operational_risks"), (CompetitiveRiskMCPTool, "analyze_competitive_risks"), (LegalRiskMCPTool, "analyze_legal_risks"), (ExitRiskMCPTool, "analyze_exit_risks"), (ProductRiskMCPTool, "analyze_product_risks"), ], ) def test_analyzer_mcp_tools_return(tool_cls, method_name): tool = tool_cls() # Bypass LLM by stubbing analyze tool.analyzer.analyze = Mock(return_value={ "category_name": getattr(tool.analyzer, "category_name", "Category"), "overall_risk_level": "medium", "category_score": 5, "indicators": [], "summary": "Stubbed" }) fn = getattr(tool, method_name) out = fn("Some organized startup text") assert out.get("overall_risk_level") in {"low", "medium", "high", "critical", "unknown"} assert isinstance(out.get("category_score"), int) def test_peer_benchmark_mcp_tool(): tool = PeerBenchmarkMCPTool() # Mock the raw LLM response that would come from the prompt tool.analyzer.llm_client = Mock() tool.analyzer.llm_client.predict = Mock(return_value={ "response": '''<JSON> { "category_name": "Peer Benchmarking", "overall_benchmark_level": "medium", "benchmark_score": 6, "peer_benchmarks": {}, "startup_metrics": {}, "comparison_table": [], "summary": "OK" } </JSON>''' }) res = tool.analyze_peer_benchmark("Startup info text") assert res["category_name"] == "Peer Benchmarking" assert "category_score" in res assert "overall_risk_level" in res assert "indicators" in res