Skip to content

TESTING.md

Latest commit

 

History

History
90 lines (62 loc) · 2.3 KB

TESTING.md

File metadata and controls

90 lines (62 loc) · 2.3 KB

Testing Semem UI and Chat API

This document provides instructions for testing the UI and Chat API of the Semem system.

Prerequisites

Before testing, make sure you have:

  1. Properly set up environment variables or a config file
  2. SPARQL endpoint running (Apache Fuseki recommended)
  3. LLM service available (Claude API access or Ollama with required models)

Starting the Server

# Start with default configuration
node ui-server.js

# Or with a custom configuration file
node ui-server.js config.json

Testing the API Directly

You can test the API endpoints directly using the provided test script:

# Make sure the script is executable
chmod +x test-chat-api.js

# Run the test script
./test-chat-api.js

This script will:

  1. Check the API health
  2. Test the standard chat endpoint
  3. Test chat with search context interjection
  4. Test the streaming chat endpoint

Testing in the Browser

  1. Open your browser and navigate to http://localhost:4100
  2. Navigate to the "Chat" tab
  3. Enter a message and press Send

Advanced Browser Testing

For more advanced testing in the browser, open the developer console and include our test script:

// Load the test script
const script = document.createElement('script');
script.src = 'test-chat.js';
document.head.appendChild(script);

// After it loads, run the test
setTimeout(() => testChat(), 500);

This will:

  1. Add the search context checkbox to the UI
  2. Submit a test message that will use search context interjection
  3. Display the results including any sources found

LLM Provider Priority

The system will try to use LLM providers in this order:

  1. Claude via hyperdata-clients (requires CLAUDE_API_KEY in environment)
  2. Ollama (requires local Ollama server running)
  3. Claude via direct API (requires CLAUDE_API_KEY in environment)
  4. OpenAI (not fully implemented)

Configuration Options

You can modify config.sample.json and save it as config.json to customize the system, including:

  • SPARQL endpoints to use (with fallback)
  • LLM providers and their priorities
  • Port and graph name settings

Troubleshooting

If tests fail, check:

  1. Environment variables (especially API keys)
  2. SPARQL endpoint availability and connection
  3. LLM service availability and connection
  4. Server logs for detailed error messages