Add OpenRouter Support to Express Sample Server

[wongchichong]

Summary

This PR adds comprehensive OpenRouter support to the Express sample server, enabling users to easily integrate with OpenRouter's AI services through the deep-chat component.

Changes

1. Added OpenRouter service endpoints in the Express server:

   • /openrouter-chat - Standard chat completion endpoint
   • /openrouter-chat-stream - Streaming chat completion endpoint
   • /openrouter-list-models - Model listing endpoint

2. Implemented OpenRouter service class with:

   • Chat completion support with both standard and streaming responses
   • Image generation capabilities
   • Model listing functionality
   • Proper error handling and forwarding to error middleware

3. Updated serviceIOFactory.ts to properly route OpenRouter requests:

   • Added case for 'openrouter' in the service routing switch statement
   • Configured OpenRouterIO to handle forwarded request bodies correctly

Service URL Pattern

The endpoint follows the pattern: http://host:port/openrouter-chat

Example configuration in deep-chat:

<deep-chat
  connect='{
    "url": "http://localhost:8080/openrouter-chat"
  }'>
</deep-chat>

Environment Variables

To use the OpenRouter service, set the following environment variables in your .env file:

• OPENROUTER_API_KEY - Your OpenRouter API key
• OPENROUTER_DEFAULT_MODEL - Default model to use (optional, defaults to 'moonshotai/kimi-k2:free')
• OPENROUTER_IMAGE_MODEL - Default image model to use (optional, defaults to 'qwen/qwen2.5-vl-72b-instruct:free')
• OPENROUTER_HTTP_REFERER - HTTP referer header (optional)
• OPENROUTER_X_TITLE - X-Title header (optional)

Features

• Full support for text-based chat completions
• Streaming response support for real-time message delivery
• Image generation capabilities
• Model listing functionality
• Proper error handling with middleware integration
• CORS support for local development

Testing

The implementation has been tested with:

• Standard chat requests
• Streaming chat requests
• Error scenarios (invalid API keys, network issues)
• Various OpenRouter models

This integration allows developers to quickly set up an OpenRouter-powered chat interface using the deep-chat component with minimal configuration.

Important Note

Before starting work on a new feature or making changes to our API, please open a new issue first.
This helps us to:

• Confirm that your idea aligns with our project's vision
• Ensure it's not something we're already working on internally
• Assess whether it's something we can realistically support or implement

Creating an issue first allows for better collaboration and helps avoid duplicated or misaligned efforts. Thank you!

[OvidijusParsiunas]

Hi @wongchichong. Thankyou for the PR. Unfortunately I am currently not accepting any more example server additions as we will soon be creating a separate repo for Deep Chat implementation examples. I will keep this PR open until this has been created. Please hang tight until it does.