API Reference

@opentiny/genui-sdk-server provides component APIs for building large language model chat HTTP services.

startServer()

Starts an Express HTTP server that provides chat completion services.

• Type

function startServer(options: IStartServerOptions): void

interface IStartServerOptions {
  /** API base URL */
  baseURL: string;
  /** API key */
  apiKey: string;
  /** Server port, default 3100 */
  port?: number;
  /** Max attempts when port is in use, default 10 */
  maxAttempts?: number;
}

• Details

Creates an Express app with CORS enabled and registers the chat route (/chat/completions). If the specified port is in use, it automatically tries the next port (up to maxAttempts times). On success, the server address is printed to the console.

• Example

import { startServer } from '@opentiny/genui-sdk-server';

startServer({
  port: 3100,
  baseURL: 'https://api.openai.com/v1',
  apiKey: '',
  maxAttempts: 10,
});

equipChatCompletions()

Equips an Express app with chat completion functionality by registering the route handler.

• Type

function equipChatCompletions(
  app: Express,
  options: IEquipChatCompletionsOptions
): void

interface IEquipChatCompletionsOptions {
  /** Route path, e.g. '/chat/completions' */
  route: string;
  /** API key */
  apiKey: string;
  /** API base URL */
  baseURL: string;
}

• Details

Creates a chat completion request instance, builds a request handler, and registers it on the specified route (POST).

• Example

import express from 'express';
import { equipChatCompletions } from '@opentiny/genui-sdk-server';

const app = express();

equipChatCompletions(app, {
  route: '/chat/completions',
  apiKey: '',
  baseURL: 'https://api.openai.com/v1',
});

app.listen(3000);

createChatCompletionHandler()

Creates a chat completion request handler that processes HTTP requests and returns streaming responses.

• Type

function createChatCompletionHandler(
  config: IChatCompletionHandlerConfig
): { handler: (req: IncomingMessage, res: ServerResponse) => Promise<void> }

interface IChatCompletionHandlerConfig {
  chatCompletions: (
    params: ChatCompletionCreateParamsBase,
    options?: IRequestOptions
  ) => Promise<Response>;
}

interface IRequestOptions {
  signal?: AbortSignal | undefined | null;
}

• Details

Parses the request body (JSON), calls the chat completion API for a streaming response, converts the stream to SSE (Server-Sent Events), handles client disconnects (automatically aborting the request), and provides unified error handling and formatting.

If the response is not streaming, an error is thrown. All errors are caught and formatted into a unified error response. If response headers have already been sent, errors are appended to the stream in SSE format.

• Example

import { createChatCompletionHandler } from '@opentiny/genui-sdk-server';
import { FetchChatCompletions } from '@opentiny/genui-sdk-chat-completions';
import http from 'http';

const chatCompletion = new FetchChatCompletions({
  apiKey: '',
  baseURL: 'https://api.openai.com/v1',
});

const { handler } = createChatCompletionHandler({
  chatCompletions: (params, options) => 
    chatCompletion.chatStream(params, options),
});

const server = http.createServer(handler);
server.listen(3000);