llmSSE.js

Go to the documentation of this file.

// llmSSE.js - SSE/streaming support for live AI output
const express = require('express');
const router = express.Router();
const { completionStream, callLlama } = require('./llamaCppService');
 
router.post('/ai/stream', async (req, res) => {
  res.set({
    'Content-Type': 'text/event-stream',
    'Cache-Control': 'no-cache',
    'Connection': 'keep-alive',
  });
  const { text, type } = req.body;
  
  try {
    // Use real streaming from llama.cpp
    let prompt;
    switch (type) {
      case 'summary':
        prompt = `Summarize the following text concisely:\n\n${text}`;
        break;
      case 'expand':
        prompt = `Expand the following bullet points into full paragraphs:\n\n${text}`;
        break;
      default:
        prompt = `Rewrite the following in a clear, professional tone. Preserve technical meaning:\n\n${text}`;
    }
    
    await completionStream(prompt, (chunk) => {
      res.write(`data: ${chunk}\n\n`);
    });
    
    res.write('event: end\ndata: [DONE]\n\n');
    res.end();
  } catch (error) {
    console.error('[SSE] Streaming error:', error.message);
    res.write(`event: error\ndata: ${error.message}\n\n`);
    res.end();
  }
});
 
module.exports = router;