server.mjs

import express from 'express';
import cors from 'cors';
import { HfInference } from '@huggingface/inference';

const app = express();
const PORT = 3000;

app.use(cors());
app.use(express.json({ limit: '10mb' }));

// Health check
app.get('/api/health', (req, res) => {
  res.json({ 
    status: 'ok', 
    service: 'ai-inference-proxy',
    hasKey: !!process.env.HF_API_KEY 
  });
});

// AI inference proxy endpoint
app.post('/api/inference', async (req, res) => {
  const apiKey = process.env.HF_API_KEY;
  const { prompt } = req.body || {};

  if (!prompt || typeof prompt !== 'string' || !prompt.trim()) {
    return res.status(400).json({ error: 'Prompt is required' });
  }

  // If no key, return your graceful fallback
  if (!apiKey) {
    console.log('No HF_API_KEY set -> using fallback');
    return res.status(200).json({ fallback: true, message: 'Using rule-based recommendations' });
  }

  try {
    console.log('Calling HF Inference Providers API (router)...');
    
    // Use direct HTTP call to router (as per docs)
    const response = await fetch(
      'https://router.huggingface.co/v1/chat/completions',
      {
        method: 'POST',
        headers: {
          'Authorization': `Bearer ${apiKey}`,
          'Content-Type': 'application/json',
        },
        body: JSON.stringify({
          model: 'meta-llama/Llama-3.2-3B-Instruct',
          messages: [
            {
              role: 'user',
              content: prompt
            }
          ],
          max_tokens: 512,
          temperature: 0.2,
          stream: false
        }),
      }
    );

    if (!response.ok) {
      const errorText = await response.text();
      console.error('HF Router API error:', response.status, response.statusText);
      console.error('Response body:', errorText);
      return res.status(200).json({
        fallback: true,
        error: 'HF Router API error',
        status: response.status,
        details: errorText
      });
    }

    const data = await response.json();
    console.log('HF Inference Providers API response received');
    
    // Return in expected format
    const generatedText = data.choices[0].message.content;
    return res.status(200).json([{ generated_text: generatedText }]);
  } catch (e) {
    // Hugging Face SDK puts details on e.cause
    const status = e?.cause?.status || e?.response?.status;
    const body = e?.cause?.response?.text ? await e.cause.response.text() :
                 e?.response?.text ? await e.response.text() : undefined;

    console.error('Proxy error:', e);
    console.error('Status:', status);
    if (body) console.error('Body:', body);

    // Graceful fallback so the UI still works
    return res.status(200).json({
      fallback: true,
      error: 'HF call failed',
      status,
      details: body || e.message || 'Unknown error'
    });
  }
});

app.listen(PORT, '0.0.0.0', () => {
  console.log(`✅ AI inference proxy running on port ${PORT}`);
  console.log(`Node version: ${process.version}`);
  console.log(`Model: meta-llama/Llama-3.2-3B-Instruct`);
  console.log(`API key configured: ${!!process.env.HF_API_KEY}`);
});