server.ts

import express from 'express';
import path from 'path';
import dotenv from 'dotenv';
import { createServer as createViteServer } from 'vite';
import { GoogleGenAI, Type } from '@google/genai';

dotenv.config();

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

app.use(express.json());

// Initialize Gemini SDK with telemetry header
const getGeminiClient = () => {
  const apiKey = process.env.GEMINI_API_KEY;
  if (!apiKey) {
    throw new Error('GEMINI_API_KEY environment variable is missing.');
  }
  return new GoogleGenAI({
    apiKey,
    httpOptions: {
      headers: {
        'User-Agent': 'aistudio-build',
      },
    },
  });
};

// API Route: Generate Sound Recipe from Text Description
app.post('/api/generate-sound', async (req, res) => {
  try {
    const { prompt, category } = req.body;
    if (!prompt || typeof prompt !== 'string') {
      res.status(400).json({ error: 'Prompt string is required' });
      return;
    }

    const ai = getGeminiClient();

    const systemInstruction = `You are a Master Sound Designer specializing in soft, warm, calm, organic UI sound effects for mobile language-learning apps (similar to Duolingo, Calm, and Headspace).
Your job is to translate a user's textual sound description into precise Web Audio synthesis parameters for short UI audio cues (under 1.0 second in duration).

IMPORTANT SOUND DESIGN PRINCIPLES FOR CALM UI SOUNDS:
1. Warmth over harshness: Cutoff frequencies should be smooth (typically 800Hz - 4500Hz).
2. Harmonic purity: Fundamental tone should be rich with gentle decay overtones (warm bell, wooden marimba, zen singing bowl, soft chime chord).
3. Soft attack: Avoid instant sharp clicks unless specifically requesting a subtle wooden tap. Use 0.008s - 0.04s attack.
4. Total duration: Must be short, under 1 second (typically 0.25s to 0.75s).
5. Chord structure: For success/unlock chimes, use pleasant harmonious ratios (e.g. Major 3rd, Perfect 5th, Octave, or Major 7th).

Output a structured JSON recipe matching the schema.`;

    const response = await ai.models.generateContent({
      model: 'gemini-3.6-flash',
      contents: `Design a short, warm, soothing UI sound effect based on this description: "${prompt}". Category preference: ${category || 'general'}.`,
      config: {
        systemInstruction,
        responseMimeType: 'application/json',
        responseSchema: {
          type: Type.OBJECT,
          properties: {
            title: { type: Type.STRING, description: 'Creative 2-4 word title for this sound effect' },
            description: { type: Type.STRING, description: 'Sensory description of how this sound feels' },
            suggestedEventType: {
              type: Type.STRING,
              description: 'Matching app event: correct, incorrect, tap, chapter_unlock, stroke_complete, or general',
            },
            durationSec: { type: Type.NUMBER, description: 'Duration in seconds between 0.1 and 0.95' },
            recipe: {
              type: Type.OBJECT,
              properties: {
                archetype: {
                  type: Type.STRING,
                  description: 'Archetype: bell, marimba, singing_bowl, wood_click, chime_chord, soft_sweep',
                },
                basePitchHz: { type: Type.NUMBER, description: 'Fundamental frequency in Hz (e.g. 261.63, 440, 523.25, 659.25)' },
                chimeRatios: {
                  type: Type.ARRAY,
                  items: { type: Type.NUMBER },
                  description: 'Frequency multipliers for chords/harmonics (e.g. [1.0, 1.25, 1.5] for major chord, or [1.0] for single note)',
                },
                envelope: {
                  type: Type.OBJECT,
                  properties: {
                    attack: { type: Type.NUMBER, description: 'Attack time in seconds (0.005 to 0.1)' },
                    decay: { type: Type.NUMBER, description: 'Decay time in seconds (0.05 to 0.8)' },
                    sustain: { type: Type.NUMBER, description: 'Sustain level 0.0 to 0.15' },
                    release: { type: Type.NUMBER, description: 'Release time in seconds (0.05 to 0.4)' },
                    curve: { type: Type.STRING, description: 'exponential or linear' },
                  },
                  required: ['attack', 'decay', 'sustain', 'release', 'curve'],
                },
                timbre: {
                  type: Type.OBJECT,
                  properties: {
                    harmonicWeights: {
                      type: Type.ARRAY,
                      items: { type: Type.NUMBER },
                      description: 'Overtone relative volumes [fund, 2nd, 3rd, 4th, 5th] e.g. [1.0, 0.35, 0.1, 0.03]',
                    },
                    fmModulationIndex: { type: Type.NUMBER, description: 'FM index 0.0 to 2.5 for bell metallic resonance' },
                    fmFrequencyRatio: { type: Type.NUMBER, description: 'FM ratio relative to base pitch e.g. 1.414, 2.0, 3.5' },
                  },
                  required: ['harmonicWeights', 'fmModulationIndex', 'fmFrequencyRatio'],
                },
                filter: {
                  type: Type.OBJECT,
                  properties: {
                    cutoffHz: { type: Type.NUMBER, description: 'Lowpass filter cutoff in Hz (400 to 5000)' },
                    resonanceQ: { type: Type.NUMBER, description: 'Filter resonance Q (0.5 to 3.0)' },
                    sweepAmount: { type: Type.NUMBER, description: 'Filter sweep down in Hz (0 to 1200)' },
                  },
                  required: ['cutoffHz', 'resonanceQ', 'sweepAmount'],
                },
                warmth: {
                  type: Type.OBJECT,
                  properties: {
                    tapeSat: { type: Type.NUMBER, description: 'Soft saturation amount (0.0 to 0.35)' },
                    lowBoostDb: { type: Type.NUMBER, description: 'Warmth bass/mid boost in dB (0 to 4.0)' },
                  },
                  required: ['tapeSat', 'lowBoostDb'],
                },
                shimmer: {
                  type: Type.OBJECT,
                  properties: {
                    vibratoRate: { type: Type.NUMBER, description: 'Vibrato frequency Hz (0 to 7)' },
                    vibratoDepthCents: { type: Type.NUMBER, description: 'Vibrato depth in pitch cents (0 to 12)' },
                    pitchBendCents: { type: Type.NUMBER, description: 'Initial pitch bend in cents (-300 to +300)' },
                  },
                  required: ['vibratoRate', 'vibratoDepthCents', 'pitchBendCents'],
                },
              },
              required: ['archetype', 'basePitchHz', 'chimeRatios', 'envelope', 'timbre', 'filter', 'warmth', 'shimmer'],
            },
          },
          required: ['title', 'description', 'suggestedEventType', 'durationSec', 'recipe'],
        },
      },
    });

    const textResult = response.text || '';
    const soundData = JSON.parse(textResult);
    res.json({ success: true, data: soundData });
  } catch (error: any) {
    console.error('Error generating sound recipe with Gemini:', error);
    res.status(500).json({
      error: 'Failed to generate sound recipe from Gemini AI.',
      details: error?.message || String(error),
    });
  }
});

// API Route: Health check
app.get('/api/health', (_req, res) => {
  res.json({ status: 'ok', time: new Date().toISOString() });
});

// Start Express + Vite server
async function start() {
  if (process.env.NODE_ENV !== 'production') {
    const vite = await createViteServer({
      server: { middlewareMode: true },
      appType: 'spa',
    });
    app.use(vite.middlewares);
  } else {
    const distPath = path.join(process.cwd(), 'dist');
    app.use(express.static(distPath));
    app.get('*', (_req, res) => {
      res.sendFile(path.join(distPath, 'index.html'));
    });
  }

  app.listen(PORT, '0.0.0.0', () => {
    console.log(`Server listening on http://0.0.0.0:${PORT}`);
  });
}

start();