maternal-app/maternal-web/hooks/useVoiceInput.ts

import { useState, useEffect, useCallback, useRef } from 'react';
import { tokenStorage } from '@/lib/utils/tokenStorage';

export interface VoiceInputResult {
  transcript: string;
  confidence: number;
  isFinal: boolean;
}

export interface VoiceInputState {
  isListening: boolean;
  isSupported: boolean;
  transcript: string;
  classification: any | null;
  error: string | null;
  usesFallback: boolean;
}

/**
 * Hook for voice input using browser Web Speech API or MediaRecorder fallback
 *
 * Provides voice recording functionality with real-time transcription.
 * Falls back to MediaRecorder + server-side transcription for iOS Safari.
 */
export function useVoiceInput() {
  const [state, setState] = useState<VoiceInputState>({
    isListening: false,
    isSupported: false,
    transcript: '',
    classification: null,
    error: null,
    usesFallback: false,
  });

  const recognitionRef = useRef<any>(null);
  const mediaRecorderRef = useRef<MediaRecorder | null>(null);
  const audioChunksRef = useRef<Blob[]>([]);
  const timeoutRef = useRef<NodeJS.Timeout | null>(null);
  const silenceTimeoutRef = useRef<NodeJS.Timeout | null>(null);
  const audioContextRef = useRef<AudioContext | null>(null);
  const analyserRef = useRef<AnalyserNode | null>(null);

  // Check if browser supports MediaRecorder (unified approach for all platforms)
  useEffect(() => {
    // Always use MediaRecorder + backend transcription for consistency
    // This gives us one flow to debug and maintain, works on all platforms
    console.log('[Voice] Checking MediaRecorder support...');

    if (navigator.mediaDevices && navigator.mediaDevices.getUserMedia) {
      console.log('[Voice] MediaRecorder supported, will use backend transcription for all platforms');
      setState(prev => ({ ...prev, isSupported: true, usesFallback: true }));
    } else {
      console.log('[Voice] MediaRecorder not supported');
      setState(prev => ({ ...prev, isSupported: false }));
    }

    return () => {
      if (recognitionRef.current) {
        try {
          recognitionRef.current.stop();
        } catch (e) {
          // Ignore errors on cleanup
        }
      }
      if (mediaRecorderRef.current && mediaRecorderRef.current.state !== 'inactive') {
        mediaRecorderRef.current.stop();
      }
      if (timeoutRef.current) {
        clearTimeout(timeoutRef.current);
      }
    };
  }, []);

  // Start listening with MediaRecorder fallback
  const startListeningWithFallback = useCallback(async () => {
    audioChunksRef.current = [];

    try {
      console.log('[Voice] Requesting microphone access...');
      const stream = await navigator.mediaDevices.getUserMedia({
        audio: {
          echoCancellation: true,
          noiseSuppression: true,
          sampleRate: 44100,
        }
      });

      console.log('[Voice] Microphone access granted, creating MediaRecorder...');

      // Set up silence detection using Web Audio API
      const audioContext = new (window.AudioContext || (window as any).webkitAudioContext)();
      const analyser = audioContext.createAnalyser();
      const microphone = audioContext.createMediaStreamSource(stream);
      analyser.fftSize = 512;
      microphone.connect(analyser);

      audioContextRef.current = audioContext;
      analyserRef.current = analyser;

      // Monitor audio levels for silence detection
      const bufferLength = analyser.frequencyBinCount;
      const dataArray = new Uint8Array(bufferLength);
      let lastSoundTime = Date.now();
      const SILENCE_THRESHOLD = 10; // Adjust based on testing
      const SILENCE_DURATION = 1500; // 1.5 seconds of silence

      const checkSilence = () => {
        analyser.getByteFrequencyData(dataArray);
        const average = dataArray.reduce((a, b) => a + b, 0) / bufferLength;

        if (average > SILENCE_THRESHOLD) {
          lastSoundTime = Date.now();
          // Clear silence timeout if sound detected
          if (silenceTimeoutRef.current) {
            clearTimeout(silenceTimeoutRef.current);
            silenceTimeoutRef.current = null;
          }
        } else {
          // Check if silence has lasted long enough
          const silenceDuration = Date.now() - lastSoundTime;
          if (silenceDuration > SILENCE_DURATION && !silenceTimeoutRef.current) {
            console.log('[Voice] Silence detected, auto-stopping...');
            if (mediaRecorderRef.current && mediaRecorderRef.current.state === 'recording') {
              mediaRecorderRef.current.stop();
            }
            return;
          }
        }

        // Continue checking if still recording
        if (mediaRecorderRef.current && mediaRecorderRef.current.state === 'recording') {
          requestAnimationFrame(checkSilence);
        }
      };

      // Start monitoring after a brief delay to avoid immediate stop
      setTimeout(() => {
        if (mediaRecorderRef.current && mediaRecorderRef.current.state === 'recording') {
          requestAnimationFrame(checkSilence);
        }
      }, 500);

      // Try different mime types for iOS Safari compatibility
      let mimeType = 'audio/webm;codecs=opus';
      if (!MediaRecorder.isTypeSupported(mimeType)) {
        console.warn('[Voice] webm not supported, trying mp4...');
        mimeType = 'audio/mp4';
        if (!MediaRecorder.isTypeSupported(mimeType)) {
          console.warn('[Voice] mp4 not supported, trying default...');
          mimeType = '';
        }
      }

      const options = mimeType ? { mimeType } : {};
      console.log('[Voice] Using MediaRecorder with options:', options);
      const mediaRecorder = new MediaRecorder(stream, options);

      mediaRecorderRef.current = mediaRecorder;

      mediaRecorder.ondataavailable = (event) => {
        if (event.data.size > 0) {
          audioChunksRef.current.push(event.data);
        }
      };

      mediaRecorder.onstop = async () => {
        console.log('[Voice] Recording stopped, processing audio...');
        const audioBlob = new Blob(audioChunksRef.current, {
          type: mimeType || 'audio/webm'
        });

        console.log('[Voice] Audio blob created, size:', audioBlob.size, 'bytes');

        // Send to backend for transcription
        try {
          const formData = new FormData();
          const extension = mimeType.includes('mp4') ? 'mp4' : 'webm';
          formData.append('audio', audioBlob, `recording.${extension}`);

          console.log('[Voice] Sending to backend for transcription...');

          // Get auth token and API base URL
          const token = tokenStorage.getAccessToken();
          const API_BASE_URL = process.env.NEXT_PUBLIC_API_URL || 'http://localhost:3020';
          const headers: HeadersInit = {};
          if (token) {
            headers['Authorization'] = `Bearer ${token}`;
          }

          const response = await fetch(`${API_BASE_URL}/api/v1/voice/transcribe`, {
            method: 'POST',
            body: formData,
            headers,
          });

          console.log('[Voice] Transcription response status:', response.status);
          const data = await response.json();
          console.log('[Voice] Transcription response data:', data);

          if (response.ok && data.success) {
            setState(prev => ({
              ...prev,
              isListening: false,
              transcript: data.transcript,
              classification: data.classification || null,
            }));
          } else {
            console.error('[Voice] Transcription failed:', data);
            setState(prev => ({
              ...prev,
              isListening: false,
              error: data.message || 'Failed to transcribe audio',
              classification: null,
            }));
          }
        } catch (error) {
          console.error('[Voice] Transcription error:', error);
          setState(prev => ({
            ...prev,
            isListening: false,
            error: 'Failed to process audio. Please try again.',
          }));
        }

        // Stop all tracks
        stream.getTracks().forEach(track => track.stop());
        console.log('[Voice] Stream tracks stopped');

        // Clean up audio context
        if (audioContextRef.current) {
          audioContextRef.current.close();
          audioContextRef.current = null;
        }
      };

      mediaRecorder.onerror = (event) => {
        console.error('[Voice] MediaRecorder error:', event);
        setState(prev => ({
          ...prev,
          isListening: false,
          error: 'Recording failed',
        }));
      };

      setState(prev => ({
        ...prev,
        isListening: true,
        transcript: '',
        error: null,
      }));

      console.log('[Voice] Starting MediaRecorder...');
      mediaRecorder.start();
      console.log('[Voice] MediaRecorder started successfully');

      // Auto-stop after 10 seconds
      timeoutRef.current = setTimeout(() => {
        if (mediaRecorderRef.current && mediaRecorderRef.current.state === 'recording') {
          console.log('[Voice] Auto-stopping after 10 seconds');
          mediaRecorderRef.current.stop();
        }
      }, 10000);
    } catch (error: any) {
      console.error('[Voice] Failed to access microphone:', error);
      console.error('[Voice] Error name:', error.name);
      console.error('[Voice] Error message:', error.message);

      let errorMessage = 'Failed to access microphone';
      if (error.name === 'NotAllowedError' || error.name === 'PermissionDeniedError') {
        errorMessage = 'Microphone permission denied. Please allow microphone access in your browser settings and try again.';
      } else if (error.name === 'NotFoundError' || error.name === 'DevicesNotFoundError') {
        errorMessage = 'No microphone found. Please check your device settings.';
      } else if (error.name === 'NotSupportedError') {
        errorMessage = 'Your browser does not support audio recording.';
      } else if (error.name === 'NotReadableError' || error.name === 'TrackStartError') {
        errorMessage = 'Microphone is already in use by another application.';
      }

      setState(prev => ({
        ...prev,
        isListening: false,
        error: errorMessage,
      }));
    }
  }, []);

  // Start listening with Web Speech API
  const startListeningWithSpeechAPI = useCallback(() => {
    const recognition = recognitionRef.current;

    // Clear previous state
    setState(prev => ({
      ...prev,
      isListening: true,
      transcript: '',
      error: null,
    }));

    let lastSpeechTime = Date.now();

    // Set up event handlers
    recognition.onstart = () => {
      console.log('[Voice] Started listening');
    };

    recognition.onresult = (event: any) => {
      let interimTranscript = '';
      let finalTranscript = '';

      for (let i = event.resultIndex; i < event.results.length; i++) {
        const transcript = event.results[i][0].transcript;
        if (event.results[i].isFinal) {
          finalTranscript += transcript;
        } else {
          interimTranscript += transcript;
        }
      }

      // Update last speech time
      lastSpeechTime = Date.now();

      // Reset silence timeout
      if (silenceTimeoutRef.current) {
        clearTimeout(silenceTimeoutRef.current);
      }

      // Set new silence timeout (1.5 seconds after last speech)
      silenceTimeoutRef.current = setTimeout(() => {
        console.log('[Voice] Silence detected, auto-stopping...');
        if (recognitionRef.current) {
          try {
            recognitionRef.current.stop();
          } catch (e) {
            // Ignore errors
          }
        }
      }, 1500);

      // Only update state with final results, show interim in console for debugging
      if (finalTranscript) {
        console.log('[Voice] Final result:', finalTranscript);
        setState(prev => ({
          ...prev,
          transcript: finalTranscript,
        }));
      } else {
        console.log('[Voice] Interim result:', interimTranscript);
      }
    };

    recognition.onerror = (event: any) => {
      console.error('[Voice] Error:', event.error, event);

      let errorMessage = 'Failed to recognize speech';
      if (event.error === 'no-speech') {
        console.warn('[Voice] No speech detected - this is often normal if user stops speaking');
        errorMessage = 'No speech detected. Please try again.';
      } else if (event.error === 'audio-capture') {
        errorMessage = 'No microphone found. Please check your settings.';
      } else if (event.error === 'not-allowed') {
        errorMessage = 'Microphone access denied. Please grant permission.';
      } else if (event.error === 'network') {
        errorMessage = 'Network error. Please check your connection.';
      }

      setState(prev => ({
        ...prev,
        isListening: false,
        error: errorMessage,
      }));
    };

    recognition.onend = () => {
      console.log('[Voice] Stopped listening');
      setState(prev => ({
        ...prev,
        isListening: false,
      }));
    };

    // Auto-stop after 10 seconds
    timeoutRef.current = setTimeout(() => {
      if (recognitionRef.current) {
        recognitionRef.current.stop();
      }
    }, 10000);

    // Start recognition
    try {
      console.log('[Voice] Calling recognition.start()');
      recognition.start();
      console.log('[Voice] recognition.start() called successfully');
    } catch (error) {
      console.error('[Voice] Failed to start:', error);
      setState(prev => ({
        ...prev,
        isListening: false,
        error: 'Failed to start voice recognition',
      }));
    }
  }, []);

  // Start listening (always uses MediaRecorder + backend transcription)
  const startListening = useCallback(() => {
    startListeningWithFallback();
  }, [startListeningWithFallback]);

  // Stop listening
  const stopListening = useCallback(() => {
    if (recognitionRef.current) {
      try {
        recognitionRef.current.stop();
      } catch (e) {
        // Ignore errors
      }
    }
    if (mediaRecorderRef.current && mediaRecorderRef.current.state === 'recording') {
      mediaRecorderRef.current.stop();
    }
    if (timeoutRef.current) {
      clearTimeout(timeoutRef.current);
      timeoutRef.current = null;
    }
  }, []);

  // Reset state
  const reset = useCallback(() => {
    setState(prev => ({
      ...prev,
      transcript: '',
      classification: null,
      error: null,
    }));
  }, []);

  return {
    ...state,
    startListening,
    stopListening,
    reset,
  };
}