ListenerApp/ListenerGS/SpeechForwarder.swift

//
//  SpeechForwarder.swift
//  ListenerGS
//
//  Created by Jeremy Rand on 2021-10-18.
//

import Foundation
import os
import Speech

class SpeechForwarder : ObservableObject {
    @Published var listening = false
    @Published var connected = false
    @Published var textHeard = ""
    
    let LISTEN_STATE_MSG = 1
    let LISTEN_TEXT_MSG = 2
    
    let port = 19026
    private var client: TCPClient?
    
    private let speechRecognizer = SFSpeechRecognizer(locale: Locale(identifier: Locale.preferredLanguages[0]))!
    
    private var recognitionRequest: SFSpeechAudioBufferRecognitionRequest?
    
    private var recognitionTask: SFSpeechRecognitionTask?
    
    private let audioEngine = AVAudioEngine()
    
    private let logger = Logger()
    
    func connect(destination : String) {
        logger.debug("Attempting to connect to \(destination)")
        client = TCPClient(address: destination, port: Int32(port))
        guard let client = client else { return }
        switch client.connect(timeout: 10) {
        case .success:
            connected = true
            logger.debug("Connected to \(destination)")
        case .failure(let error):
            client.close()
            self.client = nil
            logger.error("Failed to connect to \(destination): \(String(describing: error))")
            break
        }
    }
    
    func disconnect() {
        if (listening) {
            listen()
        }
        
        guard let client = client else { return }
        client.close()
        connected = false
    }
    
    func listen() {
        self.listening.toggle()
        if (self.listening) {
            SFSpeechRecognizer.requestAuthorization { authStatus in
                // The authorization status results in changes to the
                // app’s interface, so process the results on the app’s
                // main queue.
                OperationQueue.main.addOperation {
                switch authStatus {
                    case .authorized:
                        break
                        
                    case .denied:
                        self.listening = false
                        break

                    case .restricted:
                        self.listening = false
                        break

                    case .notDetermined:
                        self.listening = false
                        break
                        
                    default:
                        self.listening = false
                        break
                    }
                }
            }
        }
        
        guard let client = client else { return }
        if (self.listening) {
            switch (client.send(data: isListening())) {
                case .success:
                    break
                case .failure(let error):
                    self.listening = false
                    logger.error("Unable to send header: \(String(describing: error))")
            }
        }
        
        if (self.listening) {
            do {
                try startRecording()
                logger.debug("Started listening")
            }
            catch {
                self.listening = false
            }
        }
        
        if (!self.listening) {
            logger.debug("Stopped listening")
            audioEngine.stop()
            recognitionRequest?.endAudio()
            recognitionTask?.cancel()
            audioEngine.inputNode.removeTap(onBus: 0);
            audioEngine.inputNode.reset()
            switch (client.send(data: isListening())) {
                case .success:
                    break
                case .failure(let error):
                    self.listening = false
                    logger.error("Failed to send header: \(String(describing: error))")
            }
        }
    }
    
    private func isListening() -> Data {
        return pack("<hh", [LISTEN_STATE_MSG, listening ? 1 : 0])
    }

    private func send(latestText : String) {
        guard let client = client else { return }
        var commonChars = self.textHeard.count
        while (commonChars > 0) {
            if (latestText.prefix(commonChars) ==  self.textHeard.prefix(commonChars)) {
                break
            }
            commonChars -= 1
        }
        var stringToSend = ""
        if (commonChars < self.textHeard.count) {
            stringToSend = String(repeating: "\u{7f}", count: self.textHeard.count - commonChars)
        }
        stringToSend.append(contentsOf: latestText.suffix(latestText.count - commonChars).replacingOccurrences(of: "\n", with: "\r"))
        
        if (stringToSend.count > 0) {
            // TODO - Handle strings to send that are longer than 64K (doubt that would happen though)
            let nsEnc = CFStringConvertEncodingToNSStringEncoding(CFStringEncoding(CFStringBuiltInEncodings.macRoman.rawValue))
            let encoding = String.Encoding(rawValue: nsEnc) // String.Encoding
            if let bytes = stringToSend.data(using: encoding) {
                switch (client.send(data: pack("<hh", [LISTEN_TEXT_MSG, bytes.count]))) {
                    case .success:
                        switch (client.send(data: bytes)) {
                            case .success:
                                self.textHeard = latestText
                                logger.debug("Sent text \"\(stringToSend)\"")
                                break
                            case .failure(let error):
                                self.listening = false
                                logger.error("Failed to send text: \(String(describing: error))")
                        }
                    case .failure(let error):
                        self.listening = false
                        logger.error("Failed to send text: \(String(describing: error))")
                }
            }
        }
    }
    
    private func startRecording() throws {
        
        // Cancel the previous task if it's running.
        recognitionTask?.cancel()
        self.recognitionTask = nil
        
        // Configure the audio session for the app.
        let audioSession = AVAudioSession.sharedInstance()
        try audioSession.setCategory(.record, mode: .measurement, options: .duckOthers)
        try audioSession.setActive(true, options: .notifyOthersOnDeactivation)
        let inputNode = audioEngine.inputNode
        
        // Configure the microphone input.
        let recordingFormat = inputNode.outputFormat(forBus: 0)
        inputNode.installTap(onBus: 0, bufferSize: 1024, format: recordingFormat) { (buffer: AVAudioPCMBuffer, when: AVAudioTime) in
            self.recognitionRequest?.append(buffer)
        }

        // Create and configure the speech recognition request.
        recognitionRequest = SFSpeechAudioBufferRecognitionRequest()
        guard let recognitionRequest = recognitionRequest else { fatalError("Unable to create a SFSpeechAudioBufferRecognitionRequest object") }
        recognitionRequest.shouldReportPartialResults = true
        
        // Keep speech recognition data on device
        if #available(iOS 13, *) {
            recognitionRequest.requiresOnDeviceRecognition = false
        }
        
        self.textHeard = ""
        
        // Create a recognition task for the speech recognition session.
        // Keep a reference to the task so that it can be canceled.
        recognitionTask = speechRecognizer.recognitionTask(with: recognitionRequest) { result, error in
            var isFinal = false
            
            if let result = result {
                // Update the text view with the results.
                self.send(latestText: result.bestTranscription.formattedString)
                isFinal = result.isFinal
                print("Text \(result.bestTranscription.formattedString)")
            }
            
            if error != nil {
                self.logger.error("Error from recognizer: \(String(describing: error))")
            }
            
            if error != nil || isFinal {
                // Stop recognizing speech if there is a problem.
                self.audioEngine.stop()
                inputNode.removeTap(onBus: 0)

                self.recognitionRequest = nil
                self.recognitionTask = nil
                self.listening = false
                self.logger.debug("Stopped listening")
                guard let client = self.client else { return }
                switch (client.send(data: self.isListening())) {
                    case .success:
                        break
                    case .failure(let error):
                        self.logger.error("Failed to send header: \(String(describing: error))")
                }
            }
        }
        
        audioEngine.prepare()
        try audioEngine.start()
    }
}