Create working proof of concept for Sonos TTS

app/build.gradle

@@ -33,6 +33,7 @@ dependencies {
     implementation 'org.nanohttpd:nanohttpd:2.2.0'
     implementation 'androidx.localbroadcastmanager:localbroadcastmanager:1.0.0'
     implementation 'androidx.preference:preference:1.1.1'
+    implementation 'com.github.vmichalak:sonos-controller:v.0.1'
     testImplementation 'junit:junit:4.12'
     androidTestImplementation 'androidx.test.ext:junit:1.1.1'
     androidTestImplementation 'androidx.test.espresso:espresso-core:3.2.0'

app/src/main/AndroidManifest.xml

@@ -19,7 +19,8 @@
         android:label="@string/app_name"
         android:roundIcon="@mipmap/ic_launcher_round"
         android:supportsRtl="true"
-        android:theme="@style/AppTheme">
+        android:theme="@style/AppTheme"
+        android:usesCleartextTraffic="true">
         <activity
             android:name="io.bartek.ttsserver.help.HelpActivity"
             android:parentActivityName=".ttsserver.MainActivity" />

app/src/main/java/io/bartek/ttsserver/tts/TTS.kt

@@ -5,14 +5,45 @@ import android.speech.tts.TextToSpeech
 import android.speech.tts.UtteranceProgressListener
 import io.bartek.ttsserver.exception.TTSException
 import java.io.BufferedInputStream
+import java.io.File
 import java.io.FileInputStream
 import java.io.InputStream
+import java.security.MessageDigest
 import java.util.*
 
 data class SpeechData(val stream: InputStream, val size: Long)
 
-class TTS(context: Context, initListener: TextToSpeech.OnInitListener) {
+class TTS(private val context: Context, initListener: TextToSpeech.OnInitListener) {
    private val tts = TextToSpeech(context, initListener)
+   private val messageDigest = MessageDigest.getInstance("SHA-256")
+
+   fun createTTSFile(text: String, language: Locale): File {
+      val digest = hash(text, language)
+      val filename = "tts_$digest.wav"
+      val file = File(context.cacheDir, filename)
+
+      val uuid = UUID.randomUUID().toString()
+      val lock = Lock()
+      tts.setOnUtteranceProgressListener(TTSProcessListener(uuid, lock))
+
+      synchronized(lock) {
+         tts.language = language
+         tts.synthesizeToFile(text, null, file, uuid)
+         lock.wait()
+      }
+
+      if(!lock.success) {
+         throw TTSException()
+      }
+
+      return file
+   }
+
+   private fun hash(text: String, language: Locale): String {
+      val bytes = "$text$language".toByteArray()
+      val digest = messageDigest.digest(bytes)
+      return digest.fold("", { str, it -> str + "%02x".format(it) })
+   }
 
    fun fetchTTSStream(text: String, language: Locale): SpeechData {
       val file = createTempFile("tmp_tts_server", ".wav")

app/src/main/java/io/bartek/ttsserver/web/Endpoints.kt

@@ -0,0 +1,29 @@
+package io.bartek.ttsserver.web
+
+import android.content.UriMatcher
+import android.net.Uri
+
+enum class Endpoint(val uri: String, val id: Int) {
+   UNKNOWN("/", 1),
+   SAY("/say", 2),
+   WAVE("/wave", 3),
+   SONOS("/sonos", 4),
+   SONOS_CACHE("/sonos/*", 5);
+
+   companion object {
+      fun of(id: Int) = values().firstOrNull { it.id == id } ?: UNKNOWN
+   }
+}
+
+class Endpoints {
+   private val uriMatcher = UriMatcher(UriMatcher.NO_MATCH)
+
+   init {
+      Endpoint.values().forEach {
+         uriMatcher.addURI("", it.uri, it.id)
+      }
+   }
+
+   fun match(uri: String) =
+      Endpoint.of(uriMatcher.match(Uri.parse("content://$uri")))
+}

app/src/main/java/io/bartek/ttsserver/web/WebServer.kt

@@ -2,6 +2,7 @@ package io.bartek.ttsserver.web
 
 import android.content.Context
 import android.content.Intent
+import android.net.Uri
 import android.speech.tts.TextToSpeech
 import androidx.localbroadcastmanager.content.LocalBroadcastManager
 import androidx.preference.PreferenceManager
@@ -12,6 +13,9 @@ import io.bartek.ttsserver.service.ForegroundService
 import io.bartek.ttsserver.service.ServiceState
 import io.bartek.ttsserver.tts.TTS
 import org.json.JSONObject
+import java.io.BufferedInputStream
+import java.io.File
+import java.io.FileInputStream
 import java.util.*
 
 private data class TTSRequestData(val text: String, val language: Locale)
@@ -21,14 +25,17 @@ class WebServer(port: Int, private val context: Context) : NanoHTTPD(port),
    TextToSpeech.OnInitListener {
    private val preferences = PreferenceManager.getDefaultSharedPreferences(context)
    private val tts = TTS(context, this)
+   private val endpoints = Endpoints()
 
    override fun serve(session: IHTTPSession?): Response {
       try {
          session?.let {
-            return when (it.uri) {
-               "/wave" -> wave(it)
-               "/say" -> say(it)
-               else -> throw ResponseException(NOT_FOUND, "")
+            return when (endpoints.match(it.uri)) {
+               Endpoint.SAY -> say(it)
+               Endpoint.WAVE -> wave(it)
+               Endpoint.SONOS -> sonos(it)
+               Endpoint.SONOS_CACHE -> sonosCache(it)
+               Endpoint.UNKNOWN -> throw ResponseException(NOT_FOUND, "")
             }
          }
 
@@ -40,6 +47,40 @@ class WebServer(port: Int, private val context: Context) : NanoHTTPD(port),
       }
    }
 
+   private fun sonos(session: IHTTPSession): Response {
+//      SonosDiscovery.discover().firstOrNull { it.zoneGroupState.name == "Salon" }
+//         ?.play()
+
+      if (session.method != Method.POST) {
+         throw ResponseException(METHOD_NOT_ALLOWED, "")
+      }
+
+      if (session.headers[CONTENT_TYPE]?.let { it != MIME_JSON } != false) {
+         throw ResponseException(BAD_REQUEST, "")
+      }
+
+      val (text, language) = getRequestData(session)
+      val file = tts.createTTSFile(text, language)
+      return newFixedLengthResponse(file.toString())
+   }
+
+   private fun sonosCache(session: IHTTPSession): Response {
+      if (session.method != Method.GET) {
+         throw ResponseException(METHOD_NOT_ALLOWED, "")
+      }
+
+      val filename = Uri.parse(session.uri).lastPathSegment ?: throw ResponseException(BAD_REQUEST, "")
+      val file = File(context.cacheDir, filename)
+
+      if(!file.exists()) {
+         throw ResponseException(NOT_FOUND, "")
+      }
+
+      val stream = BufferedInputStream(FileInputStream(file))
+      val size = file.length()
+      return newFixedLengthResponse(OK, MIME_WAVE, stream, size)
+   }
+
    private fun wave(session: IHTTPSession): Response {
       if (!preferences.getBoolean(PreferenceKey.ENABLE_WAVE_ENDPOINT, true)) {
          throw ResponseException(NOT_FOUND, "")

build.gradle

@@ -20,7 +20,7 @@ allprojects {
     repositories {
         google()
         jcenter()
-        
+        maven { url 'https://jitpack.io' }
     }
 }