add. google cloud translation (detecting languages)

build.gradle.kts

@@ -13,21 +13,22 @@ repositories {
 
 dependencies {
     implementation("com.annimon:tgbots-module:6.5.1")
-    implementation("org.jetbrains.kotlinx:kotlinx-coroutines-core-jvm:1.7.0-Beta")
-
     implementation("com.fasterxml.jackson.module:jackson-module-kotlin:2.14.2")
     implementation("com.github.kittinunf.fuel:fuel:2.3.1")
     implementation("com.google.code.gson:gson:2.10.1")
     implementation("org.imgscalr:imgscalr-lib:4.2")
     implementation("org.json:json:20230227")
+    implementation("org.slf4j:slf4j-api:2.0.6")
+    implementation("ch.qos.logback:logback-classic:1.4.7")
 
-    implementation("org.xerial:sqlite-jdbc:3.40.1.0")
     implementation("org.jetbrains.exposed:exposed-core:0.41.1")
-    implementation("org.jetbrains.exposed:exposed-jdbc:0.41.1")
     implementation("org.jetbrains.exposed:exposed-java-time:0.41.1")
+    implementation("org.jetbrains.exposed:exposed-jdbc:0.41.1")
+    implementation("org.jetbrains.kotlinx:kotlinx-coroutines-core-jvm:1.7.0-Beta")
+    implementation("org.xerial:sqlite-jdbc:3.40.1.0")
 
-    implementation("org.slf4j:slf4j-api:2.0.6")
-    implementation("ch.qos.logback:logback-classic:1.4.7")
+    implementation("com.google.cloud:google-cloud-translate:2.23.0")
+    implementation("com.google.cloud:google-cloud-texttospeech:2.24.0")
 }
 
 application {

src/main/kotlin/com/helltar/artific_intellig_bot/BotConfig.kt

@@ -28,7 +28,6 @@ data class BotMainConfig(
 
 open class BotConfig {
 
-    val googleCloudKey: String
     val openaiKey: String
     val stableDiffusionKey: String
 
@@ -38,7 +37,6 @@ open class BotConfig {
         try {
             Properties().run {
                 load(FileReader(filename))
-                googleCloudKey = getProperty("google_cloud_key")
                 openaiKey = getProperty("openai_key")
                 stableDiffusionKey = getProperty("stable_diffusion_key")
             }

src/main/kotlin/com/helltar/artific_intellig_bot/commands/user/chat/ChatGPTCommand.kt

@@ -4,6 +4,9 @@ import com.annimon.tgbotsmodule.commands.context.MessageContext
 import com.github.kittinunf.fuel.core.extensions.jsonBody
 import com.github.kittinunf.fuel.core.isSuccessful
 import com.github.kittinunf.fuel.httpPost
+import com.google.api.gax.rpc.ApiException
+import com.google.cloud.texttospeech.v1.*
+import com.google.cloud.translate.TranslateOptions
 import com.google.gson.Gson
 import com.helltar.artific_intellig_bot.Commands.cmdChatAsVoice
 import com.helltar.artific_intellig_bot.DIR_FILES
@@ -16,13 +19,8 @@ import com.helltar.artific_intellig_bot.commands.user.chat.ChatGPTData.CHAT_ROLE
 import com.helltar.artific_intellig_bot.commands.user.chat.ChatGPTData.CHAT_ROLE_USER
 import com.helltar.artific_intellig_bot.commands.user.chat.ChatGPTData.ChatData
 import com.helltar.artific_intellig_bot.commands.user.chat.ChatGPTData.ChatMessageData
-import com.helltar.artific_intellig_bot.commands.user.chat.ChatGPTData.TextToSpeechJsonData
-import com.helltar.artific_intellig_bot.commands.user.chat.ChatGPTData.TtsAudioConfigData
-import com.helltar.artific_intellig_bot.commands.user.chat.ChatGPTData.TtsInputData
-import com.helltar.artific_intellig_bot.commands.user.chat.ChatGPTData.TtsVoiceData
 import com.helltar.artific_intellig_bot.dao.DatabaseFactory
 import com.helltar.artific_intellig_bot.localizedString
-import com.helltar.artific_intellig_bot.utils.Utils.detectLangCode
 import org.json.JSONException
 import org.json.JSONObject
 import org.slf4j.LoggerFactory
@@ -130,17 +128,15 @@ open class ChatGPTCommand(ctx: MessageContext) : BotCommand(ctx) {
             userUsageTokens[userId] = userUsageTokens[userId]!! + usageTotalTokens
             userContextMap[userId]?.add(ChatMessageData(CHAT_ROLE_ASSISTANT, answer))
 
-            deleteMessage(waitMessageId)
-
-            if (isVoiceOut) {
+            if (isVoiceOut)
                 sendVoice(answer, messageId)
-                return
-            }
-
-            if (isCommandDisabled(cmdChatAsVoice))
-                replyToMessage(answer, messageId, markdown = true)
             else
-                sendVoice(answer, messageId)
+                if (isCommandDisabled(cmdChatAsVoice))
+                    replyToMessage(answer, messageId, markdown = true)
+                else
+                    sendVoice(answer, messageId)
+
+            deleteMessage(waitMessageId)
 
         } catch (e: JSONException) {
             log.error(e.message)
@@ -149,7 +145,7 @@ open class ChatGPTCommand(ctx: MessageContext) : BotCommand(ctx) {
     }
 
     private fun sendVoice(text: String, messageId: Int) {
-        textToSpeech(text, detectLangCode(text))?.let { oggBytes ->
+        textToSpeech(text, detectLanguage(text))?.let { oggBytes ->
             // todo: tempFile
             val voice = File.createTempFile("tmp", ".ogg").apply { writeBytes(oggBytes) }
             sendVoice(voice, messageId)
@@ -167,31 +163,32 @@ open class ChatGPTCommand(ctx: MessageContext) : BotCommand(ctx) {
             .responseString()
 
     private fun textToSpeech(text: String, languageCode: String): ByteArray? {
-        var json = Gson().toJson(TextToSpeechJsonData(TtsInputData(text), TtsVoiceData(languageCode), TtsAudioConfigData()))
-
-        "https://texttospeech.googleapis.com/v1/text:synthesize?fields=audioContent&key=$googleCloudKey".httpPost()
-            .header("Content-Type", "application/json; charset=utf-8")
-            .timeout(FUEL_TIMEOUT)
-            .timeoutRead(FUEL_TIMEOUT)
-            .jsonBody(json)
-            .responseString().run {
-                json =
-                    if (second.isSuccessful)
-                        third.get()
-                    else {
-                        log.error(this.third.component2()?.message)
-                        return null
-                    }
+        TextToSpeechClient.create().use { textToSpeechClient ->
+            val input = SynthesisInput.newBuilder().setText(text).build()
+
+            val voice =
+                VoiceSelectionParams.newBuilder()
+                    .setLanguageCode(languageCode)
+                    .setSsmlGender(SsmlVoiceGender.NEUTRAL)
+                    .build()
+
+            val audioConfig = AudioConfig.newBuilder().setAudioEncoding(AudioEncoding.OGG_OPUS).build()
+
+            return try {
+                val response = textToSpeechClient.synthesizeSpeech(input, voice, audioConfig)
+                response.audioContent.toByteArray()
+            } catch (e: ApiException) {
+                log.error(e.message)
+                null
             }
-
-        return try {
-            Base64.getDecoder().decode(JSONObject(json).getString("audioContent"))
-        } catch (e: JSONException) {
-            log.error(e.message)
-            null
         }
     }
 
+    private fun detectLanguage(text: String): String {
+        val translate = TranslateOptions.getDefaultInstance().getService()
+        return translate.detect(text).language
+    }
+
     private fun getLoadingGifFileId(): String {
         if (DatabaseFactory.filesIds.exists(FILE_NAME_LOADING_GIF))
             return DatabaseFactory.filesIds.getFileId(FILE_NAME_LOADING_GIF)

src/main/kotlin/com/helltar/artific_intellig_bot/commands/user/chat/ChatGPTData.kt

@@ -16,26 +16,4 @@ object ChatGPTData {
         val role: String,
         val content: String
     )
-
-    /* https://cloud.google.com/text-to-speech/docs/reference/rest/v1/text/synthesize */
-
-    data class TtsInputData(
-        val text: String
-    )
-
-    data class TtsVoiceData(
-        val languageCode: String,
-        val ssmlGender: String = "FEMALE"
-    )
-
-    data class TtsAudioConfigData(
-        val audioEncoding: String = "OGG_OPUS",
-        val speakingRate: Float = 1.0f
-    )
-
-    data class TextToSpeechJsonData(
-        val input: TtsInputData,
-        val voice: TtsVoiceData,
-        val audioConfig: TtsAudioConfigData
-    )
 }