Update LiveModelFutures to return LiveSessionFutures instead of LiveSession (#6834)

Update the `connect` method of` LiveModelFutures` class to return
`ListenableFuture` instead of `LiveSession`.

This change reduces the burden on Java developers, as they would have to
create a `LiveSessionFutures` object anyway.

Additionally, the `startAudioConverstion` method is now annotated with
`@JvmOverloads` to maintain compatibility with Java.

Author: VinayGuthal

firebase-vertexai/CHANGELOG.md

@@ -1,4 +1,11 @@
 # Unreleased
+* [changed] **Breaking Change**: `LiveModelFutures.connect` now returns `ListenableFuture<LiveSessionFutures>` instead of `ListenableFuture<LiveSession>`.
+    * **Action Required:** Remove any transformations from LiveSession object to LiveSessionFutures object. 
+    * **Action Required:** Change type of variable handling `LiveModelFutures.connect` to `ListenableFuture<LiveSessionsFutures>`
+* [changed] **Breaking Change**: Removed `UNSPECIFIED` value for enum class `ResponseModality`
+    * **Action Required:** Remove all references to `ResponseModality.UNSPECIFIED`
+* [changed] **Breaking Change**: Renamed `LiveGenerationConfig.setResponseModalities` to `LiveGenerationConfig.setResponseModality`
+    * **Action Required:** Replace all references of `LiveGenerationConfig.setResponseModalities` with `LiveGenerationConfig.setResponseModality`
 * [feature] Added support for `HarmBlockThreshold.OFF`. See the
   [model documentation](https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/configure-safety-filters#how_to_configure_content_filters){: .external}
   for more information.

firebase-vertexai/api.txt

@@ -115,7 +115,7 @@ package com.google.firebase.vertexai.java {
   }
 
   @com.google.firebase.vertexai.type.PublicPreviewAPI public abstract class LiveModelFutures {
-    method public abstract com.google.common.util.concurrent.ListenableFuture<com.google.firebase.vertexai.type.LiveSession> connect();
+    method public abstract com.google.common.util.concurrent.ListenableFuture<com.google.firebase.vertexai.java.LiveSessionFutures> connect();
     method public static final com.google.firebase.vertexai.java.LiveModelFutures from(com.google.firebase.vertexai.LiveGenerativeModel model);
     field public static final com.google.firebase.vertexai.java.LiveModelFutures.Companion Companion;
   }
@@ -132,6 +132,7 @@ package com.google.firebase.vertexai.java {
     method public abstract com.google.common.util.concurrent.ListenableFuture<kotlin.Unit> send(String text);
     method public abstract com.google.common.util.concurrent.ListenableFuture<kotlin.Unit> sendFunctionResponse(java.util.List<com.google.firebase.vertexai.type.FunctionResponsePart> functionList);
     method public abstract com.google.common.util.concurrent.ListenableFuture<kotlin.Unit> sendMediaStream(java.util.List<com.google.firebase.vertexai.type.MediaData> mediaChunks);
+    method public abstract com.google.common.util.concurrent.ListenableFuture<kotlin.Unit> startAudioConversation();
     method public abstract com.google.common.util.concurrent.ListenableFuture<kotlin.Unit> startAudioConversation(kotlin.jvm.functions.Function1<? super com.google.firebase.vertexai.type.FunctionCallPart,com.google.firebase.vertexai.type.FunctionResponsePart>? functionCallHandler);
     method public abstract com.google.common.util.concurrent.ListenableFuture<kotlin.Unit> stopAudioConversation();
     method public abstract void stopReceiving();
@@ -597,7 +598,7 @@ package com.google.firebase.vertexai.type {
     method public com.google.firebase.vertexai.type.LiveGenerationConfig.Builder setFrequencyPenalty(Float? frequencyPenalty);
     method public com.google.firebase.vertexai.type.LiveGenerationConfig.Builder setMaxOutputTokens(Integer? maxOutputTokens);
     method public com.google.firebase.vertexai.type.LiveGenerationConfig.Builder setPresencePenalty(Float? presencePenalty);
-    method public com.google.firebase.vertexai.type.LiveGenerationConfig.Builder setResponseModalities(com.google.firebase.vertexai.type.ResponseModality? responseModalities);
+    method public com.google.firebase.vertexai.type.LiveGenerationConfig.Builder setResponseModality(com.google.firebase.vertexai.type.ResponseModality? responseModality);
     method public com.google.firebase.vertexai.type.LiveGenerationConfig.Builder setSpeechConfig(com.google.firebase.vertexai.type.SpeechConfig? speechConfig);
     method public com.google.firebase.vertexai.type.LiveGenerationConfig.Builder setTemperature(Float? temperature);
     method public com.google.firebase.vertexai.type.LiveGenerationConfig.Builder setTopK(Integer? topK);
@@ -696,7 +697,6 @@ package com.google.firebase.vertexai.type {
     field public static final com.google.firebase.vertexai.type.ResponseModality.Companion Companion;
     field public static final com.google.firebase.vertexai.type.ResponseModality IMAGE;
     field public static final com.google.firebase.vertexai.type.ResponseModality TEXT;
-    field public static final com.google.firebase.vertexai.type.ResponseModality UNSPECIFIED;
   }
 
   public static final class ResponseModality.Companion {

firebase-vertexai/src/main/kotlin/com/google/firebase/vertexai/java/LiveModelFutures.kt

@@ -19,7 +19,6 @@ package com.google.firebase.vertexai.java
 import androidx.concurrent.futures.SuspendToFutureAdapter
 import com.google.common.util.concurrent.ListenableFuture
 import com.google.firebase.vertexai.LiveGenerativeModel
-import com.google.firebase.vertexai.type.LiveSession
 import com.google.firebase.vertexai.type.PublicPreviewAPI
 import com.google.firebase.vertexai.type.ServiceConnectionHandshakeFailedException
 
@@ -32,16 +31,16 @@ import com.google.firebase.vertexai.type.ServiceConnectionHandshakeFailedExcepti
 public abstract class LiveModelFutures internal constructor() {
 
   /**
-   * Start a [LiveSession] with the server for bidirectional streaming.
-   * @return A [LiveSession] that you can use to stream messages to and from the server.
+   * Start a [LiveSessionFutures] with the server for bidirectional streaming.
+   * @return A [LiveSessionFutures] that you can use to stream messages to and from the server.
    * @throws [ServiceConnectionHandshakeFailedException] If the client was not able to establish a
    * connection with the server.
    */
-  public abstract fun connect(): ListenableFuture<LiveSession>
+  public abstract fun connect(): ListenableFuture<LiveSessionFutures>
 
   private class FuturesImpl(private val model: LiveGenerativeModel) : LiveModelFutures() {
-    override fun connect(): ListenableFuture<LiveSession> {
-      return SuspendToFutureAdapter.launchFuture { model.connect() }
+    override fun connect(): ListenableFuture<LiveSessionFutures> {
+      return SuspendToFutureAdapter.launchFuture { LiveSessionFutures.from(model.connect()) }
     }
   }
 

firebase-vertexai/src/main/kotlin/com/google/firebase/vertexai/java/LiveSessionFutures.kt

@@ -48,6 +48,12 @@ public abstract class LiveSessionFutures internal constructor() {
     functionCallHandler: ((FunctionCallPart) -> FunctionResponsePart)?
   ): ListenableFuture<Unit>
 
+  /**
+   * Starts an audio conversation with the Gemini server, which can only be stopped using
+   * [stopAudioConversation].
+   */
+  public abstract fun startAudioConversation(): ListenableFuture<Unit>
+
   /**
    * Stops the audio conversation with the Gemini Server.
    *
@@ -124,6 +130,9 @@ public abstract class LiveSessionFutures internal constructor() {
       functionCallHandler: ((FunctionCallPart) -> FunctionResponsePart)?
     ) = SuspendToFutureAdapter.launchFuture { session.startAudioConversation(functionCallHandler) }
 
+    override fun startAudioConversation() =
+      SuspendToFutureAdapter.launchFuture { session.startAudioConversation() }
+
     override fun stopAudioConversation() =
       SuspendToFutureAdapter.launchFuture { session.stopAudioConversation() }
 

firebase-vertexai/src/main/kotlin/com/google/firebase/vertexai/type/LiveGenerationConfig.kt

@@ -130,8 +130,8 @@ private constructor(
     public fun setFrequencyPenalty(frequencyPenalty: Float?): Builder = apply {
       this.frequencyPenalty = frequencyPenalty
     }
-    public fun setResponseModalities(responseModalities: ResponseModality?): Builder = apply {
-      this.responseModality = responseModalities
+    public fun setResponseModality(responseModality: ResponseModality?): Builder = apply {
+      this.responseModality = responseModality
     }
     public fun setSpeechConfig(speechConfig: SpeechConfig?): Builder = apply {
       this.speechConfig = speechConfig

firebase-vertexai/src/main/kotlin/com/google/firebase/vertexai/type/ResponseModality.kt

@@ -18,7 +18,6 @@ package com.google.firebase.vertexai.type
 
 import com.google.firebase.vertexai.common.util.FirstOrdinalSerializer
 import kotlinx.serialization.KSerializer
-import kotlinx.serialization.SerialName
 import kotlinx.serialization.Serializable
 
 /** Represents the type of content present in a response (e.g., text, image, audio). */
@@ -27,7 +26,6 @@ public class ResponseModality private constructor(public val ordinal: Int) {
 
   @Serializable(Internal.Serializer::class)
   internal enum class Internal {
-    @SerialName("MODALITY_UNSPECIFIED") UNSPECIFIED,
     TEXT,
     IMAGE,
     AUDIO;
@@ -38,21 +36,17 @@ public class ResponseModality private constructor(public val ordinal: Int) {
       when (this) {
         TEXT -> ResponseModality.TEXT
         IMAGE -> ResponseModality.IMAGE
-        AUDIO -> ResponseModality.AUDIO
-        else -> ResponseModality.UNSPECIFIED
+        else -> ResponseModality.AUDIO
       }
   }
 
   internal fun toInternal() =
     when (this) {
       TEXT -> "TEXT"
       IMAGE -> "IMAGE"
-      AUDIO -> "AUDIO"
-      else -> "UNSPECIFIED"
+      else -> "AUDIO"
     }
   public companion object {
-    /** Unspecified modality. */
-    @JvmField public val UNSPECIFIED: ResponseModality = ResponseModality(0)
 
     /** Represents a plain text response modality. */
     @JvmField public val TEXT: ResponseModality = ResponseModality(1)