feat: interaction.{output_text,output_image,output_audio,output_video}

Add SDK methods for shortcut properties on the response object to access the output generated in response to the current request, without forcing developers to iterate through the full steps timeline or filter out previous conversation history:

interaction.output_text: Concatenated last consecutive text generated by the model in response to the current request.
.output_image: The last image generated by the model in response to the current request.
.output_audio: The last audio clip generated by the model in response to the current request.
.output_video: The last video generated by the model in response to the current request.

PiperOrigin-RevId: 915804808

Author: MarkDaoust

src/interactions/resources/interactions.ts

@@ -74,13 +74,17 @@ export class BaseInteractions extends APIResource {
       ...options,
       stream: isStreaming,
       ...(needsLegacyLyriaShim && isStreaming ? { __streamClass: LegacyLyriaStream } : {}),
-    }) as APIPromise<Interaction> | APIPromise<Stream<InteractionSSEEvent>>;
-    if (needsLegacyLyriaShim && !isStreaming) {
-      return (promise as APIPromise<Interaction>)._thenUnwrap(
-        (data) => coerceLegacyInteractionResponse(data) as Interaction,
-      );
+    });
+
+    if (isStreaming) {
+      return promise as APIPromise<Stream<InteractionSSEEvent>>;
+    }
+
+    let nonStreaming = promise as APIPromise<Interaction>;
+    if (needsLegacyLyriaShim) {
+      nonStreaming = nonStreaming._thenUnwrap((data) => coerceLegacyInteractionResponse(data) as Interaction);
     }
-    return promise;
+    return nonStreaming._thenUnwrap(addOutputProperties);
   }
 
   /**
@@ -119,7 +123,9 @@ export class BaseInteractions extends APIResource {
     options?: RequestOptions,
   ): APIPromise<Interaction> {
     const { api_version = this._client.apiVersion } = params ?? {};
-    return this._client.post(path`/${api_version}/interactions/${id}/cancel`, options);
+    return (
+      this._client.post(path`/${api_version}/interactions/${id}/cancel`, options) as APIPromise<Interaction>
+    )._thenUnwrap(addOutputProperties);
   }
 
   /**
@@ -154,15 +160,75 @@ export class BaseInteractions extends APIResource {
     options?: RequestOptions,
   ): APIPromise<Interaction> | APIPromise<Stream<InteractionSSEEvent>> {
     const { api_version = this._client.apiVersion, ...query } = params ?? {};
-    return this._client.get(path`/${api_version}/interactions/${id}`, {
+    const response = this._client.get(path`/${api_version}/interactions/${id}`, {
       query,
       ...options,
       stream: params?.stream ?? false,
-    }) as APIPromise<Interaction> | APIPromise<Stream<InteractionSSEEvent>>;
+    });
+
+    if (params?.stream) {
+      return response as APIPromise<Stream<InteractionSSEEvent>>;
+    }
+
+    return (response as APIPromise<Interaction>)._thenUnwrap(addOutputProperties);
   }
 }
 export class Interactions extends BaseInteractions {}
 
+function addOutputProperties(interaction: Interaction): Interaction {
+  const steps = interaction.steps ?? [];
+
+  let firstTrailing = steps.length;
+  while (firstTrailing > 0 && steps[firstTrailing - 1]!.type === 'model_output') {
+    firstTrailing--;
+  }
+  const modelSteps = steps.slice(firstTrailing) as ModelOutputStep[];
+
+  const output = modelSteps.flatMap((step) => step.content ?? []);
+
+  const textParts: string[] = [];
+  for (let i = output.length - 1; i >= 0; i--) {
+    const content = output[i]!;
+    if (content.type !== 'text') break;
+    textParts.push(content.text ?? '');
+  }
+  const output_text = textParts.reverse().join('');
+
+  let output_image: ImageContent | undefined;
+  let output_audio: AudioContent | undefined;
+  let output_video: VideoContent | undefined;
+
+  for (let i = steps.length - 1; i >= 0; i--) {
+    const step = steps[i]!;
+    const anyStep = step as any;
+    if (anyStep.type === 'user_input') {
+      break;
+    }
+    if (anyStep.type === 'model_output' && anyStep.content) {
+      for (let j = anyStep.content.length - 1; j >= 0; j--) {
+        const content = anyStep.content[j]!;
+        if (content.type === 'image' && !output_image) {
+          output_image = content as any;
+        }
+        if (content.type === 'audio' && !output_audio) {
+          output_audio = content as any;
+        }
+        if (content.type === 'video' && !output_video) {
+          output_video = content as any;
+        }
+      }
+    }
+  }
+
+  return {
+    ...interaction,
+    ...(output_text && { output_text }),
+    ...(output_image && { output_image }),
+    ...(output_audio && { output_audio }),
+    ...(output_video && { output_video }),
+  };
+}
+
 /**
  * The configuration for allowed tools.
  */
@@ -1134,6 +1200,34 @@ export interface Interaction {
    * completes.
    */
   webhook_config?: WebhookConfig;
+
+  /**
+   * Concatenated text from the last model output in response to the current request.
+   *
+   * Note: this is added by the SDK.
+   */
+  output_text?: string;
+
+  /**
+   * The last image generated by the model in response to the current request.
+   *
+   * Note: this is added by the SDK.
+   */
+  output_image?: ImageContent;
+
+  /**
+   * The last audio generated by the model in response to the current request.
+   *
+   * Note: this is added by the SDK.
+   */
+  output_audio?: AudioContent;
+
+  /**
+   * The last video generated by the model in response to the current request.
+   *
+   * Note: this is added by the SDK.
+   */
+  output_video?: VideoContent;
 }
 
 export interface InteractionCompletedEvent {