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: 915550451

Author: MarkDaoust

google/genai/_interactions/types/interaction.py

@@ -34,6 +34,7 @@
 from .._legacy_lyria import is_legacy_lyria_response_body
 from .webhook_config import WebhookConfig
 from .document_content import DocumentContent
+from .model_output_step import ModelOutputStep
 from .dynamic_agent_config import DynamicAgentConfig
 from .text_response_format import TextResponseFormat
 from .audio_response_format import AudioResponseFormat
@@ -212,3 +213,60 @@ def _coerce_outputs_to_steps(cls, values: Any) -> Any:
             def _coerce_outputs_to_steps(cls, data: Any) -> Any:
                 coerced, _ = cls._maybe_coerce_outputs(data)
                 return coerced
+
+    @property
+    def _last_model_output_steps(self) -> List[ModelOutputStep]:
+        if not self.steps:
+            return []
+        trailing: List[ModelOutputStep] = []
+        for step in reversed(self.steps):
+            if isinstance(step, ModelOutputStep):
+                trailing.append(step)
+            else:
+                break
+        trailing.reverse()
+        return trailing
+
+    @property
+    def output_text(self) -> str:
+        """Concatenated text from the last consecutive model output steps."""
+        parts: List[str] = []
+        for step in self._last_model_output_steps:
+            if step.content:
+                for content in step.content:
+                    if isinstance(content, TextContent):
+                        parts.append(content.text)
+        return "".join(parts)
+
+    @property
+    def output_image(self) -> Optional[ImageContent]:
+        """The last image from the last consecutive model output steps."""
+        result = None
+        for step in self._last_model_output_steps:
+            if step.content:
+                for content in step.content:
+                    if isinstance(content, ImageContent):
+                        result = content
+        return result
+
+    @property
+    def output_audio(self) -> Optional[AudioContent]:
+        """The last audio from the last consecutive model output steps."""
+        result = None
+        for step in self._last_model_output_steps:
+            if step.content:
+                for content in step.content:
+                    if isinstance(content, AudioContent):
+                        result = content
+        return result
+
+    @property
+    def output_video(self) -> Optional[VideoContent]:
+        """The last video from the last consecutive model output steps."""
+        result = None
+        for step in self._last_model_output_steps:
+            if step.content:
+                for content in step.content:
+                    if isinstance(content, VideoContent):
+                        result = content
+        return result