fix: Filter out video events with inline data from being stored in session

Co-authored-by: Kathy Wu <wukathy@google.com>
PiperOrigin-RevId: 911614336

Author: wukath

src/google/adk/flows/llm_flows/contents.py

@@ -795,8 +795,8 @@ def _is_request_input_event(event: Event) -> bool:
   return _is_function_call_event(event, REQUEST_INPUT_FUNCTION_CALL_NAME)
 
 
-def _is_live_model_audio_event_with_inline_data(event: Event) -> bool:
-  """Check if the event is a live/bidi audio event with inline data.
+def _is_live_model_media_event_with_inline_data(event: Event) -> bool:
+  """Check if the event is a live/bidi media event (audio, video, image) with inline data.
 
   There are two possible cases and we only care about the second case:
   content=Content(
@@ -826,12 +826,14 @@ def _is_live_model_audio_event_with_inline_data(event: Event) -> bool:
   if not event.content or not event.content.parts:
     return False
   for part in event.content.parts:
-    if (
-        part.inline_data
-        and part.inline_data.mime_type
-        and part.inline_data.mime_type.startswith('audio/')
-    ):
-      return True
+    if part.inline_data and part.inline_data.mime_type:
+      mime = part.inline_data.mime_type.lower()
+      if (
+          mime.startswith('audio/')
+          or mime.startswith('video/')
+          or mime.startswith('image/')
+      ):
+        return True
   return False
 
 

src/google/adk/runners.py

@@ -786,17 +786,17 @@ async def _compute_artifact_delta_for_rewind(
 
   def _should_append_event(self, event: Event, is_live_call: bool) -> bool:
     """Checks if an event should be appended to the session."""
-    # Don't append audio response from model in live mode to session.
+    # Don't append media (audio/video/image) response from model in live mode to session.
     # The data is appended to artifacts with a reference in file_data in the
-    # event.
+    # event if save_live_blob is True.
     # We should append non-partial events only.For example, non-finished(partial)
     # transcription events should not be appended.
     # Function call and function response events should be appended.
     # Other control events should be appended.
-    if is_live_call and contents._is_live_model_audio_event_with_inline_data(
+    if is_live_call and contents._is_live_model_media_event_with_inline_data(
         event
     ):
-      # We don't append live model audio events with inline data to avoid
+      # We don't append live model media events with inline data to avoid
       # storing large blobs in the session. However, events with file_data
       # (references to artifacts) should be appended.
       return False

tests/unittests/test_runners.py

@@ -1188,6 +1188,35 @@ def test_should_append_event_other_event(self):
     )
     assert self.runner._should_append_event(event, is_live_call=True) is True
 
+  def test_should_not_append_event_live_model_video(self):
+    event = Event(
+        invocation_id="inv1",
+        author="model",
+        content=types.Content(
+            parts=[
+                types.Part(
+                    inline_data=types.Blob(data=b"123", mime_type="video/mp4")
+                )
+            ]
+        ),
+    )
+    assert self.runner._should_append_event(event, is_live_call=True) is False
+
+  def test_should_append_event_non_live_model_video(self):
+    event = Event(
+        invocation_id="inv1",
+        author="model",
+        content=types.Content(
+            parts=[
+                types.Part(
+                    inline_data=types.Blob(data=b"123", mime_type="video/mp4")
+                )
+            ]
+        ),
+    )
+    assert self.runner._should_append_event(event, is_live_call=False) is True
+
+
 
 @pytest.fixture
 def user_agent_module(tmp_path, monkeypatch):