Skip to content

Personal AI, On Personal Devices

Python

fix(cloud): handle parallel Gemini tool calls

Elliot Sluskyf08ad574d30a
diff --git a/src/openjarvis/engine/cloud.py b/src/openjarvis/engine/cloud.py
index fe5172f6..e4485772 100644
--- a/src/openjarvis/engine/cloud.py
+++ b/src/openjarvis/engine/cloud.py
@@ -1350,14 +1350,16 @@ class CloudEngine(InferenceEngine):
args = json.loads(args)
except (json.JSONDecodeError, TypeError):
args = {"input": args}
- function_call: Dict[str, Any] = {
- "name": tool_call.name,
- "args": args if isinstance(args, dict) else {},
+ function_call_part: Dict[str, Any] = {
+ "function_call": {
+ "name": tool_call.name,
+ "args": args if isinstance(args, dict) else {},
+ }
}
signature = self._thought_sigs.get(tool_call.id)
if signature is not None:
- function_call["thought_signature"] = signature
- parts.append({"function_call": function_call})
+ function_call_part["thought_signature"] = signature
+ parts.append(function_call_part)
contents.append({"role": "model", "parts": parts})
elif message.role.value == "assistant":
contents.append({"role": "model", "parts": [{"text": message.content}]})
@@ -1377,7 +1379,7 @@ class CloudEngine(InferenceEngine):
if tools:
config.tools = [{"function_declarations": _convert_tools_to_google(tools)}]
- tool_ids: Dict[str, int] = {}
+ tool_call_count = 0
for chunk in self._google_client.models.generate_content_stream(
model=model,
contents=contents,
@@ -1402,8 +1404,12 @@ class CloudEngine(InferenceEngine):
name = getattr(function_call, "name", "")
raw_args = getattr(function_call, "args", {})
args = dict(raw_args) if hasattr(raw_args, "items") else {}
- tool_id = f"google_{name}"
- tool_index = tool_ids.setdefault(tool_id, len(tool_ids))
+ # Gemini emits complete function-call parts, so each part is
+ # a distinct invocation. The same function may legitimately
+ # be called more than once in a parallel response.
+ tool_index = tool_call_count
+ tool_id = f"google_{name}_{tool_index}"
+ tool_call_count += 1
tool_call = {
"index": tool_index,
"id": tool_id,
@@ -1430,7 +1436,7 @@ class CloudEngine(InferenceEngine):
if text:
yield StreamChunk(content=text)
- yield StreamChunk(finish_reason="tool_calls" if tool_ids else "stop")
+ yield StreamChunk(finish_reason="tool_calls" if tool_call_count else "stop")
async def _stream_openrouter(
self,
diff --git a/tests/engine/test_cloud_stream_full.py b/tests/engine/test_cloud_stream_full.py
index 8adf235e..a57373a0 100644
--- a/tests/engine/test_cloud_stream_full.py
+++ b/tests/engine/test_cloud_stream_full.py
@@ -506,12 +506,12 @@ async def test_stream_full_google_preserves_tool_calls(monkeypatch: pytest.Monke
tool_call = result[0].tool_calls[0]
assert tool_call == {
"index": 0,
- "id": "google_get_weather",
+ "id": "google_get_weather_0",
"type": "function",
"function": {"name": "get_weather", "arguments": '{"city": "Berlin"}'},
"thought_signature": b"sig",
}
- assert engine._thought_sigs["google_get_weather"] == b"sig"
+ assert engine._thought_sigs["google_get_weather_0"] == b"sig"
assert result[-1].finish_reason == "tool_calls"
config = client.models.generate_content_stream.call_args.kwargs["config"]
assert config.tools == [
@@ -545,7 +545,7 @@ async def test_stream_full_google_preserves_mixed_and_multiple_calls(
client.models.generate_content_stream.return_value = iter(
[
_google_stream_chunk(text_part, weather_part),
- _google_stream_chunk(weather_part, calendar_part),
+ _google_stream_chunk(calendar_part),
]
)
engine = _make_cloud_engine(google_client=client)
@@ -566,7 +566,7 @@ async def test_stream_full_google_preserves_mixed_and_multiple_calls(
assert result[1].tool_calls == [
{
"index": 0,
- "id": "google_get_weather",
+ "id": "google_get_weather_0",
"type": "function",
"function": {
"name": "get_weather",
@@ -575,18 +575,9 @@ async def test_stream_full_google_preserves_mixed_and_multiple_calls(
}
]
assert result[2].tool_calls == [
- {
- "index": 0,
- "id": "google_get_weather",
- "type": "function",
- "function": {
- "name": "get_weather",
- "arguments": '{"city": "Berlin"}',
- },
- },
{
"index": 1,
- "id": "google_get_calendar",
+ "id": "google_get_calendar_1",
"type": "function",
"function": {
"name": "get_calendar",
@@ -597,6 +588,94 @@ async def test_stream_full_google_preserves_mixed_and_multiple_calls(
assert result[-1].finish_reason == "tool_calls"
+async def test_stream_full_google_keeps_parallel_same_name_calls_distinct(
+ monkeypatch: pytest.MonkeyPatch,
+):
+ """Parallel invocations of one function receive unique indexes and IDs."""
+ paris = SimpleNamespace(name="get_weather", args={"city": "Paris"})
+ london = SimpleNamespace(name="get_weather", args={"city": "London"})
+ parts = [
+ SimpleNamespace(function_call=paris, text=None, thought_signature=b"sig"),
+ SimpleNamespace(function_call=london, text=None, thought_signature=None),
+ ]
+ client = MagicMock()
+ client.models.generate_content_stream.return_value = iter(
+ [_google_stream_chunk(*parts)]
+ )
+ engine = _make_cloud_engine(google_client=client)
+ engine._thought_sigs = {}
+
+ with monkeypatch.context() as patch:
+ for name, module in _google_types_modules().items():
+ patch.setitem(sys.modules, name, module)
+ result = [
+ chunk
+ async for chunk in engine.stream_full(
+ [Message(role=Role.USER, content="Weather in Paris and London")],
+ model="gemini-3-flash-preview",
+ )
+ ]
+
+ calls = result[0].tool_calls
+ assert [(call["index"], call["id"]) for call in calls] == [
+ (0, "google_get_weather_0"),
+ (1, "google_get_weather_1"),
+ ]
+ assert [call["function"]["arguments"] for call in calls] == [
+ '{"city": "Paris"}',
+ '{"city": "London"}',
+ ]
+
+
+async def test_stream_full_google_replays_signature_on_part(
+ monkeypatch: pytest.MonkeyPatch,
+):
+ """A saved Gemini signature is replayed beside, not inside, function_call."""
+ client = MagicMock()
+ client.models.generate_content_stream.return_value = iter([])
+ engine = _make_cloud_engine(google_client=client)
+ engine._thought_sigs = {"google_get_weather_0": b"sig"}
+ messages = [
+ Message(role=Role.USER, content="weather"),
+ Message(
+ role=Role.ASSISTANT,
+ content=None,
+ tool_calls=[
+ ToolCall(
+ id="google_get_weather_0",
+ name="get_weather",
+ arguments='{"city": "Berlin"}',
+ )
+ ],
+ ),
+ Message(role=Role.TOOL, name="get_weather", content='{"temp": 20}'),
+ ]
+
+ with monkeypatch.context() as patch:
+ for name, module in _google_types_modules().items():
+ patch.setitem(sys.modules, name, module)
+ result = [
+ chunk
+ async for chunk in engine.stream_full(
+ messages, model="gemini-3-flash-preview"
+ )
+ ]
+
+ contents = client.models.generate_content_stream.call_args.kwargs["contents"]
+ assert contents[1]["parts"] == [
+ {
+ "function_call": {
+ "name": "get_weather",
+ "args": {"city": "Berlin"},
+ },
+ "thought_signature": b"sig",
+ }
+ ]
+ assert result[-1].finish_reason == "stop"
+
+
# ---------------------------------------------------------------------------
# stream_full routing tests
# ---------------------------------------------------------------------------