Python: fix(core): match empty expected tool call arguments when actual call arguments are None

python/packages/core/agent_framework/_evaluation.py

@@ -40,7 +40,7 @@
 import json
 import logging
 import warnings
-from collections.abc import Awaitable, Callable, Sequence
+from collections.abc import Awaitable, Callable, Mapping, Sequence
 from dataclasses import dataclass, field
 from enum import Enum
 from typing import (
@@ -1119,15 +1119,21 @@ def _extract_tool_calls(item: EvalItem) -> list[tuple[str, dict[str, Any] | None
         for c in msg.contents or []:
             if c.type == "function_call" and c.name:
                 args: dict[str, Any] | None = None
-                if isinstance(c.arguments, dict):
-                    args = c.arguments
+                if c.arguments is None:
+                    args = {}
+                elif isinstance(c.arguments, Mapping):
+                    args = dict(c.arguments)
                 elif isinstance(c.arguments, str):
-                    try:
-                        parsed = json.loads(c.arguments)
-                        if isinstance(parsed, dict):
-                            args = cast(dict[str, Any], parsed)
-                    except (json.JSONDecodeError, TypeError):
-                        pass
+                    stripped = c.arguments.strip()
+                    if not stripped:
+                        args = {}
+                    else:
+                        try:
+                            parsed = json.loads(stripped)
+                            if isinstance(parsed, dict):
+                                args = cast(dict[str, Any], parsed)
+                        except (json.JSONDecodeError, TypeError):
+                            pass
                 calls.append((c.name, args))
     return calls
 

python/packages/core/tests/core/test_local_eval.py

@@ -6,6 +6,7 @@
 
 import inspect
 import json
+from typing import Any
 
 import pytest
 
@@ -529,7 +530,7 @@ def check_b(response: str) -> bool:
 
 
 def _make_tool_call_item(
-    calls: list[tuple[str, dict | None]],
+    calls: list[tuple[str, Any]],
     expected: list[ExpectedToolCall] | None = None,
 ) -> EvalItem:
     """Build an EvalItem with tool calls in the conversation."""
@@ -644,6 +645,57 @@ def test_null_argument_requires_key(self, arguments, expected_passed, as_json):
 
         assert result.passed is expected_passed
 
+    def test_empty_expected_args_matches_none_actual_args(self):
+        item = _make_tool_call_item(
+            calls=[("get_weather", None)],
+            expected=[ExpectedToolCall("get_weather", {})],
+        )
+        result = tool_call_args_match(item)
+        assert result.passed is True
+
+    def test_empty_expected_args_matches_empty_actual_args(self):
+        item = _make_tool_call_item(
+            calls=[("get_weather", {})],
+            expected=[ExpectedToolCall("get_weather", {})],
+        )
+        result = tool_call_args_match(item)
+        assert result.passed is True
+
+    def test_empty_expected_args_matches_empty_string_actual_args(self):
+        item = _make_tool_call_item(
+            calls=[("get_weather", "")],
+            expected=[ExpectedToolCall("get_weather", {})],
+        )
+        result = tool_call_args_match(item)
+        assert result.passed is True
+
+    def test_empty_expected_args_does_not_match_invalid_json(self):
+        item = _make_tool_call_item(
+            calls=[("get_weather", "not-json")],
+            expected=[ExpectedToolCall("get_weather", {})],
+        )
+        result = tool_call_args_match(item)
+        assert result.passed is False
+        assert "args mismatch" in result.reason
+
+    def test_empty_expected_args_does_not_match_non_object_json(self):
+        item = _make_tool_call_item(
+            calls=[("get_weather", "[]")],
+            expected=[ExpectedToolCall("get_weather", {})],
+        )
+        result = tool_call_args_match(item)
+        assert result.passed is False
+        assert "args mismatch" in result.reason
+
+    def test_empty_expected_args_does_not_match_non_object_payload(self):
+        item = _make_tool_call_item(
+            calls=[("get_weather", [])],
+            expected=[ExpectedToolCall("get_weather", {})],
+        )
+        result = tool_call_args_match(item)
+        assert result.passed is False
+        assert "args mismatch" in result.reason
+
     def test_tool_not_called(self):
         item = _make_tool_call_item(
             calls=[("get_news", None)],