Python: Fix tool normalization and provider sample consolidation (#3953)

* Fix tool normalization and provider samples

- restore callable/single-tool normalization paths and unset tool-choice behavior\n- consolidate and expand chat/provider samples (OpenAI/Azure/Anthropic/Ollama/Bedrock)\n- migrate Bedrock lazy import surface to agent_framework.amazon and move provider samples

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* small fix in sample

* Finalize provider, samples, and core cleanup

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* Fix CopilotTool passthrough in agent

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix link

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

python/packages/core/tests/core/test_function_invocation_logic.py

@@ -56,6 +56,36 @@ async def test_base_client_with_function_calling(chat_client_base: SupportsChatG
     assert response.messages[2].text == "done"
 
 
+async def test_base_client_with_function_calling_tools_in_kwargs(chat_client_base: SupportsChatGetResponse):
+    exec_counter = 0
+
+    @tool(name="test_function", approval_mode="never_require")
+    def ai_func(arg1: str) -> str:
+        nonlocal exec_counter
+        exec_counter += 1
+        return f"Processed {arg1}"
+
+    chat_client_base.run_responses = [
+        ChatResponse(
+            messages=Message(
+                role="assistant",
+                contents=[
+                    Content.from_function_call(call_id="1", name="test_function", arguments='{"arg1": "value1"}')
+                ],
+            )
+        ),
+        ChatResponse(messages=Message(role="assistant", text="done")),
+    ]
+
+    response = await chat_client_base.get_response("hello", tools=[ai_func])
+
+    assert exec_counter == 1
+    assert len(response.messages) == 3
+    assert response.messages[1].role == "tool"
+    assert response.messages[1].contents[0].type == "function_result"
+    assert response.messages[1].contents[0].result == "Processed value1"
+
+
 @pytest.mark.parametrize("max_iterations", [3])
 async def test_base_client_with_function_calling_resets(chat_client_base: SupportsChatGetResponse):
     exec_counter = 0

python/packages/core/tests/core/test_types.py

@@ -921,8 +921,8 @@ def test_chat_options_tool_choice_validation():
     }
     assert validate_tool_mode({"mode": "none"}) == {"mode": "none"}
 
-    # None should return mode==none
-    assert validate_tool_mode(None) == {"mode": "none"}
+    # None should remain unset
+    assert validate_tool_mode(None) is None
 
     with raises(ContentError):
         validate_tool_mode("invalid_mode")

python/packages/core/tests/openai/test_openai_assistants_client.py

@@ -701,6 +701,7 @@ def test_prepare_options_basic(mock_async_openai: MagicMock) -> None:
     assert run_options["model"] == "gpt-4"
     assert run_options["temperature"] == 0.7
     assert run_options["top_p"] == 0.9
+    assert "tool_choice" not in run_options
     assert tool_results is None
 
 
@@ -733,6 +734,52 @@ def test_prepare_options_with_tool_tool(mock_async_openai: MagicMock) -> None:
     assert run_options["tool_choice"] == "auto"
 
 
+def test_prepare_options_with_tools_without_tool_choice(mock_async_openai: MagicMock) -> None:
+    """Test _prepare_options keeps tool_choice unset when not provided."""
+
+    client = create_test_openai_assistants_client(mock_async_openai)
+
+    @tool(approval_mode="never_require")
+    def test_function(query: str) -> str:
+        """A test function."""
+        return f"Result for {query}"
+
+    options = {
+        "tools": [test_function],
+    }
+
+    messages = [Message(role="user", text="Hello")]
+    run_options, _ = client._prepare_options(messages, options)  # type: ignore
+
+    assert "tools" in run_options
+    assert "tool_choice" not in run_options
+
+
+def test_prepare_options_with_single_tool_tool(mock_async_openai: MagicMock) -> None:
+    """Test _prepare_options with a single FunctionTool (non-sequence)."""
+    client = create_test_openai_assistants_client(mock_async_openai)
+
+    @tool(approval_mode="never_require")
+    def test_function(query: str) -> str:
+        """A test function."""
+        return f"Result for {query}"
+
+    options = {
+        "tools": test_function,
+        "tool_choice": "auto",
+    }
+
+    messages = [Message(role="user", text="Hello")]
+    run_options, tool_results = client._prepare_options(messages, options)  # type: ignore
+
+    assert "tools" in run_options
+    assert len(run_options["tools"]) == 1
+    assert run_options["tools"][0]["type"] == "function"
+    assert "function" in run_options["tools"][0]
+    assert run_options["tool_choice"] == "auto"
+    assert tool_results is None
+
+
 def test_prepare_options_with_code_interpreter(mock_async_openai: MagicMock) -> None:
     """Test _prepare_options with code interpreter tool."""
     client = create_test_openai_assistants_client(mock_async_openai)

python/packages/core/tests/openai/test_openai_chat_client.py

@@ -190,6 +190,21 @@ def test_unsupported_tool_handling(openai_unit_test_env: dict[str, str]) -> None
     assert result["tools"] == [dict_tool]
 
 
+def test_prepare_tools_with_single_function_tool(openai_unit_test_env: dict[str, str]) -> None:
+    """Test that a single FunctionTool is accepted for tool preparation."""
+    client = OpenAIChatClient()
+
+    @tool(approval_mode="never_require")
+    def test_function(query: str) -> str:
+        """A test function."""
+        return f"Result for {query}"
+
+    result = client._prepare_tools_for_openai(test_function)
+    assert "tools" in result
+    assert len(result["tools"]) == 1
+    assert result["tools"][0]["type"] == "function"
+
+
 @tool(approval_mode="never_require")
 def get_story_text() -> str:
     """Returns a story about Emily and David."""