refactor: clean up test examples

Author: tradeqvest

tests/models/test_anthropic.py

@@ -668,10 +668,7 @@ async def my_tool(first: str, second: str) -> int:
                 input_tokens=20,
                 output_tokens=5,
                 tool_calls=1,
-                details={
-                    'input_tokens': 20,
-                    'output_tokens': 5,
-                },
+                details={'input_tokens': 20, 'output_tokens': 5},
             )
         )
         assert tool_called

tests/test_examples.py

@@ -645,7 +645,6 @@ async def model_logic(  # noqa: C901
             parts=[ToolCallPart(tool_name='final_result', args=args, tool_call_id='pyd_ai_tool_call_id')]
         )
     elif isinstance(m, ToolReturnPart) and m.tool_name == 'do_work':
-        # Trigger a second tool call to cause tool_calls_limit to be exceeded in the docs example
         return ModelResponse(parts=[ToolCallPart(tool_name='do_work', args={}, tool_call_id='pyd_ai_tool_call_id')])
     elif isinstance(m, RetryPromptPart) and m.tool_name == 'calc_volume':
         return ModelResponse(

tests/test_usage_limits.py

@@ -253,18 +253,6 @@ async def ret_a(x: str) -> str:
     assert result.usage() == snapshot(RunUsage(requests=2, input_tokens=103, output_tokens=14, tool_calls=1))
 
 
-def test_deprecated_usage_limits():
-    with warns(
-        snapshot(['DeprecationWarning: `request_tokens_limit` is deprecated, use `input_tokens_limit` instead'])
-    ):
-        assert UsageLimits(input_tokens_limit=100).request_tokens_limit == 100  # type: ignore
-
-    with warns(
-        snapshot(['DeprecationWarning: `response_tokens_limit` is deprecated, use `output_tokens_limit` instead'])
-    ):
-        assert UsageLimits(output_tokens_limit=100).response_tokens_limit == 100  # type: ignore
-
-
 async def test_output_tool_not_counted() -> None:
     """Test that output tools are not counted in tool_calls usage metric."""
     test_agent = Agent(TestModel())
@@ -308,3 +296,15 @@ async def flaky_tool(x: str) -> str:
     # The tool was called twice (1 failure + 1 success), but only the successful call should be counted
     assert call_count == 2
     assert result.usage() == snapshot(RunUsage(requests=3, input_tokens=176, output_tokens=29, tool_calls=1))
+
+
+def test_deprecated_usage_limits():
+    with warns(
+        snapshot(['DeprecationWarning: `request_tokens_limit` is deprecated, use `input_tokens_limit` instead'])
+    ):
+        assert UsageLimits(input_tokens_limit=100).request_tokens_limit == 100  # type: ignore
+
+    with warns(
+        snapshot(['DeprecationWarning: `response_tokens_limit` is deprecated, use `output_tokens_limit` instead'])
+    ):
+        assert UsageLimits(output_tokens_limit=100).response_tokens_limit == 100  # type: ignore