Use cattrs hooks for nested recursive classes in msgspec

[digit50]

closes #795

With the msgspec converter, msgspec_attrs_unstructure_factory generated a class's unstructuring hook while its recursion probe still held classes in the working set. Field handlers that led back to a probed class were looked up during that time and came back as msgspec.to_builtins, so nested instances skipped registered hooks and lost private attributes. I changed both generation points (the private-attribute path and the one after the probe) so that, while any probe is active, the factory returns a hook that generates the real one on its first call, after the outermost probe has finished. Outside a probe the hook is generated immediately as before. Returning converter.unstructure instead is not enough: a namedtuple field caches the hooks of its element types during the probe, and a private class reached that way would then dispatch to itself forever.

The new tests in tests/preconf/test_msgspec_cpython.py cover a self-recursive class with a hooked datetime field, a mutually recursive pair where one class has a private attribute, and a private class reached through a namedtuple field. The first two fail on main with the reported output.

[codspeed]

Merging this PR will degrade performance by 11.35%

ℹ️ 64 benchmarks have no flame graph

The flamegraph could not be generated because we failed to resolve the symbols in the profiling trace. This happens when we fail to harvest information about language frames.

• test_unstructure_simple_int_enum[OrjsonConverter]
• test_unstructure_simple_int_enum[MsgspecJsonConverter]
• test_unstructure_simple_int_enum[Converter]
• test_unstructure_simple_int_enum[BaseConverter]
• test_unstruct_attrs_deep_nest[UnstructureStrategy.AS_DICT-Converter]
• test_unstructure_attrs_nested[UnstructureStrategy.AS_DICT-BaseConverter]
• test_unstructure_attrs_primitives[UnstructureStrategy.AS_DICT-Converter]
• test_structure_attrs_mappings[Converter]
• test_unstructure_attrs_nested[UnstructureStrategy.AS_DICT-Converter]
• test_unstructure_simple_enum[BaseConverter]
• …and 54 more

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

❌ 4 regressed benchmarks
✅ 60 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌ ℹ️	test_unstructure_simple_int_enum[OrjsonConverter]	16 µs	18.2 µs	-12.39%
❌ ℹ️	test_unstructure_simple_int_enum[MsgspecJsonConverter]	16.1 µs	18.2 µs	-11.76%
❌ ℹ️	test_unstructure_simple_int_enum[Converter]	16.3 µs	18.4 µs	-11.18%
❌ ℹ️	test_unstructure_simple_int_enum[BaseConverter]	17.2 µs	19.1 µs	-10.05%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing digit50:fix/msgspec-recursive-hooks (1fe118f) with main (2c6cf5b)¹}

Footnotes

1. No successful run was found on main (73fa273) during the generation of this report, so 2c6cf5b was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

[Tinche]

Pretty sure the performance regressions are unrelated. Can you add a link to the PR to the history entry?

Also I don't think this is the end of the story - for example, Event | None still won't work. I will think on a more general solution in the future.