Optimize MutableProxy dataclasses detection on hot-path

[masenf]

Type of change

• Bug fix (non-breaking change which fixes an issue)

Changes To Core Features

• Have you added an explanation of what your changes do and why you'd like us to include them?
• Have you written new tests for your core changes, as applicable?

Description

This PR optimizes the performance of state var reads through MutableProxy by improving the _is_called_from_dataclasses_internal() check, which runs on every element access in the hot-path.

Changes:

1. Cache dataclasses module filename (reflex/istate/proxy.py):

   • Store dataclasses.__file__ in a module-level constant _DATACLASSES_FILE to avoid repeated attribute lookups
   • Replace inspect.getfile(frame) with direct frame.f_code.co_filename access, avoiding the type-dispatch overhead of inspect.getfile() which dominates this per-element read hot-path
   • Updated comments to explain the performance rationale

2. Add comprehensive benchmarks (tests/benchmarks/test_state_proxy.py):

   • New benchmark suite measuring proxy read overhead across different var types:
     • Non-mutable scalar reads (baseline, no proxy)
     • Mutable list iteration (proxy wrapping each element)
     • Mutable dict indexing (proxy wrapping each value)
     • Mutable list of dataclasses (recursive proxy wrapping with dataclasses detection)
   • Parametrized fixture for easy comparison of access patterns
   • Exercises both the direct-return path (scalars) and the proxy-wrapping path (mutable types)

The optimization targets the per-element proxy read hot-path where _is_called_from_dataclasses_internal() is called repeatedly during iteration or indexing of mutable vars. By eliminating the expensive inspect.getfile() call and caching the filename, we reduce overhead on every element access.

Test Plan

• Added tests/benchmarks/test_state_proxy.py with parametrized benchmarks covering scalar, list, dict, and dataclass list access patterns
• Benchmarks can be run with uv run pytest tests/benchmarks/test_state_proxy.py -v to measure the optimization impact
• Existing unit tests continue to pass, ensuring no behavioral changes

https://claude.ai/code/session_01MWnJui2xhcgH5GsXaJqta5

[codspeed-hq]

Merging this PR will not alter performance

✅ 24 untouched benchmarks
🆕 4 new benchmarks

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
🆕	test_var_access[mutable_dataclass_list]	N/A	206.8 ms	N/A
🆕	test_var_access[mutable_dict]	N/A	88 ms	N/A
🆕	test_var_access[mutable_list]	N/A	70.9 ms	N/A
🆕	test_var_access[non_mutable_scalar]	N/A	57.4 ms	N/A

---

_{Comparing claude/gracious-johnson-6FlzO (e7d9eb2) with main (bb8b6f5)}

[greptile-apps]

Greptile Summary

This PR optimizes the _is_called_from_dataclasses_internal() hot-path check in MutableProxy by caching dataclasses.__file__ at import time and reading frame.f_code.co_filename directly instead of calling inspect.getfile(), which carries type-dispatch overhead on every proxy element access.

• reflex/istate/proxy.py: Adds _DATACLASSES_FILE module-level constant and swaps inspect.getfile(frame) == dataclasses.__file__ for frame.f_code.co_filename == _DATACLASSES_FILE, reducing per-element overhead during list/dict iteration through a proxy.
• tests/benchmarks/test_state_proxy.py: Adds a pytest-codspeed benchmark suite (scalar, list, dict, and dataclass-list access patterns) matching the conventions of the existing benchmark files in the project.

Confidence Score: 5/5

Safe to merge; the optimization is a drop-in replacement that behaves identically to the previous code in all realistic CPython deployments.

The change swaps inspect.getfile(frame) for the direct frame.f_code.co_filename attribute read. In CPython, inspect.getfile() for a frame ultimately resolves to co_filename after type-dispatch and a getsourcefile roundtrip, so the comparison result is identical in any environment where dataclasses.__file__ is already the source .py path (the normal case). In .pyc-only environments both old and new code fail the comparison the same way — no regression is introduced. The benchmark additions follow the exact import and fixture pattern already used in tests/benchmarks/test_compilation.py and rely on pytest-codspeed, which is a declared test dependency.

No files require special attention.

Important Files Changed

Filename	Overview
reflex/istate/proxy.py	Caches dataclasses.__file__ into _DATACLASSES_FILE and replaces inspect.getfile(frame) with frame.f_code.co_filename in the hot-path check; semantically equivalent in all common CPython environments.
tests/benchmarks/test_state_proxy.py	New benchmark suite following the existing project pattern; imports pytest-codspeed (already a listed dev dependency), correctly exercises both the direct-return scalar path and the MutableProxy per-element wrap path.

_{Reviews (1): Last reviewed commit: "perf(proxy): avoid inspect.getfile on Mu..." | Re-trigger Greptile}