fix(coding): write user-defined functions module as UTF-8

[genisis0x]

Summary

LocalCommandLineCodeExecutor._setup_functions writes the generated functions module to disk via:

func_file.write_text(func_file_content)

Path.write_text defaults to locale.getpreferredencoding(), which is cp1252 on most Windows installs. Any user-supplied function whose source contains a non-cp1252 glyph crashes mid-write with UnicodeEncodeError. Common triggers:

• non-Latin docstring (CJK, Cyrillic, Arabic, accented Latin)
• emoji in a docstring or string literal
• smart quotes / em-dashes in an Annotated[..., Field(description=\"…\")] parameter
• accented identifier (rare but legal in Python 3)

The first call also crashes silently in the sense that _setup_functions_complete is left at False, so the next execute_code_blocks invocation re-runs the same broken write path.

Same defect class as #1731 / #2814 / #2815, on the LocalCommandLineCodeExecutor surface.

Fix

Pass encoding=\"utf-8\" explicitly to the write_text call. Inline comment explains why so a future refactor doesn't drop the kwarg. No public API change; no new options.

Test

test/coding/test_user_defined_functions.py gets two regression tests:

• test_setup_functions_writes_module_as_utf8 drives _setup_functions with a FunctionWithRequirements.from_str source containing Café, em-dash, ☕, 例, and 🚀; reads the resulting module as raw bytes and decodes as UTF-8 explicitly.
• test_setup_functions_passes_utf8_encoding_to_write_text patches Path.write_text and asserts encoding='utf-8' is in the kwargs of the first call inside _setup_functions. This guards the contract on POSIX, where the platform default is already UTF-8 and the byte-level test would otherwise pass even if the kwarg were dropped.

Both tests tolerate the downstream subprocess syntax-check call _setup_functions runs after the write — it depends on a python executable being on PATH, not guaranteed in every environment, and the write we care about has already happened by the time the subprocess fires.

.venv/bin/python -m pytest test/coding/test_user_defined_functions.py::test_setup_functions_writes_module_as_utf8 test/coding/test_user_defined_functions.py::test_setup_functions_passes_utf8_encoding_to_write_text reports 2 passed. The byte-level test confirms the round-trip; the patch-based test confirms the kwarg, and reverting the production change makes the second one fail on POSIX.

[marklysze]

Correct fix — func_file.write_text(func_file_content, encoding="utf-8") is the minimal change needed to survive any non-cp1252 glyph in a user-supplied function's source. The parametrized test with a realistic payload (emoji, CJK, diacritics) is solid. LGTM.

[marklysze]

Clean, targeted fix. Path.write_text defaulting to locale.getpreferredencoding() is a well-known Windows footgun — good catch. The test correctly mocks locale.getpreferredencoding to simulate a cp1252 environment and verifies the write succeeds with non-cp1252 content. Approved.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
autogen/coding/local_commandline_code_executor.py	89.47% <100.00%> (-1.98%)	⬇️

... and 67 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.