fix: raise clear error on unsupported language in execute_code

[isheng-eqi]

Summary

Fixes #37874

Problem

CodeExecutor.execute_code() maps CodeLanguage to sandbox runtime via code_language_to_running_language.get(language). When the language is unsupported, .get() returns None and the request is sent with "language": null, which leads to a cryptic remote sandbox error instead of a clear local failure.

execute_workflow_code_template() already fails fast with CodeExecutionError("Unsupported language ...") when no transformer exists, but execute_code itself has no equivalent guard.

Fix

Add an explicit guard at the start of execute_code():

running_language = cls.code_language_to_running_language.get(language)
if running_language is None:
    raise CodeExecutionError(f"Unsupported language {language}")

Then use the validated running_language in the sandbox request data.

Changes

• api/core/helper/code_executor/code_executor.py: Add unsupported language guard (5 lines added, 1 changed)

Mathes the existing pattern in execute_workflow_code_template (line 145-146).

[github-actions]

Pyrefly Type Coverage

Metric	Base	PR	Delta
Type coverage	52.39%	52.39%	0.00%
Strict coverage	51.90%	51.90%	0.00%
Typed symbols	32,120	32,120	0
Untyped symbols	29,465	29,465	0
Modules	2974	2974	0