fix(scripts): stop setup-tasks text mode crashing on a legacy stdout code page

[jawwad-ali]

Problem

_check_file / _check_dir in scripts/python/setup_tasks.py hard-code the U+2713 / U+2717 glyphs and print() them to sys.stdout:

marker = "✓" if path.is_file() else "✗"
print(f"  {marker} {description}")

On Windows sys.stdout falls back to the ANSI code page whenever stdout is not a console — a pipe or a file redirect, which is exactly how an agent or a workflow step invokes these scripts. U+2713 is unencodable in cp1252, so the document listing aborts mid-report with UnicodeEncodeError.

Relationship to #3890

This is the byte-identical twin of the block in scripts/python/check_prerequisites.py. I flagged it explicitly in that PR's body rather than widening that PR's scope:

scripts/python/setup_tasks.py:58-65 carries a byte-identical block and crashes the same way. I kept this PR to one script — happy to extend it to the twin here or as a follow-up, whichever you prefer.

This is that follow-up. The two PRs touch different source files and different test files, so they are independent and can merge in either order.

Fix

Downgrade to ASCII only when stdout cannot encode the glyph, so a UTF-8 console is unaffected:

glyph = "✓" if ok else "✗"
try:
    glyph.encode(getattr(sys.stdout, "encoding", None) or "utf-8")
except (LookupError, UnicodeEncodeError):
    return "[OK]" if ok else "[FAIL]"
return glyph

[OK] / [FAIL] is the ASCII rendering these markers already have in this repo:

Site	Code
scripts/powershell/common.ps1:243	Write-Output " [OK] $Description"
scripts/powershell/common.ps1:246	Write-Output " [FAIL] $Description"
tests/parity_helpers.py:130-131	text.replace(" ✓ ", " [OK] ").replace(" ✗ ", " [FAIL] ")

The PowerShell twin emits the ASCII form natively and the shared parity helper maps the glyphs onto it, so the suite already treats the two as equivalent output.

Breaking risk: a UTF-8-capable stdout still gets the glyphs, byte-identical to today. The only case that changes is one that previously raised and truncated the report.

Verification

• Fail-before / pass-after: the new test (running the real script with PYTHONIOENCODING=cp1252) fails on unpatched source and passes with the fix. File: 1 failed → 3 passed, 14 skipped (the skips are the bash-requiring parity cases on Windows).
• Scoped regression: failure set identical to the clean-main baseline captured on 81bf741 (0 pre-existing in scope).
• uvx ruff@0.15.0 check src tests → clean

---

Written with assistance from Claude Code. Bug found, reproduced, and verified by me on current main.