Read the OpenAPI spec as UTF-8 regardless of process locale

[jeremy]

Unblocks the v0.14.0 cold-verification leg (#700).

The LC_ALL=C hardening pinned encoding: "UTF-8" on reads across scripts/
(see check-projected-examples.rb, validate-api-gaps.rb, …), but the three
Ruby generators under ruby/scripts/ were missed. Under a C locale Ruby's
default external encoding is US-ASCII, and JSON.parse(File.read(openapi_path))
dies on the spec's first multibyte character:

lib/json/common.rb:364:in 'String#encode': "\xE2" on US-ASCII (Encoding::InvalidByteSequenceError)

Surfaced by rb-check-drift failing during a cold make check on a build
server over non-interactive ssh (no locale exported). Notably, rb-check-drift
runs in no CI job — only in local make check — so green CI never
disproved this; the per-step LC_ALL: C pins in test.yml don't reach it
either. Whether the Ruby drift check should join CI (and run under LC_ALL: C
like its hardened siblings) is a real question, but a separate one from this
fix — flagged here rather than smuggled in.

Verified, both directions under LC_ALL=C: the pre-fix generate-metadata.rb
exits 1 with exactly the error above; the pinned generate-metadata.rb and
generate-types.rb exit 0 (generate-services.rb gets exercised by the full
rb-check-drift rerun on the build server once this lands).

---

Summary by cubic

Force UTF-8 when reading the OpenAPI spec in the Ruby generators to prevent JSON parse errors under LC_ALL=C. This fixes rb-check-drift failures in cold make check and unblocks the v0.14.0 cold-verification leg (#700).

• Bug Fixes
  • Read the spec as UTF-8 in ruby/scripts/generate-metadata.rb, ruby/scripts/generate-services.rb, and ruby/scripts/generate-types.rb.
  • Avoids US-ASCII default under C locale that raised Encoding::InvalidByteSequenceError on multibyte characters.

^{Written for commit aec43fa. Summary will update on new commits.}

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[Copilot]

Pull request overview

Ensures Ruby generators parse OpenAPI as UTF-8 under locale-independent builds, unblocking v0.14.0 verification.

Changes:

• Explicitly sets UTF-8 when reading OpenAPI.
• Applies the fix consistently across all three Ruby generators.

Tip

If you aren't ready for review, convert to a draft PR.
Click "Convert to draft" or run gh pr ready --undo.
Click "Ready for review" or run gh pr ready to reengage.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
ruby/scripts/generate-metadata.rb	Reads OpenAPI metadata as UTF-8.
ruby/scripts/generate-services.rb	Reads service definitions as UTF-8.
ruby/scripts/generate-types.rb	Reads schemas as UTF-8.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[jeremy]

Merging on green CI + Copilot's clean review (no comments). Codex remains over its code-review usage limit, as on #698 — absent for quota reasons, not verdict. The fix carries its own red proof: pre-fix generate-metadata.rb exits 1 under LC_ALL=C with the exact Encoding::InvalidByteSequenceError from the cold-check log; the pinned scripts exit 0.