fix(scripts): bound objectui-range --help's self-read to the header prefix

[yinlianghui]

Fixes#11952

What

objectui-range --help built its usage text from every column-0 //
line in scripts/objectui-range.mjs, not just the header block. Six
mid-file implementation comments (the pinAt() helper's one-line
rationale at old line 129, and this file's own self-test section banner
at old lines 333-337) leaked into the usage text a reader pastes.

Triage's acceptance pin (quoted verbatim in the issue): "--help
output contains the header and none of the six measured stray lines."

Fix

The header is a contiguous prefix of the file (right after the
shebang) — measured on today's origin/main (1e79aa4f8), lines 2-74
are all column-0 //, then line 75 (import { execFileSync } ...)
breaks it, and no other column-0 // line appears again until 129. So
printHelp() now walks the file's lines in order and stops at the
first non-// line once the header block has started — a
takeWhile, not a filter — the repair triage's option 1 named as
cheapest and structurally closing the "add a comment anywhere later"
hazard. The self-read itself is kept (no-drift property preserved) —
only its boundary changed.

Total column-0 // line count is unchanged at 79 (PR #11951's
byte-identity pin); the new rationale prose is a /** */ JSDoc block on
printHelp(), matching #11951's own reason for choosing that comment
style on this file.

Pin

Per the standing requirement to ship the acceptance criterion as a
mechanically-checkable test, not a PR-body claim: extended this file's
own self-test idiom (--self-test, cited from #4843) with two new
checks —

• --help still carries the real header (Usage:, the old-rev/new-rev
  line, OBJECTUI_ROOT=...).
• --help does not contain any of the six measured stray lines
  (the pinAt() rationale and the self-test banner text).

Reverse-verified: committed the fix, then mutated printHelp()
back to the old unbounded .filter() in place (confirmed the mutation
landed on disk via grep -c before running), re-ran --self-test —
the new stray-line check went red (✗ --help does NOT leak mid-file implementation comments), confirming the pin actually detects
the regression it exists to catch. git checkout -- with the target path (via a
trap ... EXIT in the mutation script) restored the fix; working tree
confirmed clean afterward.

Byte diff

Captured --help output before/after into two files and diffed them —
exactly the six stray lines are gone, nothing else moved:

74,79d73
< Resolve the objectui SHA pinned at a given framework rev (or the working tree).
< ---------------------------------------------------------------------------
< Self-test (#4843) — the repo idiom for a `scripts/` gate: build a throwaway
< git repo carrying the exact shapes measured on the real range, run the real
< code over it, assert the ARTIFACT (the markdown a maintainer pastes).
< ---------------------------------------------------------------------------

--help byte length: 4693 → 4237 (content-asserted in the pin, not
byte-count-asserted, per the standing guidance that a byte-count
assertion rots).

Scope

Only scripts/objectui-range.mjs touched. No reformatting/restructuring
elsewhere in the file — git diff origin/main -- scripts/objectui-range.mjs
is exactly two hunks: the printHelp() body + its doc-comment, and the
new self-test block.

Gates (derived via `node scripts/pm/dispatch-gates.mjs --repo

objectstack-ai/objectstackat final HEAD05f5dec`)

All 8 matched local gate families green (exit 0, captured before any
pipe; verdict line quoted from each gate's own output):

• pnpm check:agent-test-spelling — ✓ check-agent-test-spelling: 0 violations — 367 file(s) · 3681 bare '--' token(s) · 1139 launcher-rooted run(s) · 9 separator(s) JUDGED · 5 vitest-backed script name(s) derived from 80 manifest(s)
• pnpm check:cross-package-test-inputs — OK: 16 package(s) read outside themselves, all declared, and turbo.json hashes every declared glob.
• pnpm check:entry-guard — ✓ check:entry-guard: 157 scripts/ file(s) — every entry guard goes through invoked-as.mjs; 113 export bindings, 111 of them inert on import (2 known-unsafe, ⛔ SHRINK-ONLY).
• pnpm check:objectui-changeset (runs this file's own --self-test, including the two new checks) — ✓ objectui-range --self-test: all checks passed
• pnpm check:parse-guard — ✓ check:parse-guard: 156 scripts/ file(s) — every TypeScript parse goes through ts-parse.mjs.
• pnpm check:pnpm-filter-targets — ✓ check:pnpm-filter-targets: 135/168 '--filter' occurrence(s) across 26 file(s) resolve against 78 workspace package(s); ...
• node scripts/check-ci-filter-parity.mjs — OK: all 96 declared cross-package glob(s) (81 unique) are covered by 'core' or 'crosspkg', ...
• node scripts/check-cross-package-test-inputs.mjs — OK: 16 package(s) read outside themselves, all declared, and turbo.json hashes every declared glob.

Plus pnpm check:nul-bytes (any edit warrants it) — check-nul-bytes: OK (scanned 6672 text file(s) ... no raw ASCII control bytes).

pnpm --stack-size=4000 eslint scripts/objectui-range.mjs --no-inline-config --format json — 0 errors, 0 warnings, 1 file linted (scoped-run evidence: population read from eslint's own JSON output, not a guess; this repo's eslint.config.mjs never enables type-aware linting for any file — measured with a positive control in the config's own header comment — so this scoped run cannot have missed a judgment on any untouched file).

No package's typecheck/test covers this file: root tsconfig.json
excludes packages//apps//examples/ and has no allowJs, and
scripts/ is not itself a workspace package, so turbo run typecheck/test never touches it — the file's real correctness gate
is check:objectui-changeset's --self-test above, which does.

Changeset

Root scripts/ tooling only, nothing published — skip-changeset
applies. Labeled and read back — confirmed present on this PR.

---

Generated by Claude Code