[rudycelekli]

What this changes

--out and --format now require their value, and an unknown --format is rejected right after the argument loop, before the spec module is imported.

Why

Both flags read argv[++index] without checking that it exists. A flag left last became undefined and fell through the ?? chain, so the existing rejection in the same file — unknown format "..."; use one of poster, drawing, detailed or all — was bypassed by the one input that most looks like a mistake.

On main:

$ node scripts/print.mjs examples/raptor-engine/sheet.mjs --format bogus
Error: unknown format "bogus"; use one of poster, drawing, detailed or all   # exit 1

$ node scripts/print.mjs examples/raptor-engine/sheet.mjs --format
wrote raptor-engine-poster-a4.html (3202 KB) and raptor-engine-poster-a4.pdf (599 KB, 1 page)   # exit 0

$ node scripts/print.mjs examples/raptor-engine/sheet.mjs --out
wrote raptor-engine-poster-a4.html (3202 KB) and raptor-engine-poster-a4.pdf (599 KB, 1 page)   # exit 0

No shipped spec sets format, so a bare --format always prints the poster default and exits 0: the sheet is not the one asked for and nothing says so.

Separately, the unknown-format check ran after await import(specFile), so a misspelled format on a figure whose light page is not built was reported as the missing page, or as a module error, rather than as the typo it was.

The later check is kept, because it still guards a format coming from the spec itself.

This is the same fix already recorded under Unreleased for capture.mjs, and matches drive.mjs, lines.mjs, turn-check.mjs and turn-bench.mjs.

Tests

• scripts/print-options.test.mjs

It fails on main by name (# pass 0 # fail 1) and passes here (# pass 1 # fail 0). node --test scripts/print-sheet.test.mjs scripts/print-options.test.mjs is # tests 12 # pass 12 # fail 0.

Valid input still passes validation and is unaffected: --format poster|drawing|detailed|all, --out <name>, and no --format at all.

I did not run the full scripts/*.test.mjs suite (it needs Chrome and the temp react/typescript install). The change touches only main()'s argument parsing in print.mjs, which no other module imports.

Checks

• Every example rebuilds, and git status --short shows only the changes I meant. (No example or kit change; git status --short shows only scripts/print.mjs, scripts/print-options.test.mjs and CHANGELOG.md.)
• No comments in code.
• Kit changes are opt-in, and every existing example is byte-identical. Not applicable: no kit/ change, and no committed page changes.
• I built and looked at both themes. Not applicable: no visual change.
• Figures with pipes, cables or round parts. Not applicable: no figure change.
• node scripts/lines.mjs. Not applicable: no figure change.
• Shaders: not touched.
• 3D: not touched.
• Every number in the figure, the readout, the caption and the docs is true.
• A line under Unreleased in CHANGELOG.md.

Screenshots

No visual change and no committed page changes, so there is nothing to screenshot.

[rahul05ranjan]

Verified this end to end on a clean checkout.

The premise is real: at main, scripts/print.mjs swallows a bare --format (argv[++index] gives undefined, line 225 falls through), the spec loads, and line 234's options.format ?? spec.format ?? "poster" silently picks a fallback — and the misspelled-format check at line 235 only runs after both the spec import and the built-page check, so --format sepia dies with no built page before it ever mentions the bad flag. I reproduced both on main with a minimal spec: the bare flag reaches the page-build error, sepia reaches the same error.

At this branch the behavior is what the changelog promises:

• print.mjs /tmp/sheet-probe.mjs --format → --format needs a value at line 221, before the spec import.
• --format sepia → unknown format "sepia" at line 232, also before the spec import.
• node --test scripts/print-options.test.mjs passes (1/1). Nice touch that the test drives a nonexistent spec file — it actually proves the errors fire before the spec read rather than after.

The --out guard via the shared valueOf is the right shape; no other flag in that loop takes a value.

One nit: the unknown-format message suggests use one of ${FORMATS.join(", ")} or all, but all is handled before the loop at line 231 rather than being a member of FORMATS — correct as written, just worth knowing the list in the message never contains all itself, so the tail or all is doing the work.

From a security standpoint this is a fail-fast correctness fix rather than a boundary issue — print.mjs runs locally with trusted argv — but validating before the spec import also means a crafted spec module is no longer imported when the invocation itself is malformed, which is a reasonable side benefit.