fix(a11y): define preset aria-pressed semantics for typed values

[seonghobae]

Current exact boundary

• protected base: main@47c6fd27de13b0da37a7db64697b869941909351
• exact head: 67f7386bd3f10206a7265d3390ae0451d3cf3d86
• compare: ahead 7 / behind 0; merge base is exact protected main
• lifecycle: Draft / causal source repair landed / repository-wide doctrine restored / hosted + browser evidence pending
• effective range is exactly four product/test/documentation files: CHANGELOG.md, saas_web.py, tests/test_preset_accessibility_contract.py, tests/test_saas_web.py.

RED → source GREEN

The prior handler used parseInt(this.value, 10). A value such as 26214400.9 was therefore truncated to 26214400, so the 25 MiB preset could expose aria-pressed="true" even though the numeric control did not exactly equal that preset and native step validity was false.

A focused regression was committed first (45f3ed006b1cc9637d8a729a0ed15e2e5802af41). Production repair then landed as normal descendant 7c3dc5536c723179f2add38bca69418c8ae757fd:

• both single and batch inputs declare integer-byte step="1";
• both handlers use exact Number(this.value) instead of lossy integer parsing;
• a preset is pressed only when the field is non-empty, finite, currently valid, and exactly equals that preset;
• decimal/step-invalid input leaves every preset unpressed and exposes aria-invalid with an explicit whole-byte message;
• empty-state and existing required-field cleanup remain intact.

CHANGELOG.md remains code-current for the buyer-visible repair.

Single-writer / doctrine repair

Fresh effective-range inspection found that the branch still carried a .jules/palette.md entry turning this local e.isTrusted/preset implementation decision into repository-wide doctrine, while the PR body claimed no such generated doctrine remained. That is a scope/ownership repair finding, not product semantics.

The current branch was re-read immediately before mutation. Normal child 67f7386bd3f10206a7265d3390ae0451d3cf3d86 restores .jules/palette.md byte-for-byte to protected main using the protected blob and advances the branch with force=false. No product/test delta was removed. Fresh three-dot compare confirms .jules/palette.md is no longer in the effective PR range.

Remaining acceptance

The rendered-HTML regression pins the causal contract, but it is not a substitute for real browser/a11y execution. Before Ready:

• exact preset click and typed exact preset value must select exactly one matching button;
• non-preset integer, decimal/step-invalid, empty, and invalid input must leave all presets unpressed in both single and batch forms;
• keyboard and pointer activation must converge;
• browser validity state and accessibility-tree aria-pressed/aria-invalid evidence must agree;
• all live exact-head required checks/reviews must be terminal-valid.

Fresh workflows on 67f7386bd3f10206a7265d3390ae0451d3cf3d86 are all non-terminal: CodeQL PR 33956920552, SAST Semgrep 33956920566, Security Scan 33956920649, CI 33956920627, fuzz 33956920557 are queued. Current inline review-thread inventory is empty. No predecessor GREEN transfers.

Delivery Gate

• 의도성: PASS — preset state follows the admitted target-byte value rather than event provenance.
• 기능 완전성: PARTIAL — deterministic decimal-truncation source defect is repaired; real browser click/type/keyboard evidence remains.
• 콘텐츠 적합성: PASS — repository-wide generated doctrine is restored to protected authority; effective change is limited to product/test/CHANGELOG.
• 복원력: PARTIAL — single/batch invalid-state parity is source-covered; current-browser execution remains pending.
• 증거성: PARTIAL — focused regression, code-current CHANGELOG, exact protected-base compare and fresh workflows exist; hosted checks and accessibility-tree evidence are not terminal.
• 고유성: N/A — state-semantics repair, not brand-expression work.

Keep Draft until one unchanged exact head satisfies browser/a11y and live hosted/review gates. No self-approval, bypass, gate weakening, source-neutral retrigger, force-push, or destructive rebase.

[google-labs-jules]

👋 Jules, reporting for duty! I'm here to lend a hand with this pull request.

When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down.

I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job!

For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with @jules. You can find this option in the Pull Request section of your global Jules UI settings. You can always switch back!

New to Jules? Learn more at jules.google/docs.

---

For security, I will only act on instructions from the user who triggered this task.

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}