BUGFIX: template-require-valid-alt-text — reject empty-string aria-label/labelledby/alt on <input type=image>, <object>, <area>#2730
Merged
NullVoxPopuli merged 12 commits intoember-cli:masterfrom Apr 25, 2026
Conversation
…aria-labelledby/alt Before: for <input type="image">, <object>, and <area>, the rule checked only for the PRESENCE of an accessible-name fallback attribute (aria-label / aria-labelledby / alt / title). An empty-string value provides no accessible name but slipped past. Fix: add hasNonEmptyTextAttr() that requires the attribute's static value to be non-whitespace. Dynamic values (mustache, concat) remain accepted — we can't tell at lint time whether they resolve to empty. <img>'s alt handling is unchanged — alt="" is still valid there (spec-defined marker for decorative images). Nine new invalid tests cover the three elements × three fallback attrs.
Translates 41 cases from peer-plugin rules:
- jsx-a11y alt-text
- vuejs-accessibility alt-text
- lit-a11y alt-text
Fixture documents parity after this fix:
- Empty-string aria-label/aria-labelledby on <object>, <area>, and
<input type=image> is now flagged (reusing existing objectMissing /
areaMissing / inputImage messageIds).
Remaining divergences (<img alt role=presentation> accepting non-empty
alt in jsx-a11y, <img aria-label> without alt) are annotated inline.
This was referenced Apr 21, 2026
…verage (Copilot review) - JSDoc for hasNonEmptyTextAttr() rewritten: no longer overstates the guarantee for dynamic values, and notes that aria-labelledby IDREFs are not validated. - Added invalid-case coverage for whitespace-only aria-label / aria-labelledby / title — ACCNAME 1.2 §4.3.2 step 2D. - hasNonEmptyTextAttr() already trims static values, so the new whitespace-only cases flag without further rule changes.
johanrd
force-pushed
the
fix/alt-text-empty-aria-label
branch
from
c4885d4 to
82022bb
Compare
…scope, area fallback test, HBS parity
…t-a11y-behavior.md (Copilot review) That doc was never checked in. Remove the dangling reference and note that divergences are captured inline (grep for 'DIVERGENCE —') plus in PR descriptions — matches how the other peer-parity fixtures describe their own divergence records.
…ared helper (Copilot review)
Extract a new `getStaticAttrValue` util that resolves literal-valued
mustaches (`{{"foo"}}`, `{{true}}`, `{{-1}}`) and single-part concat
statements (`"{{true}}"`) to their static string value. `hasNonEmptyTextAttr`
now delegates to the helper — `aria-label={{""}}` / `aria-label="{{""}}"`
normalise to the empty string and flag the same as the text-node equivalent;
genuinely dynamic values (PathExpressions, multi-part concat) still
short-circuit to "assume truthy". Closes the bypass where authors wrapped
an empty accessible name in mustaches.
Byte-identical carrier of lib/utils/static-attr-value.js across all PRs
that land it.
…op audit fixture Upstream maintainers don't want the per-PR `tests/audit/peer-parity` pattern. Port one case that pinned distinct behavior: - `<img alt=" " />` as VALID — whitespace-only alt is currently treated as decorative; jsx-a11y agrees. All other audit cases were already covered by the regular tests on this branch (extensive existing coverage of object/area/input variants, empty aria-label/labelledby, presentation-role conflicts, etc).
| // Whitespace-only alt — pin our current behavior. Peer plugins | ||
| // (jsx-a11y) accept this; we don't trim before considering "empty alt". | ||
| '<template><img alt=" " /></template>', | ||
|
|
Contributor
There was a problem hiding this comment.
But do we want this to be valid? Is it valuable?
Contributor
There was a problem hiding this comment.
Maybe it's fine, to two cases above are also empty
NullVoxPopuli
approved these changes
Apr 25, 2026
This was referenced Apr 25, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Note
This is part of a series where Claude has audited
eslint-plugin-emberagainst jsx-a11y, vuejs-accessibility, angular-eslint, lit-a11y and html-validate,ember-template-lint, and the HTML and WCAG specs.Summary
aria-label=""contributes no accessible name per ACCNAME 1.2 §4.3.2 step 2D: "…has an aria-label attribute whose value is not undefined, not the empty string, nor, when trimmed of whitespace, is not the empty string." Emptyaria-labelledby=""likewise contributes no name per ACCNAME 1.2 §4.3.2 step 2B, which requires the attribute to "contain at least one valid IDREF" — an empty value references none. Emptytitle=""provides no useful accessible name. ACCNAME 1.2 step 2I (Tooltip) literally says "if the current node has a Tooltip attribute, return its value" — there is no emptiness check at this step, so an empty title returns the empty string as a candidate name. That empty name does not satisfy the host-language conformance requirements that follow. For<input type="image">,<object>, and<area>, WCAG SC 4.1.2 states that "all user interface components have a name and role that can be programmatically determined" — an empty fallback value leaves no programmatically determinable name.<input type="image" aria-label="" />,<object aria-labelledby="" />, and<area title="" />are accepted.aria-label,aria-labelledby, andtitleon<input type="image">,<object>, and<area>. Dynamic values (mustache, concat) stay accepted — we can't know at lint time whether they resolve to empty.Fix: add
hasNonEmptyTextAttr()that requires a static value to be non-whitespace.<img>'salt=""is unchanged — an emptyalton<img>is spec-defined as a marker for decorative images.Nine new invalid tests (3 elements × 3 fallback attrs) cover the fix.
Prior art
Verified each peer in source:
alt-textariaLabelHasValue(lines 37-46) rejectsundefinedandlength === 0. Called forobject,area, andinput[type="image"]. Does NOT trim whitespace (differs from our rule which does).alt-texthasAriaLabel(node)+getElementAttributeValuetruthiness. Empty strings are falsy → flagged.alt-textisValidObjectNode/isValidAreaNode/isValidInputNodeall return true on mere attribute-name presence (same bug class this PR fixes).alt-text!elementHasAttribute(...). Also: no<object>or<area>handler — rule only checksimg,input[type=image],role="img".