fix: compare tag and attribute names case insensitively

[arpitjain099]

HTML tag and attribute names are ASCII case insensitive, and the parser passes them through exactly as written. Two rules in defaultRuleset compare them raw, so the same violation in a different case goes unreported.

src-not-empty matches the tag with no i flag and the attribute by exact string:

((/^(img|script|embed|bgsound|iframe)$/.test(tagName) === true &&
  attr.name === 'src') ||
  (tagName === 'link' && attr.name === 'href') ||
  (tagName === 'object' && attr.name === 'data')) &&
attr.value === ''

attr-no-duplication keys its seen-set on attr.name directly.

Four spellings of one bug, plus the duplicate-attribute pair, through the CLI on main. The two casing rules are turned off in the config, which is routine: SVG's viewBox, Angular and Vue attribute names and generated markup all trip them legitimately.

L5  <img src="">              The attribute [ src ] of the tag [ img ] must have a value. (src-not-empty)
L9  <div id="a" id="b">       Duplicate of attribute name [ id ] was found. (attr-no-duplication)
L10 <div ID="a" id="b">       The id value [ a ] must be unique. (id-unique)

Scanned 1 files, found 3 errors

Lines 6, 7 and 8 are <img SRC="">, <IMG src=""> and <IMG SRC="">. None of them draws anything. Had line 5 not been in the file it would have exited clean.

After the change the same file reports 7, with all four src-not-empty cases and both attr-no-duplication cases:

L6  <img SRC="">   The attribute [ SRC ] of the tag [ img ] must have a value.
L7  <IMG src="">   The attribute [ src ] of the tag [ img ] must have a value.
L8  <IMG SRC="">   The attribute [ SRC ] of the tag [ img ] must have a value.
L10 <div ID="a" id="b">  Duplicate of attribute name [ id ] was found.

The message still echoes the attribute as the author spelled it, so the output points at the real text.

This is not a new convention for the project: grep -L toLowerCase src/core/rules/*.ts shows 28 of the 44 rules already normalise, and these two were among the 16 that did not.

Verification

Three cases added: the four img/src spellings, uppercase LINK HREF and OBJECT DATA, and href/HREF duplicated on one element. All three fail on main.

npm test goes from 370 to 373 passing with no failures, npm run lint reports no issues and prettier --check is clean on all four touched files.

dist/ is regenerated and included, since a clean build on main produces no churn there and the test suite imports ../../dist/htmlhint.js. If you would rather that be left out of contributions, say so and I will drop it.

[Copilot]

🔵 Needs a closer look

Use ASCII-only normalization and preserve the original tag spelling in diagnostics.

Pull request overview

Updates HTMLHint rules to compare tag and attribute names case-insensitively, with regression tests and regenerated distribution files.

Changes:

• Normalizes names in src-not-empty and attr-no-duplication.
• Adds mixed-case regression tests.
• Regenerates compiled JavaScript.
• Requires ASCII-only normalization and preservation of original diagnostic spelling.

File summaries

File	Summary
test/rules/src-not-empty.spec.js	Tests case-insensitive source checks
test/rules/attr-no-duplication.spec.js	Tests case-insensitive duplicate detection
src/core/rules/src-not-empty.ts	Case-insensitive tag and attribute matching
src/core/rules/attr-no-duplication.ts	Case-insensitive duplicate detection
dist/core/rules/src-not-empty.js	Regenerated rule output
dist/core/rules/attr-no-duplication.js	Regenerated rule output

Review details

Suppressed comments (3)

src/core/rules/attr-no-duplication.ts:19

• String.prototype.toLowerCase() performs Unicode case mappings, but HTML attribute names are only ASCII case-insensitive. For example, the distinct attribute names U+0130 (Latin capital I with dot) and i followed by U+0307 both normalize to the latter here and will be reported as duplicates. Fold only A–Z (and add a regression case) so non-ASCII attribute names remain distinct.

        attrName = attr.name.toLowerCase()

src/core/rules/src-not-empty.ts:11

• String#toLowerCase() applies Unicode case folding, but HTML name matching is ASCII case-insensitive. This can cause false positives for non-ASCII names (for example, JavaScript lowercases K to k, so a <div k="a" K="b"> pair can be reported as duplicate, and Unicode characters such as ſ can make a non-HTML tag match a built-in tag). Use an ASCII-only A–Z normalization for both tagName and attrName (and add a regression case) so only ASCII case variants compare equal.

      const tagName = event.tagName.toLowerCase()

src/core/rules/src-not-empty.ts:11

• The normalized tagName is also interpolated into the existing diagnostic, so uppercase markup now reports tag [ img ] instead of preserving the source spelling (tag [ IMG ]). Keep the normalized value for comparisons but use event.tagName when constructing the message, matching the existing preservation of attr.name.

      const tagName = event.tagName.toLowerCase()

• Files reviewed: 4/6 changed files
• Comments generated: 0
• Review effort level: Lite

---

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