@@ -7,25 +7,22 @@ const ignoresAndknownFalsePositives = require( '../utils/accessibility-ignores-a
  * Global accessibility scan rules.
  * Applied to all pages defined in utils/accessibility-pages.
  *
+ * See the Axe Options parameter documentation.
  * See: https://github.com/dequelabs/axe-core/blob/master/doc/API.md#options-parameter
  */
 const globalRules = {
 	runOnly: [ 'wcag2a', 'wcag2aa', 'best-practice' ],
+	absolutePaths: true, // Report the absolute CSS target selector for better exclusions mechanism.
 };
 
 /**
  * Filters violations to remove violations we want to intentionally ignore and known false positives.
  * Only returns violations for nodes that don't match excluded selectors.
  *
- * SELECTOR MATCHING: Uses substring matching. An exclusion pattern matches a violation's
- * target selector if the pattern appears as a substring. For example:
- *   - A violation reports a Target Selector: "th[aria-label="Privacy Policy"] > strong > .row-title".
- *   - Pattern ".row-title" MATCHES (substring found).
- *   - Pattern "strong > .row-title" MATCHES (substring found).
- *   - Pattern ".column-title .row-title" does NOT match (not a substring).
+ * See more details in tests/e2e/utils/accessibility-ignores-and-false-positives.js.
  *
  * @param {Object} results    Axe results object.
- * @param {Object} exclusions Known false positives config (rule ID → selectors[]).
+ * @param {Object} exclusions Explicit exclusions and known false positives config (rule ID → selectors[]).
  * @return {Object} Filtered results.
  */
 function filterIgnoresAndFalsePositives( results, exclusions ) {
@@ -39,9 +36,20 @@ function filterIgnoresAndFalsePositives( results, exclusions ) {
 		// Filter out nodes matching excluded selectors.
 		violation.nodes = violation.nodes.filter( ( node ) => {
 			const targetSelector = node.target[ 0 ];
-			return ! excludedSelectors.some( ( excluded ) =>
-				targetSelector?.includes( excluded )
-			);
+
+			// Check if target selector matches any excluded selector pattern.
+			const isExcluded = excludedSelectors.some( ( excluded ) => {
+				const selectors = excluded.trim().split( ' ' ).filter( s => s.length > 0 );
+				return selectors.every( selector => {
+					// Escape special regex chars, then replace \* with .* for wildcard matching
+					const escaped = selector.replace( /[.*+?^${}()|[\]\\]/g, '\\$&' );
+					const wildcard = escaped.replace( /\\\*/g, '.*' );
+					const pattern = `(?:^| )${ wildcard }(?=[\\s>+~\\[]|$)`;
+					return new RegExp( pattern ).test( targetSelector );
+				} );
+			} );
+
+			return ! isExcluded;
 		} );
 
 		return violation;
 
@@ -1,30 +1,48 @@
 /**
- * Ignores and known false positives after human inspection.
+ * Explicit exclusions and known false positives after human inspection.
  *
- * Maps rule IDs to an array of CSS selectors that should be excluded from violations.
+ * Maps rule IDs to CSS selectors that should be excluded from the reported violations.
  * These are instances where axe-core flags a violation, but after human review,
  * the pattern is acceptable or represents a known limitation of the automated check.
  *
- * SELECTOR MATCHING: Uses substring matching. An exclusion pattern matches a violation's
- * target selector if the pattern appears as a substring. For example:
- *   - A violation reports a Target Selector: "th[aria-label="Privacy Policy"] > strong > .row-title".
- *   - Pattern ".row-title" MATCHES (substring found).
- *   - Pattern "strong > .row-title" MATCHES (substring found).
- *   - Pattern ".column-title .row-title" does NOT match (not a substring).
+ * SELECTOR MATCHING:
+ * Uses space-separated selector tokens. An exclusion pattern matches a violation's
+ * target selector if all selector tokens appear in order (but not necessarily
+ * consecutively). Intermediate selectors can be skipped.
+ * For example:
+ *   - Violation Target Selector: "html > body > #__wp-uploader > .attachments-browser > .attachments-wrapper > li[aria-label="image-1"]".
+ *   - Pattern ".attachments-browser li" MATCHES (skips .attachments-wrapper; all tokens found in order).
+ *   - Pattern ".attachments-browser .attachments-wrapper" MATCHES (all tokens found).
+ *   - Pattern ".attachments-browserxyz" does NOT match (selector token boundary prevents partial matches).
+ *   - Attributes on target selectors are ignored: "li[aria-label="..."]" matches pattern "li".
+ *
+ * WILDCARD SUPPORT:
+ * Use an asterisk * to match any characters in a selector token. For example:
+ *   - Pattern "#wp-admin-bar-*" matches "#wp-admin-bar-site-name", "#wp-admin-bar-updates", etc.
+ *   - Pattern "#wpadminbar #wp-toolbar #wp-admin-bar-*" matches full descendant chains with variable final selectors.
+ *
+ * CONSTRAINTS:
+ * Exclusion patterns must be space-separated tokens only (e.g., '#id .class li').
+ * Do not use attribute selectors, pseudo-classes, or explicit combinators (>, +, ~) in patterns.
+ * Spaces implicitly represent descendant relationships (any number of intermediate selectors).
+ *
+ * NOTE:
+ * The passed pattern isn't meant to be a meaningful CSS selector. It's only used
+ * to match the Axe-core target selector of a violation. Just use a space-separated
+ * list of tokens that appear in the 'Target Selector' entry of the reported violation.
  */
 
 module.exports = {
 	'link-in-text-block': [
 		// Post title links in list tables may have surrounding text and are only distinguished by color.
 		// We consider this acceptable as these links can be distinguished by the context.
-		'strong > .row-title',
-		'a[aria-label="Edit',
+		'.wp-list-table strong .row-title',
 	],
 	'aria-allowed-role': [
 		// The "group" role for the admin bar list items is flagged as invalid by axe-core.
 		// This is a known false positive in axe-core and can be ignored. Note
 		// that the 'aria-allowed-role' rule is part of the 'best-practice' rules group.
-		'#wp-admin-bar-',
+		'#wp-toolbar #wp-admin-bar-*',
 	],
 	'region': [
 		// Visually hidden text that clarifies what the ARIA live regions are about.