From b634c343effefc871b4fb703c38a834ffaaa8846 Mon Sep 17 00:00:00 2001
From: Steven Lambert <steven.lambert@deque.com>
Date: Mon, 30 Sep 2019 11:29:37 -0600
Subject: [PATCH] fix(label-content-name-mismatch): ignore private space
 unicode (#1822)

* fix(label-content-name-mismatch): ignore private space unicode code characters

* add test

* fix test

* add surrogate private use

* fix comments
---
 lib/commons/text/unicode.js                      | 15 +++++++++++++++
 test/checks/label/label-content-name-mismatch.js |  8 ++++++++
 test/commons/text/unicode.js                     | 14 ++++++++++++++
 3 files changed, 37 insertions(+)

diff --git a/lib/commons/text/unicode.js b/lib/commons/text/unicode.js
index 24eb8214a6..31280c5301 100644
--- a/lib/commons/text/unicode.js
+++ b/lib/commons/text/unicode.js
@@ -48,6 +48,7 @@ text.removeUnicode = function removeUnicode(str, options) {
 	}
 	if (nonBmp) {
 		str = str.replace(getUnicodeNonBmpRegExp(), '');
+		str = str.replace(getSupplementaryPrivateUseRegExp(), '');
 	}
 	if (punctuations) {
 		str = str.replace(getPunctuationRegExp(), '');
@@ -91,6 +92,7 @@ function getUnicodeNonBmpRegExp() {
 		'\u25A0-\u25FF' + // Geometric Shapes
 		'\u2600-\u26FF' + // Misc Symbols
 		'\u2700-\u27BF' + // Dingbats
+		'\uE000-\uF8FF' + // Private Use
 			']'
 	);
 }
@@ -115,3 +117,16 @@ function getPunctuationRegExp() {
 	 */
 	return /[\u2000-\u206F\u2E00-\u2E7F\\'!"#$%&()*+,\-.\/:;<=>?@\[\]^_`{|}~]/g;
 }
+
+/**
+ * Get regular expression for supplementary private use
+ *
+ * @returns {RegExp}
+ */
+function getSupplementaryPrivateUseRegExp() {
+	/**
+	 * Reference: https://www.unicode.org/charts/PDF/UD800.pdf
+	 * https://www.unicode.org/charts/PDF/UDC00.pdf
+	 */
+	return /[\uDB80-\uDBBF][\uDC00-\uDFFD]/g;
+}
diff --git a/test/checks/label/label-content-name-mismatch.js b/test/checks/label/label-content-name-mismatch.js
index e48f3f0f82..26e2ed382f 100644
--- a/test/checks/label/label-content-name-mismatch.js
+++ b/test/checks/label/label-content-name-mismatch.js
@@ -84,6 +84,14 @@ describe('label-content-name-mismatch tests', function() {
 		assert.isTrue(actual);
 	});
 
+	it('returns true when visible text excluding private use unicode is part of accessible name', function() {
+		var vNode = queryFixture(
+			'<button id="target" aria-label="Favorites"> Favorites</button>'
+		);
+		var actual = check.evaluate(vNode.actualNode, options, vNode);
+		assert.isTrue(actual);
+	});
+
 	it('returns undefined (needs review) when visible text name is only an emoji', function() {
 		var vNode = queryFixture(
 			'<button id="target" aria-label="comet">☄️</button>'
diff --git a/test/commons/text/unicode.js b/test/commons/text/unicode.js
index cf0429830b..aea08babaf 100644
--- a/test/commons/text/unicode.js
+++ b/test/commons/text/unicode.js
@@ -187,6 +187,20 @@ describe('text.removeUnicode', function() {
 		assert.equal(actual, '');
 	});
 
+	it('returns string removing all private use unicode', function() {
+		var actual = axe.commons.text.removeUnicode('', {
+			nonBmp: true
+		});
+		assert.equal(actual, '');
+	});
+
+	it('returns string removing all supplementary private use unicode', function() {
+		var actual = axe.commons.text.removeUnicode('󰀀󿰀󿿽󰏽', {
+			nonBmp: true
+		});
+		assert.equal(actual, '');
+	});
+
 	it('returns string removing combination of unicode characters', function() {
 		var actual = axe.commons.text.removeUnicode(
 			'The ☀️ is orange, the ◓ is white.',