Merge pull request #2102 from ahuth/unicode-private-use-area-A

fix(unicode) detect supplementary private use area A
dequelabs · Mar 19, 2020 · f1739c2 · f1739c2
2 parents 7952a37 + 6a90379
commit f1739c2
Show file tree

Hide file tree

Showing 2 changed files with 25 additions and 6 deletions.
diff --git a/lib/commons/text/unicode.js b/lib/commons/text/unicode.js
@@ -19,7 +19,10 @@ text.hasUnicode = function hasUnicode(str, options) {
 		return axe.imports.emojiRegexText().test(str);
 	}
 	if (nonBmp) {
-		return getUnicodeNonBmpRegExp().test(str);
+		return (
+			getUnicodeNonBmpRegExp().test(str) ||
+			getSupplementaryPrivateUseRegExp().test(str)
+		);
 	}
 	if (punctuations) {
 		return getPunctuationRegExp().test(str);
@@ -127,9 +130,11 @@ function getPunctuationRegExp() {
  * @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;
+	// 1. High surrogate area (https://www.unicode.org/charts/PDF/UD800.pdf)
+	// 2. Low surrogate area (https://www.unicode.org/charts/PDF/UDC00.pdf)
+	// 3. Supplementary private use area A (https://www.unicode.org/charts/PDF/UF0000.pdf)
+	//
+	//             1              2                                  3
+	//      ┏━━━━━━┻━━━━━━┓┏━━━━━━┻━━━━━━┓ ┏━━━━━━━━━━━━━━━━━━━━━━━━━┻━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
+	return /[\uDB80-\uDBBF][\uDC00-\uDFFD]|(?:[\uDB80-\uDBBE][\uDC00-\uDFFF]|\uDBBF[\uDC00-\uDFFD])/g;
 }
diff --git a/test/commons/text/unicode.js b/test/commons/text/unicode.js
@@ -69,6 +69,13 @@ describe('text.hasUnicode', function() {
 			});
 			assert.isTrue(actual);
 		});
+
+		it('returns true for a string with characters in supplementary private use area A', function() {
+			var actual = axe.commons.text.hasUnicode('\uDB80\uDC19', {
+				nonBmp: true
+			});
+			assert.isTrue(actual);
+		});
 	});
 
 	describe('text.hasUnicode, characters of type Emoji', function() {
@@ -201,6 +208,13 @@ describe('text.removeUnicode', function() {
 		assert.equal(actual, '');
 	});
 
+	it('returns the string with supplementary private use area A characters removed', function() {
+		var actual = axe.commons.text.removeUnicode('\uDB80\uDC19', {
+			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.',