ComboboxControl: Simplify string normalization

WordPress · Apr 19, 2024 · 30031db · 30031db
1 parent 4b08cf1
commit 30031db
Show file tree

Hide file tree

Showing 2 changed files with 67 additions and 47 deletions.
diff --git a/packages/components/src/utils/strings.ts b/packages/components/src/utils/strings.ts
@@ -6,64 +6,22 @@ import { paramCase } from 'change-case';
 
 const ALL_UNICODE_DASH_CHARACTERS = new RegExp(
 	`[${ [
-		// - (hyphen-minus)
-		'\u002d',
 		// ~ (tilde)
 		'\u007e',
 		//  (soft hyphen)
 		'\u00ad',
-		// ֊ (armenian hyphen)
-		'\u058a',
-		// ־ (hebrew punctuation maqaf)
-		'\u05be',
-		// ᐀ (canadian syllabics hyphen)
-		'\u1400',
-		// ᠆ (mongolian todo soft hyphen)
-		'\u1806',
-		// ‐ (hyphen)
-		'\u2010',
-		// non-breaking hyphen)
-		'\u2011',
-		// ‒ (figure dash)
-		'\u2012',
-		// – (en dash)
-		'\u2013',
-		// — (em dash)
-		'\u2014',
-		// ― (horizontal bar)
-		'\u2015',
 		// ⁓ (swung dash)
 		'\u2053',
-		// superscript minus)
+		// ⁻ (superscript minus)
 		'\u207b',
-		// subscript minus)
+		// ₋ (subscript minus)
 		'\u208b',
 		// − (minus sign)
 		'\u2212',
-		// ⸗ (double oblique hyphen)
-		'\u2e17',
-		// ⸺ (two-em dash)
-		'\u2e3a',
-		// ⸻ (three-em dash)
-		'\u2e3b',
-		// 〜 (wave dash)
-		'\u301c',
-		// 〰 (wavy dash)
-		'\u3030',
-		// ゠ (katakana-hiragana double hyphen)
-		'\u30a0',
-		// ︱ (presentation form for vertical em dash)
-		'\ufe31',
-		// ︲ (presentation form for vertical en dash)
-		'\ufe32',
-		// ﹘ (small em dash)
-		'\ufe58',
-		// ﹣ (small hyphen-minus)
-		'\ufe63',
-		// － (fullwidth hyphen-minus)
-		'\uff0d',
+		// any other Unicode dash character
+		'\\p{Pd}',
 	].join( '' ) }]`,
-	'g'
+	'gu'
 );
 
 export const normalizeTextString = ( value: string ): string => {

diff --git a/packages/components/src/utils/test/strings.js b/packages/components/src/utils/test/strings.js
@@ -106,5 +106,67 @@ describe( 'normalizeTextString', () => {
 		expect( normalizeTextString( 'foo⸻bar' ) ).toBe( 'foo-bar' );
 		expect( normalizeTextString( 'foo゠bar' ) ).toBe( 'foo-bar' );
 		expect( normalizeTextString( 'foo－bar' ) ).toBe( 'foo-bar' );
+
+		const dashCharacters = [
+			// - (hyphen-minus)
+			'\u002d',
+			// ~ (tilde)
+			'\u007e',
+			//  (soft hyphen)
+			'\u00ad',
+			// ֊ (armenian hyphen)
+			'\u058a',
+			// ־ (hebrew punctuation maqaf)
+			'\u05be',
+			// ᐀ (canadian syllabics hyphen)
+			'\u1400',
+			// ᠆ (mongolian todo soft hyphen)
+			'\u1806',
+			// ‐ (hyphen)
+			'\u2010',
+			// non-breaking hyphen)
+			'\u2011',
+			// ‒ (figure dash)
+			'\u2012',
+			// – (en dash)
+			'\u2013',
+			// — (em dash)
+			'\u2014',
+			// ― (horizontal bar)
+			'\u2015',
+			// ⁓ (swung dash)
+			'\u2053',
+			// superscript minus)
+			'\u207b',
+			// subscript minus)
+			'\u208b',
+			// − (minus sign)
+			'\u2212',
+			// ⸗ (double oblique hyphen)
+			'\u2e17',
+			// ⸺ (two-em dash)
+			'\u2e3a',
+			// ⸻ (three-em dash)
+			'\u2e3b',
+			// 〜 (wave dash)
+			'\u301c',
+			// 〰 (wavy dash)
+			'\u3030',
+			// ゠ (katakana-hiragana double hyphen)
+			'\u30a0',
+			// ︱ (presentation form for vertical em dash)
+			'\ufe31',
+			// ︲ (presentation form for vertical en dash)
+			'\ufe32',
+			// ﹘ (small em dash)
+			'\ufe58',
+			// ﹣ (small hyphen-minus)
+			'\ufe63',
+			// － (fullwidth hyphen-minus)
+			'\uff0d',
+		];
+		expect( normalizeTextString( dashCharacters.join( '' ) ) ).toBe(
+			'-'.repeat( dashCharacters.length )
+		);
 	} );
 } );