Handle more edge-cases

Closes #58

index.js

@@ -4,6 +4,8 @@ import emojiRegex from 'emoji-regex';
 
 const segmenter = new Intl.Segmenter();
 
+const defaultIgnorableCodePointRegex = /^\p{Default_Ignorable_Code_Point}$/u;
+
 export default function stringWidth(string, options = {}) {
 	if (typeof string !== 'string' || string.length === 0) {
 		return 0;
@@ -33,8 +35,37 @@ export default function stringWidth(string, options = {}) {
 			continue;
 		}
 
+		// Ignore zero-width characters
+		if (
+			(codePoint >= 0x20_0B && codePoint <= 0x20_0F) // Zero-width space, non-joiner, joiner, left-to-right mark, right-to-left mark
+			|| codePoint === 0xFE_FF // Zero-width no-break space
+		) {
+			continue;
+		}
+
 		// Ignore combining characters
-		if (codePoint >= 0x3_00 && codePoint <= 0x3_6F) {
+		if (
+			(codePoint >= 0x3_00 && codePoint <= 0x3_6F) // Combining diacritical marks
+			|| (codePoint >= 0x1A_B0 && codePoint <= 0x1A_FF) // Combining diacritical marks extended
+			|| (codePoint >= 0x1D_C0 && codePoint <= 0x1D_FF) // Combining diacritical marks supplement
+			|| (codePoint >= 0x20_D0 && codePoint <= 0x20_FF) // Combining diacritical marks for symbols
+			|| (codePoint >= 0xFE_20 && codePoint <= 0xFE_2F) // Combining half marks
+		) {
+			continue;
+		}
+
+		// Ignore surrogate pairs
+		if (codePoint >= 0xD8_00 && codePoint <= 0xDF_FF) {
+			continue;
+		}
+
+		// Ignore variation selectors
+		if (codePoint >= 0xFE_00 && codePoint <= 0xFE_0F) {
+			continue;
+		}
+
+		// This covers some of the above cases, but we still keep them for performance reasons.
+		if (defaultIgnorableCodePointRegex.test(character)) {
 			continue;
 		}
 

test.js

@@ -37,6 +37,10 @@ test('ignores control characters', t => {
 
 test('handles combining characters', t => {
 	t.is(stringWidth('x\u0300'), 1);
+	t.is(stringWidth('\u0300\u0301'), 0);
+	t.is(stringWidth('e\u0301e'), 2);
+	t.is(stringWidth('x\u036F'), 1);
+	t.is(stringWidth('\u036F\u036F'), 0);
 });
 
 test('handles ZWJ characters', t => {
@@ -45,3 +49,52 @@ test('handles ZWJ characters', t => {
 	t.is(stringWidth('👩‍👩‍👦‍👦'), 2);
 	t.is(stringWidth('👨‍❤️‍💋‍👨'), 2);
 });
+
+test('handles zero-width characters', t => {
+	t.is(stringWidth('\u200B'), 0);
+	t.is(stringWidth('x\u200Bx'), 2);
+	t.is(stringWidth('\u200C'), 0);
+	t.is(stringWidth('x\u200Cx'), 2);
+	t.is(stringWidth('\u200D'), 0);
+	t.is(stringWidth('x\u200Dx'), 2);
+	t.is(stringWidth('\uFEFF'), 0);
+	t.is(stringWidth('x\uFEFFx'), 2);
+});
+
+test('handles surrogate pairs', t => {
+	t.is(stringWidth('\uD83D\uDE00'), 2); // 😀
+	t.is(stringWidth('A\uD83D\uDE00B'), 4);
+});
+
+test('handles variation selectors', t => {
+	t.is(stringWidth('\u{1F1E6}\uFE0F'), 1); // Regional indicator symbol A with variation selector
+	t.is(stringWidth('A\uFE0F'), 1);
+	t.is(stringWidth('\uFE0F'), 0);
+});
+
+test('handles edge cases', t => {
+	t.is(stringWidth(''), 0);
+	t.is(stringWidth('\u200B\u200B'), 0);
+	t.is(stringWidth('x\u200Bx\u200B'), 2);
+	t.is(stringWidth('x\u0300x\u0300'), 2);
+	t.is(stringWidth('\uD83D\uDE00\uFE0F'), 2); // 😀 with variation selector
+	t.is(stringWidth('\uD83D\uDC69\u200D\uD83C\uDF93'), 2); // 👩‍🎓
+	t.is(stringWidth('x\u1AB0x\u1AB0'), 2); // Combining diacritical marks extended
+	t.is(stringWidth('x\u1DC0x\u1DC0'), 2); // Combining diacritical marks supplement
+	t.is(stringWidth('x\u20D0x\u20D0'), 2); // Combining diacritical marks for symbols
+	t.is(stringWidth('x\uFE20x\uFE20'), 2); // Combining half marks
+});
+
+test('ignores default ignorable code points', t => {
+	t.is(stringWidth('\u2060'), 0); // Word joiner
+	t.is(stringWidth('\u2061'), 0); // Function application
+	t.is(stringWidth('\u2062'), 0); // Invisible times
+	t.is(stringWidth('\u2063'), 0); // Invisible separator
+	t.is(stringWidth('\u2064'), 0); // Invisible plus
+	t.is(stringWidth('\uFEFF'), 0); // Zero-width no-break space
+	t.is(stringWidth('x\u2060x'), 2);
+	t.is(stringWidth('x\u2061x'), 2);
+	t.is(stringWidth('x\u2062x'), 2);
+	t.is(stringWidth('x\u2063x'), 2);
+	t.is(stringWidth('x\u2064x'), 2);
+});