fix(utils): fix extractLeadingEmoji() edge cases (#12100)

packages/docusaurus-theme-common/src/utils/__tests__/emojiUtils.test.ts

@@ -64,4 +64,65 @@ describe('extractLeadingEmoji', () => {
       rest: 'Hello World 😀',
     });
   });
+
+  it('does not extract single digit', () => {
+    expect(extractLeadingEmoji('1 Hello World')).toEqual({
+      emoji: null,
+      rest: '1 Hello World',
+    });
+  });
+
+  it('does not extract multiple digits', () => {
+    expect(extractLeadingEmoji('11 Hello World')).toEqual({
+      emoji: null,
+      rest: '11 Hello World',
+    });
+  });
+
+  it('extracts real keycap emoji (digit + VS16 + combining keycap)', () => {
+    // 1️⃣ is digit "1" followed by U+FE0F (Variation Selector-16) and
+    // U+20E3 (Combining Enclosing Keycap) — a multi-codepoint grapheme
+    // that should still be detected as emoji
+    expect(extractLeadingEmoji('1\u{FE0F}\u{20E3} Something')).toEqual({
+      emoji: '1️⃣',
+      rest: ' Something',
+    });
+  });
+
+  it('extracts VS16 emoji - ⚠️ (text-default + variation selector)', () => {
+    // ⚠ (U+26A0) has Emoji=true, Emoji_Presentation=false.  With VS16
+    // (U+FE0F) it becomes a multi-codepoint grapheme that IS an emoji
+    expect(extractLeadingEmoji('⚠\u{FE0F} Warning')).toEqual({
+      emoji: '⚠\u{FE0F}',
+      rest: ' Warning',
+    });
+  });
+
+  it('does not detect text-presentation symbols - ©', () => {
+    expect(extractLeadingEmoji('© 2026')).toEqual({
+      emoji: null,
+      rest: '© 2026',
+    });
+  });
+
+  it('does detect ©️', () => {
+    expect(extractLeadingEmoji('©️ 2026')).toEqual({
+      emoji: '©️',
+      rest: ' 2026',
+    });
+  });
+
+  it('does not detect text-presentation symbols - ™', () => {
+    expect(extractLeadingEmoji('™ Brand')).toEqual({
+      emoji: null,
+      rest: '™ Brand',
+    });
+  });
+
+  it('does not detect text-presentation symbols - ♠', () => {
+    expect(extractLeadingEmoji('♠ Cards')).toEqual({
+      emoji: null,
+      rest: '♠ Cards',
+    });
+  });
 });

packages/docusaurus-theme-common/src/utils/emojiUtils.ts

@@ -29,11 +29,10 @@ export function extractLeadingEmoji(input: string): {
     return {emoji: null, rest: input};
   }
 
-  // Leading grapheme contains an emoji (covers flags/ZWJ/skin tones)
-  if (
-    !/\p{Extended_Pictographic}/u.test(grapheme) &&
-    !/\p{Emoji}/u.test(grapheme)
-  ) {
+  // Leading grapheme contains an emoji
+  // Covers flags/ZWJ/skin tones, excludes digits
+  // See https://github.com/facebook/docusaurus/pull/12072
+  if (!/^\p{RGI_Emoji}$/v.test(grapheme)) {
     return {emoji: null, rest: input};
   }
 

project-words.txt

@@ -133,6 +133,8 @@ Kaszubowski
 Katex
 katex
 Kato
+Keycap
+keycap
 Keytar
 keytar
 Kinsta