Provides legacy fallback for browser that don't support regex flag v (#35659)

app/javascript/mastodon/features/emoji/constants.ts

@@ -15,17 +15,6 @@ export const SKIN_TONE_CODES = [
   0x1f3ff, // Dark skin tone
 ] as const;
 
-// TODO: Test and create fallback for browsers that do not handle the /v flag.
-export const UNICODE_EMOJI_REGEX = /\p{RGI_Emoji}/v;
-// See: https://www.unicode.org/reports/tr51/#valid-emoji-tag-sequences
-export const UNICODE_FLAG_EMOJI_REGEX =
-  /\p{RGI_Emoji_Flag_Sequence}|\p{RGI_Emoji_Tag_Sequence}/v;
-export const CUSTOM_EMOJI_REGEX = /:([a-z0-9_]+):/i;
-export const ANY_EMOJI_REGEX = new RegExp(
-  `(${UNICODE_EMOJI_REGEX.source}|${CUSTOM_EMOJI_REGEX.source})`,
-  'gv',
-);
-
 // Emoji rendering modes. A mode is what we are using to render emojis, a style is what the user has selected.
 export const EMOJI_MODE_NATIVE = 'native';
 export const EMOJI_MODE_NATIVE_WITH_FLAGS = 'native-flags';

app/javascript/mastodon/features/emoji/render.ts

@@ -9,7 +9,6 @@ import {
   EMOJI_TYPE_UNICODE,
   EMOJI_TYPE_CUSTOM,
   EMOJI_STATE_MISSING,
-  ANY_EMOJI_REGEX,
 } from './constants';
 import {
   searchCustomEmojisByShortcodes,
@@ -32,7 +31,12 @@ import type {
   LocaleOrCustom,
   UnicodeEmojiToken,
 } from './types';
-import { emojiLogger, stringHasAnyEmoji, stringHasUnicodeFlags } from './utils';
+import {
+  anyEmojiRegex,
+  emojiLogger,
+  stringHasAnyEmoji,
+  stringHasUnicodeFlags,
+} from './utils';
 
 const log = emojiLogger('render');
 
@@ -207,7 +211,7 @@ export function tokenizeText(text: string): TokenizedText {
 
   const tokens = [];
   let lastIndex = 0;
-  for (const match of text.matchAll(ANY_EMOJI_REGEX)) {
+  for (const match of text.matchAll(anyEmojiRegex())) {
     if (match.index > lastIndex) {
       tokens.push(text.slice(lastIndex, match.index));
     }

app/javascript/mastodon/features/emoji/utils.ts

@@ -1,23 +1,32 @@
 import debug from 'debug';
 
-import {
+import { emojiRegexPolyfill } from '@/mastodon/polyfills';
-  CUSTOM_EMOJI_REGEX,
-  UNICODE_EMOJI_REGEX,
-  UNICODE_FLAG_EMOJI_REGEX,
-} from './constants';
 
 export function emojiLogger(segment: string) {
   return debug(`emojis:${segment}`);
 }
 
 export function stringHasUnicodeEmoji(input: string): boolean {
-  return UNICODE_EMOJI_REGEX.test(input);
+  return new RegExp(EMOJI_REGEX, supportedFlags()).test(input);
 }
 
 export function stringHasUnicodeFlags(input: string): boolean {
-  return UNICODE_FLAG_EMOJI_REGEX.test(input);
+  if (supportsRegExpSets()) {
+    return new RegExp(
+      '\\p{RGI_Emoji_Flag_Sequence}|\\p{RGI_Emoji_Tag_Sequence}',
+      'v',
+    ).test(input);
+  }
+  return new RegExp(
+    // First range is regional indicator symbols,
+    // Second is a black flag + 0-9|a-z tag chars + cancel tag.
+    // See: https://en.wikipedia.org/wiki/Regional_indicator_symbol
+    '(?:\uD83C[\uDDE6-\uDDFF]){2}|\uD83C\uDFF4(?:\uDB40[\uDC30-\uDC7A])+\uDB40\uDC7F',
+  ).test(input);
 }
 
+// Constant as this is supported by all browsers.
+const CUSTOM_EMOJI_REGEX = /:([a-z0-9_]+):/i;
 export function stringHasCustomEmoji(input: string) {
   return CUSTOM_EMOJI_REGEX.test(input);
 }
@@ -25,3 +34,23 @@ export function stringHasCustomEmoji(input: string) {
 export function stringHasAnyEmoji(input: string) {
   return stringHasUnicodeEmoji(input) || stringHasCustomEmoji(input);
 }
+
+export function anyEmojiRegex() {
+  return new RegExp(
+    `${EMOJI_REGEX}|${CUSTOM_EMOJI_REGEX.source}`,
+    supportedFlags('gi'),
+  );
+}
+
+function supportsRegExpSets() {
+  return 'unicodeSets' in RegExp.prototype;
+}
+
+function supportedFlags(flags = '') {
+  if (supportsRegExpSets()) {
+    return `${flags}v`;
+  }
+  return flags;
+}
+
+const EMOJI_REGEX = emojiRegexPolyfill?.source ?? '\\p{RGI_Emoji}';

app/javascript/mastodon/polyfills/index.ts

@@ -20,5 +20,16 @@ export function loadPolyfills() {
     loadIntlPolyfills(),
     // eslint-disable-next-line @typescript-eslint/no-unnecessary-condition -- those properties might not exist in old browsers, even if they are always here in types
     needsExtraPolyfills && importExtraPolyfills(),
+    loadEmojiPolyfills(),
   ]);
 }
+
+// In the case of no /v support, rely on the emojibase data.
+async function loadEmojiPolyfills() {
+  if (!('unicodeSets' in RegExp.prototype)) {
+    emojiRegexPolyfill = (await import('emojibase-regex/emoji')).default;
+  }
+}
+
+// Null unless polyfill is needed.
+export let emojiRegexPolyfill: RegExp | null = null;

package.json

@@ -69,6 +69,7 @@
     "emoji-mart": "npm:emoji-mart-lazyload@latest",
     "emojibase": "^16.0.0",
     "emojibase-data": "^16.0.3",
+    "emojibase-regex": "^16.0.0",
     "escape-html": "^1.0.3",
     "fast-glob": "^3.3.3",
     "fuzzysort": "^3.0.0",

yarn.lock

@@ -2659,6 +2659,7 @@ __metadata:
     emoji-mart: "npm:emoji-mart-lazyload@latest"
     emojibase: "npm:^16.0.0"
     emojibase-data: "npm:^16.0.3"
+    emojibase-regex: "npm:^16.0.0"
     escape-html: "npm:^1.0.3"
     eslint: "npm:^9.23.0"
     eslint-import-resolver-typescript: "npm:^4.2.5"
@@ -6565,6 +6566,13 @@ __metadata:
   languageName: node
   linkType: hard
 
+"emojibase-regex@npm:^16.0.0":
+  version: 16.0.0
+  resolution: "emojibase-regex@npm:16.0.0"
+  checksum: 10c0/8ee5ff798e51caa581434b1cb2f9737e50195093c4efa1739df21a50a5496f80517924787d865e8cf7d6a0b4c90dbedc04bdc506dcbcc582e14cdf0bb47af0f0
+  languageName: node
+  linkType: hard
+
 "emojibase@npm:^16.0.0":
   version: 16.0.0
   resolution: "emojibase@npm:16.0.0"