Revert "Revert "Switch kRegex to use RegExp with u directly"" (#1…

…5644) * Revert "Revert "Switch `kRegex` to use RegExp with `u` directly" (#15643)" This reverts commit 52375f5.
prettier · Jun 14, 2024 · ccde7ca · ccde7ca
1 parent 57253de
commit ccde7ca
Show file tree

Hide file tree

Showing 3 changed files with 8 additions and 11 deletions.
diff --git a/scripts/build/build-javascript-module.js b/scripts/build/build-javascript-module.js
@@ -258,6 +258,10 @@ function getEsbuildOptions({ file, files, shouldCollectLicenses, cliOptions }) {
     outfile: path.join(DIST_DIR, cliOptions.saveAs ?? file.output.file),
     // https://esbuild.github.io/api/#main-fields
     mainFields: file.platform === "node" ? ["module", "main"] : undefined,
+    supported: {
+      // https://github.com/evanw/esbuild/issues/3471
+      "regexp-unicode-property-escapes": true,
+    },
   };
 
   if (file.platform === "universal") {

diff --git a/src/language-markdown/constants.evaluate.js b/src/language-markdown/constants.evaluate.js
@@ -24,11 +24,6 @@ const CJK_REGEXP = new RegExp(
   `(?:${cjkCharset.toString()})(?:${variationSelectorsCharset.toString()})?`,
 );
 
-const K_REGEXP = new Charset(
-  unicodeRegex({ Script: ["Hangul"] }),
-  unicodeRegex({ Script_Extensions: ["Hangul"] }),
-).toRegExp();
-
 // http://spec.commonmark.org/0.25/#ascii-punctuation-character
 const asciiPunctuationCharset =
   /* prettier-ignore */ new Charset(
@@ -57,4 +52,4 @@ const PUNCTUATION_REGEXP = new Charset(
   unicodePunctuationCharset,
 ).toRegExp();
 
-export { CJK_REGEXP, K_REGEXP, PUNCTUATION_REGEXP };
+export { CJK_REGEXP, PUNCTUATION_REGEXP };
diff --git a/src/language-markdown/utils.js b/src/language-markdown/utils.js
@@ -1,10 +1,6 @@
 import assert from "node:assert";
 
-import {
-  CJK_REGEXP,
-  K_REGEXP,
-  PUNCTUATION_REGEXP,
-} from "./constants.evaluate.js";
+import { CJK_REGEXP, PUNCTUATION_REGEXP } from "./constants.evaluate.js";
 import { locEnd, locStart } from "./loc.js";
 
 const INLINE_NODE_TYPES = new Set([
@@ -40,6 +36,8 @@ const KIND_CJ_LETTER = "cj-letter";
 const KIND_K_LETTER = "k-letter";
 const KIND_CJK_PUNCTUATION = "cjk-punctuation";
 
+const K_REGEXP = /\p{Script_Extensions=Hangul}/u;
+
 /**
  * @typedef {" " | "\n" | ""} WhitespaceValue
  * @typedef { KIND_NON_CJK | KIND_CJ_LETTER | KIND_K_LETTER | KIND_CJK_PUNCTUATION } WordKind