fix: cache list indentation regexes (#3969)

astelmach01 · UziTech · styfle · web-flow · commit a37983f188d6 · 2026-05-18T23:13:54.000-06:00
* Cache list indentation regexes

* Handle zero indent in cached regexes

* Update src/rules.ts

Co-authored-by: Steven &lt;steven@ceriously.com&gt;

---------

Co-authored-by: Tony Brix &lt;tony@brix.ninja&gt;
Co-authored-by: Steven &lt;steven@ceriously.com&gt;
diff --git a/src/rules.ts b/src/rules.ts
@@ -1,5 +1,18 @@
 const noopTest = { exec: () => null } as unknown as RegExp;
 
+function cachedIndentRegex(createRegex: (indent: number) => RegExp) {
+  const cache: RegExp[] = [];
+  return (indent: number) => {
+    const cacheIndex = Math.max(0, Math.min(3, indent - 1));
+    let regex = cache[cacheIndex];
+    if (!regex) {
+      regex = createRegex(cacheIndex);
+      cache[cacheIndex] = regex;
+    }
+    return regex;
+  };
+}
+
 function edit(regex: string | RegExp, opt = '') {
   let source = typeof regex === 'string' ? regex : regex.source;
   const obj = {
@@ -78,12 +91,12 @@ export const other = {
   notSpaceStart: /^\S*/,
   endingNewline: /\n$/,
   listItemRegex: (bull: string) => new RegExp(`^( {0,3}${bull})((?:[\t ][^\\n]*)?(?:\\n|$))`),
-  nextBulletRegex: (indent: number) => new RegExp(`^ {0,${Math.min(3, indent - 1)}}(?:[*+-]|\\d{1,9}[.)])((?:[ \t][^\\n]*)?(?:\\n|$))`),
-  hrRegex: (indent: number) => new RegExp(`^ {0,${Math.min(3, indent - 1)}}((?:- *){3,}|(?:_ *){3,}|(?:\\* *){3,})(?:\\n+|$)`),
-  fencesBeginRegex: (indent: number) => new RegExp(`^ {0,${Math.min(3, indent - 1)}}(?:\`\`\`|~~~)`),
-  headingBeginRegex: (indent: number) => new RegExp(`^ {0,${Math.min(3, indent - 1)}}#`),
-  htmlBeginRegex: (indent: number) => new RegExp(`^ {0,${Math.min(3, indent - 1)}}<(?:[a-z].*>|!--)`, 'i'),
-  blockquoteBeginRegex: (indent: number) => new RegExp(`^ {0,${Math.min(3, indent - 1)}}>`),
+  nextBulletRegex: cachedIndentRegex((indent: number) => new RegExp(`^ {0,${indent}}(?:[*+-]|\\d{1,9}[.)])((?:[ \t][^\\n]*)?(?:\\n|$))`)),
+  hrRegex: cachedIndentRegex((indent: number) => new RegExp(`^ {0,${indent}}((?:- *){3,}|(?:_ *){3,}|(?:\\* *){3,})(?:\\n+|$)`)),
+  fencesBeginRegex: cachedIndentRegex((indent: number) => new RegExp(`^ {0,${indent}}(?:\`\`\`|~~~)`)),
+  headingBeginRegex: cachedIndentRegex((indent: number) => new RegExp(`^ {0,${indent}}#`)),
+  htmlBeginRegex: cachedIndentRegex((indent: number) => new RegExp(`^ {0,${indent}}<(?:[a-z].*>|!--)`, 'i')),
+  blockquoteBeginRegex: cachedIndentRegex((indent: number) => new RegExp(`^ {0,${indent}}>`)),
 };
 
 /**
diff --git a/test/unit/Lexer.test.js b/test/unit/Lexer.test.js
@@ -26,6 +26,26 @@ function expectInlineTokens({ md, options, tokens, links = {} }) {
 }
 
 describe('Lexer', () => {
+  describe('rules', () => {
+    it('should create indentation regexes for zero indent', () => {
+      const {
+        nextBulletRegex,
+        hrRegex,
+        fencesBeginRegex,
+        headingBeginRegex,
+        htmlBeginRegex,
+        blockquoteBeginRegex,
+      } = new Lexer().tokenizer.rules.other;
+
+      assert.ok(nextBulletRegex(0).test('- item\n'));
+      assert.ok(hrRegex(0).test('---\n'));
+      assert.ok(fencesBeginRegex(0).test('```'));
+      assert.ok(headingBeginRegex(0).test('# heading'));
+      assert.ok(htmlBeginRegex(0).test('<div>'));
+      assert.ok(blockquoteBeginRegex(0).test('> quote'));
+    });
+  });
+
   describe('paragraph', () => {
     it('space between paragraphs', () => {
       expectTokens({
