Improve perf of blockSkip on modern engines

ziadkh0 · ziadkh0 · commit 70e5d9404f54 · 2025-10-30T13:18:55.000+01:00
RegExp lookbehind assertions are supported in Safari &gt;= 16.5. But the
lookbehind-less version has a performance penalty. So having a fallback keeps
support for older Safari versions while maintaining performance for modern
engines.
diff --git a/src/Lexer.ts b/src/Lexer.ts
@@ -320,12 +320,10 @@ export class _Lexer<ParserOutput = string, RendererOutput = string> {
     }
 
     // Mask out other blocks
+    let offset;
     while ((match = this.tokenizer.rules.inline.blockSkip.exec(maskedSrc)) != null) {
-      if (match[2]) {
-        maskedSrc = maskedSrc.slice(0, match.indices![2][0]) + '[' + 'a'.repeat(match[2].length - 2) + ']' + maskedSrc.slice(this.tokenizer.rules.inline.blockSkip.lastIndex);
-      } else {
-        maskedSrc = maskedSrc.slice(0, match.index) + '[' + 'a'.repeat(match[0].length - 2) + ']' + maskedSrc.slice(this.tokenizer.rules.inline.blockSkip.lastIndex);
-      }
+      offset = match[2] ? match[2].length : 0;
+      maskedSrc = maskedSrc.slice(0, match.index + offset) + '[' + 'a'.repeat(match[0].length - offset - 2) + ']' + maskedSrc.slice(this.tokenizer.rules.inline.blockSkip.lastIndex);
     }
 
     // Mask out blocks from extensions
diff --git a/src/rules.ts b/src/rules.ts
@@ -16,6 +16,16 @@ function edit(regex: string | RegExp, opt = '') {
   return obj;
 }
 
+const supportsLookbehind = (() => {
+try {
+  // eslint-disable-next-line prefer-regex-literals
+  return !!new RegExp('(?<=1)(?<!1)');
+// eslint-disable-next-line @typescript-eslint/no-unused-vars
+} catch(error) {
+  return false;
+}
+})();
+
 export const other = {
   codeRemoveIndent: /^(?: {1,4}| {0,3}\t)/gm,
   outputLinkReplace: /\\([\[\]])/g,
@@ -267,9 +277,10 @@ const _punctuationOrSpaceGfmStrongEm = /(?!~)[\s\p{P}\p{S}]/u;
 const _notPunctuationOrSpaceGfmStrongEm = /(?:[^\s\p{P}\p{S}]|~)/u;
 
 // sequences em should skip over [title](link), `code`, <html>
-const blockSkip = edit(/link|code|html/, 'dg')
+const blockSkip = edit(/link|precode-code|html/, 'g')
   .replace('link', /\[(?:[^\[\]`]|(?<a>`+)[^`]+\k<a>(?!`))*?\]\((?:\\[\s\S]|[^\\\(\)]|\((?:\\[\s\S]|[^\\\(\)])*\))*\)/)
-  .replace('code', /(?:^^|[^`])((?<b>`+)[^`]+\k<b>(?!`))/)
+  .replace('precode-', supportsLookbehind ? '(?<!`)()' : '(^^|[^`])')
+  .replace('code', /(?<b>`+)[^`]+\k<b>(?!`)/)
   .replace('html', /<(?! )[^<>]*?>/)
   .getRegex();
 

Original file line number	Diff line number	Diff line change
`@@ -320,12 +320,10 @@ export class _Lexer<ParserOutput = string, RendererOutput = string> {`
`320`	`320`	`}`
`321`	`321`
`322`	`322`	`// Mask out other blocks`
	`323`	`+ let offset;`
`323`	`324`	`while ((match = this.tokenizer.rules.inline.blockSkip.exec(maskedSrc)) != null) {`
`324`		`- if (match[2]) {`
`325`		`- maskedSrc = maskedSrc.slice(0, match.indices![2][0]) + '[' + 'a'.repeat(match[2].length - 2) + ']' + maskedSrc.slice(this.tokenizer.rules.inline.blockSkip.lastIndex);`
`326`		`- } else {`
`327`		`- maskedSrc = maskedSrc.slice(0, match.index) + '[' + 'a'.repeat(match[0].length - 2) + ']' + maskedSrc.slice(this.tokenizer.rules.inline.blockSkip.lastIndex);`
`328`		`- }`
	`325`	`+ offset = match[2] ? match[2].length : 0;`
	`326`	`+ maskedSrc = maskedSrc.slice(0, match.index + offset) + '[' + 'a'.repeat(match[0].length - offset - 2) + ']' + maskedSrc.slice(this.tokenizer.rules.inline.blockSkip.lastIndex);`
`329`	`327`	`}`
`330`	`328`
`331`	`329`	`// Mask out blocks from extensions`