v8.0.0

Compare Source

Major Changes

• 450d2bb: Added ast option to compiler to expose the parsed AST directly. When ast: true, the compiler returns the AST structure (ASTNode[]) instead of rendered JSX.

  Breaking Changes:

  • The internal type ParserResult has been renamed to ASTNode for clarity. If you were accessing this type directly (e.g., via module augmentation or type manipulation), you'll need to update references from MarkdownToJSX.ParserResult to MarkdownToJSX.ASTNode.

  First time the AST is accessible to users! This enables:

  • AST manipulation and transformation before rendering
  • Custom rendering logic without parsing
  • Caching parsed AST for performance
  • Linting or validation of markdown structure

  Usage:

  import { compiler } from 'markdown-to-jsx'
  import type { MarkdownToJSX } from 'markdown-to-jsx'
  
  // Get the AST structure
  const ast: MarkdownToJSX.ASTNode[] = compiler('# Hello world', {
    ast: true,
  })
  
  // Inspect/modify AST
  console.log(ast) // Array of parsed nodes
  
  // Render AST to JSX using createRenderer (not implemented yet)

  The AST format is MarkdownToJSX.ASTNode[]. When footnotes are present, the returned value will be an object with ast and footnotes properties instead of just the AST array.

• 3fa0c22: Refactored inline formatting parsing to eliminate ReDoS vulnerabilities and improve performance. The previous regex-based approach was susceptible to exponential backtracking on certain inputs and had several edge case bugs with nested formatting, escaped characters, and formatting inside links. The new implementation uses a custom iterative scanner that runs in O(n) time and is immune to ReDoS attacks.

  This also consolidates multiple formatting rule types into a single unified rule with boolean flags, reducing code duplication and bundle size. Performance has improved measurably on simple markdown strings:

  Breaking Changes:

  The following RuleType enum values have been removed and consolidated into a single RuleType.textFormatted:

  • RuleType.textBolded
  • RuleType.textEmphasized
  • RuleType.textMarked
  • RuleType.textStrikethroughed

  If you're using these rule types directly (e.g., for custom AST processing or overrides), you'll need to update your code to check for RuleType.textFormatted instead and inspect the node's boolean flags (bold, italic, marked, strikethrough) to determine which formatting is applied.

Minor Changes

• a421067: fix: overhaul HTML block parsing to eliminate exponential backtracking

  Replaced the complex nested regex HTML_BLOCK_ELEMENT_R with an efficient iterative depth-counting algorithm that maintains O(n) complexity. The new implementation uses stateful regex matching with lastIndex to avoid exponential backtracking on nested HTML elements while preserving all existing functionality.

  Performance improvements:

  • Eliminates O(2^n) worst-case exponential backtracking
  • Linear O(n) time complexity regardless of nesting depth

Patch Changes

• e6b1e14: Fix renderer crash on extremely deeply nested markdown content

  Previously, rendering markdown with extremely deeply nested content (e.g., thousands of nested bold markers like ****************...text...****************) would cause a stack overflow crash. The renderer now gracefully handles such edge cases by falling back to plain text rendering instead of crashing.

  Technical details:

  • Added render depth tracking to prevent stack overflow
  • Graceful fallback at 2500 levels of nesting (way beyond normal usage)
  • Try/catch safety net as additional protection for unexpected errors
  • Zero performance impact during normal operation
  • Prevents crashes while maintaining O(n) parsing complexity

  This fix ensures stability even with adversarial or malformed inputs while having no impact on normal markdown documents.

• fe95c02: Remove unnecessary wrapper when footnotes are present.