What's Changed

This release aligns HTML parsing with the WHATWG spec Almost all changes are to HTML mode only — XML mode is unaffected unless noted.

Raw-text & RCDATA tags

• <iframe>, <noembed>, <noframes>, and <plaintext> are now raw-text tags, their content is no longer parsed as HTML
• <textarea> now decodes entities like <title> already did
• Self-closing <script/>, <style/>, etc. now enter their raw-text state (the / is ignored per spec) unless recognizeSelfClosing is enabled

SVG & MathML

• Tag names inside <svg> are case-adjusted per spec (foreignObject, clipPath, etc.)
• CDATA sections inside foreign content are treated as text
• Special-tag detection is disabled inside foreign content
• Stray </svg> / </math> no longer corrupt the parser's context tracking

Comments & declarations

• <!-->, <!--->, <!->, <!> now parse as valid comments per spec
• <?…> and non-DOCTYPE <!…> in HTML mode emit bogus comments instead of being silently dropped
• <!DOCTYPEhtml> (no space) is recognized as a DOCTYPE
• Unclosed comments, <!DOCTYPE, <?…, <![CDATA[… at EOF emit the correct token type

Implicit open/close

• <h1>–<h6> implicitly close other headings
• <a> closes a previous <a>
• Nested <form> is ignored when one is already open
• <image> is rewritten to <img> outside foreign content
• </> is silently ignored instead of emitted as text

Other fixes

• Fixed reset() not clearing attribute state, which could leak data across parseComplete() calls

#2387

Full Changelog: v11.0.0...v12.0.0

Breaking Changes

• The module is now ESM only #2381
  • CommonJS require() is not supported in legacy environment anymore. Use import instead.
  • The minimum Node.js version is now 20.19.0.
• Dependencies have been bumped to their latest major versions: domhandler v6, domutils v4, domelementtype v3, entities v8.

Features

• Added WebWritableStream for the Web Streams API, enabling direct piping from fetch() response bodies into the parser #2376

Bug Fixes

• Comments now accept --!> as a closing sequence per the HTML spec, and <!--> is recognized as an empty comment in HTML mode #2382
• XML processing instructions (<?xml ... ?>) now require the full ?> closing sequence instead of just > #2382
• Fixed reset() not clearing isSpecial and sequenceIndex state, which could cause incorrect parsing after reuse #2382
• Fixed XML comment parsing: <!--> is no longer treated as a complete comment in xmlMode #2383

Other Changes

• Expanded README with full API reference, parser options, events, and practical examples #2384

New Contributors

• @vimzh made their first contribution in #2376

Full Changelog: v10.1.0...v11.0.0

What's Changed

• entities was bumped from 6.0.1 to 7.0.1, bringing size & speed improvements #2215
• Test files are no longer shipped in the published module 72da671

New Contributors

• @KTibow made their first contribution, bumping us to eslint 9 in #2204

Full Changelog: v10.0.0...v10.1.0

• Breaking: Simplify writable stream import path d5882db
• feat: Support xmp tag parsing (#1790 by @nati-elmaliach) ecdb071
• Run tests with vitest (#1845) aa0c781
• Dependency upgrades

Fixes

• Fixed onattribend's endIndex (#1540 by @DimaIT)
• Treat textarea as special tag (#1719 by @DimaIT)

Features

• Export QuoteType (#1543 by @DimaIT) and Handler interface (#1690 by @benkroeger)

Breaking Changes

• The tokenizer now uses the EntityDecoder from the entities module #1480
  • Parsing of entities in attributes is now aligned with the HTML spec, and some inputs will produce different results. Eg. in <a href='&amp=boo'> the attribute value won't be modified any more.
  • The ontextentity tokenizer callback now has an endIndex argument; if you use the tokenizer directly, make sure indices are still the same.
• Stacks inside the parser have been reversed. #1511

Features

• Added a createDocumentStream function, analogous to createDomStream (which is now deprecated) #1510

Full Changelog: v8.0.2...v9.0.0

Bug Fixes

• Reset tokenizer baseState after closing tag name by @KillyMXI in #1460

Other changes

• Dependency version bumps
• GitHub Workflows security hardening by @sashashura in #1365
• refactor(lint): Add eslint-plugin-n and -unicorn by @fb55 in #1352
• chore(test): Move from JSON tests to specs by @fb55 in #1354
• docs(readme): Use GitHub Actions CI badge by @fb55 in #1374

New Contributors

• @sashashura made their first contribution in #1365
• @KillyMXI made their first contribution in #1460

Full Changelog: v8.0.1...v8.0.2

• Added missing WritableStream export in the package.json 6923fca

v8.0.0...v8.0.1

Breaking

• The deprecated FeedHandler class has been removed #1166
  • See #1166 for how to migrate.
• Typescript >= 4.5 is now required; see #1242
• The types from domhandler and domutils have changed, the deprecated normalizeWhitespace option was removed #1164
• The parser was updated to no longer concatenate strings. This led to several changes of internal interfaces. #1045
  • This reduces the memory overhead when parsing streams, and avoids copying memory.
  • Breaking if you were previously extending internals.
• Parser.write() and Parser.end() now only accept string arguments. If you were previously
  passing Buffer, convert it to a string first (e.g. parser.write(buffer.toString())), or use
  WritableStream which handles decoding for you.

Features

• htmlparser2 is now a dual CommonJS & ESM module #1165

Other changes

• Updated for entities' updated decoding tree structure #1146
• Highlight special close-implies-open logic by @vassudanagunta in #1047
• Update Events/07 test to clarify interpretation of tag end slashes by @vassudanagunta in #1046
• Suggest parse5 for HTML compliance by @vassudanagunta in #1147

New Contributors

• @vassudanagunta made their first contribution in #1047

Full Changelog: v7.2.0...v8.0.0

What's Changed

Fixes:

• Decode entities after < by @fb55 in #1008
• Stringify non-string chunks by @fb55 in #1010

Docs

• docs(readme): make parseDocument() example clearer by @cameronsteele in #998

Refactors:

• Introduce sequences & fast forwarding by @fb55 in #1007
• Emit text before entities once entity is confirmed by @fb55 in #1009

The refactors lead to a combined ~5% speed-up.

New Contributors

• @cameronsteele made their first contribution in #998

Full Changelog: v7.1.2...v7.2.0