TypeScript icon, indicating that this package has built-in type declarations

5.0.1 • Public • Published


Build Coverage Downloads Size Sponsors Backers Chat

hast utility to make trees safe.


What is this?

This package is a utility that can make a tree that potentially contains dangerous user content safe for use. It defaults to what GitHub does to clean unsafe markup, but you can change that.

When should I use this?

This package is needed whenever you deal with potentially dangerous user content.

The plugin rehype-sanitize wraps this utility to also sanitize HTML at a higher-level (easier) abstraction.


This package is ESM only. In Node.js (version 16+), install with npm:

npm install hast-util-sanitize

In Deno with esm.sh:

import {sanitize} from 'https://esm.sh/hast-util-sanitize@5'

In browsers with esm.sh:

<script type="module">
  import {sanitize} from 'https://esm.sh/hast-util-sanitize@5?bundle'


import {h} from 'hastscript'
import {sanitize} from 'hast-util-sanitize'
import {toHtml} from 'hast-util-to-html'
import {u} from 'unist-builder'

const unsafe = h('div', {onmouseover: 'alert("alpha")'}, [
    {href: 'jAva script:alert("bravo")', onclick: 'alert("charlie")'},
  u('text', '\n'),
  h('script', 'alert("charlie")'),
  u('text', '\n'),
  h('img', {src: 'x', onerror: 'alert("delta")'}),
  u('text', '\n'),
  h('iframe', {src: 'javascript:alert("echo")'}),
  u('text', '\n'),
  h('math', h('mi', {'xlink:href': 'data:x,<script>alert("foxtrot")</script>'}))

const safe = sanitize(unsafe)



<div onmouseover="alert(&#x22;alpha&#x22;)"><a href="jAva script:alert(&#x22;bravo&#x22;)" onclick="alert(&#x22;charlie&#x22;)">delta</a>
<img src="x" onerror="alert(&#x22;delta&#x22;)">
<iframe src="javascript:alert(&#x22;echo&#x22;)"></iframe>
<math><mi xlink:href="data:x,<script>alert(&#x22;foxtrot&#x22;)</script>"></mi></math></div>



<img src="x">



This package exports the identifiers defaultSchema and sanitize. There is no default export.


Default schema (Schema).

Follows GitHub style sanitation.

sanitize(tree[, options])

Sanitize a tree.


New, safe tree (Node).


Schema that defines what nodes and properties are allowed.

The default schema is defaultSchema, which follows how GitHub cleans. If any top-level key is missing in the given schema, the corresponding value of the default schema is used.

To extend the standard schema with a few changes, clone defaultSchema like so:

import deepmerge from 'deepmerge'
import {h} from 'hastscript'
import {defaultSchema, sanitize} from 'hast-util-sanitize'

// This allows `className` on all elements.
const schema = deepmerge(defaultSchema, {attributes: {'*': ['className']}})

const tree = sanitize(h('div', {className: ['foo']}), schema)

// `tree` still has `className`.
// {
//   type: 'element',
//   tagName: 'div',
//   properties: {className: ['foo']},
//   children: []
// }

Whether to allow comment nodes (boolean, default: false).

For example:

allowComments: true

Whether to allow doctype nodes (boolean, default: false).

For example:

allowDoctypes: true

Map of tag names to a list of tag names which are required ancestors (Record<string, Array<string>>, default: defaultSchema.ancestors).

Elements with these tag names will be ignored if they occur outside of one of their allowed parents.

For example:

ancestors: {
  tbody: ['table'],
  // …
  tr: ['table']

Map of tag names to allowed property names (Record<string, Array<[string, ...Array<RegExp | boolean | number | string>] | string>, default: defaultSchema.attributes).

The special key '*' as a tag name defines property names allowed on all elements.

The special value 'data*' as a property name can be used to allow all data properties.

For example:

attributes: {
  a: [
    'ariaDescribedBy', 'ariaLabel', 'ariaLabelledBy', /* … */, 'href'
  // …
  '*': [
    // …

Instead of a single string in the array, which allows any property value for the field, you can use an array to allow several values. For example, input: ['type'] allows type set to any value on inputs. But input: [['type', 'checkbox', 'radio']] allows type when set to 'checkbox' or 'radio'.

You can use regexes, so for example span: [['className', /^hljs-/]] allows any class that starts with hljs- on spans.

When comma- or space-separated values are used (such as className), each value in is checked individually. For example, to allow certain classes on spans for syntax highlighting, use span: [['className', 'number', 'operator', 'token']]. This will allow 'number', 'operator', and 'token' classes, but drop others.


List of property names that clobber (Array<string>, default: defaultSchema.clobber).

For example:

clobber: ['ariaDescribedBy', 'ariaLabelledBy', 'id', 'name']

Prefix to use before clobbering properties (string, default: defaultSchema.clobberPrefix).

For example:

clobberPrefix: 'user-content-'

Map of property names to allowed protocols (Record<string, Array<string>>, default: defaultSchema.protocols).

This defines URLs that are always allowed to have local URLs (relative to the current website, such as this, #this, /this, or ?this), and only allowed to have remote URLs (such as https://example.com) if they use a known protocol.

For example:

protocols: {
  cite: ['http', 'https'],
  // …
  src: ['http', 'https']

Map of tag names to required property names with a default value (Record<string, Record<string, unknown>>, default: defaultSchema.required).

This defines properties that must be set. If a field does not exist (after the element was made safe), these will be added with the given value.

For example:

required: {
  input: {disabled: true, type: 'checkbox'}

👉 Note: properties are first checked based on schema.attributes, then on schema.required. That means properties could be removed by attributes and then added again with required.


List of tag names to strip from the tree (Array<string>, default: defaultSchema.strip).

By default, unsafe elements (those not in schema.tagNames) are replaced by what they contain. This option can drop their contents.

For example:

strip: ['script']

List of allowed tag names (Array<string>, default: defaultSchema.tagNames).

For example:

tagNames: [
  // …


This package is fully typed with TypeScript. It exports the additional type Schema.


Projects maintained by the unified collective are compatible with maintained versions of Node.js.

When we cut a new major release, we drop support for unmaintained versions of Node. This means we try to keep the current release line, hast-util-sanitize@^5, compatible with Node.js 16.


By default, hast-util-sanitize will make everything safe to use. Assuming you understand that certain attributes (including a limited set of classes) can be generated by users, and you write your CSS (and JS) accordingly. When used incorrectly, deviating from the defaults can open you up to a cross-site scripting (XSS) attack.

Use hast-util-sanitize after the last unsafe thing: everything after it could be unsafe (but is fine if you do trust it).


See contributing.md in syntax-tree/.github for ways to get started. See support.md for ways to get help.

This project has a code of conduct. By interacting with this repository, organization, or community you agree to abide by its terms.


MIT © Titus Wormer

Package Sidebar


npm i hast-util-sanitize

Weekly Downloads






Unpacked Size

46.7 kB

Total Files


Last publish


  • wooorm
  • kmck