hast-util-from-html
hast utility that turns HTML into a syntax tree.
Contents
What is this?
This package is a utility that takes HTML input and turns it into a hast syntax tree.
When should I use this?
If you want to handle syntax trees manually, use this.
Use parse5 instead when you just want to parse HTML and don’t care about hast. You can also use hast-util-from-parse5 and parse5 yourself, or use the rehype plugin rehype-parse, which wraps this utility to also parse HTML at a higher-level (easier) abstraction. xast-util-from-xml can be used if you are dealing with XML instead of HTML.
If you might run in a browser and prefer a ligher alternative, while not caring about positional info, parse errors, and consistency across browsers, use hast-util-from-html-isomorphic, which wraps this in Node and uses browser APIs otherwise.
Finally you can use the utility hast-util-to-html for the inverse of this utility. It turns hast into HTML.
Install
This package is ESM only. In Node.js (version 16+), install with npm:
npm install hast-util-from-html
In Deno with esm.sh:
import {fromHtml} from 'https://esm.sh/hast-util-from-html@2'
In browsers with esm.sh:
<script type="module">
import {fromHtml} from 'https://esm.sh/hast-util-from-html@2?bundle'
</script>
Use
import {fromHtml} from 'hast-util-from-html'
const tree = fromHtml('<h1>Hello, world!</h1>', {fragment: true})
console.log(tree)
Yields:
{
type: 'root',
children: [
{
type: 'element',
tagName: 'h1',
properties: {},
children: [Array],
position: [Object]
}
],
data: { quirksMode: false },
position: {
start: { line: 1, column: 1, offset: 0 },
end: { line: 1, column: 23, offset: 22 }
}
}
API
This package exports the identifier fromHtml. There is no default export.
fromHtml(value[, options])
Turn serialized HTML into a hast tree.
Parameters
value (Compatible) — serialized HTML to parseoptions (Options, optional) — configuration
Returns
Tree (Root).
ErrorCode
Known names of parse errors (TypeScript type).
Types
type ErrorCode =
| 'abandonedHeadElementChild'
| 'abruptClosingOfEmptyComment'
| 'abruptDoctypePublicIdentifier'
// … see readme on `options[key in ErrorCode]` above.
ErrorSeverity
Error severity (TypeScript type).
Types
export type ErrorSeverity =
// Turn the parse error off:
| 0
| false
// Turn the parse error into a warning:
| 1
| true
// Turn the parse error into an actual error: processing stops.
| 2
OnError
Function called when encountering HTML parse errors.
Parameters
Returns
Nothing (void).
Options
Configuration (TypeScript type).
Fields
options.space
Which space the document is in ('html' or 'svg', default: 'html').
When an <svg> element is found in the HTML space, hast-util-from-html already automatically switches to and from the SVG space when entering and exiting it.
👉 Note: this is not an XML parser. It supports SVG as embedded in HTML. It does not support the features available in XML. Passing SVG files might break but fragments of modern SVG should be fine. Use xast-util-from-xml to parse XML.
👉 Note: make sure to set fragment: true if space: 'svg'.
options.verbose
Add extra positional info about attributes, start tags, and end tags (boolean, default: false).
options.fragment
Whether to parse as a fragment (boolean, default: false). The default is to expect a whole document. In document mode, unopened html, head, and body elements are opened.
options.onerror
Function called when encountering HTML parse errors (OnError, optional).
options[key in ErrorCode]
Specific parse errors can be configured by setting their identifiers (see ErrorCode) as keys directly in options to an ErrorSeverity as value.
The list of parse errors:
Examples
Example: fragment versus document
The following example shows the difference between parsing as a document and parsing as a fragment:
import {fromHtml} from 'hast-util-from-html'
const doc = '<title>Hi!</title><h1>Hello!</h1>'
console.log(fromHtml(doc))
console.log(fromHtml(doc, {fragment: true}))
…yields (positional info and data omitted for brevity):
{
type: 'root',
children: [
{type: 'element', tagName: 'html', properties: {}, children: [Array]}
]
}
{
type: 'root',
children: [
{type: 'element', tagName: 'title', properties: {}, children: [Array]},
{type: 'element', tagName: 'h1', properties: {}, children: [Array]}
]
}
👉 Note: observe that when a whole document is expected (first example), missing elements are opened and closed.
Example: whitespace around and inside <html>
The following example shows how whitespace is handled when around and directly inside the <html> element:
import {fromHtml} from 'hast-util-from-html'
import {inspect} from 'unist-util-inspect'
const doc = `<!doctype html>
<html lang=en>
<head>
<title>Hi!</title>
</head>
<body>
<h1>Hello!</h1>
</body>
</html>`
console.log(inspect(fromHtml(doc)))
…yields:
root[2] (1:1-9:8, 0-119)
│ data: {"quirksMode":false}
├─0 doctype (1:1-1:16, 0-15)
└─1 element<html>[3] (2:1-9:8, 16-119)
│ properties: {"lang":"en"}
├─0 element<head>[3] (3:3-5:10, 33-72)
│ │ properties: {}
│ ├─0 text "\n " (3:9-4:5, 39-44)
│ ├─1 element<title>[1] (4:5-4:23, 44-62)
│ │ │ properties: {}
│ │ └─0 text "Hi!" (4:12-4:15, 51-54)
│ └─2 text "\n " (4:23-5:3, 62-65)
├─1 text "\n " (5:10-6:3, 72-75)
└─2 element<body>[3] (6:3-8:10, 75-111)
│ properties: {}
├─0 text "\n " (6:9-7:5, 81-86)
├─1 element<h1>[1] (7:5-7:20, 86-101)
│ │ properties: {}
│ └─0 text "Hello!" (7:9-7:15, 90-96)
└─2 text "\n \n" (7:20-9:1, 101-112)
👉 Note: observe that the line ending before <html> is ignored, the line ending and two spaces before <head> is moved inside it, and the line ending after </body> is moved before it.
This behavior is described by the HTML standard (see the section 13.2.6.4.1 “The ‘initial’ insertion mode” and adjacent states) which we follow.
The changes to this meaningless whitespace should not matter, except when formatting markup, in which case rehype-format can be used to improve the source code.
Example: parse errors
The following example shows how HTML parse errors can be enabled and configured:
import {fromHtml} from 'hast-util-from-html'
const doc = `<!doctypehtml>
<title class="a" class="b">Hello…</title>
<h1/>World!</h1>`
fromHtml(doc, {
onerror: console.log,
missingWhitespaceBeforeDoctypeName: 2, // Mark one as a fatal error.
nonVoidHtmlElementStartTagWithTrailingSolidus: false // Ignore one.
})
…yields:
[1:10-1:10: Missing whitespace before doctype name] {
ancestors: undefined,
cause: undefined,
column: 10,
fatal: true,
line: 1,
place: {
start: { line: 1, column: 10, offset: 9 },
end: { line: 1, column: 10, offset: 9 }
},
reason: 'Missing whitespace before doctype name',
ruleId: 'missing-whitespace-before-doctype-name',
source: 'hast-util-from-html',
note: 'Unexpected `h`. Expected ASCII whitespace instead',
url: 'https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-whitespace-before-doctype-name'
}
[2:23-2:23: Unexpected duplicate attribute] {
ancestors: undefined,
cause: undefined,
column: 23,
fatal: false,
line: 2,
place: {
start: { line: 2, column: 23, offset: 37 },
end: { line: 2, column: 23, offset: 37 }
},
reason: 'Unexpected duplicate attribute',
ruleId: 'duplicate-attribute',
source: 'hast-util-from-html',
note: 'Unexpectedly double attribute. Expected attributes to occur only once',
url: 'https://html.spec.whatwg.org/multipage/parsing.html#parse-error-duplicate-attribute'
}
🧑🏫 Info: messages in unified are warnings instead of errors. Other linters (such as ESLint) almost always use errors. Why? Those tools only check code style. They don’t generate, transform, and format code, which is what we focus on, too. Errors in unified mean the same as an exception in your JavaScript code: a crash. That’s why we use warnings instead, because we can continue doing work.
Syntax
HTML is parsed according to WHATWG HTML (the living standard), which is also followed by browsers such as Chrome and Firefox.
Types
This package is fully typed with TypeScript. It exports the additional types ErrorCode, ErrorSeverity, OnError, and Options.
Compatibility
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-from-html@^2, compatible with Node.js 16.
Security
Parsing HTML is safe but using user-provided content can open you up to a cross-site scripting (XSS) attack. Use hast-util-santize to make the hast tree safe.
Contribute
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.
License
MIT © Titus Wormer