unified is an interface for processing text using syntax trees. It’s what powers remark, retext, and rehype, but it also allows for processing between multiple syntaxes.
The website for unified, unified.js.org, provides a less
technical and more practical introduction to unified. Make sure to visit it
and try its introductory Guides.
npm:
npm install unifiedvar unified = require('unified')
var markdown = require('remark-parse')
var remark2rehype = require('remark-rehype')
var doc = require('rehype-document')
var format = require('rehype-format')
var html = require('rehype-stringify')
var report = require('vfile-reporter')
unified()
.use(markdown)
.use(remark2rehype)
.use(doc)
.use(format)
.use(html)
.process('# Hello world!', function(err, file) {
console.error(report(err || file))
console.log(String(file))
})Yields:
no issues found
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
</head>
<body>
<h1>Hello world!</h1>
</body>
</html>unified is an interface for processing text using syntax trees. Syntax trees are a representation understandable to programs. Those programs, called plugins, take these trees and modify them, amongst other things. To get to the syntax tree from input text there’s a parser. To get from that back to text there’s a compiler. This is the process of a processor.
| ....................... process() ......................... |
| ......... parse() ..... | run() | ..... stringify() ....... |
+--------+ +----------+
Input ->- | Parser | ->- Syntax Tree ->- | Compiler | ->- Output
+--------+ | +----------+
X
|
+--------------+
| Transformers |
+--------------+
Every processor implements another processor. To create a new processor invoke another processor. This creates a processor that is configured to function the same as its ancestor. But when the descendant processor is configured in the future it does not affect the ancestral processor.
When processors are exposed from a module (for example, unified itself) they should not be configured directly, as that would change their behaviour for all module users. Those processors are frozen and they should be invoked to create a new processor before they are used.
The syntax trees used in unified are Unist nodes: plain
JavaScript objects with a type property. The semantics of those types are
defined by other projects.
There are several utilities for working with these nodes.
The following projects process different syntax trees. They parse text to their respective syntax tree and they compile their syntax trees back to text. These processors can be used as-is, or their parsers and compilers can be mixed and matched with unified and other plugins to process between different syntaxes.
The below plugins work with unified, unrelated to what flavour the syntax tree is in:
unified-diff— Ignore messages for unchanged lines in Travis
See remark, rehype, and retext for lists of their plugins.
When processing documents metadata is often gathered about that document. VFile is a virtual file format which stores data and handles metadata and messages for unified and its plugins.
There are several utilities for working with these files.
To configure a processor invoke its use method, supply it a
plugin, and optionally settings.
unified can integrate with the file-system through
unified-engine. On top of that, CLI apps can be created with
unified-args, Gulp plugins with unified-engine-gulp, and
Atom Linters with unified-engine-atom.
A streaming interface is provided through unified-stream.
The API gives access to processing metadata (such as lint messages) and supports multiple passed through files:
var unified = require('unified')
var markdown = require('remark-parse')
var styleGuide = require('remark-preset-lint-markdown-style-guide')
var remark2retext = require('remark-retext')
var english = require('retext-english')
var equality = require('retext-equality')
var remark2rehype = require('remark-rehype')
var html = require('rehype-stringify')
var report = require('vfile-reporter')
unified()
.use(markdown)
.use(styleGuide)
.use(
remark2retext,
unified()
.use(english)
.use(equality)
)
.use(remark2rehype)
.use(html)
.process('*Emphasis* and _importance_, you guys!', function(err, file) {
console.error(report(err || file))
console.log(String(file))
})Yields:
1:16-1:28 warning Emphasis should use `*` as a marker emphasis-marker remark-lint
1:34-1:38 warning `guys` may be insensitive, use `people`, `persons`, `folks` instead gals-men retext-equality
⚠ 2 warnings
<p><em>Emphasis</em> and <em>importance</em>, you guys!</p>The processors can be combined in two modes.
Bridge mode transforms the syntax tree from one flavour (the origin) to another (the destination). Then, transformations are applied on that tree. Finally, the origin processor continues transforming the original syntax tree.
Mutate mode also transforms the syntax tree from one flavour to another. But then the origin processor continues transforming the destination syntax tree.
In the previous example (“Programming interface”), remark-retext is used in
bridge mode: the origin syntax tree is kept after retext is done; whereas
remark-rehype is used in mutate mode: it sets a new syntax tree and discards
the original.
Object describing how to process text.
Function — New unfrozen processor which is configured to
function the same as its ancestor. But when the descendant processor is
configured in the future it does not affect the ancestral processor.
The following example shows how a new processor can be created (from the remark processor) and linked to stdin(4) and stdout(4).
var remark = require('remark')
var concat = require('concat-stream')
process.stdin.pipe(concat(onconcat))
function onconcat(buf) {
var doc = remark()
.processSync(buf)
.toString()
process.stdout.write(doc)
}Configure the processor to use a plugin and optionally configure that plugin with options.
processor.use(plugin[, options])processor.use(preset)processor.use(list)
plugin(Plugin)options(*, optional) — Configuration forpluginpreset(Object) — Object with an optionalplugins(set tolist), and/or an optionalsettingsobjectlist(Array) — List of plugins, presets, and pairs (pluginandoptionsin an array)
processor — The processor on which use is invoked.
use cannot be called on frozen processors. Invoke the processor
first to create a new unfrozen processor.
There are many ways to pass plugins to .use(). The below example gives an
overview.
var unified = require('unified')
unified()
// Plugin with options:
.use(plugin, {})
// Plugins:
.use([plugin, pluginB])
// Two plugins, the second with options:
.use([plugin, [pluginB, {}]])
// Preset with plugins and settings:
.use({plugins: [plugin, [pluginB, {}]], settings: {position: false}})
// Settings only:
.use({settings: {position: false}})
function plugin() {}
function pluginB() {}Parse text to a syntax tree.
file(VFile) — Or anything which can be given tovfile()
Node — Syntax tree representation of input.
parse freezes the processor if not already frozen.
parse does not apply transformers from the run phase to the
syntax tree.
The below example shows how the parse function can be used to create a
syntax tree from a file.
var unified = require('unified')
var markdown = require('remark-parse')
var tree = unified()
.use(markdown)
.parse('# Hello world!')
console.log(tree)Yields:
{ type: 'root',
children:
[ { type: 'heading',
depth: 1,
children: [Array],
position: [Position] } ],
position:
{ start: { line: 1, column: 1, offset: 0 },
end: { line: 1, column: 15, offset: 14 } } }Function handling the parsing of text to a syntax tree. Used in the
parse phase in the process and invoked with a string and
VFile representation of the document to parse.
Parser can be a normal function in which case it must return a
Node: the syntax tree representation of the given file.
Parser can also be a constructor function (a function with keys in its
prototype) in which case it’s invoked with new. Instances must have a
parse method which is invoked without arguments and must return a
Node.
Compile a syntax tree to text.
string — String representation of the syntax tree file.
stringify freezes the processor if not already frozen.
stringify does not apply transformers from the run phase
to the syntax tree.
The below example shows how the stringify function can be used to generate a
file from a syntax tree.
var unified = require('unified')
var html = require('rehype-stringify')
var h = require('hastscript')
var tree = h('h1', 'Hello world!')
var doc = unified()
.use(html)
.stringify(tree)
console.log(doc)Yields:
<h1>Hello world!</h1>Function handling the compilation of syntax tree to a text. Used in the
stringify phase in the process and invoked with a
Node and VFile representation of the document to stringify.
Compiler can be a normal function in which case it must return a string:
the text representation of the given syntax tree.
Compiler can also be a constructor function (a function with keys in its
prototype) in which case it’s invoked with new. Instances must have a
compile method which is invoked without arguments and must return a string.
Transform a syntax tree by applying plugins to it.
node(Node)file(VFile, optional) — Or anything which can be given tovfile()done(Function, optional)
Promise if done is not given. Rejected with an error, or
resolved with the resulting syntax tree.
run freezes the processor if not already frozen.
Invoked when transformation is complete. Either invoked with an error or a syntax tree and a file.
The below example shows how the run function can be used to transform a
syntax tree.
var unified = require('unified')
var references = require('remark-reference-links')
var u = require('unist-builder')
var tree = u('root', [
u('paragraph', [
u('link', {href: 'https://example.com'}, [u('text', 'Example Domain')])
])
])
unified()
.use(references)
.run(tree, function(err, tree) {
if (err) throw err
console.log(tree)
})Yields:
{ type: 'root',
children:
[ { type: 'paragraph', children: [Array] },
{ type: 'definition',
identifier: '1',
title: undefined,
url: undefined } ] }Transform a syntax tree by applying plugins to it.
If asynchronous plugins are configured an error is thrown.
Node — The given syntax tree.
runSync freezes the processor if not already frozen.
Process the given representation of a file as configured on the processor. The
process invokes parse, run, and stringify internally.
Promise if done is not given. Rejected with an error or
resolved with the resulting file.
process freezes the processor if not already frozen.
The below example shows how the process function can be used to process a
file whether plugins are asynchronous or not with Promises.
var unified = require('unified')
var markdown = require('remark-parse')
var remark2rehype = require('remark-rehype')
var doc = require('rehype-document')
var format = require('rehype-format')
var html = require('rehype-stringify')
unified()
.use(markdown)
.use(remark2rehype)
.use(doc)
.use(format)
.use(html)
.process('# Hello world!')
.then(
function(file) {
console.log(String(file))
},
function(err) {
console.error(String(err))
}
)Yields:
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
</head>
<body>
<h1>Hello world!</h1>
</body>
</html>Invoked when the process is complete. Invoked with a fatal error, if any, and
the VFile.
err(Error, optional) — Fatal errorfile(VFile)
The below example shows how the process function can be used to process a
file whether plugins are asynchronous or not with a callback.
var unified = require('unified')
var parse = require('remark-parse')
var stringify = require('remark-stringify')
var github = require('remark-github')
var report = require('vfile-reporter')
unified()
.use(parse)
.use(github)
.use(stringify)
.process('@mention', function(err, file) {
console.error(report(err || file))
console.log(String(file))
})Yields:
no issues found
[**@mention**](https://github.com/blog/821)Process the given representation of a file as configured on the processor. The
process invokes parse, run, and stringify internally.
If asynchronous plugins are configured an error is thrown.
file(VFile)value(string) — String representation of a file
VFile — Virtual file with modified contents.
processSync freezes the processor if not already frozen.
The below example shows how the processSync function can be used to process a
file if all plugins are known to be synchronous.
var unified = require('unified')
var markdown = require('remark-parse')
var remark2rehype = require('remark-rehype')
var doc = require('rehype-document')
var format = require('rehype-format')
var html = require('rehype-stringify')
var processor = unified()
.use(markdown)
.use(remark2rehype)
.use(doc)
.use(format)
.use(html)
console.log(processor.processSync('# Hello world!').toString())Yields:
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
</head>
<body>
<h1>Hello world!</h1>
</body>
</html>Get or set information in an in-memory key-value store accessible to all phases of the process. An example is a list of HTML elements which are self-closing, which is needed when parsing, transforming, and compiling HTML.
key(string) — Identifiervalue(*, optional) — Value to set. Omit if gettingkey
processor— If setting, the processor on whichdatais invoked*— If getting, the value atkey
Setting information with data cannot occur on frozen processors.
Invoke the processor first to create a new unfrozen processor.
The following example show how to get and set information:
var unified = require('unified')
unified()
.data('alpha', 'bravo')
.data('alpha') // => 'bravo'Freeze a processor. Frozen processors are meant to be extended and not to be configured or processed directly.
Once a processor is frozen it cannot be unfrozen. New processors functioning just like it can be created by invoking the processor.
It’s possible to freeze processors explicitly, by calling .freeze(), but
.parse(), .run(), .stringify(), and
.process() call .freeze() to freeze a processor too.
Processor — The processor on which freeze is invoked.
The following example, index.js, shows how rehype prevents
extensions to itself:
var unified = require('unified')
var parse = require('rehype-parse')
var stringify = require('rehype-stringify')
module.exports = unified()
.use(parse)
.use(stringify)
.freeze()The below example, a.js, shows how that processor can be used and configured.
var rehype = require('rehype')
var format = require('rehype-format')
// ...
rehype()
.use(format)
// ...The below example, b.js, shows a similar looking example which operates on
the frozen rehype interface. If this behaviour was allowed it
would result in unexpected behaviour so an error is thrown. This is
invalid:
var rehype = require('rehype')
var format = require('rehype-format')
// ...
rehype
.use(format)
// ...Yields:
~/node_modules/unified/index.js:440
throw new Error(
^
Error: Cannot invoke `use` on a frozen processor.
Create a new processor first, by invoking it: use `processor()` instead of `processor`.
at assertUnfrozen (~/node_modules/unified/index.js:440:11)
at Function.use (~/node_modules/unified/index.js:172:5)
at Object.<anonymous> (~/b.js:6:4)unified plugins change the way the applied-on processor works in the following ways:
- They modify the processor: such as changing the parser, the compiler, or linking it to other processors
- They transform syntax tree representation of files
- They modify metadata of files
Plugins are a concept. They materialise as attachers.
move.js:
module.exports = move
function move(options) {
var expected = (options || {}).extname
if (!expected) {
throw new Error('Missing `extname` in options')
}
return transformer
function transformer(tree, file) {
if (file.extname && file.extname !== expected) {
file.extname = expected
}
}
}index.js:
var unified = require('unified')
var parse = require('remark-parse')
var remark2rehype = require('remark-rehype')
var stringify = require('rehype-stringify')
var vfile = require('to-vfile')
var report = require('vfile-reporter')
var move = require('./move')
unified()
.use(parse)
.use(remark2rehype)
.use(move, {extname: '.html'})
.use(stringify)
.process(vfile.readSync('index.md'), function(err, file) {
console.error(report(err || file))
if (file) {
vfile.writeSync(file) // Written to `index.html`.
}
})An attacher is the thing passed to use. It configures the processor
and in turn can receive options.
Attachers can configure processors, such as by interacting with parsers and compilers, linking them to other processors, or by specifying how the syntax tree is handled.
The context object is set to the invoked on processor.
options(*, optional) — Configuration
transformer — Optional.
Attachers are invoked when the processor is frozen: either when
.freeze() is called explicitly, or when .parse(), .run(),
.stringify(), or .process() is called for the first
time.
Transformers modify the syntax tree or metadata of a file. A transformer is a
function which is invoked each time a file is passed through the transform
phase. If an error occurs (either because it’s thrown, returned, rejected, or
passed to next), the process stops.
The transformation process in unified is handled by trough, see
it’s documentation for the exact semantics of transformers.
Error— Can be returned to stop the processNode— Can be returned and results in further transformations andstringifys to be performed on the new treePromise— If a promise is returned, the function is asynchronous, and must be resolved (optionally with aNode) or rejected (optionally with anError)
If the signature of a transformer includes next (third argument), the
function may finish asynchronous, and must invoke next().
err(Error, optional) — Stop the processnode(Node, optional) — New syntax treefile(VFile, optional) — New virtual file
Presets provide a potentially sharable way to configure processors. They can contain multiple plugins and optionally settings as well.
preset.js:
exports.settings = {bullet: '*', fences: true}
exports.plugins = [
require('remark-preset-lint-recommended'),
require('remark-comment-config'),
require('remark-preset-lint-markdown-style-guide'),
[require('remark-toc'), {maxDepth: 3, tight: true}],
require('remark-github')
]index.js:
var remark = require('remark')
var vfile = require('to-vfile')
var report = require('vfile-reporter')
var preset = require('./preset')
remark()
.use(preset)
.process(vfile.readSync('index.md'), function(err, file) {
console.error(report(err || file))
if (file) {
vfile.writeSync(file)
}
})unified is built by people just like you! Check out
contributing.md for ways to get started.
This project has a Code of Conduct. By interacting with this repository, organisation, or community you agree to abide by its terms.
Want to chat with the community and contributors? Join us in Gitter!
Have an idea for a cool new utility or tool? That’s great! If you want
feedback, help, or just to share it with the world you can do so by creating
an issue in the unifiedjs/ideas repository!
Preliminary work for unified was done in 2014 for
retext and inspired by ware. Further incubation
happened in remark. The project was finally externalised
in 2015 and published as unified. The project was authored by
@wooorm.
Although unified since moved it’s plugin architecture to trough,
thanks to @calvinfo,
@ianstormtaylor, and others for their
work on ware, which was a huge initial inspiration.