A simple and unified notation for both object and markup data. (beta)

View project on GitHub

Mark Notation

npm version Build Status codecov Analytics

Mark Notation or simply Mark, is a new unified notation for both object and markup data. The notation is a superset of what can be represented by JSON, HTML and XML, but overcomes many limitations of these popular data formats, yet still having a very clean syntax and simple data model.

  • It has clean syntax with fully-type data model (like JSON or even better)
  • It is generic and extensible (like XML or even better)
  • It has built-in mixed content support (like HTML5 or even better)
  • It supports high-order composition (like S-expressions or even better)
Clean syntax yes yes no verbose yes yes (only for basic usage)
Fully-typed yes yes no no (when no schema) yes yes
Generic yes yes no yes yes yes
Mixed content support yes hard yes yes hard (poor map support) hard
High-order composition yes possible no verbose yes possible
Wide adoption not (at the moment) yes yes yes limited limited

Mark Syntax

The major syntax extension Mark makes to JSON is the introduction of a Mark object. It is a JSON object extended with a type name and a list of content items, similar to element in HTML and XML.

For example, a HTML registration form:

  <div class="form-group">
    <label for="email">Email address:</label>
    <input type="email" id="email">
  <div class="form-group">
    <label for="pwd">Password</label>
    <input type="password" id="pwd">
  <button class='btn btn-info'>Submit</button>

Could be represented in Mark as:

{form                                   // object with type-name 'form'
  (?comment?)                           // Mark pragma, like HTML comment
  {div class:"form-group"               // nested Mark object
    {label for:email                    // property 'for' and its value, both unquoted
      "Email address:"                  // text needs to be quoted
    {input type:email id:email}         // object without any contents
  {div class:form-group
    {label for:pwd "Password"}
    {input type:password id:pwd}        // comma is optional 
  {button class:['btn' btn-info]        // property with complex values
    'Submit'                            // text quoted with single quote

You can refer to the syntax spec for details.

Mark Data Model

Mark has a very simple and fully-typed data model.

Each Mark object has 3 facets of data:

  • Type name, which is mapped to under JavaScript.
  • Properties, which is a collection of key-value pairs, stored as normal JavaScript named properties.
  • Contents, which is a list of content objects, stored as indexed properties inside the same JavaScript object.

Mark utilizes a novel feature in JavaScript that a plain JS object is actually array-like, it can contain both named properties and indexed properties.

So each Mark object is mapped to just one plain JavaScript object, which is more compact and efficient comparing to other JSON-based DOM models (e.g. JsonML, virtual-dom, MicroXML), and is more intuitive to used under JS.

Roughly speaking, data models of JSON, XML, HTML are subsets of Mark data model, and Mark data model is a subset of JS data model.

You can refer to the data model spec for details.

Mark vs. JSON

Mark is a superset of JSON. It extends JSON notation with a type-name, and a list of content objects.

Comparing to JSON, Mark has the following advantages:

  • It has a type-name, which is important in identifying what the data represents; whereas JSON is actually an anonymous object, missing the type name.
  • It has built-in mixed-content support, which is common in all markup formats, and thus allows Mark to convenient represent document-oriented data, which is awkward for JSON.
  • It incorporates some syntax enhancements to JSON ~(e.g. allowing comments, property name and value without quotes, optional trailing comma or between properties and array values)~, thus making the format more friendly for human.

Some disadvantages of Mark, comparing to JSON would be:

  • It is no longer a subset of JavaScript in syntax, although a Mark object is still a simple POJO in data model.
  • It does not yet have wide support, like JSON, at the moment.

Mark vs. HTML

Comparing to HTML, Mark has the following advantages:

  • Mark is a generic data format, whereas HTML is a specialized format for web content.
  • It does not have whitespace ambiguity, as the text objects are quoted explicitly. Thus Mark can be minified or prettified without worrying about changing the underlying content.
  • Its properties can have complex values, like JSON, not just quoted string values as in HTML.
  • It has a very clean syntax, whereas HTML5 parsing can be challenging even with HTML5 spec.
  • It is always properly closed; whereas HTML self-closing tag syntax is non-extensible and error-prone.
  • The DOM produced under Mark model, is just a hierarchy of POJO objects, which can be easily processed using the built-in JS functions or 3rd party libraries, making Mark an ideal candidate for virtual DOM and other application usages.

Mark vs. XML

Comparing to XML, Mark has the following advantages:

  • Mark properties can have complex object as value; whereas XML attribute values always need to be quoted and cannot have complex object as value, which is not flexible in syntax and data model.
  • Mark syntax is much cleaner than XML. It does not have whitespace ambiguity. It does not have all the legacy things like DTD.
  • The data model produced by Mark is fully typed, like JSON; whereas XML is only semi-typed without schema.

Mark vs. S-expressions

S-expression from Lisp gave rise to novel ideas like high-order composition, self-hosting program. The clean and flexible syntax of Mark make it ideal for many such applications (e.g. Mark Template, a new JS template engine using Mark for its template syntax), just like s-expression.

The advantage of Mark over S-expressions is that it takes a more modern, JS-first approach in its design, and can be more readily used in web and node.js environments.


mark.js is the JS library to work with data in Mark format. It consists of 4 modules:

  • The core module mark.js, which provides parse() and stringify() functions, like JSON, and a direct Mark object construction function Mark(), and some functional APIs to work with the object content.
  • Sub-module mark.mutate.js, which provides mutative APIs to change the Mark object data model.
  • Sub-module mark.convert.js, which provides conversion between Mark format and other formats like HTML and XML.
  • Sub-module mark.selector.js, which provides CSS selector based query interface on the Mark object model, like jQuery.


Install from NPM:

npm install mark-js --save

Then in your node script, use it as:

const Mark = require('mark-js');
var obj = Mark.parse(`{div {span 'Hello World!'}}`);
console.log("Greeting from Mark: " + Mark.stringify(obj));

To use the library in browser, you can include the mark.js under /dist directory into your html page, like:

<script src='mark.js'></script>
var obj = Mark(`{div {span 'Hello World!'}}`);  // using a shorthand
console.log("Greeting from Mark: " + Mark.stringify(obj));

Note: /dist/mark.js has bundled all sub-modules and all dependencies with it, and is meant to run in browser. The entire script is about 14K after gzip. It supports latest browsers, including Chrome, Safari, Firefox, Edge.

If you just want the core functional API, without the sub-modules, you can also use mark.core.js, which is only 7K after gzip. You can also refer to the package.json to create your own custom bundle with the sub-modules you need.

If you need to support legacy browsers, like IE11, which do not have proper ES6 support, you need to use /dist/mark.es5.js. IE < 11 are not supported.


Tools, Extensions and Applications of Mark

  • Mark Template: a JS template engine inspired by JSX and XSLT, using Mark for the template syntax.
  • Mark VSC Extension: Mark Notation support for Visual Studio Code.


Thanks to the following platforms or services that support the open source development of Mark: NPM, GitHub, Travis CI, Codecov, BrowserStack, jsDelivr, Plunker,