node package manager


Internal DSL for generating HTML


An internal DSL for generating HTML in JavaScript.


Basic elements

html.div('#main', [
    html.h1(null, 'Hello, world!'),
    html.img({src: 'foo.jpg'})
<div id="main">
    <h1>Hello, world!</h1>
    <img src="foo.jpg"/>

Loops etc.

Using Underscore.js or similar:

function todoItem(item) {
    return{rel:}, [
        html.div('.title', item.title),
        html.button('.destroy', 'delete')
function todoList(list) {
    return html.ul('.todo-list',, todoItem));
    {id: 1, title: 'item one'},
    {id: 2, title: 'item two'},
    {id: 3, title: 'item three'}
<ul class="todo-list">
    <li rel="1">
        <div class="title">item one</div>
        <button class="destroy">delete</button>
    <li rel="2">
        <div class="title">item two</div>
        <button class="destroy">delete</button>
    <li rel="3">
        <div class="title">item three</div>
        <button class="destroy">delete</button>

Why use an internal DSL?

  • It's a more convenient and safer alternative to string contatenation
  • Very flexible, you can use all the power of JavaScript functions and control structures
  • For small bits of HTML you might not want to switch contexts from code to a template
  • Easier to debug than a templating engine
  • You get full tool-chain support:
    • editor support: syntax highlighting, code tools etc etc
    • code analyzers: jslint, jshint
    • testing/coverage tools

When to use?

  • Consider using where you might currently use string concatenation
  • Avoid using for large HTML documents or in places where speed is critical
  • Good for small snippets used for client-side page updates
  • Bad for generating huge amounts of HTML on the server


I like to alias the 'pithy' library as 'html':

var html = require('pithy');

You can then just use html.tagname as a function to create the appropriate element. Please note, you actually get a html.SafeString object back, not a native JavaScript String. This might mess up your isString() tests. If you have a workaround please send a pull-request.

There is also a html.escape() function for escaping HTML (returns a html.SafeString). It will not escape a value that is already a html.SafeString object.