Skip to content
WLang is a general-purpose (code generation / templating) engine.
Find file
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Build Status

WLang is a powerful code generation and templating engine, implemented on top of temple and much inspired by the excellent mustache.

WLang2 is still work in progress.



  • Tunable templating engine. You can define your own tags, and their behavior.
  • Powerful logic-less HTML concretization to render web pages à la Mustache with extra.
  • Compiled templates for speedy generation.
  • Tilt and Sinatra integration.

WLang 2.0 also has a few remaining issues.

  • It does not support rubinius so far, due to an incompatibility with the Citrus parser generator.
  • It also have some issues with spacing; not a big issue for HTML rendering but might prevent certain generation tasks.

Tunable templating engine

WLang is a templating engine, written in ruby. In that, it is similar to ERB, Mustache and the like:

WLang::Html.render 'Hello to ${who}!', who: 'you & the world'
# => "Hello you & the world!"

To output HTML pages, WLang does not provides you with killer features or extraordinary shortcus. It supports escaping, as shown above, but many other templating engines do. For such HTML tasks, WLang does a pretty good job but many other engines perform faster and have nicer features. See the examples folder that documents WLang::Html.

WLang is designed to help you for other uses cases, user-defined ones in particular, such as generating code or whatever text generation task for which other engines quickly become inappropriate. WLang helps there because it allows you to create your own dialect, that is, you can define your own tags and their behavior. For instance,

class Highlighter < WLang::Dialect

  def highlight(buf, fn)
    var_name  = render(fn)
    var_value = evaluate(var_name)
    buf << var_value.to_s.upcase

  tag '$', :highlight
Highlighter.render('Hello ${who}!'), who: 'you & the world'
# => "Hello YOU & THE WORLD !"

WLang already provides a few useful dialects, such as WLang::Html (inspired by Mustache but a bit more powerful in my opinion). If they don't match your needs, it is up to you to define you own dialect for making your generation task easy. Have a look at the implementation of WLang's ones, it's pretty simple to get started!

Abstract semantics

WLang has a powerful semantics in terms of concatenation of strings and high-order functions (i.e. functions that take other functions as parameters). Let take the following template as an example:

Hello ${who} !

The functional semantics of this template is as follows:

(fn (concat "Hello", ($ (fn "who")), " !"))

That is, the compilation of this template yields a function that concatenates the string "Hello" with the result of the higher-order function ($ ) (that itself takes another function as a parameter, corresponding to the sub-template in its brackets delimited blocks) and then the string " !". Providing a concrete semantics to those high-order functions yields so called WLang dialects, as we've seen before.

Having a well-defined semantics allows wlang to properly compile your user-defined dialect and its instantiation engine so as to preserve decent performances. The WLang architecture is a typical compiler chain. This means that, provided some additional coding, you could even define your own language/syntax and reuse the compilation mechanism, provided that you preserve the semantics above.

Higher-order constructs

A feature that distinguishes WLang from most templating engines is the fact that higher-level constructions are permitted. In addition to tag functions that accept multiple arguments, thus multiple blocks in the source text, those blocks may be complex templates themselves.

For instance, the following behavior is perfectly implementable:

HighLevel.render 'Hello *{ ${collection} }{ ${self} }{ and } !',
                 collection: 'whos', whos: [ "you", "wlang", "world" ]
# => "Hello you and wlang and world"

An implementation of HighLevel might be as follows:

class HighLevel < WLang::Dialect

  def join(buf, expr, main, between)
    evaluate(expr).each_with_index do |val,i|
      buf << render(between, val) unless i==0
      buf << render(main, val).strip

  def print(buf, fn)
    buf << evaluate(fn).to_s

  tag '*', :join
  tag '$', :print

Use at your own risk, though, as it might lead to dialects that are difficult to understand and/or use and present serious injections risks! Otherwise, higher-order constructions provides you with very powerful tools.

Tilt integration

WLang 2.0 has built-in support for Tilt facade to templating engines. In order to use that API:

require 'tilt'         # needed in your bundle, not a wlang dependency
require 'wlang'        # loads Tilt support provided Tilt has already been required

template ="path/to/a/template.wlang")   # suppose 'Hello ${who}!'
template.render(:who => "world")
# => Hello world!

template =, :dialect => Highlighter)
template.render(:who => "world")
# => Hello WORLD!

Please note that you should require tilt first, then wlang. Otherwise, you'll have to require wlang/tilt explicitely.

Something went wrong with that request. Please try again.