CoffeeDoc

An API documentation generator for CoffeeScript

CoffeeDoc is a simple API documentation generator for CoffeeScript. It reads python-style docstrings in your CoffeeScript class and function definitions, passes them through Markdown and outputs the result as easy to read HTML.

Thanks to apgwoz, CoffeeDoc can also generate wiki pages for Github!

CoffeeDoc can also return your documentation as JSON, so you can run it through an external documentation system such as Sphinx.

CoffeeDoc is inspired by the excellent Docco, and is intended for projects that require more structured API documentation.

The docstring convention CoffeeDoc uses is inspired by Python, and looks like this:

###
# CoffeeDoc example documentation #

This is a module-level docstring, and will be displayed at the top of the module documentation.
Documentation generated by [CoffeeDoc](https://github.com/omarkhan/coffeedoc)
###

class MyClass extends Superclass
    ###
    This docstring documents MyClass. It can include *Markdown* syntax,
    which will be converted to html.
    ###
    constructor: (@args) ->
        ### Constructor documentation goes here. ###

    method: (args) ->
        ### This is a method of MyClass ###

myFunc = (arg1, arg2, args...) ->
    ###
    This function will be documented by CoffeeDoc
    ###
    doSomething()

The documentation generated from the above script can be seen here. For a more interesting example, here is the result of running coffeedoc against src/coffeedoc.coffee.

Installation

CoffeeDoc requires Node.js, CoffeeScript, eco, and optimist. Install using npm with the following command:

sudo npm install -g coffeedoc

The -g option installs CoffeeDoc globally, adding the coffeedoc executable to your PATH. If you would rather install locally, omit the -g option.

You can also install from source using cake. From the source directory, run:

sudo cake install

Usage

CoffeeDoc can be run from the command line:

Usage: coffeedoc [options] [targets]

Options:
  --output, -o      Set output directory                                            [default: "docs"]
  --ignore, -i      Files or directories to ignore
  --stdout          Direct all output to stdout instead of files
  --hide-private    Do not document methods beginning with an underscore
  --parser          Parser to use. Built-in parsers: commonjs, requirejs            [default: "commonjs"]
  --renderer        Renderer to use. Built-in renderers: html, gfm, json            [default: "html"]
  --indexTemplate   Override the default index template for the selected renderer
  --moduleTemplate  Override the default module template for the selected renderer
  --help, -h        Show this help and exit

If [targets] is a directory, CoffeeDoc will recursively document all .coffee files found under that directory.

If you wish to document several modules, make sure you generate all the docs with a single command -- this ensures that they will all appear in the index.html file.

Note on Markdown headers

Markdown uses # characters for headers, e.g.

# Header 1
## Header 2
### Header 3
#### Header 4
##### Header 5
###### Header 6

As using a sequence of 3 or more # characters within a CoffeeScript block comment would end the comment block, CoffeeDoc allows for the \# escape sequence in docstrings. So instead of ### Header, use \#\#\# Header or ##\# Header. Ugly, but it works.

How it works

CoffeeDoc uses the CoffeeScript parser to generate a parse tree for the given source files. It then extracts the relevant information from the parse tree: class and function names, class member functions, function argument lists and docstrings.

Docstrings are defined as the first herecomment block following the class or function definition. Note that regular single line comments will be ignored.

The resulting documentation information is then passed to an eco template to generate the html output.

TODO

• Doctests

Alternatives

• Docco for literate programming style docs.
• Codo for something more ruby than python.

Licence

CoffeeDoc is © 2012 Omar Khan, released under the MIT licence. Use it, fork it.