Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Simple, blog aware infrastructure to generate static sites
JavaScript CSS

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.

blake - generate site

The blake Node.js module provides a simple, blog aware infrastructure to generate static sites. For unrestricted choice of input formats and template languages, blake confines itself to IO and template routing; it delegates artifact generation to user-written functions.

Build Status David DM

CLI Usage

blake source_directory target_directory
blake source_directory target_directory source_file ...

In the first form, blake writes all files generated from input data in the source_directory to the target_directory. In the second synopsis form, output is generated from the specified source files only.

Library Usage

Generate from directory

var blake = require('blake')
  , source = 'blake-site'
  , target = '/tmp/blake-site'
  , join = require('path').join
  , Reader = require('fstream').Reader
  , props = { path:join(source, 'data') }
  , cop = require('cop')

new Reader(props)
  .pipe(blake(source, target))
  .pipe(cop(function (filename) { return filename + '\n' }))

Copy static resources and generate from directory

var blake = require('blake')
  , join = require('path').join
  , source = join(process.cwd(), './blake-site')
  , target = '/tmp/blake-site'
  , Reader = require('fstream').Reader
  , props = { path:join(source, 'data') }
  , cop = require('cop')
  , copy = require('../lib/copy.js')

copy(join(source, 'resources'), target)
  .on('error', console.error)
  .on('end', function () {
    new Reader(props)
      .pipe(blake(source, target))
      .pipe(cop(function (filename) { return filename + '\n' }))

Generate from files

var blake = require('blake')
  , cop = require('cop')
  , readArray = require('event-stream').readArray
  , filenames = ['first/file', 'second/file', 'third/file']
  , source = 'source_directory'
  , target = 'target_directory'

  .pipe(blake(source, target))
  .pipe(cop(function (filename) { return filename + '\n' }))

Generate and push to S3

Since the blake function returns a Transform stream that emits the paths of the generated artifacts, we can pipe to pushup, and upload the files directly to S3.

var resolve = require('path').resolve
  , cop = require('cop')
  , getProps = require('pushup/lib/getProps')
  , blake = require('blake')
  , pushup = require('pushup')
  , Reader = require('fstream').Reader
  , sep = require('path').sep
  , source = 'source_directory'
  , target = '/tmp/target_directory'
  , reader = new Reader({ path:resolve(source, 'data') })
  , props = getProps()


  .pipe(blake(source, target))

function adjustPath (p) {
  return p.split(sep).slice(3).join(sep)


blake(source, target)

The blake module exports a single function that returns a Transform stream. While writing source filenames to it, you can read target filenames (of written arfifacts) from it.

  • source The source directory.
  • target The target directory.


blake requires a configuration module at source_directory/config.js, which exports paths, and views (a map of generator functions):

exports.paths = {
  data: 'data' 
, templates: 'templates'
, resources: 'resources'
, posts: 'data/posts'

exports.views = {
  'rss.jade': require('./rss.js')
, 'article.jade': require('./article.js')
, 'home.jade': require('./home.js')
, 'about.jade': require('./about.js')
, 'archive.jade': require('./archive.js')

The paths object defines input paths, with two required directories: data and templates. From data blake loads general input data, templates is the directory for templates. The two optional directories are resources and posts. The content of resources is copied to the target_directory' unchanged. Theposts` directory hosts blog posts.

The views object is a map of user-written functions that implement the actual generation of output artifacts. Here, these functions are mapped by template name.


At the top of each input file blake expects a JSON string that is interpreted as header providing transformation parameters. Besides it can contain additional user defined data—the item parameter, passed to the view functions, provides a reference to the raw header. Input data for a blog entry could look like so:

  "title": "Example",
  "description": "An example article",
  "template": "article.jade",
  "date": "2012-03-21"

Your highness, when I said that you are like a stream of bat's piss, 
I only mean that you shine out like a shaft of gold when all around 
it is dark.

The end of the header is marked by an empty line. Everything that follows is interpreted as content and is passed to the views untouched.


JSON at the top of an input file:

  "title": "Example",
  "description": "An example article",
  "template": "article.jade",
  "date": "2012-03-21",
  "path": "2012/03",
  "name": "example"
  • title is the title of the page (optional)
  • description is the description of the page or rather the post (optional)
  • templateis the filename of the template to use (required)
  • date is the publish date, if not provided it's set to NOW (optional)
  • path is the output path, if not provided the path of the input file is used (optional)
  • name is used as filename of the output file, if not provided the filename of the input file is used (optional)

The source object, passed to the views, provides a reference to the raw header object. Thus, the header is extendable with arbritrary fields, which can be interpreted by the generators (written by you).

If you decide to mirror the input paths in your output, you can omit path and name. In that case a typical header of a blog post might look like the following:

  "title": "Example",
  "description": "An example article",
  "template": "article.jade",
  "date": "2012-03-21",

Input data with this header, located at source_directory/data/posts/2012/03/, would produce 2012/03/article.html.

An input file can consist of just a header (without content) to generate, for example, an RSS feed.

  "title": "Blog",
  "description": "Stuff I write",
  "link": "",
  "template": "rss.jade",
  "name": "rss.xml"


The views—alternative naming would be: transformers, generators, or bakers—are the functions that generate your artifacts; they have the following signature:

function (item, callback)

The passed in 'item' provides the input data to generate the artifact (or most likely the page).

Here, for example, an item representing a blog post:

{ header: 
  { title: 'Static Websites',
    description: '...',
    template: 'article.jade',
    data: Thu May 17 2012 02:00:00 GMT +0200 (CEST),
    path: '2012/05',
    name: 'static-websites.html' }
  body: '...', 
  { target: '/tmp/michaelnisi-site',
    resources: '/Users/michael/workspace/michaelnisi/resources',
    data: '/Users/michael/workspace/michaelnisi/data',
    templates: '/Users/michael/workspace/michaelnisi/templates',
  posts: '/Users/michael/workspace/michaelnisi/data/posts' },
  title: 'Static Websites',
  name: 'static-websites.html',
  date: Thu May 17 2012 02:00:00 GMT+0200 (CEST),
  templatePath: '/Users/michael/workspace/michaelnisi/templates/article.jade',
  path: '/tmp/michaelnisi-site/2012/05/static-websites.html',
  link: '2012/05/static-websites',
  dateString: 'Thu May 17 2012',
  bake: [Function],
  template: <Buffer 0a 20 20 20 20 64 69 76 ...> }

To see a simple example:

git clone git:// 
cd blake/example
npm install
node generate.js
open /tmp/blake-site/index.html

To evaluate a more elaborate example, you might generate my blog, for which I use Jade and Markdown:

npm install -g blake
git clone git://
cd troubled
npm install
blake . /tmp/troubled-site


Of course you can build your site locally, and upload it to your webserver manually; but I recommend to run blake on a server, using post-receive hooks to automatically generate your site, post to each push your input data repository receives.



To use the command-line interface:

npm install -g blake


MIT License

Something went wrong with that request. Please try again.