a lightweight static website generator

There are many static website generators out there, but most of them are quite opionated: they demand your posts be put in this directory and your posts ordered in that way, and they often include some tricky custom-defined formats.

Static websites are really quite simple. They're just a bunch of HTML files in a directory. A static website generator simply transforms text files into HTML files, and includes a few goodies like stylesheets and javascript modules.

embellish is a lightweight generator, that can build a website from any directory, transforming any text files it finds into HTML, allowing you to organically grow your website. It's also very small, and quite easy to modify.

As embellish is designed around the Python ecosystem, it carefully uses standard Python-like indented-text formats, that have been well-tested, namely:

Example Websites Built with Embellish

Quick & Dirty Websites

Other generators dictate your website to be structured their way. embellish does not.

Go to a directory with a markdown file (e.g. the embellish directory):

> embellish .


readme.html is generated and opened in a browser.

Now edit readme.txt, save, and refresh the browser.


The project is hosted at github.

Have node installed? Then:

  > npm install -g embellishjs

Or install manually using the zip file:

> npm link

All the following examples can be found in the examples directory of the zip file.

Blog with Categories

Any litmus test for a static website generator is to make a blog. Here's a quick guide using embellish where markdown is used for the text, YAML for metadata, and haml to write the HTML.

embellish will recognize files with extensions .txt, .md, .mkd and .markdown as post/article pages.

Article page

embellish expects a text page with an optional YAML front-matter defined by --- such as in post.mkd .

post.mkd - example of a Markdown Post
title: The Title of this Post
template: post.pug
# Great Post in Markdown

## Markdown Sub-headers!

This is the _body_ of the text 
written in markdown

The front-matter is optional. embellish reads the header as YAML metadata. There are a few fields used by embellish but you can add anything you want. This metadata will be made available to you in the template as a dictionary called page.

To generate an HTML page from this article post, a template is required. In this case, the template referred to in the header is post.pug. If no template is given, a default one is used that is located in the embllish directory under defaults/default.pug.

Templates and Style Sheets

Templates are used to generate HTML files from text files and their attendant header information.

embellish uses pug templates. A simple example is post.pug , which was referred in post.txt above.

post.pug - example of a simple HAML template to generate an article HTML
  link(href='styles.css' rel='stylesheet')
    | #{ page.title }
    | category: 
    | #{ page.category } ·
    | published: 
    | #{ page.date }#
  | !{ page.content }

For details on how templates work in embellish, see the Template Guide below.

Notice that there is a link to a style sheet styles.css . Now, you could simply creat your own styles.css and put it in the directory, but I strongly recommend you to write SASS style sheets such as styles.sass , and use sass to compile into styles.css. note in previous versions, this was done for you but it was rather fragile and difficult to debug. Now, you should just compile to css yourself.

style.sass - a SASS file to generate style.css
  background-color: #FFF
  font-family: Helvetica
  width: 450px
  margin: 40px auto
  text-transform: uppercase
  font-weight: 900
  font-size: 18px
  letter-spacing: 0.1em
  font-style: italic
  font-size: 12px

With the post, the template, and the style sheet done, we are now ready to make the blog:

> embellish .

Open post.html in your browser.

Index Pages

To really get the blog going, though, we will need to build an index page to list a bunch of posts. First, we create a text file that will generate an index of posts index.mkd .

index.mkd - example of a markdown file that generates a blog index file
template: index.pug
index: True
sortKey: date
sortReverse: True
maxSubpages: 5
title: Blog Posts
Some words to describe the following list of 
posts that are arranged in reverse chronological 

In the header of the post, the key flag is:

index: True

which tells embellish to send a list of posts into the template during compilation of the HTML file. Since in this case, no category was given, all the posts found by embellish will be used.

The posts will be sorted by date as indicated by:

sortKey: date

Any other field could have been used, but let's start with date. We'd also like to do it in reverse order as fits blogging convention:

sortReverse: True

And we don't want too many posts, so

maxSubpages: 5

Templating Index Pages

Okay now the text file is done, let's make the template for an index file index.pug .

index.pug - example of a HAML template that generates the HTML code for an index markdown file
  link(href='styles.css' rel='stylesheet')
    | #{ page.title }
  | !{ page.content }
    each subpage in page.subpages
          a(href= `${subpage.relSiteUrl}/${subpage.url}`)= subpage.title
        i= subpage.excerpt
          if subpage.category
            | category: #{ subpage.category }
          | date: #{subpage.date}

How it works is that during compilation, embellish will package all the information for the page into a dictionary page and send it into the template. As well another dictionary site will also be sent that includes the site configuration information. These will be described in detail below in the Templating Guide.

The header of the template is pretty similar to post.pug above, with the big difference in that index.pug can handle a special field called page.subpages.

page.subpages holds a list of all the pages found by embellish that belong to this index page. If a category was given, then only pages with the same category will be kept in the list. Here, since no category was given, all pages are given.

To handle the display of all the page.subpages, there is a pug control loop:

each subpage in page.subpages
    | #{ subpage.title }
    | #{ subpage.excerpt }

Note: it's important you use subpage as the looping variable, as page would otherwise clash with the main page dictionary.

You might have noticed that there is no pagination. I hate pagination, so I did not implement it. Just put all posts in one page.

Referring to URLs

It's important to get the URL's right to show up in the template.

First we need the site URL, and there are two ways to get it:

  1. The site's absolute URL, which is #{ site.url }
  2. The site's relative URL which depends on the location of the page in question #{page.relSiteUrl}

The URL of a page thus combines the site's URL with the page's URL.

If you want the absolute URL, you can construct the url using this interpolated string (notice the use of back-ticks):


For the relative URL of a page, which is more flexible, but more prone to breaking during deployment :


In index.pug , relative links are used for the subpage in the looping through the collected posts, so the link is:


Archiving With Categories

Now with blog posts, you might want to archive posts under different categories.

Let's say you have a post about vampires vampire.mkd . You give it a category in the header by:

category: vampire
vampire.mkd - a post with a category of vampire.
title: The Vampire Post
template: post.pug
category: vampire
# This is Vampire Post

## There's Vampire as Category!

This is also in __markdown__.
archive.mkd - a post with a category of vampire.
template: index.pug
index: True
sortKey: date
category: vampire
sortReverse: True
title: Archive of Posts about Vampires

Then, to collect all your vampire posts in an archive, you can create an index file archive.mkd , where the only difference from index.mkd , is that a category is given:

category: vampire

And then you run:

> embellish .

Which will generate an index post of all pages in index.html and an archive page for vampire posts archive.html.

Larger Websites

Configuration Files

Obviously for a more complicated website, you want to separate your content, templates, assets, and output into different directories.

To tell embellish where the directories are, you create a configuration file in the YAML format, such as site.yaml :

site.yaml a YAML configuration file for a larger website
url: http://boscoh.com # if  then use relative urls
contentDir: content  # look for markdown files
templateDir: templates  # look for templates
cachedPages: site.cache
outputDir: site  # generated files and static files put here
recursive: True # recursively searches through directories
mediaDir: media  # files to be copied directly into the output directory
writeExt: .html

To whit:

To run embellish against this configuration file:

embellish site.yaml

And the result will be in the outputDir.

Note: when using embellish with a directory as the argument, such as:

 > embellish .

Internally, embellish sets these fields in site - contentDir, templateDir, mediaDir, outputDir - to the current directory. embellish will skip the copying of files, and thus the website is made in place!

Recursive Directory Processing

Since embellish is meant to be a quick-and-dirty website generator, the default is to search only in the directory specified for content files.

This avoids accidentally telling embellish to search through a very large forest of subdirectories.

But you can tell embellish to exhaustively process all subdirectories with the -r option:

> embellish -r .

Templating guide

To write custom templates in jinja2-haml, you'll have to understand the data model embellish uses to compile templates.

During compilation, two dictionaries are passed into every template:

  1. site: contains all the settings for the entire site. This includes all the source directories, site url, and is essentially, the information in the site.yaml configuration file.

  2. page: contains all the information of a given page, including subpages when the page is an index.

Page Metadata

Every page dictionary comes equiped with a set of default metadata. The best way to show this is from the Python source code :

page - the python dictionary that is piped into every template
let defaultPage = {
  template: 'default.pug',  // name of template file
  filename: null,  // name of markdown file
  checksum: '', // checksum used to check final output to avoid redundant writes
  excerpt: '', // text to put in excerpt, no tags please!
  content: '',  // main text of article
  title: '',  // title for indexing and for large display
  category: '',  // category of article for indexing
  relSiteUrl: '',  // the top site directory relative to this page
  dateFormatString: null, // dateFormat formatting string
  date: null,  // published date
  slug: null,  // url-safe name of article used to make url and files
  url: '',   // relative url used for links in index files
  target: '',    // target filename, maybe different to url due to redirection
  index: false,   // indicates if this is an indexing page
  sortKey: null,  // the field on which to sort under in indexing
  sortReverse: true,  // ascending or descing order for sorting
  subpages: [],   // in indexing, pages belonging to the index placed here
  maxSubpages: null,  // a maximum limit of files to put in subpages

This relevant fields in the page dictionary is then overriden by the YAML header of the corresponding source file.

In a pug template, these fields are accessible in the form of:

#{ page.title }

In particular, dates are converted to standard Python datetime objects, and can be passed into the jinja templates by calling the datetime string method:

#{ page.date }

If excerpts are not given, page.excerpt is set to the first 50 non-tag words in the main text.

Filenames and Directory Structure

Two design principles of embellish is that it lets you determine the URL and file placement as much as possible, and that relative URLs should work. No matter what, page.target will contain the filename of the target HTML file, which is written in the site.outputDir directory. If page.url or page.target are specified, these will be used.

Most of the time page.url will match page.target. If page.url is not specified, then it is assumed that the path of the markdown file with respect to the site.contentDir represents the page.url and the page.target. In flat-file mode, this means the .html file will appear in the same directory as the markdown file.

A common exception to this are index.html files in subdirectories. Because browsers default to dir/ for dir/index.html, a useful motif in the header for index.html files is:

url: archive/
target: archive/index.html

Two other fields need to be mentioned. Sometimes legitimate filenames cannot serve as URLs, so a conversion is done to turn the basename of the filename into a page.slug. This can be directly overriden if the page.slug field is given.

Finally, the extension of the output files are normally assumed to be .html but this can be overriden in site.ext in the configuration file.

Absolute & Relative URLs

To reiterate, the site's URL is given in:

  1. an absolute form #{site.url}
  2. a relative form, relative to the page #{page.relSiteurl}

Where for instance if embellish was run in the current directory, and so is the file, then:

page.relSiteurl = ''

The page's URL is thus a combination of the site URL and the page's relative URL:

  1. absolute: `${site.url }/${ page.url}`
  2. relative `${page.relSiteurl}/${ page.url}`