Jerkll: A tiny static site generator that runs in your browser

While playing with RequireJS for another project, I was impressed to see that browsers can recursively fetch files at load time, and still feel fast. This made me think:

Could I maintain my website as simple Markdown and template files and have the HTML generation done at request time in the browser?

The idea seemed novel to me, so I decided to implement it. It ended up being surprisingly trivial, so I code golfed it and reduced it to the following 14 lines:

(onhashchange = function(e, d) {
    var p = location.hash ? location.hash.slice(1) : 'index';
    d = d || {'template': 'pages/' + p + ".md"};
    var r = new XMLHttpRequest(), f = /^---\n((.|\n)*)\n---/;
    r.onload = function() {
        d.content = marked(r.response.replace(f, function(f, j) {
            for(i in s = JSON.parse(j)){ d[i] = d[i] || s[i]; };
            d.template = s.template;
            return '';
        }).replace(/{{ (\S*) }}/g, function(m, w) { return d[w] || m; }));
        d.template ? onhashchange(e, d) : document.body.innerHTML = d.content;
    };
    r.open("get", d.template);
    r.send();
})();

This code, together with the marked.js Markdown library, reimplements the major functionality of a static site generator like Jekyll without any need for server side software.

When you navigate to a new page it recursively fetches the template files and replaces the DOM with the generated HTML.

In Action

You can see a demo here or find the code on GitHub.

Why?

This approach has a couple interesting advantages over the standard static site generator model:

1. Lower barrier to entry. Everybody has a browser. Not everybody has Ruby/Node/Python/PHP and the associated package manager.

2. A more elaborate version could build the entire site in the browser and offer you a .zip file of the generated HTML files which you would deploy to your server. (I actually have a version of this working).

3. Templating is done at request time, so results are always up to date.

The Catch

There are a few limitations that make this approach not terribly feasible for a real project:

1. It can't discover files, so you can't have a page that lists all your blog posts without maintaining all the links yourself.

2. While it runs entirely in your browser, The Cross Origin rule means you need an actual web server to use it. This could be as simple as: python -m SimpleHTTPServer, but you still need to use the command line.