magnars.stasis

https://github.com/magnars/stasis.git

git clone 'https://github.com/magnars/stasis.git'

(ql:quickload :magnars.stasis)
★244

Stasis Build Status

A Clojure library of tools for developing static web sites.

Breaking change in 2.0

Stasis exports paths without a file extension as directories with an index.html file. Most web servers will respond to the slash-less request with a redirect to the URL including a slash. This redirect is entirely avoidable by just linking to the right URL in the first place.

This change should help you avoid these needless redirects, increasing the speed of your site further.

Install

Add [stasis "2.5.0"] to :dependencies in your project.clj.

Please note that this project uses Semantic Versioning. There will be no breaking changes without a major version increase. There's also a change log.

Another static site framework? Why?

Well, that's exactly it. I didn't want to use a framework. I don't like the restrained feeling I get when using them. I prefer coding things over messing around with configuration files.

I want to

Statis is a collection of functions that are useful when creating static web sites.

No more. There are no batteries included.

If you want a framework that makes it really quick and easy to create a blog, you should take a look at these:

They generally come with a folder where you put your blog posts in some templating language, and a set of configuration options about how to set up your blog. They often generate code for you to tweak.

Usage

The core of Stasis is two functions: serve-pages and export-pages. Both take a map from path to contents:

(def pages {"/index.html" "<h1>Welcome!</h1>"})

The basic use case is to serve these live on a local server while developing - and then exporting them as static pages to deploy on some server.

Serving live pages locally

Stasis can create a Ring handler to serve your pages.

(ns example
  (:require [stasis.core :as stasis]))

(def app (stasis/serve-pages pages))

Add Ring as a dependecy and Lein-Ring as a plugin, and point Ring to your app in project.clj.

:ring {:handler example/app}

and start it with lein ring server-headless.

Exporting the pages

To export, just give Stasis some pages and a target directory:

(defn export []
  (stasis/empty-directory! target-dir)
  (stasis/export-pages pages target-dir))

In this example we're also emptying the target-dir first, to ensure old pages are removed.

When you've got this function, you can create an alias for leiningen:

:aliases {"build-site" ["run" "-m" "example/export"]}

and run lein build-site on the command line. No need for a lein plugin.

Livelier live pages

Let's say you want to dynamically determine the set of pages - maybe based on files in a folder. You'll want those to show up without restarting.

To be fully live, instead pass serve-pages a get-pages function:

(defn get-pages []
  (merge {"/index.html" "<h1>Welcome!</h1>"}
         (get-product-pages)
         (get-people-pages)))

(def app (stasis/serve-pages get-pages))

Do I have to build every single page for each request?

No. That's potentially quite a lot of parsing for a large site.

(def pages
  {"/index.html" (fn [context] (str "<h1>Welcome to " (:uri context) "!</h1>"))})

Since we're dynamically building everything for each request, having a function around the contents means you don't have to build out the entire site every time.

Stasis passes a context to each page. When it's served live as a Ring app the context is actually the Ring request, and as such contains the given :uri. Stasis' export-pages makes sure to add :uri to the context too.

You can also pass in configuration options that are included on the context:

(defn my-config {:some "options"})

(def app (stasis/serve-pages get-pages my-config))

(stasis/export-pages pages target-dir my-config)

These are then available when rendering your page.

Finally, some Ring middlewares put values on the request to be used in rendering. This supports that. Read on:

But what about stylesheets, images and javascript?

Yeah, Stasis doesn't really concern itself with that, since it doesn't have to.

In its simplest form, you can add some JavaScript and CSS to the map of pages. It'll be served and exported just fine. Which is good if you want to dynamically create some JSON, for instance.

But for your CSS, JavaScript and images, I recommend a frontend optimization library. You can use any asset lib that hooks into Ring and lets you export the optimized assets to disk.

I use Optimus. To get you started, here's an example:

(ns example
  (:require [ring.middleware.content-type :refer [wrap-content-type]]
            [stasis.core :as stasis]
            [optimus.prime :as optimus]
            [optimus.assets :as assets]
            [optimus.optimizations :as optimizations]
            [optimus.strategies :refer [serve-live-assets]]
            [optimus.export]
            [example.app :refer [get-pages target-dir]]))

(defn get-assets []
  (assets/load-assets "public" ["/styles/all.css"
                                #"/photos/.*\.jpg"]))

(def app (-> (stasis/serve-pages get-pages)
             (optimus/wrap get-assets optimizations/all serve-live-assets)
             wrap-content-type))

(defn export []
  (let [assets (optimizations/all (get-assets) {})
        pages (get-pages)]
    (stasis/empty-directory! target-dir)
    (optimus.export/save-assets assets target-dir)
    (stasis/export-pages pages target-dir {:optimus-assets assets})))

I create a function to get all the assets, and then add the Optimus Ring middleware to my app. I want to serve assets live, but still have them optimized - this lets my dev environment be as similar to prod as possible.

Then I simply tell Optimus to export its assets into the same target dir as Stasis.

Notice that I add :optimus-assets to the config map passed to stasis/export-pages, which will then be available on the context map passed to each page-generating function. This mirrors what optimus/wrap does on the live Ring server, and allows for linking to assets by their original path.

That's all the detail I'll go into here, but you can read more about all the ways Optimus helps you with frontend performance optimization in its extensive README.

So, what else does Stasis have to offer?

This is about everything you need to start building static sites. But Stasis does come with a few more tools.

slurp-directory

You'll probably create a folder to hold a list of pages, posts, products or people at some point. Read them all in with slurp-directory.

(def articles (slurp-directory "resources/articles/" #"\.md$"))

This gives us a map {"/relative/path.md" "file contents"}. The relative path can be useful if we're creating URLs based on file names.

Here's another example:

(def products (->> (slurp-directory "resources/products/" #"\.edn$")
                   (vals)
                   (map read-string)))

This matches all edn-files in resources/products/, slurps in their contents and transforms it to a list of Clojure data structures.

Fun fact: This is a valid code:

(def app (serve-pages (slurp-directory "resources/pages/" #"\.html$")))

It would serve all .html files in that folder, matching the URL structure to files on disk.

Like slurp, slurp-directory can also receive optional arguments such as :encoding:

(slurp-directory "resources/articles/" #"\.md$" :encoding "UTF-8")

slurp-resources

Just like slurp-directory, except it reads off the class path instead of directly from disk. For performance reasons the .m2 folder is excluded. Open an issue if that causes you pain.

merge-page-sources

You might have several sources for pages that need to be merged into the final pages map. Wouldn't it be nice if someone told you about conflicting URLs?

(defn create-pages [content]
  (merge-page-sources
   {:person-pages (create-person-pages (:people content))
    :article-pages (create-article-pages (:articles content))
    :general-pages (create-general-pages content)}))

(defn get-pages []
  (create-pages (load-content)))

So merge-page-sources takes a map. The values are the page-maps to merge. The keys in the map are only used for error reporting:

URL conflicts between :article-pages and :general-pages: #{"/about.html"}

report-differences

This entirely side-effecty function takes an old and a new map of strings, compares them and reports the differences to standard out.

Here's a code example from one of my projects:

(def export-dir "./dist")

(defn- load-export-dir []
  (stasis/slurp-directory export-dir #"\.[^.]+$"))

(defn export
  "Export the entire site as flat files to the export-dir."
  []
  (let [old-files (load-export-dir)] ;; 1.
    (stasis/empty-directory! export-dir)
    (stasis/export-pages (get-pages) export-dir)
    (println)
    (println "Export complete:")
    (stasis/report-differences old-files (load-export-dir)) ;; 2.
    (println)))

  1. We slurp the old directory into memory before emptying it

  2. After we're done exporting, pass in old and new files to print the report.

This prints something along these lines (in glorious ansi color):

Export complete:
- 260 unchanged files.
- 2 changed files:
    - /index.html
    - /troubleshooting/index.html
- 1 removed file:
    - /verified-hash/index.html
- 1 added file:
    - /verified-hashes/index.html

Dependent pages

Instead of returning a string of html for your pages, you may return a map:

{:contents "the page contents",
 :dependent-pages {"/uri" "dependent page contents"}}

These dependent pages are then also served and exported.

Why would I need that?

Most often, you don't.

However: When creating a page, you may find that you want to extract pieces of the page into yet new pages. Say you want to do some optimizations and extract inline JavaScript from pages and serve them as cacheable separate files.

Dependent pages gives you the opportunity to do this without cluttering up your page-creation logic. When the page containing JavaScript is loaded, you may alter the generated page to use a script tag that links to a separate file, and then return a page map where :dependent-pages contains the page (of JavaScript) that contains the extracted inline source.

Q & A

Can I avoid the .html endings on my pages?

Yes. Stasis will handle URLs like /projects/clojure/ by creating projects and clojure folders, and placing an index.html in it.

Beware of the implications tho. You suddenly have multiple valid URLs to the same page, which is not good for your standing with Google. There's also the case of /projects/clojure vs /projects/clojure/ that might trip you up.

It'll probably take some wrangling of your web server to get this in pristine shape.

Are there any full fledged examples to look at?

Christian Johansen wrote Building static sites in Clojure with Stasis, which is an excellent starting point.

If you just want some code to look at, check these out:

Uses Enlive for templating, and Optimus for frontend optimization.

Uses a combination of Hiccup, Enlive and Markdown (via Cegdown) to build the pages, Optimus for frontend optimization, and Pygments (via Clygments) for code highlighting.

Uses a home brewed HTML based templating system, Pygments (via Clygments) for syntax highlighting, and Optimus for the asset pipeline.

Parses markdown table syntax to create metadata and tags for its blog posts. Clever! Uses many of the same technologies listed above.

Uses Selmer and Hiccup for building the pages, and Optimus for building assets.

Parses Markdown as well as hiccup defined in edn files. Edn pages and posts can define rendering functions that are aware of the entire site for maximum flexibility.

Got an open source site written in Stasis? Do let me know, and I'll add it here!

Why won't my Enlive templates update when I edit them?

Your template definitions are reloaded only when you touch the code in that namespace. Ring is unaware of the dependency on the template files.

There are some tricks in this thread. I prefer to place my templates in a separate namespace, and do this:

(defn reload-templates []
  (require 'example.templates :reload))

And then call that in my get-pages function.

How do I create an RSS feed for my blog?

No worries, it's just a bit of XML generation. Here's a working snippet from whattheemacsd.com to create an Atom feed:

(ns what-the-emacsd.rss
  (:require [clojure.data.xml :as xml]))

(defn- entry [post]
  [:entry
   [:title (:name post)]
   [:updated (:date post)]
   [:author [:name "Magnar Sveen"]]
   [:link {:href (str "http://whattheemacsd.com" (:path post))}]
   [:id (str "urn:whattheemacsd-com:feed:post:" (:name post))]
   [:content {:type "html"} (:content post)]])

(defn atom-xml [posts]
  (xml/emit-str
   (xml/sexp-as-element
    [:feed {:xmlns "http://www.w3.org/2005/Atom"}
     [:id "urn:whattheemacsd-com:feed"]
     [:updated (-> posts first :date)]
     [:title {:type "text"} "What the .emacs.d!?"]
     [:link {:rel "self" :href "http://whattheemacsd.com/atom.xml"}]
     (map entry posts)])))

If this seems like too much, well, maybe you're using the wrong static site library. But anyway, there's even a library to create RSS for you here: clj-rss.

Is Stasis stable?

Yes. This project uses Semantic Versioning. There will be no breaking changes without a major version increase. And we've got plenty of tests to keep us in line.

Again, why not use one of the existing frameworks?

I think the existing frameworks are great if they fit your style. Stasis imposes no styles. There are very few decisions made for you - no markdown vs asciidoc, no enlive vs hiccup. No configuration options. You have to - and get to - make those yourself.

So, yeah … I think Stasis would be a great starting point if you want to create the 6th static site framework to go in that list at the top. :-)

Change log

From 2.4 to 2.5

From 2.3 to 2.4

From 2.2 to 2.3

From 2.1 to 2.2

From 2.0 to 2.1

From 1.1 to 2.0

Stasis exports paths without a file extension as directories with an index.html file. Most web servers will respond to the slash-less request with a redirect to the URL including a slash. This redirect is entirely avoidable by just linking to the right URL in the first place.

This change should help you avoid these needless redirects, increasing the speed of your site further.

From 1.0 to 1.1

Contributors

Thanks!

Contribute

Yes, please do. And add tests for your feature or fix, or I'll certainly break it later.

Running the tests

lein midje will run all tests.

lein midje namespace.* will run only tests beginning with “namespace.”.

lein midje :autotest will run all the tests indefinitely. It sets up a watcher on the code files. If they change, only the relevant tests will be run again.

License

Copyright © 2014 Magnar Sveen

Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.