This repo is a step by step tutorial on setting up a static site with Powerpack, a static site toolkit for Clojure.
The code in this repository represents the result of following the full tutorial. You can check out the commit for each step to follow along.
You might wonder why there is no generator that can generate a new project for you. I personally don't like wading through a bunch of generated code, and I think generators can lead to poorer API design, and have deliberately chosen to not make one. Instead, Powerpack aims to get up and running with minimal code. Punching it in manually will cost you next to nothing in the grand scheme of things, and it will give you a better understanding of your app.
Create a new project:
mkdir powerblog
cd powerblog
mkdir -p src/powerblog
mkdir -p dev/powerblog
mkdir content
mkdir -p resources/publicAdd deps.edn to the root with the following content:
{:paths ["src" "resources"]
:deps {org.clojure/clojure {:mvn/version "1.11.1"}
no.cjohansen/powerpack {:mvn/version "2023.12.21"}}
:aliases {:dev {:extra-paths ["dev"]}}}Powerpack has reasonable defaults for most things, so we'll start with a minimal configuration and build it out as we go.
Add this to src/powerblog/core.clj:
(ns powerblog.core)
(defn render-page [context page]
"<h1>Hello world</h1>")
(def config
{:site/title "The Powerblog"
:powerpack/render-page #'render-page})Add an empty Datomic schema to the default location, resources/schema.edn:
[]If you need, you can change the location of the schema with the
:datomic/schema-file config option:
(def config
{:datomic/schema-file "resources/schema.edn"
:site/title "The Powerblog"
:powerpack/render-page #'render-page})We will need a namespace to run the site in development. Add the following to
dev/powerblog/dev.clj:
(ns powerblog.dev
(:require [powerblog.core :as blog]
[powerpack.dev :as dev]))
(defmethod dev/configure! :default []
blog/config) ;; 1
(comment
(dev/start) ;; 2
(dev/stop) ;; 3
(dev/reset) ;; 4
(dev/get-app) ;; 5
)So, what's all this?
- This is how Powerpack gets a hold of your configuration in development.
- Evaluate this form to start the site
- Evaluate this form to stop the site
- Evaluate this form to reload all your code, rebuild the database and start the site.
- Evaluate this form to grab a copy of the app instance. You can inspect it to find what configuration is being used, etc. You do not need this, it's just for the curious.
Before we start up the app, I recommend a small convenience that ensures that
your code is loaded automatically when the REPL starts up. Add the following to
dev/user.clj:
(ns user
(:require powerblog.dev))If you're planning to run REPLs from Emacs, which I warmly recommend, you will
also want to add the following to .dir-locals.el at the root of the project:
((nil
(cider-clojure-cli-aliases . "-A:dev")
(cider-preferred-build-tool . clojure-cli)))This makes sure that CIDER includes the :dev alias when starting your REPL.
Powerpack doesn't come with content, so we'll have to add at least one page in
order to have something to test. Add the following to content/test.md:
# Hello world!
Hello there, thanks for stopping by.Powerpack will read all files in content (or whatever you set
:powerpack/content-dir to) into Datomic. It understands markdown and EDN out
of the box, and you can teach it how to parse other file formats as well.
By default, markdown files will be considered a page, and will receive a
:page/uri that corresponds to its relative path under
:powerpack/content-dir. In other words, content/test.md will be available as
/test/.
Start a REPL, and evaluate (dev/start) from the dev namespace. With the
default configuration, you should see a message about Powerpack being available
on http://localhost:5050/. The frontpage will greet
you with a 404 for now, but your test page is available at
http://localhost:5050/test/.
You probably noticed that the test page did not render the markdown content.
That's because our rendering function in powerblog.core looks like this:
(defn render-page [context page]
"<h1>Hello world</h1>")This function receives two parameters, a context, and a page. The context
contains :uri, just like a Ring request, a Datomic database value in :app/db
and :powerpack/app, your full Powerpack app. It is also possible to add
additional keys to the context per request, like secondary data sources, custom
configuration and whatever. We'll get back to that.
The more interesting parameter is the page, it is an entity map from Datomic.
It will contain whatever information you have added to the database under this
specific URL. Since we haven't added our own schema yet, it only contains the
URL and the contents of the markdown file:
{:page/uri "/test/",
:page/body "# Hello world!\n\nHello there, thanks for stopping by.\n"}Let's update the test page to render this as Markdown. Powerpack comes with a
markdown utility. Be sure to keep your browser open, visiting the test page,
then update src/powerblog/core.clj:
(ns powerblog.core
(:require [powerpack.markdown :as md]))
(defn render-page [context page]
(md/render-html (:page/body page)))
;; ...You will now be introduced to Powerpack's development experience: the browser automatically refreshes to render the updated version. Powerpack live reloads your page whenever you change the content, the code that renders it, or any assets (CSS, images, etc).
Powerpack understands mapdown, a small extension to Markdown that allows you to put key/value pairs into a markdown file. Let's use it to create a frontpage.
Add the following to content/index.md:
:page/uri /
:page/body
# The Powerblog
You have reached the Powerblog, the highly fictitious blog that simply exists to
showcase [Powerpack](https://github.com/cjohansen/powerpack).Because this file contains the :page/uri key, Powerpack will not give it a URL
based on its path. Instead, this will be our frontpage. You should be able to
see it on http://localhost:5050/.
To demonstrate yet another development feature, try changing :page/body to
:page-body in the markdown file and save it. Powerpack will encounter an
error that is displayed in a HUD at the bottom of the page until you fix it.
Change it back and save the file to see the error go away.
Since we're building a blog, let's add a schema to store blog posts and authors.
Change resources/schema.edn so it contains the following:
[{:db/ident :blog-post/author
:db/valueType :db.type/ref
:db/cardinality :db.cardinality/one}
{:db/ident :blog-post/tags
:db/valueType :db.type/keyword
:db/cardinality :db.cardinality/many}
{:db/ident :person/id
:db/valueType :db.type/keyword
:db/cardinality :db.cardinality/one
:db/unique :db.unique/identity}
{:db/ident :person/full-name
:db/valueType :db.type/string
:db/cardinality :db.cardinality/one}]Whenever you change the schema, Powerpack will reboot your app and refresh the browser.
Refer to Datomic's schema documentation for details on the schema.
Next up, we will add an author. Since we're not expecting large amounts of markup, we'll stick this information in EDN files rather than markdown. You can have multiple authors in one file, or one file per author - it's up to you. Let's just create a file with a single author in it.
Add the following to content/authors/christian.edn:
{:person/id :christian
:person/full-name "Christian Johansen"}Unlike markdown files, EDN files are not treated as pages by Powerpack. The
content is just read into Datomic. If you want EDN files to create one or more
pages, it must contain map(s) with :page/uri on them.
Create a blog post in content/blog-posts/first-post.md:
:page/title On the wonders of climbing
:blog-post/tags [:climbing :nature]
:blog-post/author {:person/id :christian}
:page/body
# On the wonders of climbing
[Climbing](https://en.wikipedia.org/wiki/Climbing), a primal instinct ingrained
in our evolutionary history, takes on a playful and acrobatic twist when
observed in the world of monkeys. As we delve into the realm of these agile and
nimble creatures, we uncover a captivating tapestry of tree-bound adventures,
showcasing their unparalleled mastery of the vertical realm.This is mapdown once again. Because we didn't give it a specific :page/uri, it
will be available at its path,
/blog-posts/first-post/. Notice
the use of :page/title -- another built-in schema attribute. Also notice that
the :blog-post/tags is parsed as a collection of keywords to match the
database schema.
The structured information about the blog post and the author is currently
nowhere to be seen on the site. You can verify that it's present by interacting
with the database. Update the comment-block in powerblog.dev to the
following and evaluate the forms:
(comment
(set! *print-namespace-maps* false)
(dev/start)
(dev/stop)
(dev/reset)
(def app (dev/get-app))
(require '[datomic.api :as d])
(def db (d/db (:datomic/conn app)))
(->> (d/entity db [:page/uri "/blog-posts/first-post/"])
:blog-post/author
(into {}))
;;=> {:person/id :christian
;; :person/full-name "Christian Johansen"}
)This example shows you that the page is indeed a blog post, and you can navigate to the author from the entity map.
We are currently rendering all pages the same way. We can add a dispatching
mechanism in render-page to render the frontpage differently from the blog
posts and other pages.
We'll handle the frontpage by dispatching on the URL, and then we'll see a more robust mechanism that includes that page kind in the database.
Update render-page in powerblog.core to the following:
(defn render-page [context page]
(cond
(= "/" (:page/uri page))
(render-frontpage context page)
:else
(md/render-html (:page/body page))))Add this new function above it:
(defn render-frontpage [context page]
[:html
[:head
[:title "The Powerblog"]]
[:body
(md/render-html (:page/body page))
[:h2 "Blog posts"]]])This page returns hiccup, which Powerpack knows how to render.
With this function in place, we can now render the frontpage differently from the rest of the pages. Let's spice up the frontpage with a list of blog posts.
The database is available on the context parameter under :app/db. We can use
it to query for any page that contains the blog-post/author key:
(defn get-blog-posts [db]
(->> (d/q '[:find [?e ...]
:where
[?e :blog-post/author]]
db)
(map #(d/entity db %))))
(defn render-frontpage [context page]
[:html
[:head
[:title "The Powerblog"]]
[:body
(md/render-html (:page/body page))
[:h2 "Blog posts"]
[:ul
(for [blog-post (get-blog-posts (:app/db context))]
[:li [:a {:href (:page/uri blog-post)} (:page/title blog-post)]])]]])When you save this, the frontpage should update immediately and show your one blog post. Clicking the link should take you to the blog post.
Having added a title to the frontpage, it becomes clear that the blog post is a
little bare bones. It doesn't even have a body tag. When that is the case,
Powerpack assumes the page is a fragment (perhaps for fetching via JavaScript)
and does no post processing. If we update the render function to wrap the page
in an HTML document, Powerpack adds :page/title as the head title if you
don't provide one yourself:
(defn render-page [context page]
(cond
(= "/" (:page/uri page))
(render-frontpage context page)
:else
[:html [:body (md/render-html (:page/body page))]]))Now the blog posts also have a DOCTYPE and the document title reads:
On the wonders of climbing | The Powerblog
E.g. the page title, with the :site/title global configuration option added.
As demonstrated so far, Powerpack will do its best to ingest content from files
to Datomic automatically. But automatic based on assumptions only takes us so
far. For instance, it would be nice if the blog posts had a URL that started
with /blog/ instead of the current /blog-posts/, and if they had a more
stable marker than the presence of :blog-post/author. We can fix this by
processing data on the way from files to the database.
You can give Powerpack a function with :powerpack/create-ingest-tx that will
be called every time content has been read from a file to ingest. The function
receives the file name and the parsed content. The content is always a vector --
even if the source is an EDN file with a single map, Powerpack will wrap it in a
vector. This is done so that the data can be transacted into Datomic
automatically, even when there is no :powerpack/create-ingest-tx. Let's see an
example. Head over to powerblog.core and add the option:
(defn create-tx [file-name txes]
(cond->> txes
(re-find #"^blog-posts/" file-name)
(map #(assoc % :page/kind :page.kind/blog-post))))
(def config
{:site/title "The Powerblog"
:powerpack/render-page #'render-page
:powerpack/create-ingest-tx #'create-tx})This adds the built-in attribute :page/kind to all the blog posts. The
attribute takes any keyword, and is well suited for separating different
kinds of pages.
When you change the main configuration, Powerpack will automatically reboot.
While we're at it, let's place the rendering functions in powerblog.pages and
the ingest function in powerblog.ingest -- gotta keep 'em separated. The
resulting powerblog.core namespace looks like this:
(ns powerblog.core
(:require [powerblog.ingest :as ingest]
[powerblog.pages :as pages]))
(def config
{:site/title "The Powerblog"
:powerpack/render-page #'pages/render-page
:powerpack/create-ingest-tx #'ingest/create-tx})As your site grows, you might continue this line and add separate namespaces for individual page types as well. We'll get there eventually.
Before we fix the render dispatch from before, lets expand the ingest function
to add :page/kind to all pages, that is all transaction entries that have a
:page/uri:
(defn get-page-kind [file-name]
(cond
(re-find #"^blog-posts/" file-name)
:page.kind/blog-post
(re-find #"^index\.md" file-name)
:page.kind/frontpage
(re-find #"\.md$" file-name)
:page.kind/article))
(defn create-tx [file-name txes]
(let [kind (get-page-kind file-name)]
(for [tx txes]
(cond-> tx
(and (:page/uri tx) kind)
(assoc :page/kind kind)))))Let's now revisit powerblog.pages/render-page and use :page/kind for
dispatch:
(defn layout [{:keys [title]} & content]
[:html
[:head
(when title [:title title])]
[:body
content]])
(def header
[:header [:a {:href "/"} "Powerblog"]])
(defn render-frontpage [context page]
(layout {:title "The Powerblog"}
(md/render-html (:page/body page))
[:h2 "Blog posts"]
[:ul
(for [blog-post (get-blog-posts (:app/db context))]
[:li [:a {:href (:page/uri blog-post)} (:page/title blog-post)]])]))
(defn render-article [context page]
(layout {}
header
(md/render-html (:page/body page))))
(defn render-blog-post [context page]
(render-article context page))
(defn render-page [context page]
(case (:page/kind page)
:page.kind/frontpage (render-frontpage context page)
:page.kind/blog-post (render-blog-post context page)
:page.kind/article (render-article context page)))This is now starting to look like something to build upon. The structure that's
emerging is one where data processing happens at ingest, and page rendering is
about converting data from the database to markup. Notice that while Powerpack
caters to this sort of structure (e.g. by providing :page/kind), you are free
to find your own approach.
Most modern websites have some colors and typography that deviates from the bare browser defaults (unfortunately, not every day is CSS naked day). Let's add some of our own.
Powerpack has no opinion on how you do CSS. For this demonstration, we'll start small with a single CSS file. Add the following to resources/public/styles.css:
html {
font-size: 20px;
font-family: Helvetica, arial, sans-serif;
}
body {
background: #18181b;
color: #f0f0f0;
max-width: 800px;
margin: 20px auto;
line-height: 1.5;
}
a:link, a:visited {
color: #2563eb;
text-decoration: underline;
}
a:hover {
text-decoration: none;
}Then update the main Powerpack configuration in powerblog.core by adding the
CSS file as a bundle:
(def config
{:site/title "The Powerblog"
:powerpack/render-page #'pages/render-page
:powerpack/create-ingest-tx #'ingest/create-tx
:optimus/bundles {"app.css"
{:public-dir "public"
:paths ["/styles.css"]}}})Powerpack uses Optimus to serve assets. This way you will have perfectly optimized assets for use in production.
:optimus/bundles will automatically be included in any HTML response. CSS
bundles go in head, and JavaScript bundles go in the end of body. So,
without any further ado, the site should now look a little bit more smashing
than before.
Whenever you update the CSS file, Powerpack will hot reload it for you.
Let's try our hand at a slightly more involved asset setup by adding TailwindCSS.
Install and initialize Tailwind:
npm install -D tailwindcss
npm install -D @tailwindcss/typography
npx tailwindcss initNext we'll configure Tailwind. It will be able to glean what classes we're using
from the Clojure source code and update the CSS file accordingly. Put the
following in tailwind.config.js:
/** @type {import('tailwindcss').Config} */
module.exports = {
content: ["./src/**/*.clj"],
theme: {
extend: {
typography: theme => ({
DEFAULT: {
css: {
a: {
color: theme('colors.blue.600')
},
'a:hover': {
color: theme('colors.blue.500')
}
}
},
invert: {}
})
}
},
plugins: [
require('@tailwindcss/typography')
]
}We'll need to run the Tailwind CLI to generate the CSS file. I like Make, so
put the following in a Makefile:
tailwind:
npx tailwindcss -i ./src/main.css -o ./resources/public/styles.css --watch
.PHONY: tailwind
Add the source CSS file in src/main.css:
@tailwind base;
@tailwind components;
@tailwind utilities;
@layer base {
html {
font-size: 20px;
}
}Now run make tailwind in a terminal, and Tailwind will keep
resources/public/styles.css up to date for us. We can then add some Tailwind
utility classes to our page setup:
(defn layout [{:keys [title]} & content]
[:html.dark:bg-zinc-900
[:head
(when title [:title title])]
[:body.py-8
content]])
(def header
[:header.mx-auto.dark:prose-invert.prose.mb-8
[:a {:href "/"} "Powerblog"]])
(defn render-frontpage [context page]
(layout {:title "The Powerblog"}
[:article.prose.dark:prose-invert.mx-auto
(md/render-html (:page/body page))
[:h2 "Blog posts"]
[:ul
(for [blog-post (get-blog-posts (:app/db context))]
[:li [:a {:href (:page/uri blog-post)} (:page/title blog-post)]])]]))
(defn render-article [context page]
(layout {}
header
[:article.prose.dark:prose-invert.mx-auto
(md/render-html (:page/body page))]))
(defn render-blog-post [context page]
(render-article context page))And just like that, all of Tailwind is at our hand.
A good development environment is all fun and games, but not worth alot if you can't put something in production. So let's put something in production.
Create a new namespace in src/powerblog/export.clj with the following content:
(ns powerblog.export
(:require [powerblog.core :as blog]
[powerpack.export :as export]))
(defn ^:export export! [& args]
(-> blog/config
(assoc :site/base-url "https://www.example.com")
export/export!))As you can tell, this namespace is an excellent place to make adjustments to the
Powerpack configuration that are more suitable for the production environment.
For example, the :site/base-url property is used to qualify open graph URLs,
or - if you didn't add any - add an og:url meta tag to your pages,
differentiate internal and external links, and qualify asset and image urls.
Add a :build alias to deps.edn:
{,,,
:aliases
{:dev {:extra-paths ["dev"]}
:build {:exec-fn powerblog.export/export!}}}With this alias you can export the site like so:
$ clojure -X:build
[powerpack.export] Creating app
[powerpack.app] Created database in 245ms
[powerpack.export] ... complete in 249ms
[powerpack.ingest] Ingested authors/christian.edn
[powerpack.ingest] Ingested blog-posts/first-post.md
[powerpack.ingest] Ingested index.md
[powerpack.ingest] Ingested test.md
[powerpack.app] Ingested all data in 37ms
[powerpack.export] Rendering, validating and exporting 3 pages
[powerpack.export] ... complete in 89ms
[powerpack.export] Exporting assets
[powerpack.export] ... complete in 2ms
[powerpack.export] Export complete
[powerpack.export] Exported 3 pages
[powerpack.export] Ran Powerpack export in 2081msNow you have a static copy of your site in target/powerpack (change the
destination by setting :powerpack/build-dir in the Powerpack configuration).
These files can be served as is by any static website host: nginx, AWS S3, etc.
Powerpack comes with image manipulation skills courtesy of imagine. Imagine allow you to configure various image aliases that perform transformations such as duotone/greyscale, crop, fit, scale, and transform any image accordingly.
We will use an image from the Wikipedia article on climbing, by Heinz Zak to demonstrate:
mkdir resources/public/images
wget https://upload.wikimedia.org/wikipedia/commons/2/26/Heinz_Zak%2C_Separate_Reality_5%2C11d%2C_Free_Solo%2C_Yosemite-Nationalpark%2C_Kalifornien%2C_USA.jpg \
-O resources/public/images/climbing.jpgLet's add the image to our blog post. Update content/blog-posts/first-post.md
to the following:
:page/title On the wonders of climbing
:blog-post/author {:person/id :christian}
:page/body
# On the wonders of climbing
[Climbing](https://en.wikipedia.org/wiki/Climbing), a primal instinct ingrained
in our evolutionary history, takes on a playful and acrobatic twist when
observed in the world of monkeys. As we delve into the realm of these agile and
nimble creatures, we uncover a captivating tapestry of tree-bound adventures,
showcasing their unparalleled mastery of the vertical realm.
When you do this, Powerpack will complain loudly. It does not like serving assets that are not configured through either Optimus or Imagine. We'll explore both options.
To serve the image through Optimus, add :optimus/assets to the Powerpack
configuration:
(def config
{:site/title "The Powerblog"
:powerpack/render-page #'pages/render-page
:powerpack/create-ingest-tx #'ingest/create-tx
:optimus/bundles {"app.css"
{:public-dir "public"
:paths ["/styles.css"]}}
:optimus/assets [{:public-dir "public"
:paths [#".*\.jpg"]}]})Now your image is available, and more importantly -- will be exported with the rest of the site.
To serve the image with Imagine, we will add transformation configuration (refer to the Imagine readme for more details on that:
(def config
{:site/title "The Powerblog"
:powerpack/render-page #'pages/render-page
:powerpack/create-ingest-tx #'ingest/create-tx
:optimus/bundles {"app.css"
{:public-dir "public"
:paths ["/styles.css"]}}
:optimus/assets [{:public-dir "public"
:paths [#".*\.jpg"]}]
:imagine/config {:prefix "image-assets"
:resource-path "public"
:disk-cache? true
:transformations
{:preview-small
{:transformations [[:fit {:width 184 :height 184}]
[:crop {:preset :square}]]
:retina-optimized? true
:retina-quality 0.4
:width 184}}}})We can now prefix the image URL with the transformation name preview-small to
serve the image as a retina optimized 184x184 square image. In
content/blog-posts/first-post.md:
...
Powerpack comes with i18n support through m1p, a tiny library for i18n, theming, and other "content flavoring".
To use m1p, start by telling Powerpack where to find your dictionaries. In
powerblog.core:
(def config
{;; ...
:m1p/dictionaries {:nb ["src/powerblog/i18n/nb.edn"]
:en ["src/powerblog/i18n/en.edn"]}})At their simplest, m1p dictionaries are just maps. Once again we'll start small. Add these two files:
;; src/powerblog/i18n/nb.edn
{:powerblog.pages/blog-posts "Blogginnlegg"}
;; src/powerblog/i18n/en.edn
{:powerblog.pages/blog-posts "Blog posts"}Next up, we'll need to know what locale a page is supposed to be in. The
built-in Powerpack schema includes :page/locale, which takes a keyword. We'll
use this to create two frontpages: one in Norwegian, and one in English.
Start by updating content/index.md like so:
:page/uri /
:page/locale :en
:page/body
# The Powerblog
You have reached the Powerblog, the highly fictitious blog that simply exists to
showcase [Powerpack](https://github.com/cjohansen/powerpack).Then add content/index-nb.md:
:page/uri /nb/
:page/locale :nb
:page/body
# Powerbloggen
Du har nådd Powerbloggen, den høyst fiktive bloggen som kun eksisterer for å demonstrere
[Powerpack](https://github.com/cjohansen/powerpack).Update the ingest function in powerblog.ingest so it marks the Norwegian
version as a frontpage as well:
(defn get-page-kind [file-name]
(cond
(re-find #"^blog-posts/" file-name)
:page.kind/blog-post
(re-find #"^index(-nb)?\.md" file-name)
:page.kind/frontpage
(re-find #"\.md$" file-name)
:page.kind/article))Now we're all set to localize the rendering function for the frontpage. Remember
that it currently contains the heading [:h2 "Blog posts"]. We can now replace
it with a reference to the key in our dictionaries, e.g.
:powerblog.pages/blog-posts:
(defn render-frontpage [context page]
(layout {:title "The Powerblog"}
[:article.prose.dark:prose-invert.mx-auto
(md/render-html (:page/body page))
[:h2 [:i18n :powerblog.pages/blog-posts]]
[:ul
(for [blog-post (get-blog-posts (:app/db context))]
[:li [:a {:href (:page/uri blog-post)} (:page/title blog-post)]])]]))Since we were thoughtful enough to use the same namespace for the key as the namespace this code lives in, we can even do this:
(defn render-frontpage [context page]
(layout {:title "The Powerblog"}
[:article.prose.dark:prose-invert.mx-auto
(md/render-html (:page/body page))
[:h2 [:i18n ::blog-posts]]
[:ul
(for [blog-post (get-blog-posts (:app/db context))]
[:li [:a {:href (:page/uri blog-post)} (:page/title blog-post)]])]]))We can also make the dictionaries a little more convenient to edit by using Clojure's namespaced maps feature:
;; src/powerblog/en.edn
#:powerblog.pages
{:blog-posts "Blog posts"}
;; src/powerblog/nb.edn
#:powerblog.pages
{:blog-posts "Blogginnlegg"}As usual: when you edit your dictionaries, Powerpack automatically refreshes the web page for you.
As the site grows and you have several namespaces, you can put multiple namespaced maps in a vector in dictionary files -- or create multiple dictionary files. It's up to you.
Simple key/value lookup is a little limited. Let's interpolate the number of blog posts into the heading. Update the dictionaries:
#:powerblog.pages
{:blog-posts [:fn/str "Blog posts ({{:n}})"]}:fn/str is a dictionary
function,
a m1p feature. Feed it a value like so:
(defn render-frontpage [context page]
(let [blog-posts (get-blog-posts (:app/db context))]
(layout {:title "The Powerblog"}
[:article.prose.dark:prose-invert.mx-auto
(md/render-html (:page/body page))
[:h2 [:i18n ::blog-posts {:n (count blog-posts)}]]
[:ul
(for [blog-post blog-posts]
[:li [:a {:href (:page/uri blog-post)} (:page/title blog-post)]])]])))To install custom dictionary functions, like the pluralization helper from the m1p docs, provide them with the main Powerpack configuration:
(ns powerblog.core
(:require [m1p.core :as m1p]
[powerblog.ingest :as ingest]
[powerblog.pages :as pages]))
(defn pluralize [opt n & plurals]
(-> (nth plurals (min (if (number? n) n 0) (dec (count plurals))))
(m1p/interpolate-string {:n n} opt)))
(def config
{;; ...
:m1p/dictionaries {:nb ["src/powerblog/i18n/nb.edn"]
:en ["src/powerblog/i18n/en.edn"]}
:m1p/dictionary-fns {:fn/plural #'pluralize}})We can use it in dictionaries like this:
#:powerblog.pages
{:blog-posts [:fn/plural
"No blog posts yet"
"My blog post"
"Blog posts ({{:n}})"]}Now the heading will read "No blog posts yet" when there are no blog posts, "My blog post" when there is only one blog post, and "Blog posts (3)" when there are 3 blog posts.
Powerpack comes with some built-in keys you can use to auto-generate open graph meta tags on your pages:
:open-graph/title:open-graph/description:open-graph/image
These can be added directly to our blog post as such:
:page/title On the wonders of climbing
:blog-post/author {:person/id :christian}
:open-graph/title Climbing
:open-graph/description An interesting piece about climbing
:open-graph/image /preview-small/images/climbing.jpg
:page/body
# On the wonders of climbing
[Climbing](https://en.wikipedia.org/wiki/Climbing), a primal instinct ingrained
in our evolutionary history, takes on a playful and acrobatic twist when
observed in the world of monkeys. As we delve into the realm of these agile and
nimble creatures, we uncover a captivating tapestry of tree-bound adventures,
showcasing their unparalleled mastery of the vertical realm.
With these keys on the page entity, the markup will automatically include sensible open graph tags:
<!DOCTYPE html>
<html lang="en" prefix="og: http://ogp.me/ns#" class="dark:bg-zinc-900">
<head>
<title>On the wonders of climbing | The Powerblog</title>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<meta property="og:description" content="An interesting piece about climbing">
<meta property="og:title" content="Climbing">
<meta property="og:image" content="/image-assets/preview-small/fb6a746aee13f753872432da49c32a1cd019a334/images/climbing.jpg">
<meta property="og:image:width" content="184">
<meta property="og:image:height" content="184">
<link rel="stylesheet" href="/af8dc61fd222/styles.css">
</head>
<body class="py-8">
<!-- ... -->
</body>
</html>As you can see, Powerpack adds not only open graph metas, but some other useful
ones as well. It will only add these meta tags if they are not already present.
If you add your own <meta name="viewport" ...> with different content,
Powerpack will not touch it.
You will notice that the image URL is different from the one we set on the page. The generated URL includes a cache buster, which allows you to set a far future expires header on images for best performance. The URL will change whenever the underlying image changes, but not otherwise.
Some pages don't have any content that's natural to put in a separate markdown or EDN file. In order for Powerpack to recognize them as pages, they need to exist in the database. Let's look at two options to add these to the database.
The simplest approach is to have an EDN file with your static pages in the
content directory. Add the following to content/static-pages.edn:
[{:page/uri "/blog/"
:page/kind :page.kind/blog-listing
:page/locale :en}
{:page/uri "/blogg/"
:page/kind :page.kind/blog-listing
:page/locale :nb}]As the number of pages is growing, we'll take the opportunity to split pages into separate namespaces.
Move the layout to a separate namespace in src/powerblog/layout.clj:
(ns powerblog.layout)
(defn layout [{:keys [title]} & content]
[:html.dark:bg-zinc-900
[:head
(when title [:title title])]
[:body.py-8
content]])
(def header
[:header.mx-auto.dark:prose-invert.prose.mb-8
[:a {:href "/"} "Powerblog"]])
Move the frontpage to src/powerblog/pages/frontpage.clj:
(ns powerblog.pages.frontpage
(:require [datomic.api :as d]
[powerblog.layout :as layout]
[powerpack.markdown :as md]))
(defn get-blog-posts [db]
(->> (d/q '[:find [?e ...]
:where
[?e :blog-post/author]]
db)
(map #(d/entity db %))))
(defn render-page [context page]
(let [blog-posts (get-blog-posts (:app/db context))]
(layout/layout {:title "The Powerblog"}
[:article.prose.dark:prose-invert.mx-auto
(md/render-html (:page/body page))
[:h2 [:i18n ::blog-posts (count blog-posts)]]
[:ul
(for [blog-post blog-posts]
[:li [:a {:href (:page/uri blog-post)} (:page/title blog-post)]])]])))
Add the following to src/powerblog/pages/blog_listing.clj:
(ns powerblog.pages.blog-listing
(:require [powerblog.layout :as layout]
[powerblog.pages.frontpage :as frontpage]))
(defn render-page [context page]
(layout/layout {:title [:i18n ::page-title]}
[:article.prose.dark:prose-invert.mx-auto
[:h1 [:i18n ::page-title]]
[:ul
(for [blog-post (frontpage/get-blog-posts (:app/db context))]
[:li [:a {:href (:page/uri blog-post)} (:page/title blog-post)]])]]))The updated powerblog.pages now looks like:
(ns powerblog.pages
(:require [powerblog.pages.article :as article]
[powerblog.pages.blog-listing :as blog-listing]
[powerblog.pages.blog-post :as blog-post]
[powerblog.pages.frontpage :as frontpage]))
(defn render-page [context page]
(case (:page/kind page)
:page.kind/frontpage (frontpage/render-page context page)
:page.kind/blog-post (blog-post/render-page context page)
:page.kind/blog-listing (blog-listing/render-page context page)
:page.kind/article (article/render-page context page)))And the i18n dictionaries should look like this:
;; src/powerblog/en.edn
[#:powerblog.pages.frontpage
{:blog-posts [:fn/plural
"No blog posts yet"
"My blog post"
"Blog posts ({{:n}})"]}
#:powerblog.pages.blog-listing
{:page-title "Blog posts"}]
;; src/powerblog/nb.edn
[#:powerblog.pages.frontpage
{:blog-posts [:fn/plural
"Ingen blogginnlegg enda"
"Mitt blogginnlegg"
"Blogginnlegg ({{:n}})"]}
#:powerblog.pages.blog-listing
{:page-title "Blogginnlegg"}]Obviously, the blog listing is very similar to the frontpage at this point, but the main point was how to seed the database with more pages, and how to think about code organization in a growing project.
The second option is to add a post-ingest hook to Powerpack to ingest the pages.
This approach is particularly useful if you want to generate pages with code. As
an example, let's create a page for each tag in the database. To do this, we'll
want to associate a tag with each tag page, and for that we need a new schema
attribute. Update resources/schema.edn to this:
[{:db/ident :blog-post/author
:db/valueType :db.type/ref
:db/cardinality :db.cardinality/one}
{:db/ident :blog-post/tags
:db/valueType :db.type/keyword
:db/cardinality :db.cardinality/many}
{:db/ident :person/id
:db/valueType :db.type/keyword
:db/cardinality :db.cardinality/one
:db/unique :db.unique/identity}
{:db/ident :person/full-name
:db/valueType :db.type/string
:db/cardinality :db.cardinality/one}
;; Add a custom attribute for our tag pages
{:db/ident :tag-page/tag
:db/valueType :db.type/keyword
:db/cardinality :db.cardinality/one}]Now add :powerpack/on-ingested to the Powerpack configuration:
(def config
{:site/title "The Powerblog"
:powerpack/render-page #'pages/render-page
:powerpack/create-ingest-tx #'ingest/create-tx
:powerpack/on-ingested #'ingest/on-ingested
,,,
})This function will be called with the powerpack app instance and the result from
Datomic's transactions after every time it has updated the database (e.g. when
it initially boots up, and whenever you change content files, etc). This means
we can read the contents from the database and use it to transact more pages
into it. Here's the on-ingested function from powerblog.ingest:
(defn on-ingested [powerpack-app results]
(->> (for [tag (d/q '[:find [?tag ...]
:where
[_ :blog-post/tags ?tag]]
(d/db (:datomic/conn powerpack-app)))]
{:page/uri (str "/tag/" (name tag) "/")
:page/kind :page.kind/tag
:tag-page/tag tag})
(d/transact (:datomic/conn powerpack-app))
deref))There is a similar hook called :powerpack/on-started that only runs once after
the app boots up (it only receives the powerpack app instance).
We run a Datomic query to find
all the tags in use, and create a page entity for each one. Next, let's create
src/powerblog/pages/tag.clj with:
(ns powerblog.pages.tag
(:require [datomic.api :as d]
[powerblog.layout :as layout]))
(defn get-blog-posts [db tag]
(->> (d/q '[:find [?e ...]
:in $ ?tag
:where
[?e :blog-post/tags ?tag]]
db tag)
(map #(d/entity db %))))
(defn render-page [context page]
(let [title (str "Blog posts about " (name (:tag-page/tag page)))]
(layout/layout
{:title title}
[:article.prose.dark:prose-invert.mx-auto
[:h1 title]
[:ul
(for [blog-post (get-blog-posts (:app/db context) (:tag-page/tag page))]
[:li [:a {:href (:page/uri blog-post)} (:page/title blog-post)]])]])))The keen observer will notice that the tag pages are not localized. Doing so is left as an exercise for the reader (all time best cop-out).
Finally, the updated function in powerblog.pages:
(defn render-page [context page]
(case (:page/kind page)
:page.kind/frontpage (frontpage/render-page context page)
:page.kind/blog-post (blog-post/render-page context page)
:page.kind/blog-listing (blog-listing/render-page context page)
:page.kind/tag (tag/render-page context page)
:page.kind/article (article/render-page context page)))