ElixirDose

Journal of the overlooked tiny bits in young programming language. Will update weekly or two, as a way to build the habit of writing something on a regular basis. Authored by Riza and Augie as Editor In Chief.


Cowboy Tutorial Part 2: Creating Flat File Blog

If you follow ElixirDose blog, we did tried to create a very simple web framework over here. In that article we did serve some markdown file as the example. In this article, we take the markdown example a little bit further by creating a drop dead simple flat file blogging engine. Because the first article is little bit outdated, we will learn together how to get started with cowboy, a small, fast, and modular HTTP server written in Erlang.

Let's talk more detail about blogging engine that we will create. Our blogging engine will read through one folder that have several markdown files. No database whatsoever and it should be blazingly fast as a blog. The idea is to edit your content with your favorite text editor using markdown formatting, then you put on destination folder and viola! You've published your content. Quick and easy. We also support theming if someday we want to redesign the blog.

By creating this blogging engine, we will learn more about certain topics:

What we need to accomplish this project is:

Ready, steady, go!

Creating Elixir Project

We begin our journey by creating elixir project using mix tools.

$> mix new dds_blog --sup

We named our project dds_blog after Drop Dead Simple Blogging Engine. And we add --sup argument to let mix know that we wanted to create an OTP supervisor and OTP application callback in our main DdsBlog module. You may learn more about OTP here.

Don't forget to change directory to our project folder that we've created.

$> cd dds_blog

Now's the time to summon the cowboy.

Add Cowboy To The Project

To make our application running, we'll need a server that can speak HTTP. We will use Cowboy for this case because it's awesome.

To add Cowboy to our project, simply add Cowboy as the dependency inside mix.exs file.

defp deps do
  [
    {:cowboy, "1.0.0"}
  ]
end

We need also add :cowboy under applications function while we're editing mix.exs.

def application do
    [applications: [:logger, :cowboy],
    mod: {DdsBlog, []}]
end

Now run mix deps.get to pull all deps needed (Cowboy and it's deps as well).

$ mix deps.get Running dependency resolution Unlocked: cowboy Dependency resolution completed successfully ranch: v1.0.0 cowlib: v1.0.1 cowboy: v1.0.0

[..]

As you can see, we're pulling not only Cowboy package but also cowlib and ranch as packages that Cowboy's dependent on.

Create Cowboy Routes

Ok now we'll create a helper function that defines a set of routes for our project.

def run do
  routes = [
    {"/", DdsBlog.Handler, []}
  ]

  dispatch = :cowboy_router.compile([{:_, routes}])

  opts = [port: 8000]
  env = [dispatch: dispatch]

  {:ok, _pid} = :cowboy.start_http(:http, 100, opts, [env: env])
end

Oh yeah, we also calling this run function as soon this application started. This step optional by the way. You still able to run the application but you have to calling it manually. Then our code bacame something like this:

defmodule DdsBlog do
  use Application

  def start(_type, _args) do
    import Supervisor.Spec, warn: false

    children = [
      worker(__MODULE, [], function: :run)
    ]

    opts = [strategy: :one_for_one, name: DdsBlog.Supervisor]
    Supervisor.start_link(children, opts)
  end

  def run do
    routes = [
      {"/", DdsBlog.Handler, []}
    ]

    dispatch = :cowboy_router.compile([{:_, routes}])

    opts = [port: 8000]
    env = [dispatch: dispatch]

    {:ok, _pid} = :cowboy.start_http(:http, 100, opts, [env: env])
  end
end

Now when we run our application, it will respond to all requests to http://localhost:8000/.

Creating Handler

As you can see on the code above, we did pointing out route "/" into DdsBlog.Handler. Now we need to create that module to return some responses for any requests received. Let's create a new file in lib/dds_blog/handler.ex and put this code below.

defmodule DdsBlog.Handler do
  def init({:tcp, :http}, req, opts) do
    headers = [{"content-type", "text/plain"}]
    body = "Hello program!"
    {:ok, resp} = :cowboy_req.reply(200, headers, body, req)
    {:ok, resp, opts}
  end

  def handle(req, state) do
    {:ok, req, state}
  end

  def terminate(_reason, _req, _state) do
    :ok
  end
end

init function doing a lot of works. First, it tells Cowboy of what kind of connectins we wish to handle (HTTP via TCP). Then we use :cowboy_req.reply with status code of 200, a list of headers, a response body and the request itself. We will not touch handle and terminate for now. Let's consider it as boilerplate code for now.

Running For The First Time

Now it's the time for us to run this application for the first time.

$> iex -S mix
Erlang/OTP 17 [erts-6.2] [source] [64-bit] [smp:4:4] [async-threads:10] [hipe] [kernel-poll:false] [dtrace]

==> ranch (compile)
Compiled src/ranch_transport.erl
Compiled src/ranch_sup.erl
Compiled src/ranch_tcp.erl
Compiled src/ranch_ssl.erl
Compiled src/ranch_protocol.erl
Compiled src/ranch_listener_sup.erl
Compiled src/ranch_app.erl
Compiled src/ranch_acceptors_sup.erl
Compiled src/ranch_acceptor.erl
Compiled src/ranch.erl
Compiled src/ranch_server.erl
Compiled src/ranch_conns_sup.erl
==> cowlib (compile)
Compiled src/cow_qs.erl
Compiled src/cow_spdy.erl
Compiled src/cow_multipart.erl
Compiled src/cow_http_te.erl
Compiled src/cow_http_hd.erl
Compiled src/cow_date.erl
Compiled src/cow_http.erl
Compiled src/cow_cookie.erl
Compiled src/cow_mimetypes.erl
==> cowboy (compile)
Compiled src/cowboy_sub_protocol.erl
Compiled src/cowboy_middleware.erl
Compiled src/cowboy_websocket_handler.erl
Compiled src/cowboy_sup.erl
Compiled src/cowboy_static.erl
Compiled src/cowboy_spdy.erl
Compiled src/cowboy_router.erl
Compiled src/cowboy_websocket.erl
Compiled src/cowboy_protocol.erl
Compiled src/cowboy_loop_handler.erl
Compiled src/cowboy_http_handler.erl
Compiled src/cowboy_rest.erl
Compiled src/cowboy_handler.erl
Compiled src/cowboy_clock.erl
Compiled src/cowboy_bstr.erl
Compiled src/cowboy_app.erl
Compiled src/cowboy_http.erl
Compiled src/cowboy.erl
Compiled src/cowboy_req.erl
Compiled lib/dds_blog/handler.ex
Compiled lib/dds_blog.ex
Generated dds_blog.app

Interactive Elixir (1.0.1) - press Ctrl+C to exit (type h() ENTER for help)
iex(1)>

If didn't add run function as a worker child in the supervisor tree, just run in in the console. Otherwise you should be ok. Open up your browser and pointing out to http://localhost:8000/ then you'll see the most beautiful message in the programming world :)

Hello world

Static Cowboy

Let's add Cowboy to recognise static files: images, css, javascripts, etc. Everytime we hit http://localhost:8000/static/ it will relate to static folders that we will create shortly. But first, open up lib/dds_blog.ex and add one route for static files.

  def run do
    routes = [
      {"/:something", DdsBlog.Handler, []},
      {"/static/[...]", :cowboy_static, {:priv_dir, :dds_blog, "static_files"}}
    ]

    dispatch = :cowboy_router.compile([{:_, routes}])

    opts = [port: 8000]
    env = [dispatch: dispatch]

    {:ok, _pid} = :cowboy.start_http(:http, 100, opts, [env: env])
  end

Don't forget to add priv/static_files. All our static files will be in this directory.

$> mkdir -p priv/static_files

Re-run iex -S mix command and then let's try to add static file inside static_files folder for sanity check. Then try to access it on the browser.

lily

Easy, right?!

Serve Markdown File

This is where the fun begin. The idea is this: when user entry http://localhost:8000/some-markdown-file our application will look through file named some-markdown-file.md inside priv/contents/ folder, read through it, convert into html format and return it so the user will received html contents in their browser as the response.

First thing first, let's change our Cowboy route to accomodate that.

def run do
  routes = [
    {"/:filename", DdsBlog.Handler, []},
    {"/static/[...]", :cowboy_static, {:priv_dir, :dds_blog, "static_files"}}
  ]

  dispatch = :cowboy_router.compile([{:_, routes}])

  opts = [port: 8000]
  env = [dispatch: dispatch]

  {:ok, _pid} = :cowboy.start_http(:http, 100, opts, [env: env])
end

This route will accept anything user input in their urls. Then we also need to change our handler function in dds_blog/handler.ex. In Cowboy term, this called bindings. At first, I though it's query strings, but I was wrong. Query will accept http://localhost:8000/?query=yes kind of format. But we want to achieve http://localhost:8000/some-file so we use bindings.

Now let's open lib/dds_blog/handler.ex and we will handle file that requested by user via urls. But first we open just one markdown file for sanity check, as usual.

def init({:tcp, :http}, req, opts) do
  {:ok, req, opts}
end

def handle(req, state) do
  {method, req} = :cowboy_req.method(req)
  {param, req} = :cowboy_req.binding(:filename, req)
  IO.inspect param
  {:ok, req} = get_file(method, param, req)
  {:ok, req, state}
end

def get_file("GET", :undefined, req) do
  headers = [{"content-type", "text/plain"}]
  body = "Ooops. Article not exists!"
  {:ok, resp} = :cowboy_req.reply(200, headers, body, req)
end

def get_file("GET", param, req) do
  headers = [{"content-type", "text/html"}]
  {:ok, file} = File.read "priv/contents/filename.md"
  body = file
  {:ok, resp} = :cowboy_req.reply(200, headers, body, req)
end

And then we put some markdown file, just rename it to filename.md and put it to priv/contenst folder. Make it if the folder doesn't exist yet. What we've done is first we do use :cowboy.bindings to get a filename from urls. Then we passed the parameter into helper function called get_file. get_file function have two type: one with parameter and one more without parameter (a.k.a :undefined parameter). If user didn't include filename in the url, our app will call get_file :undefined and return message "Article doesn't exist". Otherwise, it will call get_file with param. After the app received parameter, it will read a file we put inside priv/contents directory, for this moment, we just read the exact file called filename.md. Then we print it as response to the user.

Markdown raw

It prints out markdown as raw. To make it return html we need to convert markdown to html first. To do that, we need a package to handle that. Add markdown package into mix.exs file.

defp deps do
  [
    {:cowboy, "1.0.0"},
    {:markdown, github: "devinus/markdown"}
  ]
end

Then get the dependencies with mix.

$> mix deps.get

After that, now we can access Markdown module and use to_html to convert markdown into html format. Let's do that now in lib/dds_blog/handler.ex file inside get_file function.

def get_file("GET", :undefined, req) do
  headers = [{"content-type", "text/plain"}]
  body = "Ooops. Article not exists!"
  {:ok, resp} = :cowboy_req.reply(200, headers, body, req)
end

def get_file("GET", param, req) do
  headers = [{"content-type", "text/html"}]
  {:ok, file} = File.read "priv/contents/filename.md"
  body = Markdown.to_html file
  {:ok, resp} = :cowboy_req.reply(200, headers, body, req)
end

That's it! Quit and restart iex -S mix and refersh your browser. It's html now.

Imgur

Very cool! Ok, let's take a break. It's exhausting...

meme

Now, get back to work!! We need to read the param variable and load markdown file inside priv/contents/ directory. To do that we just use concatenate string then read the file. After file loaded, we convert it into html then return it.

def get_file("GET", :undefined, req) do
  headers = [{"content-type", "text/plain"}]
  body = "Ooops. Article not exists!"
  {:ok, resp} = :cowboy_req.reply(200, headers, body, req)
end

def get_file("GET", param, req) do
  headers = [{"content-type", "text/html"}]
  {:ok, file} = File.read "priv/contents/" <> param <> ".md"
  body = Markdown.to_html file
  {:ok, resp} = :cowboy_req.reply(200, headers, body, req)
end

Let's copy one more markdown file into priv/contents/ folder then restart the iex -S mix command. Then we call the file from the browser. We copy file called basicfp.md and calling it with http://localhost:8000/basicfp. And, viola!

basicfp

Really cool, right?!

Now we need index page. Index page will show you a glipse of all contents we have. We need to iterate through priv/contents folder, we get all markdown file then we print it in the index file.

def get_file("GET", :undefined, req) do
  headers = [{"content-type", "text/html"}]
  file_lists = File.ls! "priv/contents/"
  content = print_articles file_lists, ""
  {:ok, resp} = :cowboy_req.reply(200, headers, content, req)
end

def get_file("GET", param, req) do
  headers = [{"content-type", "text/html"}]
  {:ok, file} = File.read "priv/contents/" <> param <> ".md"
  body = Markdown.to_html file
  {:ok, resp} = :cowboy_req.reply(200, headers, body, req)
end


def print_articles [h|t], index_contents do
  {:ok, article} = File.read "priv/contents/" <> h
  sliced = String.slice article, 0, 1000
  marked = Markdown.to_html sliced
  filename = String.slice(h, 0, String.length(h) - 3)
  more = "<a class='button' href='#{filename}'>More</a><hr />"
  print_articles t, index_contents <> marked <> more
end

def print_articles [], index_contents do
  index_contents
end

When the routes didn't received any param, we will print index page. We show all the files inside priv/contents using the power of functional programming: recursive. Take a look at two print_articles function. First of we start the recursive by calling print_articles function with list of files we've got from File.ls command followed by empty string. print_articles will loop through (read file and concatenate) all files until it reached an empty list then it will call the second print_articles function that simply do the termination point of the recursive. Then we also did truncated it so it's not too long and also we add more button linked to full article.

Re-run iex -S mix command and let check if it's worked. Now pointing out our browser to http://localhost:8000/ and we will see all the articles, truncated and with More button followed with <hr /> tag. Very nice, right?!

truncated

And if you click More, the link also working well, redirect us into full article view.

If you noticed, the markdown file didn't truncated properly. If you know how to truncated markdown properly, please let me know.

Templates And Themes

When you view source our page, both index page or detail page, it will prints out markdown file immedietely so the page is not html format properly. We need to taken care of it by prints out the markdown content inside some divs. This is what we will do now, includes adding some css framework to make our page beauty.

We will use Skeleton, a responsive css boilerplate. You free to use any other css framework out there. Download the css files, and images (and or javascript as well, if includes in the framework) then copy it into our static file directory priv/static_files.

priv/static_files/
├── css
│   ├── normalize.css
│   └── skeleton.css
└── images
    └── favicon.png

We also create new folder priv/themes to put our template file there. Let's add one file called index.html.eex. Notice that we add eex extension to html file because we want to binds some variables to the html file using EEx templating engine provided by Elixir.

priv/themes/
└── index.html.eex

Now we edit that index.html.eex file and replace static file url.to meet our Cowboy static url settings.

<!DOCTYPE html>
<html lang="en">
<head>

  <!-- Basic Page Needs
  –––––––––––––––––––––––––––––––––––––––––––––––––– -->
  <meta charset="utf-8">
  <title><%= title %></title>
  <meta name="description" content="">
  <meta name="author" content="">

  <!-- Mobile Specific Metas
  –––––––––––––––––––––––––––––––––––––––––––––––––– -->
  <meta name="viewport" content="width=device-width, initial-scale=1">

  <!-- FONT
  –––––––––––––––––––––––––––––––––––––––––––––––––– -->
  <link href="//fonts.googleapis.com/css?family=Raleway:400,300,600" rel="stylesheet" type="text/css">

  <!-- CSS
  –––––––––––––––––––––––––––––––––––––––––––––––––– -->
  <link rel="stylesheet" href="/static/css/normalize.css">
  <link rel="stylesheet" href="/static/css/skeleton.css">

  <!-- Favicon
  –––––––––––––––––––––––––––––––––––––––––––––––––– -->
  <link rel="icon" type="image/png" href="static/images/favicon.png">

</head>
<body>

  <!-- Primary Page Layout
  –––––––––––––––––––––––––––––––––––––––––––––––––– -->
  <div class="container">
    <div class="row">
      <%= content %>
    </div>
  </div>

<!-- End Document
  –––––––––––––––––––––––––––––––––––––––––––––––––– -->
</body>
</html>

As you can see, we also adding <%= content %> to bind content variable inside our container>row divs. Then we can compile the eex file and return html inside our handler.ex file. Let's do that right now. We also add <%= title %> to add html title changes if we move through pages.

def get_file("GET", :undefined, req) do
  headers = [{"content-type", "text/html"}]
  file_lists = File.ls! "priv/contents/"
  content = print_articles file_lists, ""
  title = "Welcome to DDS Blog"
  body = EEx.eval_file "priv/themes/index.html.eex", [content: content, title: title]
  {:ok, resp} = :cowboy_req.reply(200, headers, body, req)
end

def get_file("GET", param, req) do
  headers = [{"content-type", "text/html"}]
  {:ok, file} = File.read "priv/contents/" <> param <> ".md"
  content = Markdown.to_html file
  title = String.capitalize(param)
  body = EEx.eval_file "priv/themes/index.html.eex", [content: content, title: title]
  {:ok, resp} = :cowboy_req.reply(200, headers, body, req)
end

That's it! Restart the iex -S mix command and see what happen in your browser by refresh it.

Beauty

If we click More button we also see one beautiful detail page. We're pretty much finish here, but before we warp up, let's add a header, footer and make More button more appealing to click.

Let's do the More button first. Just add class button button-primary to the a tag.

def print_articles [h|t], index_contents do
  {:ok, article} = File.read "priv/contents/" <> h
  sliced = String.slice article, 0, 1000
  marked = Markdown.to_html sliced
  filename = String.slice(h, 0, String.length(h) - 3)
  more = "<a class='button button-primary' href='#{filename}'>More</a><hr />"
  print_articles t, index_contents <> marked <> more
end

We should refactor this thing a little bit. By moving out the html thingy to themes folder then we just eval eex into more variable.

def print_articles [h|t], index_contents do
  {:ok, article} = File.read "priv/contents/" <> h
  sliced = String.slice article, 0, 1000
  marked = Markdown.to_html sliced
  filename = String.slice(h, 0, String.length(h) - 3)
  more = EEx.eval_file "priv/themes/more_button.html.eex", [filename: filename]
  print_articles t, index_contents <> marked <> more
end

And now we create new template named priv/themes/more_button.html.eex wit just one line of button and hr tag. Then we will binding a filename into that.

<a class='button button-primary' href='<%= filename %>'>More</a><hr />

Refresh the browser, just to see everything is ok. And we're done.

Conclusion

We did a great job pulling it out together this simple flat file blogging engine. We built this engine just use two packages: Cowboy and markdown. How cool is that?! I know, I know, some portion of the code maybe a little be naive but we finish our mission and that's the important thing, right?! We can always improve anything else later.

This is the full code. You can always send us some issues and pull request there for inputs, feedbacks and some contributions. That's it for me and see you next time!

References

Back