We hope to offer a small but comprehensible tutorial here. At the end of this tutorial, you should be able to write all sorts of HTML templates using BlazeHtml.

Please note that you should at least know some basic Haskell before starting this tutorial. Real World Haskell and Learn You a Haskell are two good starting points.


The installation of BlazeHtml should be painless using cabal install:

[jasper@alice ~]$ cabal install blaze-html

Overloaded strings

The OverloadedStrings is not necessarily needed to work with BlazeHtml, but it is highly recommended, as it allows you to insert string literals in your HTML templates without having boilerplate function calls.

> {-# LANGUAGE OverloadedStrings #-}


This tutorial is a literate haskell file, thus we should begin by importing the modules we are going to use. To avoid name clashes, we just import everything twice – once with an alias and once without.

> import Control.Monad (forM_)
> import Text.Blaze.Html5
> import Text.Blaze.Html5.Attributes
> import qualified Text.Blaze.Html5 as H
> import qualified Text.Blaze.Html5.Attributes as A

As you can see, we imported the Html5 modules. Alternatively, you can choose to import Text.Blaze.Html4.Strict. More HTML versions are likely to be added in the future.

A first simple page

The main representation type in BlazeHtml is Html. Therefore, your “templates” will usually have the type signature ArgumentType1 → ArgumentType2 → Html. We will now write a small page that just contains a list of natural numbers up to a given n.

> numbers :: Int -> Html

Note how these templates are pure. It is therefore not recommended to mix them with IO code, or complicated control paths, generally – you should separate your “view” code from your “logic” code – but you already knew that, right?

> numbers n = docTypeHtml $ do
>     H.head $ do
>         H.title "Natural numbers"
>     body $ do
>         p "A list of natural numbers:"
>         ul $ forM_ [1 .. n] (li . toHtml)

We use the docTypeHtml combinator which is basically the doctype followed by the <html> tag.


We also provide combinators to set attributes on elements. Attribute setting is done using the ! operator.

> simpleImage :: Html
> simpleImage = img ! src "foo.png"

Oh, wait! Shouldn’t images have an alternate text attribute as well, according to the recommendations?

> image :: Html
> image = img ! src "foo.png" ! alt "A foo image."

As you can see, you can chain multiple arguments using the ! operator. Setting an attribute on an element with context also uses the ! operator:

> parentAttributes :: Html
> parentAttributes = p ! class_ "styled" $ em "Context here."

As expected, the attribute will only be added to the <p> tag, and not to the <em> tag. This is an alternative definition, equivalent to parentAttributes, but arguably less readable:

> altParentAttributes :: Html
> altParentAttributes = (p $ em "Context here.") ! class_ "styled"

Nesting & composing

It is very common to nest, compose and combine multiple templates, snippets or partials – use whatever terminology you prefer here. Again, a small example. Say we have a simple datastructure:

> data User = User
>     { getUserName :: String
>     , getPoints   :: Int
>     }

If the user is logged in, we want to have a snippet that displays the user’s current status.

> userInfo :: Maybe User -> Html
> userInfo u = H.div ! A.id "user-info" $ case u of
>     Nothing ->
>         a ! href "/login" $ "Please login."
>     Just user -> do
>         "Logged in as "
>         toHtml $ getUserName user
>         ". Your points: "
>         toHtml $ getPoints user

Once we have this, we can easily embed it somewhere else.

> somePage :: Maybe User -> Html
> somePage u = html $ do
>     H.head $ do
>         H.title "Some page."
>     body $ do
>         userInfo u
>         "The rest of the page."

In the previous example, the user would probably be pulled out of a reader monad instead of given as an argument in a realistic application.

Getting the goods

Now that we have constructed a value of of the type Html, we need to do something with it, right? You can extract your data using the renderHtml function which has the type signature Html → L.ByteString. This function can be found in the Text.Blaze.Renderer.Utf8 module.

A lazy ByteString is basically a list of byte chunks. The list of byte chunks the renderHtml is your HTML page, encoded in UTF-8. Furthermore, all chunks will be nicely-sized, so the overhead is minimal.

There are other renderers as well – for example there is a prettifying renderer called Text.Blaze.Renderer.Pretty, and if you just want a String, use Text.Blaze.Renderer.String.

The blaze-from-html tool

There is also a tool called blaze-from-html which is used to convert HTML pages to Haskell code using the BlazeHtml library. It needs to be installed seperately using cabal:

[jasper@alice ~]$ cabal install blaze-from-html

The reason that we use a seperate package is because we want to keep the set of blaze-html dependencies small. blaze-from-html usage is pretty straightforward. An example:

[jasper@alice ~]$ curl -S http://jaspervdj.be/blaze | blaze-from-html

will output the Haskell code that would be needed to produce this page. By default, blaze-from-html will use HTML5. You can use other variants as well:

[jasper@alice ~]$ blaze-from-html -v html4-transitional index.html

To include the imports as well, use the -s flag – this will give you a piece of standalone code that can be compiled directly. The -e flag causes blaze-from-html to ignore a lot of errors, which might come in handy if the page you are trying to convert has some faults in it.

Further examples

This tutorial should have given you a good idea of how BlazeHtml works. We have also provided some more real-world examples, you can find them all in this directory on github.

Go forth, and generate some HTML!