lucid-2.8.1: Clear to write, read and edit DSL for HTML

Safe HaskellNone
LanguageHaskell98

Lucid

Contents

Description

Clear to write, read and edit DSL for writing HTML

See Lucid.Html5 for a complete list of Html5 combinators. That module is re-exported from this module for your convenience.

See Lucid.Base for lower level functions like makeElement, makeAttribute, termRaw, etc.

Synopsis

Intro

HTML terms in Lucid are written with a postfix ‘_’ to indicate data rather than code. Some examples:

p_, class_, table_, style_

Note: If you're testing in the REPL you need to add a type annotation to indicate that you want HTML. In normal code your top-level declaration signatures handle that.

Plain text is written using the OverloadedStrings and ExtendedDefaultRules extensions, and is automatically escaped:

>>> "123 < 456" :: Html ()
123 &lt; 456

Except some elements, like script_:

>>> script_ "alert('Hello!' > 12)" :: Html ()
<script>alert('Hello!' > 12)</script>

Elements nest by function application:

>>> table_ (tr_ (td_ (p_ "Hello, World!"))) :: Html ()
<table><tr><td><p>Hello, World!</p></td></tr></table>

Elements are juxtaposed via monoidal append:

>>> p_ "hello" <> p_ "sup" :: Html ()
<p>hello</p><p>sup</p>

Or monadic sequencing:

>>> div_ (do p_ "hello"; p_ "sup") :: Html ()
<div><p>hello</p><p>sup</p></div>

Attributes are set by providing an argument list:

>>> p_ [class_ "brand"] "Lucid Inc" :: Html ()
<p class="brand">Lucid Inc</p>
>>> p_ [data_ "zot" "foo",checked_] "Go!" :: Html ()
<p data-zot="foo" checked>go</p>

Attribute and element terms are not conflicting:

>>> style_ [style_ "inception"] "Go deeper." :: Html ()
<style style="inception">Go deeper.</style>

Here is a fuller example of Lucid:

table_ [rows_ "2"]
       (tr_ (do td_ [class_ "top",colspan_ "2",style_ "color:red"]
                    (p_ "Hello, attributes!")
                td_ "yay!"))

Elements (and some attributes) are variadic and overloaded, see the Term class for more explanation about that.

For proper rendering you can easily run some HTML immediately with:

>>> renderText (p_ "Hello!")
> "<p>Hello!</p>"
>>> renderBS (p_ [style_ "color:red"] "Hello!")
"<p style=\"color:red\">Hello!</p>"

For ease of use in GHCi, there is a Show instance, as demonstrated above.

renderText :: Html a -> Text

Render the HTML to a lazy Text.

This is a convenience function defined in terms of execHtmlT, runIdentity and toLazyByteString, and decodeUtf8. Check the source if you're interested in the lower-level behaviour.

renderBS :: Html a -> ByteString

Render the HTML to a lazy ByteString.

This is a convenience function defined in terms of execHtmlT, runIdentity and toLazyByteString. Check the source if you're interested in the lower-level behaviour.

renderTextT :: Monad m => HtmlT m a -> m Text

Render the HTML to a lazy Text, but in a monad.

This is a convenience function defined in terms of execHtmlT and toLazyByteString, and decodeUtf8. Check the source if you're interested in the lower-level behaviour.

renderBST :: Monad m => HtmlT m a -> m ByteString

Render the HTML to a lazy ByteString, but in a monad.

This is a convenience function defined in terms of execHtmlT and toLazyByteString. Check the source if you're interested in the lower-level behaviour.

renderToFile :: FilePath -> Html a -> IO ()

Render the HTML to a lazy ByteString.

This is a convenience function defined in terms of execHtmlT, runIdentity and toLazyByteString. Check the source if you're interested in the lower-level behaviour.

Running

If the above rendering functions aren't suited for your purpose, you can run the monad directly and use the more low-level blaze Builder, which has a plethora of output modes in Blaze.ByteString.Builder.

execHtmlT

Arguments

:: Monad m 
=> HtmlT m a

The HTML to generate.

-> m Builder

The a is discarded.

Build the HTML. Analogous to execState.

You might want to use this is if you want to do something with the raw Builder. Otherwise for simple cases you can just use renderText or renderBS.

evalHtmlT

Arguments

:: Monad m 
=> HtmlT m a

HTML monad to evaluate.

-> m a

Ignore the HTML output and just return the value.

Evaluate the HTML to its return value. Analogous to evalState.

Use this if you want to ignore the HTML output of an action completely and just get the result.

For using with the Html type, you'll need runIdentity e.g.

>>> runIdentity (evalHtmlT (p_ "Hello!"))
()

runHtmlT :: HtmlT m a -> m (HashMap Text Text -> Builder -> Builder, a)

This is the low-level way to run the HTML transformer, finally returning an element builder and a value. You can pass mempty for both arguments for a top-level call. See evalHtmlT and execHtmlT for easier to use functions.

Types

type Html = HtmlT Identity

Simple HTML builder type. Defined in terms of HtmlT. Check out that type for instance information.

Simple use-cases will just use this type. But if you want to transformer over Reader or something, you can go and use HtmlT.

data HtmlT m a

A monad transformer that generates HTML. Use the simpler Html type if you don't want to transform over some other monad.

Instances

MonadTrans HtmlT

Used for lift.

(Monad m, (~) * a ()) => TermRaw Text (HtmlT m a)

Given children immediately, just use that and expect no attributes.

Monad m => Monad (HtmlT m)

Basically acts like Writer.

Monad m => Functor (HtmlT m)

Just re-uses Monad.

Monad m => Applicative (HtmlT m)

Based on the monad instance.

MonadIO m => MonadIO (HtmlT m)

If you want to use IO in your HTML generation.

(Monad m, ToHtml f, (~) * a ()) => TermRaw [Attribute] (f -> HtmlT m a)

Given attributes, expect more child input.

(Monad m, (~) * f (HtmlT m a)) => Term [Attribute] (f -> HtmlT m a)

Given attributes, expect more child input.

(~) (* -> *) m Identity => Show (HtmlT m a)

Just calls renderText.

(Monad m, (~) * a ()) => IsString (HtmlT m a)

We pack it via string. Could possibly encode straight into a builder. That might be faster.

((~) * a (), Monad m) => Monoid (HtmlT m a)

Monoid is right-associative, a la the Builder in it.

Monad m => With (HtmlT m a -> HtmlT m a)

For the contentful elements: div_

Monad m => With (HtmlT m a)

For the contentless elements: br_

Monad m => Term (HtmlT m a) (HtmlT m a)

Given children immediately, just use that and expect no attributes.

data Attribute

A simple attribute. Don't use the constructor, use makeAttribute.

Instances

Eq Attribute 
Show Attribute 
TermRaw Text Attribute

Some termRaws (like style_, title_) can be used for attributes as well as elements.

Term Text Attribute

Some terms (like style_, title_) can be used for attributes as well as elements.

(Monad m, ToHtml f, (~) * a ()) => TermRaw [Attribute] (f -> HtmlT m a)

Given attributes, expect more child input.

(Monad m, (~) * f (HtmlT m a)) => Term [Attribute] (f -> HtmlT m a)

Given attributes, expect more child input.

Classes

To support convenient use of HTML terms, HTML terms are overloaded. Here are the following types possible for an element term accepting attributes and/or children:

p_ :: Term arg result => arg -> result
p_ :: Monad m => [Attribute] -> HtmlT m () -> HtmlT m ()
p_ :: Monad m => HtmlT m () -> HtmlT m ()

The first is the generic form. The latter two are the possible types for an element.

Elements that accept no content are always concrete:

input_ :: Monad m => [Attribute] -> HtmlT m ()

And some attributes share the same name as attributes, so you can also overload them as attributes:

style_ :: TermRaw arg result => arg -> result
style_ :: Monad m => [Attribute] -> Text -> HtmlT m ()
style_ :: Monad m => Text -> HtmlT m ()
style_ :: Text -> Attribute

class Term arg result | result -> arg where

Used to construct HTML terms.

Simplest use: p_ = term "p" yields p_.

Very overloaded for three cases:

  • The first case is the basic arg of [(Text,Text)] which will return a function that wants children.
  • The second is an arg which is HtmlT m (), in which case the term accepts no attributes and just the children are used for the element.
  • Finally, this is also used for overloaded attributes, like style_ or title_. If a return type of (Text,Text) is inferred then an attribute will be made.

The instances look intimidating but actually the constraints make it very general so that type inference works well even in the presence of things like OverloadedLists and such.

Minimal complete definition

termWith

Methods

term

Arguments

:: Text

Name of the element or attribute.

-> arg

Either an attribute list or children.

-> result

Result: either an element or an attribute.

Used for constructing elements e.g. term "p" yields p_.

termWith

Arguments

:: Text

Name.

-> [Attribute]

Attribute transformer.

-> arg

Some argument.

-> result

Result: either an element or an attribute.

Use this if you want to make an element which inserts some pre-prepared attributes into the element.

Instances

Term Text Attribute

Some terms (like style_, title_) can be used for attributes as well as elements.

(Monad m, (~) * f (HtmlT m a)) => Term [Attribute] (f -> HtmlT m a)

Given attributes, expect more child input.

Monad m => Term (HtmlT m a) (HtmlT m a)

Given children immediately, just use that and expect no attributes.

class ToHtml a where

Can be converted to HTML.

Methods

toHtml :: Monad m => a -> HtmlT m ()

toHtmlRaw :: Monad m => a -> HtmlT m ()

Instances

ToHtml String 
ToHtml Text 
ToHtml Text 

class With a where

With an element use these attributes. An overloaded way of adding attributes either to an element accepting attributes-and-children or one that just accepts attributes. See the two instances.

Methods

with

Arguments

:: a

Some element, either Html a or Html a -> Html a.

-> [Attribute] 
-> a 

With the given element(s), use the given attributes.

Instances

Monad m => With (HtmlT m a -> HtmlT m a)

For the contentful elements: div_

Monad m => With (HtmlT m a)

For the contentless elements: br_

Re-exports