Pure version of parseM.

Run a ParserT on an XML fragment body.

Notice that this function doesn't enforce that all input is consumed. If you want that behavior, then please use pEndOfInput in the given ParserT.

Parser a is a type synonym for Parser Identity a.

XML parser for a value of type a.

This parser runs on top of some Monad m, making ParserT a suitable monad transformer.

You can build a ParserT using pElement, pAnyElement, pName, pAttr, pAttrs, pChildren, pText, pEndOfInput, any of the Applicative, Alternative or Monad combinators, or you can use parserT directly.

Run a ParserT using parse, parseM or runParserT

parserT is the most general way or building a ParserT.

Notice that ParserState's internals are not exported, so you won't be able to do much with it other than pass it around.

runParserT . parserT  ==  id

runParserT is the most general way or running a ParserT.

As a simpler alternative to runParserT, consider using parseM, or even parse if you don't need transformer functionality.

Notice that ParserState's internals are not exported, so you won't be able to do much with it other than pass it around.

runParserT . parserT  ==  id

Internal parser state.

Construct an initial ParserState to use with runParserT from zero or more top-level Nodes.

pElement "foo" p runs a ParserT p inside a Element node named "foo". This parser fails if such element does not exist at the current position.

Leading whitespace is ignored. If you need to preserve that whitespace for some reason, capture it using pText before using pElement.

Consumes the matched element from the parser state.

pAnyElement p runs a ParserT p inside the Element node at the current position, if any. Otherwise, if no such element exists, this parser fails.

You can recover the name of the matched element using pName inside the given ParserT. However, if you already know beforehand the name of the element that you want to match, it's better to use pElement rather than pAnyElement.

Leading whitespace is ignored. If you need to preserve that whitespace for some reason, capture it using pText before using pAnyElement.

Consumes the matched element from the parser state.

Returns the name of the currently selected Element.

This parser fails if there's no currently selected Element (see pElement, pAnyElement).

Doesn't modify the parser state.

Return the value of the requested attribute, if defined, as strict Text. Returns an empty strict Text in case the attribute is defined but no value was given to it.

This parser fails if there's no currently selected Element (see pElement, pAnyElement).

Consumes the matched attribute from the parser state.

Returns all of the available element attributes.

Returns empty strict Text as values in case an attribute is defined but no value was given to it.

This parser fails if there's no currently selected Element (see pElement, pAnyElement).

Consumes all the attributes for this element from the parser state.

Returns all of the immediate children of the current element.

If parsing top-level nodes rather than a particular element (that is, if pChildren is not being run inside pElement), then all of the top level Nodes will be returned.

Consumes all the returned nodes from the parser state.

Returns the contents of a text node as a strict Text.

Surrounidng whitespace is not removed, as it is considered to be part of the text node.

If there is no text node at the current position, then this parser fails. This implies that pText never returns an empty strict Text, since there is no such thing as a text node without text.

Please note that consecutive text nodes are always concatenated and returned together.

parse pText (text "Ha" <> text "sk" <> text "ell")
    == Right (text Haskell)

Consumes the text from the parser state. This implies that if you perform two consecutive pText calls, the second will always fail.

parse (pText >> pText) (text "Ha" <> text "sk" <> text "ell")
    == Left "Missing text node"

Like pText, but returns a lazy Text.

Succeeds if all of the elements, attributes and text nodes have been consumed.

Encodes a list of XML Nodes, representing an XML fragment body, to an UTF8-encoded and XML-escaped bytestring.

This function doesn't render self-closing elements. Instead, all elements have a corresponding closing tag.

Also, it doesn't render CDATA sections. Instead, all text is escaped as necessary.

Element attributes are rendered in alphabetical order.

Either a text or an element node in an XML fragment body.

Construct with text, textLazy or element.

Destruct with Text, TextLazy or Element.

Case analysis for a Node.

Destruct an element Node.

case n :: Node of
  Element t as cs -> ...
  _ -> ...

Construct a XML fragment body containing a single Element Node, if possible.

This function will return empty list if it is not possible to construct the Element with the given input. To learn more about why it was not possible to construct it, use element instead.

Using element' rather than element is recommended, so that you are forced to acknowledge a failing situation in case it happens. However, element is at times more convenient to use, whenever you know the input is valid.

Construct an Element Node.

Returns Left if the Element Node can't be created, with an explanation of why.

Destruct a text Node into a strict Text.

case n :: Node of
  Text t -> ...
  _ -> ...

Construct a XML fragment body containing a single text Node, if given Text not empty.

This function will return empty list if it is not possible to construct the Text with the given input. To learn more about why it was not possible to construct it, use text' instead.

Using text' rather than text is recommended, so that you are forced to acknowledge a failing situation in case it happens. However, text is at times more convenient to use. For example, when you know statically the input is valid.

Construct a text Node, if given Text not empty.

Returns Left if the Text Node can't be created, with an explanation of why.

Destruct a text Node into a lazy Text.

case n :: Node of
  TextLazy tl -> ...
  _ -> ...

A version of text working with lazy Text.

A version of text' working with lazy Text.

Post-order depth-first replacement of Node and all of its children.

This function works like fix, but the given function is trying to find a fixpoint for the individual children nodes, not for the root node.

For example, the following function renames every node named "w" to "y", and every node named "y" to "z". It accomplishes this by first renaming "w" nodes to "x", and then, by using k recursively to further rename all "x" nodes (including the ones that were just created) to "y" in a post-order depth-first manner. After renaming an "x" node to "y", the recursion stops (i.e., k is not used), so our new "y" nodes won't be further renamed to "z". However, nodes that were named "y" initially will be renamed to "z".

In our example we only replace one node with another, but a node can be replaced with zero or more nodes, depending on the length of the resulting list.

foo :: Node -> [Node]
foo = dfpos $ \k -> \case
    Element "w" as cs -> element "x" as cs >>= k
    Element "x" as cs -> element "y" as cs
    Element "y" as cs -> element "z" as cs >>= k

See dfpre for pre-orderd depth-first replacement.

WARNING If you call k in every branch, then dfpos will never terminate. Make sure the recursion stops at some point by simply returning a list of nodes instead of calling k.

Monadic version of dfpos.

Pre-order depth-first replacement of Node and all of its children.

This is just like dfpos but the search proceeds in a different order.

Monadic version of dfpre.