Don't use this, it's slow, broken, and incomplete.

hautobot is a Haskell approximation of Christophe Grand's Enlive (https://github.com/cgrand/enlive).

It depends on the Text.XmlHtml package (http://hackage.haskell.org/package/xmlhtml-0.1.5.2), which it uses for its internal datastructures and lower-level APIs.

Apologies for the name.

Details

It is likely terribly inefficient; currently each transform results in a full traversal of the document tree (although this can be somewhat mitigated by use of the 'at' function to traverse only sub-trees.)

Selectors are currently pretty basic; a selector is just a list of lists of predicates. All predicates in a step (inner list) must match, and multiple steps represent nesting/descendants. For example:

[[hasTag "section", hasId "main"], [hasTag "article", hasClass "story]]

…would match all <article class="story"> inside <section id="main">. In future this library will support a string-based CSS-like selector syntax, so you could do "section#main article.story" for the above. Obviously this would be a marked improvement.

Transforms are (list-monadic) functions of Node -> [Node], and can be composed with >=> and <=<.

Use

Look at testDoc and doTest at the bottom of the main source file. In brief:

let document = parseHTML "a name" htmlAsByteString
  selector = [[hasTag "body"], [hasTag "article"]]
  transFun = addClass "classname" >=> 
             append [TextNode (T.pack "Some Content")] >=> 
             setAttr "data-foo" "bar"
in render $
  -- like this
  transform selector transFun document
  -- or like this
  runTransforms [(selector, transFun), …] document

License

Copyright (c) 2011, Matthew Wilson.

Distributed under the BSD-3 License.