I'm having a little trouble with HXT: I am trying to locate all the nodes in a document that match some criteria, and I'm trying to combine the lenses/XPaths as predicates in an OR-like fashion, using Control.Arrow.<+>, as this guide suggests. However, when I try to "run" the arrow on my document, I am getting duplicate results. Is there an easy way to remove the duplicates, or to combine the tests in a more meaningful way? Here is my code:
run :: App -> IO ()
run a = do
inputContents <- readFile (input a)
let doc = readString [withParseHTML yes, withWarnings no] inputContents
links <- runX . xshow $ doc >>> indentDoc //> cssLinks
mapM_ putStrLn links
cssLinks = links >>> (rels <+> hrefs <+> types)
where
links = hasName "link"
rels = hasAttrValue "rel" (isInfixOf "stylesheet")
hrefs = hasAttrValue "href" (endswith ".css")
types = hasAttrValue "type" (== "text/css")
Yet every time I run this (on any web page), I get duplicated results / nodes. I noticed that <+> is part of the ArrowPlus typeclass, which mimics a monoid, and ArrowXML is an instance of both ArrowList and ArrowTree, which gives me a lot to work with. Would I have to construct ArrowIf predicates? Any help with this would be wonderful :)
You may get the arrow result as a [XmlTree], then apply List.nub, then get the string rep.
import "hxt" Text.XML.HXT.DOM.ShowXml as SX
...
links <- runX $ doc >>> indentDoc //> cssLinks
-- first remove duplicates (List.nub) then apply SX.xshow
putStrLn (SX.xshow . L.nub $ links)