haskell: xml filtering a subtree

Question

I am struggling to remove a Element with all its children with haskell. The task is to strip all table-tags from a given xml document (maybe I have not understood the concept of a cursor or it is something else I am missing).

I have tried three different approaches:

• lenses with traversing/filtering and setting the filtered value with a new element - here only the tag is replaced but not the contents
• accessing the table element with a cursor - resetting the content there and obtaining the document root again by traversing the cursor up to the document root - nothing is filtered
• filtering the children of the document root recursively - nothing is filtered

Tools

• xml-conduit
• xml-lens
• ghc-8.0.1

Input (test.xml)/output

 INPUT                                     EXPECTED OUTPUT (for the filtered cases)
<?xml version="1.0"?>                 |  <?xml version="1.0"?>                   
<root>                                |  <root>                                  
    <a>                               |      <a>                                 
        ...                           |          ...                             
    </a>                              |      </a>                                
    <b>                               |      <b>                                 
        <table>                       |          <bb>                            
            <!--table entries-->      |              ...                         
        </table>                      |          </bb>                           
        <bb>                          |      </b>                                
            ...                       |      <c>                                 
        </bb>                         |          <cc>                            
    </b>                              |              ...                         
    <c>                               |          </cc>                           
        <cc>                          |      </c>                                
            ...                       |  </root>                                 
        </cc>
    </c>
</root>

Minimal-non-working-example

{-# LANGUAGE OverloadedStrings #-}

module Minimal where

import           Control.Lens
import           Data.Conduit.Text as CT
import           Data.Default
import qualified Data.Text.Lazy.IO as TIO
import           Text.XML
import           Text.XML.Cursor
import qualified Text.XML.Lens     as L
import           Data.Maybe (isNothing, isJust)

main :: IO ()
main = do test <- Text.XML.readFile def "./test.xml"
          pput $ filterDocument test

          let cursor = fromDocument test

          pput $ docUp $ elemUp $ getRoot ((head $ cursor $// checkName (== "table")) {child = []} )

          pput $ docUp $ elemUp $ filterChildren (checkName (/= "table")) cursor
          return ()


filterChildren :: Axis -> Cursor -> Cursor
filterChildren pred c = c {child = map (filterChildren pred) (c $/ pred) }

filterDocument :: Document -> Document
filterDocument doc = doc & (L.root.L.entire.filtered (\e -> isJust $ e^?L.named "table") .~ emptyElemt)
  where emptyElemt = Element "empty" mempty []

-- helper functions --

docUp :: Element -> Document
docUp e = Document {documentRoot = e, documentPrologue = Prologue [] Nothing [], documentEpilogue = [] }

elemUp :: Cursor -> Element
elemUp cursor = Element {elementName = "DOC", elementAttributes = mempty , elementNodes = [node cursor]}

elemUp' :: [Cursor] -> Element
elemUp' cursors = Element {elementName = "DOC", elementAttributes = mempty , elementNodes = map node cursors}

getRoot :: Cursor -> Cursor
getRoot c = let p = (c $| parent)
            in if null p then c else getRoot $ head p

pput :: Document -> IO ()
pput = TIO.putStrLn . renderText pretty
  where pretty = def {rsPretty = True}

Output

> stack ghci
. . .
Ok, modules loaded: Minimal.
λ > main
<?xml version="1.0" encoding="UTF-8"?>
<root>
    <a>
        ...
    </a>
    <b>
        <empty>
            <!-- table entries -->
        </empty>
        <bb>
            ...
        </bb>
    </b>
    <c>
        <cc>
            ...
        </cc>
    </c>
</root>

<?xml version="1.0" encoding="UTF-8"?>
<DOC>
    <root>
        <a>
            ...
        </a>
        <b>
            <table>
                <!-- table entries -->
            </table>
            <bb>
                ...
            </bb>
        </b>
        <c>
            <cc>
                ...
            </cc>
        </c>
    </root>
</DOC>

<?xml version="1.0" encoding="UTF-8"?>
<DOC>
    <root>
        <a>
            ...
        </a>
        <b>
            <table>
                <!-- table entries -->
            </table>
            <bb>
                ...
            </bb>
        </b>
        <c>
            <cc>
                ...
            </cc>
        </c>
    </root>
</DOC>

Answer 1

This code seems to do what you want, based on xml-conduit. I started from the yesod web book example and implemented the transformation by a simple recursive function.

{-# LANGUAGE OverloadedStrings #-}
import qualified Data.Map        as M
import           Prelude         hiding (readFile, writeFile)
import           Text.XML

main :: IO ()
main = do
    Document prologue root epilogue <- readFile def "test.xml"

    let root' = transform root

    writeFile def
        { rsPretty = True
        } "output.html" $ Document prologue root' epilogue

transform :: Element -> Element
transform (Element _name attrs children) = 
  Element _name attrs (filterChildren children)

filterChildren :: [Node] -> [Node]
filterChildren = concatMap kickTable
  where
    kickTable :: Node -> [Node]
    kickTable (NodeElement (Element "table" attrs children)) = -- Drop it
      [  ]
    kickTable (NodeElement (Element n attrs children)) = -- Recurse on
      [ NodeElement (Element n attrs (filterChildren children)) ]
    kickTable n = -- ok - whatever
      [ n ]

My lens-foo is not strong enough to tell why your solution does not work, but from the docs - you have to be careful with filtered not to violate the traversal laws, although I don't know what happens when you violate them.

Hope that helps.

Answer 2

I don't know about Text.XML, but here is a solution with Text.XML.Light :

module Minimal where

import Data.Maybe(catMaybes)
import Text.XML.Light.Input
import Text.XML.Light.Output
import Text.XML.Light.Types

main :: IO ()
main = do
  test <- parseXML <$> readFile "./test.xml"
  mapM_ (putStrLn . ppContent) . catMaybes $ map cutTables test

cutTables :: Content -> Maybe Content
cutTables (Elem e) = if qName (elName e) == "table" then Nothing else
  Just . Elem $ e { elContent = catMaybes . map cutTables $ elContent e }
cutTables x = Just x