XML parser reverses attribute order

[scabug]

When parsing XML and then writing it out again, the order of attributes is reversed.

This prevents certain use cases where one wants to modify an XML file and have a minimal amount of changes. E.g. adding a new element to a file causes all existing attributes to get reversed.

The cause is here:

https://github.com/scala/scala/blob/81e3121eb09010375783d13e8c4c853686a34ffd/src/library/scala/xml/parsing/MarkupParser.scala#L302

aMap: MetaData is built by prepending new attributes to the linked list, causing the reversal.

The attached patch adds a failing test case. (It failed when I wrote it originally, now I can't test it because the build fails).

[scabug]

Imported From: https://issues.scala-lang.org/browse/SI-6341?orig=1
Reporter: robinst
Affected Versions: 2.10.0-M7
Attachments:

• xml-attribute-order-failing-test.patch (created on Sep 8, 2012 6:13:04 PM UTC, 922 bytes)

[scabug]

@som-snytt said:
This is a duplicate of #3441 and #2735, except that there are a few different parsers at issue: the compiler's MarkupParser for literals, the library's MarkupParser, and the utility methods that use xerces, IIUC.

  def main(args: Array[String]): Unit = {
    val input = """<elem one="test" two="test2" three="test3"></elem>"""
    val doc = XML.loadString(input)
    println(doc)
    
    val cpa = ConstructingParser.fromSource(io.Source.fromString(input), preserveWS = true)
    val cpadoc = cpa.document()
    val ppr = new PrettyPrinter(80,5)
    println(ppr.format(cpadoc.docElem))
  }

I remember this issue from the first Scala code I ever wrote, but I haven't used XML since then, so bear with me. My solution was a hook in PrettyPrinter for formatting attributes:

      // in leafTag and startTag; but also Utility.toXML/serialize
      //n.attributes buildString sb
      formatAttributes(n, sb)

I realize now this was my first fold ever:

  override protected def formatAttributes(n: Node, sb: StringBuilder) {
    //n.attributes.toList.reverse.foldLeft(sb) { (a,b) => a.append(' '); b.toString1(a); a }
    (sb /: n.attributes.toList.reverse) { (r,v) => r.append(' '); v.toString1(r); r }
  }

And maybe the last time I used /: except as an emoticon. I see toString1 is protected now, so cut/paste that, too. I don't know if such a formatAttributes hook is worth it?

[scabug]

robinst said:
Yes, this is about the library's MarkupParser, see description. #2735 was about the compiler's literal parser, so this is not a duplicate. But it's a similar issue.

The above is merely a workaround, but thanks anyway.

It should really be fixed so that round-tripping does not change things around needlessly.

[scabug]

@SethTisue said:
The scala-xml library is now community-maintained. Issues with it are now tracked at https://github.com/scala/scala-xml/issues instead of here in the Scala JIRA.

Interested community members: if you consider this issue significant, feel free to open a new issue for it on GitHub, with links in both directions.

[scabug]

Graeme Phillipson (cogpp) said:
I have raised a copy of this in the new space as scala/scala-xml#65