ansaurus

Question

How do I parse xml into "messages" and print them out in scala using stream parsing?

Answer 1

+4 A:

This is for 2.8.

The typical way to process events is to use a match statement. In my case, i always had the need to store the parents as I process elements (to know for instance in what tag the text is located):

import scala.xml.pull._
import scala.io.Source
import scala.collection.mutable.Stack

val src = Source.fromString(xml)
val er = new XMLEventReader(src)
val stack = Stack[XMLEvent]()
def iprintln(s:String) = println((" " * stack.size) + s.trim)
while (er.hasNext) {
  er.next match {
    case x @ EvElemStart(_, label, _, _) =>
      stack push x
      iprintln("got <" + label + " ...>")
    case EvElemEnd(_, label) => 
      iprintln("got </" + label + ">")
      stack pop;
    case EvText(text) => 
      iprintln(text) 
    case EvEntityRef(entity) => 
      iprintln(entity) 
    case _ => // ignore everything else
  }
}

Because entity are events, you will probably need to convert to text and combine them with the surrounding text.

In the example above I only used label, but you can also use EvElemStart(pre, label, attrs, scope) to extract more stuff and you can add an if guard to match for complex conditions.

Also if you're using 2.7.x, I don't know if http://lampsvn.epfl.ch/trac/scala/ticket/2583 was back-ported so, you may have issues to process text with entities.

More to the point, just dealing with from and to for brevity (though I would not call that the Scala way):

class Message() {
  var to:String = _
  var from:String = _
  override def toString(): String = 
    "from %s to %s".format(from, to)
}

var message:Message = _
var sb:StringBuilder = _

while (er.hasNext) {
  er.next match {
    case x @ EvElemStart(_, "message", _, _) =>
      message = new Message
    case x @ EvElemStart(_, label, _, _) if
        List("to", "from") contains label =>
      sb = new StringBuilder 
    case EvElemEnd(_, "to") => 
      message.to = sb.toString
    case EvElemEnd(_, "from") => 
      message.from = sb.toString
      sb = new StringBuilder 
    case EvElemEnd(_, "message") => 
      println(message)
    case EvText(text) if sb != null => 
      sb ++= text
    case EvEntityRef(entity) => 
      sb ++= unquote(entity) // todo
    case _ => // ignore everything else
  }
}

huynhjl 2010-05-28 15:07:21

This is helpful, but still not quite what I'm looking for. I might be able to get something working using this.

ScArcher2 2010-05-28 15:39:30

I'm trying to discover the "scala" way to build out a "message" object, or possibly even just a tuple and then print that out all at once instead of just printing when i encounter each tag and text element.

ScArcher2 2010-05-28 15:44:15

Thank you for the example! This is exactly the information I was looking for. I am still interested in the "scala" way of doing it if there is a more functional way. Thanks again!

ScArcher2 2010-05-28 18:07:26

ansaurus

tags:

views:

answers:

How do I parse xml into "messages" and print them out in scala using stream parsing?

related questions