src/Pure/General/xml.scala
author wenzelm
Fri, 22 Aug 2008 21:25:19 +0200
changeset 27947 b6dc0a396857
parent 27942 5ac9a0f9fad0
child 27948 2638b611d3ce
permissions -rw-r--r--
tuned comments; added document object model (DOM);
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
27931
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
     1
/*  Title:      Pure/General/xml.scala
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
     2
    ID:         $Id$
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
     3
    Author:     Makarius
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
     4
27947
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
     5
Simple XML tree values.
27931
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
     6
*/
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
     7
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
     8
package isabelle
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
     9
27947
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    10
import org.w3c.dom.{Node, Document}
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    11
import javax.xml.parsers.DocumentBuilderFactory
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    12
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    13
27931
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
    14
object XML {
27947
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    15
  /* datatype representation */
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    16
27931
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
    17
  type Attributes = List[(String, String)]
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
    18
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
    19
  abstract class Tree
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
    20
  case class Elem(name: String, attributes: Attributes, body: List[Tree]) extends Tree
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
    21
  case class Text(content: String) extends Tree
27941
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    22
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    23
27942
5ac9a0f9fad0 tuned comment;
wenzelm
parents: 27941
diff changeset
    24
  /* iterate over content */
27941
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    25
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    26
  private type State = Option[(String, List[Tree])]
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    27
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    28
  private def get_next(tree: Tree): State = tree match {
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    29
    case Elem(_, _, body) => get_nexts(body)
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    30
    case Text(content) => Some(content, Nil)
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    31
  }
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    32
  private def get_nexts(trees: List[Tree]): State = trees match {
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    33
    case Nil => None
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    34
    case t :: ts => get_next(t) match {
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    35
      case None => get_nexts(ts)
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    36
      case Some((s, r)) => Some((s, r ::: ts))
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    37
    }
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    38
  }
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    39
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    40
  def content(tree: Tree) = new Iterator[String] {
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    41
    private var state = get_next(tree)
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    42
    def hasNext() = state.isDefined
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    43
    def next() = state match {
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    44
      case Some((s, rest)) => { state = get_nexts(rest); s }
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    45
      case None => throw new NoSuchElementException("next on empty iterator")
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    46
    }
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    47
  }
b4656b671cce added iterator over content;
wenzelm
parents: 27931
diff changeset
    48
27947
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    49
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    50
  /* document object model (DOM) */
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    51
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    52
  def DOM(tree: Tree) = {
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    53
    val doc = DocumentBuilderFactory.newInstance.newDocumentBuilder.newDocument
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    54
    def dom_tree(tr: Tree): Node = tr match {
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    55
      case Elem(name, atts, ts) => {
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    56
        val node = doc.createElement(name)
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    57
        for ((name, value) <- atts) node.setAttribute(name, value)
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    58
        for (t <- ts) node.appendChild(dom_tree(t))
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    59
        node
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    60
      }
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    61
      case Text(txt) => doc.createTextNode(txt)
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    62
    }
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    63
    val root_elem = tree match {
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    64
      case Elem(_, _, _) => dom_tree(tree)
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    65
      case Text(_) => dom_tree(Elem("root", Nil, List(tree)))
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    66
    }
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    67
    doc.appendChild(root_elem)
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    68
    doc
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    69
  }
b6dc0a396857 tuned comments;
wenzelm
parents: 27942
diff changeset
    70
27931
b533a9de87a7 Minimalistic XML tree values.
wenzelm
parents:
diff changeset
    71
}