Manipulating Documents: The Document Object Model (DOM)

This chapter is from the book

XML Primer Plus

Learn More Buy

This chapter is from the book

This chapter is from the book 

XML Primer Plus

Learn More Buy

Creating New Content

At this point we've read and modified simple information such as attributes and text nodes. Modifying the content of an element is a bit more complex, and requires adding or removing nodes.

In this section, we'll start by creating a new person element, giving it an attribute, and adding it to the persons element for each activity. We'll then create a new document and populate it with the participant information for each activity.

Finally, we'll look at some of the ways that current implementations persist (save) a Document object.

Creating an Element

Before you can add an element to a document, you must first create it. Creating elements and other nodes is the responsibility of the Document object.

In Listing 3.17, we'll create a method for adding the new person, and execute it before displaying the activity information so the number of available spaces is correct.

Listing 3.17 Creating an Element

...
   NodeList activitiesList =
         activitiesElement.getElementsByTagName("activity");
   for (int i=0; i < activitiesList.getLength(); i++) {

     Node thisNode = activitiesList.item(i);
     addNewParticipant(thisNode, "P4");
     displayActivity(thisNode);
   }
  }
...
  public static void addNewParticipant(Node thisAct, String thisPerson) {

   Element thisActElement = (Element)thisAct;

   Node thisPersons = thisActElement.getElementsByTagName("persons").item(0);
   Element thisPersonsElement = (Element)thisPersons;

   Document doc = thisPersonsElement.getOwnerDocument();

   Element thisPersonElement = doc.createElement("person");
   thisPersonElement.setAttribute("personid", thisPerson);
  }
}

First, we establish the parent node for the new child node. In this case, that's the persons element within each activity element. Next, we need to create the actual element node, which means that we need the Document object. As before, we need to get a reference from the element itself because we're out of the original Document object's scope.

The Document creates the element (and we assign it a personid attribute) but at this moment, the element is in limbo. It's not yet a part of the tree.

Adding the Element

To make a newly created element part of the tree, we have to add it to the structure. There are several ways to do this. The first, and most common, is to simply append it to the content of the parent element, as shown in Listing 3.18.

Listing 3.18 Appending a Child

...
  public static void addNewParticipant(Node thisAct, String thisPerson) {

   Element thisActElement = (Element)thisAct;

   Node thisPersons = thisActElement.getElementsByTagName("persons").item(0);
   Element thisPersonsElement = (Element)thisPersons;

   Document doc = thisPersonsElement.getOwnerDocument();
   Element thisPersonElement = doc.createElement("person");
   thisPersonElement.setAttribute("personid", thisPerson);

   thisPersonsElement.appendChild(thisPersonElement);
  }
}

In this case, we simply add the new element to the content that already exists. Before, the persons element may have looked like the following:

      <persons>
        <person personid="P2"/>
        <person personid="P1"/>
      </persons>

Now it looks like this:

      <persons>
        <person personid="P2"/>
        <person personid="P1"/>
        <person personid="P4"/>
      </persons>

The Node interface also provides two other construction-type methods.

The first, insertBefore(), takes two arguments: the node being added, and the node that it's to precede. So if P1Element represents the person with a personid of P1, the statement

   thisPersonsElement.insertBefore(thisPersonElement, P1Element);

results in a structure of

      <persons>
        <person personid="P2"/>
        <person personid="P4"/>
        <person personid="P1"/>
      </persons>

Note that a node may only appear in a Document once, so if thisPersonElement already exists somewhere else, it's automatically removed, and then placed in its new position.

The second additional method is replaceChild(), where a call to

  Element replacedChild =
        thisPersonsElement.replaceChild(thisPersonElement, P1Element);

results in a structure of

      <persons>
        <person personid="P2"/>
        <person personid="P4"/>
      </persons>

In this case, replacedChild is also equal to P1Element, as replaceChild() returns the node that was replaced.

Creating a New Document

Now we're ready to create a new document. Overall, there's just one new technique: actually creating the document. Again, this is not the standard way of doing things; in the case of the Java implementation, the DocumentBuilder creates a new Document object.

The rest of Listing 3.19 simply shows another use for the techniques we covered earlier in this chapter.

Listing 3.19 Creating the New Document

...
public class ActivityListing {

  public static void println(String arg) {
   System.out.println(arg);
  }

  public static void main (String args[]) {
   File docFile = new File("activities.xml");
   Document doc = null;
   Document newDoc = null;

   try {

     DocumentBuilderFactory dbf = DocumentBuilderFactory.newInstance();
     DocumentBuilder db = dbf.newDocumentBuilder();
     doc = db.parse(docFile);

     newDoc = db.newDocument();

   } catch (Exception e) {
     System.out.println("Problem parsing the file.");
   }

   Element newRoot = newDoc.createElement("roster");
   newDoc.appendChild(newRoot);

   Element root = doc.getDocumentElement();

   Node firstChild = root.getFirstChild();
   Node secondChild = firstChild.getNextSibling();

   Element activitiesElement = (Element)secondChild;
   NodeList activitiesList =
         activitiesElement.getElementsByTagName("activity");
   for (int i=0; i < activitiesList.getLength(); i++) {

     Node thisNode = activitiesList.item(i);
     addNewParticipant(thisNode, "P4");
     displayActivity(thisNode);

     Element thisActivity = (Element)thisNode;
     String thisActivityId = thisActivity.getAttribute("activityid");

     Element newActivity = newDoc.createElement("activity");
     newActivity.setAttribute("activityid", thisActivityId);

     Element personsElement =
        (Element)thisActivity.getElementsByTagName("persons").item(0);
     NodeList participants = personsElement.getElementsByTagName("person");

     for (int j=0; j < participants.getLength(); j++) {
       Element thisPerson = (Element)participants.item(j);
       String thisPersonId = thisPerson.getAttribute("personid");

       Element thisNewPerson = newDoc.createElement("person");
       Node thisPersonText = newDoc.createTextNode(thisPersonId);
       thisNewPerson.appendChild(thisPersonText);

       newActivity.appendChild(thisNewPerson);
     }

     newRoot.appendChild(newActivity);

   }
  }
...

(If you're working in one of the other languages, stay with me for a moment. We'll cover the creation of a document in the next section.)

A document isn't well-formed XML until it has a root element, so next we create a new Element node with the name roster and append it to the document itself.

Now we're ready to start adding new nodes to the document. We simply pull information from the old document just as we did before, then add it to the new document. For example, we retrieve the activityid using the getAttribute() method, then create a new element and add the attribute to it.

The person elements are a bit different, because we're changing the personid from an attribute to the text of the element. Because of this, creating the new person element involves creating a new text node, then appending it as a child to the element. As each person element is completed, append it to the activity.

Finally, append each activity to the new root element.

Saving the Document

The DOM Level 2.0 doesn't specify a means for saving an XML document (though that's planned for Level 3.0), so it's up to the implementation to decide how to do so. The following subsections provide some examples.

Java

The JAXP specification doesn't include a standard way to save documents, but it's recommended that you use the package's XSL transformation abilities. We'll cover transformation in detail in Chapter 10, "Transformations Within an Application." For now, understand that we're transforming a document, but we're not providing a style sheet, so the document comes through unchanged. Listing 3.20a demonstrates the process.

Listing 3.20a Saving the Document

...
import javax.xml.transform.TransformerFactory;
import javax.xml.transform.Transformer;
import javax.xml.transform.dom.DOMSource;
import javax.xml.transform.stream.StreamResult;

public class ActivityListing {
...
  public static void main (String args[]) {
...
     newRoot.appendChild(newActivity);

   }

   try {

     TransformerFactory transformerFactory =
             TransformerFactory.newInstance();
     Transformer transformer =
            transformerFactory.newTransformer();

     DOMSource origDocSource = new DOMSource(doc);
     StreamResult origResult = new StreamResult("updated.xml");
     transformer.transform(origDocSource, origResult);

     DOMSource newDocSource = new DOMSource(newDoc);
     StreamResult newResult = new StreamResult("roster.xml");
     transformer.transform(newDocSource, newResult);

   } catch (Exception e) {
     println("Could not save document.");
   }
  }

  public static void displayActivity(Node thisAct) {
...

When transforming documents, a Transformer takes the source and sends it to the result after applying the instructions specified by the style sheet with which the Transformer was created. Because there is no style sheet in this case, the Transformer makes no changes and simply sends the original document to the result, which in this case is a file.

C++

In C++, creating an XML document is as simple as calling the save function with the destination filename.

Listing 3.20b Saving the Document in C++

#include "stdafx.h"
#import "C:\windows\system32\msxml2.dll"
using namespace MSXML2;

int _tmain(int argc, _TCHAR* argv[])
{
  ::CoInitialize(NULL);
  try
  {
    HRESULT hr;
    CComPtr<MSXML2::IXMLDOMDocument> pDomDocument;

    // We create the initial empty DOMDocument
    hr = pDomDocument.CoCreateInstance(__uuidof(MSXML2::DOMDocument));

    // We load the XML file into the DOMDocument
    hr = pDomDocument->load("activities.xml");

    // Here we can place code that makes modifications to the DOMDocument

    // Now, we save the DOMDocument as some other XML file
    hr = pDomDocument->save("new_activities.xml");
  }

  catch(...)
  {
    wprintf(L"Caught the exception");
  }
  ::CoUninitialize();
  return 0;
}

Visual Basic

The VB process is as simple as in VC++, as Listing 3.20c shows.

Listing 3.20c Saving the Document in VB

  Private Sub Button1_Click(ByVal sender As System.Object,
               ByVal e As System.EventArgs) _
               Handles Button1.Click

    'We create the initial empty DOMDocument
    Dim xmldoc As New MSXML2.DOMDocument40()
    Dim hr As Boolean

    'We load the XML file into the DOMDocument
    xmldoc.async = False
    hr = xmldoc.load("activities.xml")

    'Here we can place code that makes modifications to the DOMDocument

    'Now, we save the DOMDocument as some other XML file
    xmldoc.save("new_activities.xml")
  End Sub

PHP

Creating a new document and writing DOM structures to disk are both easy using PHP, but at the time of this writing the current version of PHP, 4.2.1, has two bugs to watch out for. First, formatting the output removes whitespace nodes, and second, the PHP version of getElementById(), get_element_by_id(), doesn't actually search for an element with an attribute of type ID. Instead, it searches for an element named ID. This behavior will likely be fixed in future versions, but I've included a workaround as part of Listing 3.20d.

Listing 3.20d Saving the Document in PHP

<?php

...

$new_doc = domxml_new_doc('1.0');
$new_root = $new_doc->create_element('roster');

$first_child   =& $root->first_child();
$activities_el  =& $first_child->next_sibling();
$activities_list = $activities_el->get_elements_by_tagname("activity");
$count_activities = count($activities_list);

for ($i = 0; $i < $count_activities; $i++) {
 $act =& $activities_list[$i];
 add_new_participant($act,"P4");
 display_activity($act);

 $act_id = $act->get_attribute("activityid");
 $new_act = $new_doc->create_element("activity");
 $new_act->set_attribute("activityid",$act_id);

 $persons_el_list = $act->get_elements_by_tagname("persons");
 $persons_el   =& $persons_el_list[0];
 $person_list   = $persons_el->get_elements_by_tagname("person") ;
 $count_persons  = count($person_list);
 for ($j = 0; $j < $count_persons; $j++) {
  $person     = $person_list[$j];
  $person_id    = $person->get_attribute("personid");
  $new_person   = $new_doc->create_element("person");
  $new_person_text = $new_doc->create_text_node($person_id);
  $new_person->append_child($new_person_text);
  $new_act->append_child($new_person);
 }

 $new_root->append_child($new_act);
}

$new_doc->append_child($new_root);

$doc->dump_file("updated.xml",false,false);
$new_doc->dump_file("roster.xml",false,false);

...

function add_new_participant (&$act, $person)
{
 $act_persons_list = $act->get_elements_by_tagname("persons");
 $act_persons_el  =& $act_persons_list[0];

 $doc      =& $act_persons_el->owner_document();
 $new_person_el = $doc->create_element("person");
 $new_person_el->set_attribute("personid",$person);
 $act_persons_el->append_child($new_person_el);
}

function & get_element_by_id(&$doc, $tagname, $attr_name, $value)
{
 $tags    = $doc->get_elements_by_tagname($tagname);
 $count_tags = count($tags);

 for ($i = 0; $i < $count_tags; $i++) {
  $el =& $tags[$i];
  if ($el->get_attribute($attr_name) == $value) {
   return $el;
  }
 }

 return false;
}
?>

Perl

Creating a Document object in Perl is a bit more complex than it is in Java. Instead of using a document builder or factory object, with XML::Xerces you use the DOMImplementation object to create a new, empty document type based on a DTD. (The DTD document doesn't have to exist, but you do have to put a DTD name in the command.) Then you can create an instance of that DocumentType, get the root Document object, and add elements.

On the other hand, saving the file is much easier. Instead of using an XSLT transform, just open a FileHandle for writing, format the document, and print it to the open FileHandle (see Listing 3.20e).

Listing 3.20e Saving the Document in Perl

use FileHandle;
use XML::Xerces;
use XML::Xerces::DOMParse;

my $file  = XML::Xerces::LocalFileInputSource->new("activities.xml");
my $parser = XML::Xerces::DOMParser->new();
eval { $parser->parse($file); };
XML::Xerces::error($@) if $@;

my $doc = $parser->getDocument();
my $root = $doc->getDocumentElement();

my $first_child  = $root->getFirstChild();
my $activities_el = $first_child->getNextSibling();

my $impl   = XML::Xerces::DOM_DOMImplementation::getImplementation();
my $new_dt  = $impl->createDocumentType( 'roster', '', 'roster.dtd' );
my $new_doc = $impl->createDocument( 'roster', 'roster', $new_dt );
my $new_root = $new_doc->getDocumentElement();
for my $act ( $activities_el->getElementsByTagName("activity") ) {
  add_new_participant( $act, "P4" );
  display_activity($act);

  ...

}

my $updated = new FileHandle( "updated.xml", "w" );
$XML::Xerces::DOMParse::INDENT = " ";
XML::Xerces::DOMParse::format($doc);
XML::Xerces::DOMParse::print( $updated, $doc );
$updated->close;

my $roster = new FileHandle( "roster.xml", "w" );
$XML::Xerces::DOMParse::INDENT = " ";
XML::Xerces::DOMParse::format($new_doc);
XML::Xerces::DOMParse::print( $roster, $new_doc );
$roster->close;
...

sub add_new_participant {
  my $act  = shift;
  my $person = shift;

  my $act_persons_el = $act->getElementsByTagName("persons")->item(0);

  my $doc      = $act_persons_el->getOwnerDocument();
  my $new_person_el = $doc->createElement("person");
  $new_person_el->setAttribute( "personid", $person );
  $act_persons_el->appendChild($new_person_el);
}

The Result

The result of all the work we've done in this chapter is an XML document with activity and person elements, as shown in Listing 3.21. (I've added whitespace to make the document a bit more readable.)

Listing 3.21 The Resulting Document

<?xml version="1.0" encoding="UTF-8"?>
<roster>
  <activity activityid="A1">
   <person>P2</person>
   <person>P1</person>
   <person>P4</person>
  </activity>
  <activity activityid="A2">
   <person>P4</person>
  </activity>
</roster>