The DOMDocumentFragment class

Ricki (2011-03-29 06:00:51)

DOMDocumentFragment only appears useful if created from a parent DOMDocument eg.
1. $dom = new DOMDocument("1.0","UTF-8");
2. $docFrag = $dom->createDocumentFragment();
3. Now append items to $docFrag
4. Graft $docFrag contents back onto $dom at the desired location
Conversely taking this approach:
1. $dom = new DOMDocument("1.0","UTF-8");
2. $docFrag = new DOMDocumentFragment();
3. Now append items to $docFrag
...will fail on step 3 with a "read only" error as $docFrag is not created as a child of DOMDocument.
I'm not sure of the reason for this: on the web people have cited security, and others have cited poor design however whatever the reason, it is really limiting when wanting to encapsulating generic independent DocumentFragments into classes for easy grafting to the desired tree. The only workarounds i have seen look expensive from a performance perspective and cumbersome from a coding perspective ie. create a dummy $dom for temporary use.
(This is valid as of PHP 5.3) I've put this here as i've wasted a lot of time finding it out - I hope this saves others some heartache.
Using new DOMDocumentFramt

matthijs at stdin dot nl (2011-03-29 03:06:26)

Note that DOMDocumentFragment is a bit special when it's added to another node. When that happens, not the fragment itself is added as a child, but all of the children of the fragment are moved over to the new parent node.

For example, consider this script:

<?php

/* Create a document and a fragment containing a single node */
$doc = new DOMDocument();
$fragment = $doc->createDocumentFragment();
$fragment->appendChild($doc->createElement('foo'));

/* Now, the foo node is a child of the fragment */
var_dump($fragment->firstChild);

/* After appending the fragment to another node, the children of the
 * fragment will have been transfered to that node (and the fragment is
 * not present in the children list!) */
$doc->appendChild($fragment);
/* So the fragment has no children anymore */
var_dump($fragment->firstChild);
/* But $doc has a single child, which is the foo element, not the
 * fragment */
var_dump($doc->childNodes->length);
var_dump($doc->firstChild);

?>

This produces the following output:

object(DOMElement)#3 (0) {
}
NULL
int(1)
object(DOMElement)#3 (0) {
}

mary dot kalinosky at thieme dot com (2009-08-12 15:43:09)

I found DOMDocument very useful for extracting raw XML from a field in a MySQL database. 

The following always gave me character entities for the XML tag brackets (< >) in the output (whether I used the utf8_encode function or htmlspecialchars function or neither of them on $row['text']):

<?php
$next_elem = $doc->createElement( $row['node_type'], utf8_encode($row['text']) );
$section_elem->appendChild($next_elem);
?>

I changed the code to use a DOMDocumentFragment and now the XML I stored in my database comes out as XML in my output with proper tag brackets instead of html character entities:

<?php
$next_elem = $doc->createDocumentFragment();
$next_elem->appendXML($row['text']);
$section_elem->appendChild($next_elem);
?>