DOMDocument parsing data inside a CDATA

Question

I have this xml (had to cut/paste via HTML).

<tr>
    <td>http://www.example.co.uk/the-view-from-22/feed/</td>
    <td>Example Blogs » The View from 22 » Example Blogs</td>
    <td>http://blogs.example.co.uk/</td>
    <td><![CDATA[Listen: The Example&rsquo;s verdict on the debate]]></td>
    <td>http://blogs.example.co.uk/coffeehouse/2015/04/podcast-special-the-debate/</td>
</tr>

It is being loaded in to an XML dom document

   $dom = new DOMDocument();
   $dom->preserveWhiteSpace = false;
   $dom->formatOutput = true;
   $dom->loadXML( $xml->asXML() );
   return $dom->saveXML();

But this throws an error about the ’ entity not being defined.

Warning: DOMDocument::loadXML() [domdocument.loadxml]: Entity 'rsquo' not defined in Entity,...

As it is in a CDATA section I expected the DOMDocument to treat it as text and ignore it... but it doesn't... is there a way around this?

The Data is being pulled directly out of a mysql database in a view, so there isn't much scope for 'fixing it up' first - I added the CDATA in the select clause for the view, that was my attempt at a fix!

edit Traced it back following suggestions below (cheers!)

The data is being loaded using $xml->addChild( $key, $value ) but the $value is in the form so is being encoded as you surmised.

So am just trying this...

How to write CDATA using SimpleXmlElement?

And it works - I am now loading up the orignal doc with:-

 if (strpos(strtoupper($value),'<![CDATA[') === 0 && strpos(strrev($value),'>]]') === 0) {
                $child = $xml->addChild( $key );
                $node = dom_import_simplexml($child);
                $no   = $node->ownerDocument;
                $node->appendChild($no->createCDATASection(substr($value,9,strlen($value)-12)));

                //simple key/value child pair
            } else {
                $xml->addChild( $key, $value );
            }

Answer 1

You could try to replace it, if it's only one ’ and not a mass of special chars.

 $dom = new DOMDocument();
 $dom->preserveWhiteSpace = false;
 $dom->formatOutput = true;
 $xml = $xml->asXML();

 $xml = str_replace('’', '’', $xml);

 $dom->loadXML($xml);
 return $dom->saveXML();

The real question is, how did the ’ get into your database. Fix it, before insertion...then you can pull well-formed XML. https://stackoverflow.com/a/3142636/1163786

Or make rsquo a valid entity:

<!DOCTYPE ROOT_XML_ELEMENT [ <!ENTITY rsquo "&#8217;"> ]>

If your content is UTF-8 simply replace it with: `

---

(I think) the original issue is this one:

Warning: Entity 'rsquo' not defined in Entity, line: ...

<?php

$xml = <<<XML
<tr>
    <td>Listen: The Example&rsquo;s verdict on the debate></td>
</tr>
XML;

$doc = new DOMDocument();
$doc->presverWhitespace = false;
$doc->formatOutput = true;
$doc->loadXML($xml);
echo $doc->saveXML();

Because entity 'rsquo' is not valid XML the error pops up. Now pperrin addressed it by adding a "CDATA fix". That's how i understand the question.

You don't need CDATA - if you

• define the entity at the root or
• add it to the DTD to make it valid or
• replace it manually (see above)
• or simply fix it before it goes to the database.

Answer 2

As I have demonstrated with my example code I was not able to reproduce your problem. Therefore I came to the conclusion that you must have a double-encoding and the double-encoded data is where the XML parser chokes and rightfully gives you the warnings. It's just that due to the double encoding this was not immediately visible.

Decode the data once so that it is properly XML encoded. Then DOMDocument can easily load it.

---

Old Answer (can still be useful for users coming here via search engines):

I suspect your problem is with $xml->asXML() as the CDATA section does not produce that error.

There is a better way to convert into a DOMDocument first:

$dom = dom_import_simplexml($xml)->documentElement;

This should also preserve the encoding with the CDATA section (not 100% sure). For your formatting it might then be you need to reload the document but maybe you don't need. Try

$dom->preserveWhiteSpace = false;
$dom->formatOutput = true;
$result = $dom->saveXML();

If the result is not yet the expected pretty-print format you're looking for, you can reload the document from dom:

$dom->preserveWhiteSpace = false;
$dom->formatOutput = true;
$dom->loadXML($dom->saveXML());
$result = $dom->saveXML();

I hope as this is DOMDocument, there is not problem with the previous CDATA encoded characters that resemble an entity.

The conversion function dom_import_simplexml() is in the manual and as SimpleXML and DOM share commons behing their interfaces, using it should be the preferred way if you want to switch between DOM and SimpleXML or vice-versa.