Question

A strange thing happened after a supplier changed the XML header a bit. I used to be able to read stuff using xpath, but now I can’t even get a reply with

$xml->xpath('/'); 

They changed it from this…

<?xml version='1.0' encoding='iso-8859-1'?> <!DOCTYPE NewsML SYSTEM 'http://www.newsml.org/dl.php?fn=NewsML/1.2/specification/NewsML_1.2.dtd' [ <!ENTITY % nitf SYSTEM 'http://www.nitf.org/IPTC/NITF/3.4/specification/dtd/nitf-3-4.dtd'> %nitf; ]> <NewsML> ... 

to this:

<?xml version='1.0' encoding='iso-8859-1'?> <NewsML   xmlns='http://iptc.org/std/NewsML/2003-10-10/'   xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance'   xsi:schemaLocation='http://iptc.org/std/NewsML/2003-10-10/ http://www.iptc.org/std/NewsML/1.2/specification/NewsML_1.2.xsd http://iptc.org/std/NITF/2006-10-18/   http://contentdienst.pressetext.com/misc/nitf-3-4.xsd' > ... 

Answer 1

Most likely this is because they’ve introduced a default namespace (xmlns='http://iptc.org/std/NewsML/2003-10-10/') into their document. SimpleXML’s support for default namespaces is not very good, to put it mildly.

Can you try to explicitly register a namespace prefix:

$xml->registerXPathNamespace('n', 'http://iptc.org/std/NewsML/2003-10-10/'); $xml->xpath('/n:NewsML'); 

You would have to adapt your XPath expressions to use the 'n:' prefix on every element. Here is some additional info: http://people.ischool.berkeley.edu/~felix/xml/php-and-xmlns.html.

EDIT: As per the spec:

The registerXPathNamespace() function creates a prefix/ns context for the next XPath query.

This means it would have to be called before every XPath query, thus a function to wrap XPath queries would be the natural thing to do:

function simplexml_xpath_ns($element, $xpath, $xmlns) {     foreach ($xmlns as $prefix_uri)     {         list($prefix, $uri) = explode('=', $prefix_uri, 2);         $element->registerXPathNamespace($prefix, $uri);     }     return $element->xpath($xpath); } 

Usage:

$xmlns = ['n=http://iptc.org/std/NewsML/2003-10-10/']; $result = simplexml_xpath_ns($xml, '/n:NewsML', $xmlns);