使用XPath指定自定义前缀时如何解决命名冲突？

Question

I'm running PHP 5.4.14, and I'm trying to use XPath to search inside an XML document. I can't know which namespace prefix will be used for each namespace, so I define my own prefixes using registerXPathNamespace().

The problem is that if I register a new prefix for XPath (e.g. C) and the document was already using it, the XPath query won't use my prefix but the original one.

Let me show you a sample code to show this behaviour:

<?php
$body = <<<EOF                                                              
<multistatus xmlns="DAV:" xmlns:VC="urn:ietf:params:xml:ns:carddav" xmlns:C="urn:ietf:params:xml:ns:caldav" xmlns:C1="http://calendarserver.org/ns/">
 <response>                                                                 
  <href>/caldav.php/jorge/contacts/</href>                                  
  <propstat>                                                                
   <prop>                                                                   
    <current-user-principal>                                                
     <href>/caldav.php/jorge/</href>                                        
    </current-user-principal>                                               
    <resourcetype>                                                          
     <collection/>                                                          
     <VC:addressbook/>                                                      
    </resourcetype>                                                         
    <displayname/>                                                          
    <VC:addressbook-home-set>                                               
     <href>/caldav.php/jorge/</href>                                        
    </VC:addressbook-home-set>                                              
   </prop>                                                                  
   <status>HTTP/1.1 200 OK</status>                                         
  </propstat>                                                               
 </response>                                                                
</multistatus>                                                              
EOF;                                                                        

$xml = new SimpleXMLElement($body);                                         

$xml->registerXPathNamespace('C', 'urn:ietf:params:xml:ns:carddav');        
$xpresult = $xml->xpath('//C:addressbook-home-set');                        

var_dump($xpresult);                                                        

If you run it you'll see the query returns no results.

Surprisingly, if I change the registered prefix from C to any other prefix that isn't already defined, such as X, then the query works as expected:

$xml->registerXPathNamespace('X', 'urn:ietf:params:xml:ns:carddav');        
$xpresult = $xml->xpath('//X:addressbook-home-set');    

Am I doing anything wrong? Given that I can't know what prefixes will be used in advance, is it possible to set a custom prefix and make sure it won't conflict with other prefixes in the document besides using an ugly MYPROGRAMUNUSEDPREFIXC prefix?

I know I can get the prefixes and namespaces used in the document, but I'd like to have fixed XPath query strings.

FWIW DOMXPath has this problem too (checked it).

Answer 1

Unless you really want to go into the theoretical possibility that any prefix you choose might already be in use, just use an improbable to guess one. Like for example the prefix: SJDUWISKJDU