You need a class name on either the a
elements you want to include or exclude. If you don't want to do this in your server side code or documents, you could add the classes with javascript as the page is loaded. With the selection logic wrapped up elsewhere, your rule could just be:
a.external_link
{
background: url(../../pics/remote.gif) right top no-repeat;
padding-right: 10px;
white-space: nowrap;
}
It would be possible with XPath to create a pattern like yours that would also exclude a
elements that had img
children, however this facility has been repeatedly (2002, 2006, 2007) proposed and rejected for CSS, largely on the grounds it goes against the incremental layout principles.
So, while it is possible to do neat conditional content additions as you have with a contextual selector and a prefix match on the href
attribute, CSS is considerably weaker than a general purpose programming language. To do more complex things you need to move the logic up a level and write out simpler instructions for the style engine to handle.