DOMXPath::evaluate

Note that this method does not provide any means to distinguish between a successful result that returns FALSE and an error.

For example, this will succeed and return FALSE:

<?php $xpath->evaluate("1 = 0"); ?>

One workaround when you know you are expecting a Boolean is to wrap the result with string(). e.g.

<?php $xpath->evaluate("string(1 = 0)"); ?>

This will return a string "false" on success, or the Boolean FALSE on error.

If your expression returns a node set, you will get a DOMNodeList instead of a typed result. Instead, try modifying your expression from "//node[1]" to "string(//node[1])".

this class can substitute the method evaluate while it is not validated. Made for Yuri Bastos and Jo�o Gilberto Magalh�es. 

<?php

    class XPtahQuery
    {
        // function returns a DOMNodeList from a relative xPath
        public static function selectNodes($pNode, $xPath) 
        {

            $pos = strpos(self::getFullXpath($pNode),"/",1);
            $xPathQuery = substr(self::getFullXpath($pNode),$pos);//to paste  /#document[1]/
            $xPathQueryFull = $xPathQuery. $xPath;
            $domXPath = new DOMXPath($pNode->ownerDocument);
            $rNodeList = $domXPath->query($xPathQueryFull);

                return $rNodeList;
        }
        // function returns a DOMNode from a xPath from other DOMNode
        public static function selectSingleNode($pNode, $xPath) 
        {

            $pos = strpos(self::getFullXpath($pNode),"/",1);
            $xPathQuery = substr(self::getFullXpath($pNode),$pos);//to paste  /#document[1]/
            $xPathQueryFull = $xPathQuery. $xPath;
            $domXPath = new DOMXPath($pNode->ownerDocument);
            $rNode = $domXPath->query($xPathQueryFull)->item(0);

                return $rNode;
        }
        //utilitaries functions off selectSingleNode
        private function getNodePos($pNode, $nodeName)
        {
            if($pNode == null)
                {
                        return 0;
            }
            else
            {
                $var = 0;
                    if ($pNode->previousSibling != null)
                    {
                    if ($pNode->previousSibling->nodeName == $nodeName)
                    {
                        $var = 1;
                    }
                    }
                    return self::getNodePos($pNode->previousSibling, $nodeName) + $var;
            }
        }
        //utilitaries functions off selectSingleNode
        private function getFullXpath($pNode)
        {
            if($pNode == null)
                {
                        return "";
            }
            else
            {

                return self::getFullXpath($pNode->parentNode) . "/" . $pNode->nodeName . "[" .strval(self::getNodePos($pNode, $pNode->nodeName)+1) . "]";//+1 to get the real xPath index

            }
        }
    }
?>

To query DOMNodes by their HTML classes, use such snippet
<?php

// CssClassXPathSelector
function ccxs($class) {
    return '[contains(concat(" ", normalize-space(@class), " "), " ' . $class . ' ")]';
}

 // then just
 $domitems = $this->xpath("//*[@id='searchResultsRows']//a" . ccxs('listing_row'));
?>

The only way how to distinguish FALSE returned value from syntax error FALSE is to re-run the XPath expression wrapped in string() function. If must return empty string. If it returns FALSE again then it is an error.

<?php

        $ret=$this->xp->evaluate($eval, $context);

        // Error detection: DOMXPath::evaluate() returns FALSE on error 
        // so does DOMXPath::evaluate("boolean(/nothing)") 
        // @workaround webdevelopers.eu
        if ($ret === false && $this->xp->evaluate("string($eval)", $context) === false) { 
            throw new Exception("Invalid XPath expression ".json_encode($eval), 3491);
        }
?>