SimpleXML与DOMDocument性能

Question

I am building an RSS parser using the SimpleXML Class and I was wondering if using the DOMDocument class would improve the speed of the parser. I am parsing an rss document that is at least 1000 lines and I use almost all of the data from those 1000 lines. I am looking for the method that will take the least time to complete.

Answer 1

SimpleXML and DOMDocument both use the same parser (libxml2), so the parsing difference between them is negligible.

This is easy to verify:

function time_load_dd($xml, $reps) {
    // discard first run to prime caches
    for ($i=0; $i < 5; ++$i) { 
        $dom = new DOMDocument();
        $dom->loadXML($xml);
    }
    $start = microtime(true);
    for ($i=0; $i < $reps; ++$i) { 
        $dom = new DOMDocument();
        $dom->loadXML($xml);
    }
    $stop = microtime(true) - $start;
    return $stop;
}
function time_load_sxe($xml, $reps) {
    for ($i=0; $i < 5; ++$i) { 
        $sxe = simplexml_load_string($xml);
    }
    $start = microtime(true);
    for ($i=0; $i < $reps; ++$i) { 
        $sxe = simplexml_load_string($xml);
    }
    $stop = microtime(true) - $start;
    return $stop;
}


function main() {
    // This is a 1800-line atom feed of some complexity.
    $url = 'http://feeds.feedburner.com/reason/AllArticles';
    $xml = file_get_contents($url);
    $reps = 10000;
    $methods = array('time_load_dd','time_load_sxe');
    echo "Time to complete $reps reps:
";
    foreach ($methods as $method) {
        echo $method,": ",$method($xml,$reps), "
";
    }
}
main();

On my machine I get basically no difference:

Time to complete 10000 reps:
time_load_dd: 17.725028991699
time_load_sxe: 17.416455984116

The real issue here is what algorithms you are using and what you are doing with the data. 1000 lines is not a big XML document. Your slowdown will not be in memory usage or parsing speed but in your application logic.