html scraping and css queries

Question

Hello, what are the advantages and disadvantages of the following libraries?

• PHP Simple HTML DOM Parser
• QP
• phpQuery

From the above i've used QP and it failed to parse invalid HTML, and simpleDomParser, that does a good job, but it kinda leaks memory because of the object model. But you may keep that under control by calling $object->clear(); unset($object); when you dont need an object anymore.

Are there any more scrapers? What are your experiences with them? I'm going to make this a community wiki, may we'll build a useful list of libraries that can be useful when scraping.

---

i did some tests based Byron's answer:

    <?
    include("lib/simplehtmldom/simple_html_dom.php");
    include("lib/phpQuery/phpQuery/phpQuery.php");


    echo "<pre>";

    $html = file_get_contents("http://stackoverflow.com/search?q=favorite+programmer+cartoon");
    $data['pq'] = $data['dom'] = $data['simple_dom'] = array();

    $timer_start = microtime(true);

    $dom = new DOMDocument();
    @$dom->loadHTML($html);
    $x = new DOMXPath($dom);

    foreach($x->query("//a") as $node)
    {
         $data['dom'][] = $node->getAttribute("href");
    }

    foreach($x->query("//img") as $node)
    {
         $data['dom'][] = $node->getAttribute("src");
    }

    foreach($x->query("//input") as $node)
    {
         $data['dom'][] = $node->getAttribute("name");
    }

    $dom_time =  microtime(true) - $timer_start;
    echo "dom: \t\t $dom_time . Got ".count($data['dom'])." items \n";






    $timer_start = microtime(true);
    $doc = phpQuery::newDocument($html);
    foreach( $doc->find("a") as $node)
    {
       $data['pq'][] = $node->href;
    }

    foreach( $doc->find("img") as $node)
    {
       $data['pq'][] = $node->src;
    }

    foreach( $doc->find("input") as $node)
    {
       $data['pq'][] = $node->name;
    }
    $time =  microtime(true) - $timer_start;
    echo "PQ: \t\t $time . Got ".count($data['pq'])." items \n";









    $timer_start = microtime(true);
    $simple_dom = new simple_html_dom();
    $simple_dom->load($html);
    foreach( $simple_dom->find("a") as $node)
    {
       $data['simple_dom'][] = $node->href;
    }

    foreach( $simple_dom->find("img") as $node)
    {
       $data['simple_dom'][] = $node->src;
    }

    foreach( $simple_dom->find("input") as $node)
    {
       $data['simple_dom'][] = $node->name;
    }
    $simple_dom_time =  microtime(true) - $timer_start;
    echo "simple_dom: \t $simple_dom_time . Got ".count($data['simple_dom'])." items \n";


    echo "</pre>";

and got

dom:         0.00359296798706 . Got 115 items 
PQ:          0.010568857193 . Got 115 items 
simple_dom:  0.0770139694214 . Got 115 items 

Answer 1

I used to use simple html dom exclusively until some bright SO'ers showed me the light hallelujah.

Just use the built in DOM functions. They are written in C and part of the PHP core. They are faster more efficient than any 3rd party solution. With firebug, getting an XPath query is muey simple. This simple change has made my php based scrapers run faster, while saving my precious time.

My scrapers used to take ~ 60 megabytes to scrape 10 sites asyncronously with curl. That was even with the simple html dom memory fix you mentioned.

Now my php processes never go above 8 megabytes.

Highly recommended.

EDIT

Okay I did some benchmarks. Built in dom is at least an order of magnitude faster.

Built in php DOM: 0.007061
Simple html  DOM: 0.117781

<?
include("../lib/simple_html_dom.php");

$html = file_get_contents("http://stackoverflow.com/search?q=favorite+programmer+cartoon");
$data['dom'] = $data['simple_dom'] = array();

$timer_start = microtime(true);

$dom = new DOMDocument();
@$dom->loadHTML($html);
$x = new DOMXPath($dom); 

foreach($x->query("//a") as $node) 
{
     $data['dom'][] = $node->getAttribute("href");
}

foreach($x->query("//img") as $node) 
{
     $data['dom'][] = $node->getAttribute("src");
}

foreach($x->query("//input") as $node) 
{
     $data['dom'][] = $node->getAttribute("name");
}

$dom_time =  microtime(true) - $timer_start;

echo "built in php DOM : $dom_time\n";

$timer_start = microtime(true);
$simple_dom = new simple_html_dom();
$simple_dom->load($html);
foreach( $simple_dom->find("a") as $node)
{
   $data['simple_dom'][] = $node->href;
}

foreach( $simple_dom->find("img") as $node)
{
   $data['simple_dom'][] = $node->src;
}

foreach( $simple_dom->find("input") as $node)
{
   $data['simple_dom'][] = $node->name;
}
$simple_dom_time =  microtime(true) - $timer_start;

echo "simple html  DOM : $simple_dom_time\n";