Question

Here is a basic example of how simple_html_dom works in a standalone php file.

test.php:

include ('simple_html_dom.php');

$url = "http://www.google.com";
$html = new simple_html_dom();
$html->load_file($url);

print $html;

If I execute it with the command: php test.php

It dumps correctly the html of the website (in this example, google.com)

Now let’s take a look at a basic example of code using a Symfony task:

class parserBasic extends sfBaseTask {
  public function configure()
  {
    $this->namespace = 'parser';
    $this->name      = 'basic';
  }

  public function execute($arguments = array(), $options = array())
  {
    $url = "http://www.google.com";
    $html = new simple_html_dom();
    $html->load_file($url);
    print $html;
  }
}

This file it’s located under: <appname>/lib/task

I don’t need to include the library in the file because being under the lib/task folder, it gets automatically loaded.

I execute the task using this command: php symfony parser:basic

And I get the following error message:

PHP Fatal error:  
Call to a member function innertext() on a non-object in
/home/<username>/<appname>/lib/task/simple_html_dom.php on line 1688

Any suggestions?

Answer 1

The problem comes from Symfony.

In fact, if an error occur when loading the file with simple_html_dom, it won’t say anything but returning false.

For example, if you perform this in your task:

$url = "http://www.google.com";
$html = new simple_html_dom();
$res = $html->load_file($url);
if (false === $res)
{
    throw new Exception("load_file failed.");
}
print $html;

You will get an exception. If you tweak simple_html_dom to display en error when loading a file, around line 1085:

// load html from file
function load_file()
{
    $args = func_get_args();
    $this->load(call_user_func_array('file_get_contents', $args), true);
    // Throw an error if we can't properly load the dom.

    if (($error=error_get_last())!==null) {
        // I added this line to see any errors
        var_dump($error);

        $this->clear();
        return false;
    }
}

You will see:

array(4) {
  ["type"]=>
  int(8)
  ["message"]=>
  string(79) "ob_end_flush(): failed to delete and flush buffer. No buffer to delete or flush"
  ["file"]=>
  string(62) "/home/.../symfony.1.4/lib/command/sfCommandApplication.class.php"
  ["line"]=>
  int(541)
}

I usually got this error (which is, in fact, a notice) when using task. The problem is here, in sfCommandApplication, with ob_end_flush:

/**
 * Fixes php behavior if using cgi php.
 *
 * @see http://www.sitepoint.com/article/php-command-line-1/3
 */
protected function fixCgi()
{
  // handle output buffering
  @ob_end_flush();
  ob_implicit_flush(true);

To fix that, I comment the line with @ob_end_flush();. And every thing goes fine. I know, it’s an ugly fix, but it works. An other way to fix that, is to disable notice from PHP (in php.ini), like :

// Report all errors except E_NOTICE
error_reporting = E_ALL ^ E_NOTICE