Support url_pattern_regex (one plugin for more sites)

[webaddicto]

Useful to create a plugin that handles more websites (i.e abc.com, abc.de, abc.eu, etc) or similar.

[webaddicto]

Example usage to match abc.com, abc.de and abc.pl:

namespace Proxy\Plugin;

use Proxy\Plugin\AbstractPlugin;
use Proxy\Event\ProxyEvent;

use Proxy\Html;

class MultiSiteMatchPlugin extends AbstractPlugin {

	protected $url_pattern_regex = '#^abc\.(com|de|pl)$#is';
	
	public function onCompleted(ProxyEvent $event){
	
		$response = $event['response'];
		$html = $response->getContent();
		// do your stuff here...
		$response->setContent($html);
	}
}

[Athlon1600]

What happens if both $url_pattern and $url_pattern_regex are set? I'd rather just have one $url_pattern that can be used to match both as regex and as non-regex. But first you would have to somehow detect whether $url_pattern contains regex or not and that can be tricky...

[webaddicto]

I would write in the help file or readme that user should use $url_pattern or $url_pattern_regex (not both).

An alternative option would be like this:

$url_pattern = "abc.com"; => match string
$url_pattern = "regex:'#^abc\.(com|de|pl)$#is'"; => extract regex:(.+?) and use preg_match()

What do you think?

[Athlon1600]

Why not just have it like this:
$url_pattern = 'abc.com' => treat it like a regular strpos match
$url_pattern = '/abc.(com|net)/' => treat it like a preg_match match

The laziest way of accomplishing this is just to check the first character of $url_pattern. If it's /, then you have a regex pattern.

[webaddicto]

I wrote 4 alternatives:

Alternative 1:

		// url filter provided and current request url does not match it
		if($this->url_pattern && strpos($url, $this->url_pattern) === false){
			return;
		}
		// url filter (regex) provided and current request url does not match it
		elseif($this->url_pattern_regex && !preg_match($this->url_pattern_regex, $url)){
			return;
		}

Alternative 2:

		// url filter provided and current request url does not match it
		if($this->url_pattern){
		    if(stripos($this->url_pattern, 'r:') === 0){
			    if(!preg_match(substr($this->url_pattern, 2), $url)) 
				return;
			} 
			else
			{
			    if(strpos($url, $this->url_pattern) === false) 
				return;
			}
		}

Alternative 3:

		// url filter provided and current request url does not match it
		if($this->url_pattern){
		    if(!preg_match('/^[a-zA-Z0-9]{1}/is', $this->url_pattern){
			    if(!preg_match($this->url_pattern, $url)) 
				return;
			} 
			else
			{
			    if(strpos($url, $this->url_pattern) === false) 
				return;
			}
		}

Alternative 4:

		// url filter provided and current request url does not match it
		if($this->url_pattern){
		    if(strpos($this->url_pattern, '/') === 0){
			    if(!preg_match($this->url_pattern, $url)) 
				return;
			} 
			else
			{
			    if(strpos($url, $this->url_pattern) === false) 
				return;
			}
		}

I would vote for 3 because it allows user to use any special character on preg_match:

** May not be good for unicode domain names? **

$url_pattern = '#^abc\.(com|de|pl)$#is'; => regex
$url_pattern = '/^abc\.(com|de|pl)$/is'; => regex
$url_pattern = '@^abc\.(com|de|pl)$@is'; => regex
$url_pattern = 'abc.com'; => string

Or 4 is fine too, but we need to write that user must use / character.

What do you think?

[Athlon1600]

Option 3 would also match 'abc.com' even when it was intended to be a regular match... I would go with 4 because few people use delimiters other than '/'. It's a default regex deliminator on every tutorial online.

[webaddicto]

Perfect, I have updated the PR with a new commit according to your request :)
0595954

[Athlon1600]

Yup, looks good to me now!

[webaddicto]

Good, I have added info about the $url_pattern on this new PR:
#55

Also this PR may help to detect the latest php-proxy version in case users report issues
#53