PHP: URL wrangler class

URL.php

<?php

declare( encoding = 'UTF-8' );
namespace shurlook\net
{

class URL
{
	const PROTOCOL = 'scheme';
	const HOST = 'host';
	const PORT = 'port';
	const USERNAME = 'user';
	const PASSWORD = 'pass';
	const PATH = 'path';
	const QUERY = 'query';
	const FRAGMENT = 'fragment';
	
	// http://www.iana.org/assignments/uri-schemes.html
	protected static $defaultPorts = array(
		'http' => 80,
		'https' => 443,
	);
	
	protected $parts = array();
	protected $cache = null;
	
	public function __construct( $parts=null )
	{
		if( is_string( $parts ) )
			$this->parts = parse_url( trim($parts) );

		else if( is_array( $parts ) )
			$this->parts = $parts;
		
	}
	
	public function __toString()
	{
		if( $this->cache )
			return $this->cache;
		
		$t = array();

		if( !empty( $this->parts[ self::PROTOCOL ] ) )
			$t[] = $this->parts[ self::PROTOCOL ] .'://';

		if( !empty( $this->parts[ self::USERNAME ] ) )
		{
			$t[] = $this->parts[ self::USERNAME ];

			if( !empty( $this->parts[ self::PASSWORD ] ) )
				$t[] = ':'. $this->parts[ self::PASSWORD ];			

			$t[] = '@';
		}
			
		if( !empty( $this->parts[ self::HOST ] ) )
			$t[] = $this->parts[ self::HOST ];
			
		if( !empty( $this->parts[ self::PATH ] ) )
			$t[] = $this->parts[ self::PATH ];
			
		if( !empty( $this->parts[ self::QUERY ] ) )
			$t[] = '?'. $this->parts[ self::QUERY ];
		
		if( !empty( $this->parts[ self::FRAGMENT ] ) )
			$t[] = '#'. $this->parts[ self::FRAGMENT ];
		
		return $this->cache = join( $t );
	}
	
	public function setProtocol( $v )
	{
		$this->cache = null;
		$this->parts[ self::PROTOCOL ] = $v;
		return $this;
	}
	public function getProtocol()
	{
		return !empty( $this->parts[ self::PROTOCOL ] ) ? $this->parts[ self::PROTOCOL ] : null;
	}
	
	public function setHost( $v )
	{
		$this->cache = null;
		$this->parts[ self::HOST ] = $v;
		return $this;
	}
	public function getHost()
	{
		return !empty( $this->parts[ self::HOST ] ) ? $this->parts[ self::HOST ] : null;
	}
	
	public function setPort( $v )
	{
		$this->cache = null;
		$this->parts[ self::PORT ] = $v;
		return $this;
	}
	public function getPort()
	{
		if( !empty( $this->parts[ self::PORT ] ) )
			return $this->parts[ self::PORT ];
	}
	public function getProtocolDefaultPort()
	{
		if( empty( $this->parts[ self::PROTOCOL ] ) )
			return null;

		$p = $this->parts[ self::PROTOCOL ];
		if( empty( self::$defaultPorts[ $p ] ) )
			return null;
		
		return self::$defaultPorts[ $p ];
	}
	public function getPortOrDefault()
	{
		if( $p = $this->getPort() )
			return $p;
		
		return $this->getProtocolDefaultPort();
	}
	
	public function setUsername( $v )
	{
		$this->cache = null;
		$this->parts[ self::USERNAME ] = $v;
		return $this;
	}
	public function getUsername()
	{
		return !empty( $this->parts[ self::USERNAME ] ) ? $this->parts[ self::USERNAME ] : null;
	}
	
	public function setPassword( $v )
	{
		$this->cache = null;
		$this->parts[ self::PASSWORD ] = $v;
		return $this;
	}
	public function getPassword()
	{
		return !empty( $this->parts[ self::PASSWORD ] ) ? $this->parts[ self::PASSWORD ] : null;
	}
	
	public function setPath( $v )
	{
		$this->cache = null;
		$this->parts[ self::PATH ] = $v;
		return $this;
	}
	public function getPath()
	{
		return !empty( $this->parts[ self::PATH ] ) ? $this->parts[ self::PATH ] : null;
	}
	public function getPathDirectory()
	{
		if( empty( $this->parts[ self::PATH ] ) )
			return null;
		
		if( mb_substr( $this->parts[ self::PATH ], -1 ) == '/' )
			return $this->parts[ self::PATH ];
			
		$t = dirname( $this->parts[ self::PATH ] );
		return $t . ($t != '/' ? '/' : '');
	}
	public function getPathFile()
	{
		if( empty( $this->parts[ self::PATH ] ) )
			return null;
		
		if( mb_substr( $this->parts[ self::PATH ], -1 ) == '/' )
			return null;
			
		return basename( $this->parts[ self::PATH ] );
	}
	public function isRelativePath()
	{
		if( !($p = $this->getPath()) )
			return null;
		
		return substr( $p, 0, 1 ) != '/';
	}
	
	public function setQuery( $v )
	{
		$this->cache = null;
		$this->parts[ self::QUERY ] = $v;
		return $this;
	}
	public function getQuery()
	{
		return !empty( $this->parts[ self::QUERY ] ) ? $this->parts[ self::QUERY ] : null;
	}
	public function addQuery( $key, $value )
	{
		$this->cache = null;
		$t = $key .'='. urlencode( $value );
		if( empty( $this->parts[ self::QUERY ] ) )
		{
			$this->parts[ self::QUERY ] = $t;
			return $this;
		}
		
		$this->parts[ self::QUERY ] .= '&'. $t;
		return $this;
	}
	public function getQueryArray()
	{
		if( empty( $this->parts[ self::QUERY ] ) )
			return array();

		$t = array();
		parse_str( $this->parts[ self::QUERY ], $t );
		return $t;
	}
	
	public function setFragment( $v )
	{
		$this->cache = null;
		$this->parts[ self::FRAGMENT ] = $v;
		return $this;
	}
	public function getFragment()
	{
		return !empty( $this->parts[ self::FRAGMENT ] ) ? $this->parts[ self::FRAGMENT ] : null;
	}
	
	public function getResource()
	{
		if( $t = $this->getQuery() )
			return $this->getPath() .'?'. $this->getQuery();
		
		return $this->getPath();
	}
	
	public function hostIsIp()
	{
		if( empty( $this->parts[ self::HOST ] ) )
			return null;
			
		return preg_match( '#\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}/#iuS', $this->parts[ self::HOST ] );
	}
	
	public function getDomain()
	{
		if( empty( $this->parts[ self::HOST ] ) || $this->hostIsIp() )
			return null;
		
		$matches = array();
		if( !preg_match( '#[^\.]*\.[^\.]*$#uS', $this->parts[ self::HOST ], $matches ) )
			return $this->parts[ self::HOST ];
			
		return $matches[0];
	}
	
	
	public function isRelativeURL()
	{
		return !$this->getHost();
	}
	
	// http://tools.ietf.org/html/rfc3986#section-5
	public function absoluteTo( $base )
	{
		// abort if this is not a relative URL
		if( !$this->isRelativeURL() )
			return clone $this;
		
		if( is_string( $base ) )
			$base = new URL( $base );
		
		// abort if $base is a relative URL
		if( $base->isRelativeURL() )
			throw new \Exception( '"'. $base .'" is a relative URL and thus not suited as the base URL for relative-to-absolute-URL-translation' );
		
		// don't modify the current object
		$url = clone $this;

		$url->setProtocol( $base->getProtocol() );
		$url->setUsername( $base->getUsername() );
		$url->setPassword( $base->getPassword() );
		$url->setHost( $base->getHost() );
		$url->setPort( $base->getPort() );
				
		if( $url->isRelativePath() )
		{
			$p = $base->getPath();
			$basePath = ( !$p || $p == '/' ) ? '/' : $base->getPathDirectory();
			$url->setPath( $basePath . $url->getPath() );
		}
		
		return $url;
	}

}

if( false )
{
	$urls = array(
		'mailto:[email protected]',
		'http://www.google.de/path/to/file?some=query#hash',
		'http://xyz.www.google.de/path/to/file?some=query#hash',
		'www.google.de/path/to/file?some=query#hash',
		'/path/to/file?some=query#hash',
		'path/to/file?some=query#hash',
	);


	foreach( $urls as $url )
	{
		$u = new URL( $url );
		var_dump( 
			$url,
			$u,
			$u->getProtocol(),
			$u->getDomain(),
			$u->getPathDirectory(),
			$u->getPathFile()
		); 
		echo '<hr>';
	}

	$urls = array(
		'relative/path?blubber#hash1' => 'http://www.google.de/path/to/file?some=query#hash',
		'/absolute/path?blubber#hash1' => 'http://www.google.de/path/to/file?some=query#hash',
	
		'relative/path?blubber#hash2' => 'www.google.de/path/to/file?some=query#hash',
		'/absolute/path?blubber#hash2' => 'www.google.de/path/to/file?some=query#hash',
	
		'relative/path?blubber#hash3' => 'http://user:[email protected]:1234/path/to/file?some=query#hash',
		'/absolute/path?blubber#hash3' => 'http://user:[email protected]:1234/path/to/file?some=query#hash',
	
		'relative/path?blubber#hash4' => 'http://www.google.de/path/to/',
		'/absolute/path?blubber#hash4' => 'http://www.google.de/path/to/',
		
		'relative/path?blubber#hash5' => 'http://www.google.de/path/to',
		'/absolute/path?blubber#hash5' => 'http://www.google.de/path/to',

		'relative/path?blubber#hash6' => 'http://www.google.de/',
		'/absolute/path?blubber#hash6' => 'http://www.google.de/',
		
		'relative/path?blubber#hash7' => 'http://www.google.de/?some=query',
		'/absolute/path?blubber#hash7' => 'http://www.google.de/?some=query',
		
		'relative/path?blubber#hash8' => 'http://www.google.de',
		'/absolute/path?blubber#hash8' => 'http://www.google.de',
		
		'relative/path?blubber#hash9' => 'http://www.google.de?some=query',
		'/absolute/path?blubber#hash9' => 'http://www.google.de?some=query',
	);

	foreach( $urls as $rel => $base )
	{
		try
		{
			$relURL = new URL( $rel );
			$baseURL = new URL( $base );
			$url = $relURL->absoluteTo( $baseURL );
			var_dump( $rel, $base, $baseURL, $url->__toString() );
		}
		catch( \Exception $e )
		{
			var_dump( $rel, $base, $e->getMessage() );
		}
		echo '<hr>';
	}
}
}

?>