Maintain custom query variables

So I think I kind of figured a pretty clean hack out, I edited the kpmg_permalink_fixer and have pasted the edited function below. But basically what I;m doing is, if there is a query string in the url, save it as a separate variable and append to all the redirect urls (wp_redirect).

function kpg_permalink_fixer() {
	$options=kpg_pf_get_options();
	extract($options);
	// fix request_uri on IIS
	if (!array_key_exists('REQUEST_URI',$_SERVER)) {
		$_SERVER['REQUEST_URI'] = substr($_SERVER['PHP_SELF'],1 );
		if (isset($_SERVER['QUERY_STRING'])) {
			$_SERVER['REQUEST_URI'].='?'.$_SERVER['QUERY_STRING'];
		}
	}
	$plink = $_SERVER['REQUEST_URI'];
	$query='';
	if (strpos($plink,'?')!==false)
		{
		$query=substr($plink,strpos($plink,'?'));
		$plink=substr($plink,0,strpos($plink,'?'));
		}
	if (strpos($plink,'#')!==false)  $plink=substr($plink,0,strpos($plink,'#'));
	$flink = $plink; // flink has the page that was 404'd - not the basename
	if (strpos($plink."\t","/\t")!==false)  $plink=substr($plink,0,strpos($plink."\t","/\t"));

	//$plink=basename($plink); // plink now is the permalink part of the request.
	// often I found this is wrong, I want to use the wholw taxonomy in the search

	// remove the indexes if exist
	if (strpos(strtolower($plink)."\t","/index.html\t")!==false) $plink=substr($plink."\t",0,strpos(strtolower($plink)."\t","/index.html\t"));
	if (strpos(strtolower($plink)."\t","/index.htm\t")!==false) $plink=substr($plink."\t",0,strpos(strtolower($plink)."\t","/index.htm\t"));
	if (strpos(strtolower($plink)."\t","/index.shtml\t")!==false) $plink=substr($plink."\t",0,strpos(strtolower($plink)."\t","/index.shtml\t"));
	if (strpos(strtolower($plink)."\t","/default.asp\t")!==false) $plink=substr($plink."\t",0,strpos(strtolower($plink)."\t","/default.asp\t"));
		// set up stats

	// remove the server
	$plink=str_ireplace(home_url(),'',$plink);
	// now get rid of the slashes
	$reason=$plink;
	$plink=trim($plink);
	$plink=trim($plink,'/');

	$plink=str_replace('/','-',$plink); // this way the taxonomy becomes part of the search
	$plink=str_replace('%20','-',$plink); // spaces are wrong

	$ref='';
	if (array_key_exists('HTTP_REFERER',$_SERVER)) $ref=$_SERVER['HTTP_REFERER'];
	$ref=esc_url_raw($ref);
	$ref=strip_tags($ref);
	$ref=remove_accents($ref);
	$ref=kpg_pf_really_clean($ref);

	$agent='';
	if (array_key_exists('HTTP_USER_AGENT',$_SERVER)) $agent=$_SERVER["HTTP_USER_AGENT"];
	$agent=strip_tags($agent);
	$agent=remove_accents($agent);
	$agent=kpg_pf_really_clean($agent);
	$agent=htmlentities($agent);
	$request=$flink;
	$request=esc_url_raw($request);
	$request=strip_tags($request);
	$request=remove_accents($request);
	$request=kpg_pf_really_clean($request);

	// set up stats
	$r404=array();
	$r404[0]=date('m/d/Y H:i:s',time() + ( get_option( 'gmt_offset' ) * 3600 ));
	$r404[1]=$request;
	$r404[2]=$ref;
	$r404[3]=$agent;
	$r404[4]=$_SERVER['REMOTE_ADDR'];
	$r404[6]='';
	// testing an ignore for the category
	if (strpos($plink,"/category/")!==false) {
		$r404[6]='/category/ is not redirected.';
		kpg_find_permalink_error_log($options,$e404,$r404,$stats);
		return;
	}

	// do not mess with robots trying to find wp-login.php and wp-signup.php
	if (strpos($plink."\t","/wp-login.php\t")!==false||strpos($plink."\t","/wp-signup.php\t")!==false||strpos($plink."\t","/feed\t")!==false){
		$r404[6]='$plink is probably a robot looking for exploits.';
		kpg_find_permalink_error_log($options,$e404,$r404,$stats);
		return;
	}

	// check for bypassed or generated files
	if ($chkrobots=='Y'&&strpos(strtolower($plink)."\t","robots.txt\t")!==false) {
		// looking for a robots.txt
		// header out the .txt file
		$r404[6]='display tobots.txt';
		kpg_find_permalink_error_log($options,$e404,$r404,$stats);
		header('HTTP/1.1 200 OK');
		header('Content-Type: text/plain');
		echo $robots;
		exit();
	}

	if ($chkcrossdomain=='Y'&&strpos(strtolower($plink)."\t","crossdomain.xml\t")!==false) {
		// looking for a robots.txt
		// header out the .txt file
		$r404[6]='display crossdomain.xml';
		kpg_find_permalink_error_log($options,$e404,$r404,$stats);
		header('HTTP/1.1 200 OK');
		header('Content-Type: application/xml');
		echo '<'.'?xml version="1.0"?'.">\r\n"; // because of ? and stuff need to echo this separate
		?>
<!DOCTYPE cross-domain-policy SYSTEM "https://www.macromedia.com/xml/dtds/cross-domain-policy.dtd">
<cross-domain-policy>
<allow-access-from domain="<?php echo $_SERVER["HTTP_HOST"]; ?>" />
</cross-domain-policy>
		<?php
		exit();
	}

	if ($chkicon=='Y'&&strpos(strtolower($plink)."\t","favicon.ico\t")!==false) {
		// this only works if the favicon.ico is being redirected to wordpress on a 404
		$f=dirname(__FILE__)."/includes/favicon.ico";
		if (file_exists($f)) {
			if (function_exists('header_remove'))header_remove();
			ini_set('zlib.output_compression','Off');
			header('HTTP/1.1 200 OK');
			$r404[6]='display favicon.ico';
			kpg_find_permalink_error_log($options,$e404,$r404,$stats);
			header('Content-Type: image/x-icon');
			header('Content-Disposition: attachment; filename="favicon.ico"');
			readfile($f);
			exit();
		}
	}
/*
	apple-touch-icon-57x57-precomposed.png
	apple-touch-icon-57x57.png
	apple-touch-icon-precomposed.png
	apple-touch-icon.png
*/
	if ($chkicon=='Y'&&(strpos(strtolower($plink)."\t","apple-touch-icon.png\t")!==false
			||strpos(strtolower($plink)."\t","apple-touch-icon-57x57.png\t")!==false
			||strpos(strtolower($plink)."\t","apple-touch-icon-precomposed.png\t")!==false
			||strpos(strtolower($plink)."\t","apple-touch-icon.png\t")!==false
		)
	) {
		// this only works if the favicon.ico is being redirected to wordpress on a 404
		$f=dirname(__FILE__)."/includes/apple-touch-icon.png";
		if (file_exists($f)) {
			if (function_exists('header_remove'))header_remove();
			ini_set('zlib.output_compression','Off');
			$r404[6]='display apple-touch-icon.png';
			kpg_find_permalink_error_log($options,$e404,$r404,$stats);
			header('HTTP/1.1 200 OK');
			header('Content-Type: image/png');
			readfile($f);
			exit();
		}
	}

	if ($chksitemap=='Y'&&strpos(strtolower($plink)."\t","sitemap.xml\t")!==false) {
		// if there is no sitemap, return the last 20 entries made
		$r404[6]='display sitemap.xml';
		kpg_find_permalink_error_log($options,$e404,$r404,$stats);
		header('HTTP/1.1 200 OK');
		header('Content-Type: application/xml');
		$sitemap=kpg_pf_sitemap();
		exit();
	}
	if ($chkdublin=='Y'&&strpos(strtolower($plink)."\t","dublin.rdf\t")!==false) {
		// dublin.rdf is a little used method for robots to get more info about your site
		$r404[6]='display dublin.rdf';
		kpg_find_permalink_error_log($options,$e404,$r404,$stats);
		header('HTTP/1.1 200 OK');
		header('Content-Type: application/xml');
		echo '<'.'?xml version="1.0"?'.'>'; // because of ? and stuff need to echo this separate
	?>
 <rdf:RDF xmlns:rdf="https://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:dc= "https://purl.org/dc/elements/1.1/">
 <rdf:Description rdf:about="<?php echo get_home_url(); ?>">
 <dc:contributor><?php echo get_bloginfo('name'); ?></dc:contributor>
 <dc:date><?php echo date('Y-m-d',time() + ( get_option( 'gmt_offset' ) * 3600 )); ?></dc:date>
 <dc:description><?php echo get_bloginfo('description'); ?></dc:description>
 <dc:language><?php echo get_bloginfo('language'); ?></dc:language>
 <dc:publisher></dc:publisher>
 <dc:source><?php echo get_home_url(); ?></dc:source>
 </rdf:Description>
 </rdf:RDF>

	<?php
		exit();
	}
	if ($chkopensearch=='Y'&&(strpos(strtolower($plink)."\t","opensearch.xml\t")!==false||strpos(strtolower($plink)."\t","search.xml\t")!==false)) {
		// search.xml may hel people search your site.
		$r404[6]='display opensearch.xml';
		kpg_find_permalink_error_log($options,$e404,$r404,$stats);
		header('HTTP/1.1 200 OK');
		header('Content-Type: application/xml');
		echo '<'.'?xml version="1.0"?'.">\r\n"; // because of ? and stuff need to echo this separate
	?>
 <OpenSearchDescription xmlns="https://a9.com/-/spec/opensearch/1.1/">
 <ShortName><?php echo get_bloginfo('name'); ?></ShortName>
 <Description>Search this site</Description>
 <Image>favicon.ico</Image>
 <Url type="text/html" template="<?php echo get_home_url(); ?>/seach"/>
 </OpenSearchDescription>

	<?php
		exit();
	}
	// some file types should not be included. these files are true 404s and WordPress can't fix that.
	$ignoreTypes=array(
	'jpg',
	'gif',
	'png',
	'pdf',
	'txt',
	'asp',
	'php',
	'cfm',
	'js',
	'xml',
	'php',
	'mp3',
	'wmv',
	'css'
	);
    foreach ($ignoreTypes as $it) {
		if(strpos(strtolower($plink)."\t",'.'.$it."\t")!==false) {
			$r404[6]="request for non WP file:.$it";
			kpg_find_permalink_error_log($options,$e404,$r404,$stats);
			return;
		}
	}

	// santize to get rid of all odd characters, including cross browser scripts.
	$plink=strtolower($plink); // make it case insensitive
	// do some more cleanup
	$plink=urldecode($plink);
	$plink=strip_tags($plink);
	$plink=remove_accents($plink);
	$plink=kpg_pf_really_clean($plink);
	$plink=str_replace('_','-',$plink);
	$plink=str_replace(' ','-',$plink);
	$plink=str_replace('%20','-',$plink);
	$plink=str_replace('%22','-',$plink);
	$plink=str_replace('/archive/','-',$plink);
	$plink=sanitize_title_with_dashes($plink); // gets rid of some words that wordpress things are unimportant
	// check if the incoming line needs a blogger fix
	if (empty($plink)) {
		// redirect back to siteurl
		$flink=home_url();
		$r404[5]=$flink;

		$r404[6]="empty search, send to home";
		kpg_find_permalink_fixed_log($options,$f404,$r404,$stats);
		wp_redirect($flink.$query,(int)$kpg_pf_301); // let wp do it - more compatable.
		exit();
	}

	if ($labels=='Y') {
		if (strpos($flink,'/labels/')>0) {
			$flink=str_replace('/labels/','/category/',$flink);
			$flink=str_replace('.html','',$flink); // get dir of html and shtml at the end - don't need to search for these
			$flink=str_replace('.shtml','',$flink);
			$flink=str_replace('.htm','',$flink);
			$flink=str_replace('_','-',$flink); // underscores should be dashes
			$flink=str_replace('.','-',$flink); // periods should be dashes
			$flink=str_replace(' ','-',$flink); // spaces are wrong
			$flink=str_replace('%20','-',$flink); // spaces are wrong
			$flink=str_replace('%22','-',$flink); // spaces are wrong
			$flink=str_replace('"','-',$flink); // spaces are wrong
			$r404[5]=$flink;
			$r404[6]="Redirect /label/ to /category/";
			kpg_find_permalink_fixed_log($options,$f404,$r404,$stats);
			wp_redirect($flink.$query,(int)$kpg_pf_301); // let wp do it - more compatable.
			exit();
		}
	}

	// check to see if the user is coming in on a base default

	// now figure if we need to fix a permalink
	//echo "\r\n\r\n<!-- step 2 $find -->\r\n\r\n";
	if ($find>0) {
		$plink=str_replace('.html','',$plink); // get dir of html and shtml at the end - don't need to search for these
		$plink=str_replace('.shtml','',$plink);
		$plink=str_replace('.htm','',$plink);
		$plink=str_replace('.php','',$plink);
		// first check for the original slug - use the wordpress slug fixer on it.
		if (strpos(strtolower($flink)."\t","/index.html\t")!==false) $flink=substr($flink."\t",0,strpos(strtolower($flink)."\t","/index.html\t"));
		if (strpos(strtolower($flink)."\t","/index.htm\t")!==false) $flink=substr($flink."\t",0,strpos(strtolower($flink)."\t","/index.htm\t"));
		if (strpos(strtolower($flink)."\t","/index.shtml\t")!==false) $flink=substr($flink."\t",0,strpos(strtolower($flink)."\t","/index.shtml\t"));
		if (strpos(strtolower($flink)."\t","/default.asp\t")!==false) $flink=substr($flink."\t",0,strpos(strtolower($flink)."\t","/default.asp\t"));
		$flink=basename($flink);
		$flink=str_replace('.html','',$flink); // get dir of html and shtml at the end - don't need to search for these
		$flink=str_replace('.shtml','',$flink);
		$flink=str_replace('.htm','',$flink);
		$flink=str_replace('_','-',$flink); // underscores should be dashes
		$flink=str_replace('.','-',$flink); // periods should be dashes
		$flink=str_replace(' ','-',$flink); // spaces are wrong
		$flink=str_replace('%20','-',$flink); // spaces are wrong
		$flink=str_replace('https://','',$flink);
		$flink=str_replace('https://','',$flink);
		$flink=sanitize_url($flink);
		$flink=str_replace('https://','',$flink);
		$flink=str_replace('https://','',$flink);
		$flink=str_replace('%22','-',$flink); // spaces are wrong
		$flink=str_replace('"','-',$flink); // spaces are wrong

		// check for matches to slugs

		$reason="Found exact slug $flink";
		$ID=kpg_find_permalink_post_direct($flink);
		// check - exact matches on flink
		if ($ID==false) {
			$ansa=kpg_find_permalink_post_exact($flink,$find ,$kpg_pf_numbs ,$kpg_pf_common ,$kpg_pf_short);
			$ID=$ansa[0];
			$cnt=$ansa[1];
			$reason="Found $cnt exact word matches to slug $flink";
		}
		if ($ID==false&&$chkloose=='Y') {
			$ansa=kpg_find_permalink_post_loose($flink,$find ,$kpg_pf_numbs ,$kpg_pf_common ,$kpg_pf_short);
			$ID=$ansa[0];
			$cnt=$ansa[1];
			$reason="Found $cnt loose word matches to $flink";
		}
		if ($ID==false&&$chkfullurl=='Y') {
			$ansa=kpg_find_permalink_post_exact($plink,$find ,$kpg_pf_numbs ,$kpg_pf_common ,$kpg_pf_short);
			$ID=$ansa[0];
			$cnt=$ansa[1];
			$reason="Found $cnt exact word matches to $plink";
		}
		if ($ID==false&&$chkloose=='Y'&&$chkfullurl=='Y') {
			$ansa=kpg_find_permalink_post_loose($plink,$find ,$kpg_pf_numbs ,$kpg_pf_common ,$kpg_pf_short);
			$ID=$ansa[0];
			$cnt=$ansa[1];
			$reason="Found $cnt loose word matches to $plink";
		}

		if( $ID==false && $chkmetaphone=='Y')  {
			// missed on regular words - try a metaphone search?? Only do it on original slug
			$ansa=kpg_find_permalink_post_metaphone( $flink,$find ,$kpg_pf_numbs ,$kpg_pf_common ,$kpg_pf_short);
			$ID=$ansa[0];
			$cnt=$ansa[1];
			if ($ID!==false)
				$reason="Found $cnt metaphone 'sounds-like' word matches to $flink";
			else
				$reason="failed all searches";
		}
		if( $ID==false && $chkmetaphone=='Y' && $chkfullurl=='Y')  {
			// missed on regular words - try a metaphone search?? Only do it on original slug
			$ansa=kpg_find_permalink_post_metaphone( $plink,$find ,$kpg_pf_numbs ,$kpg_pf_common ,$kpg_pf_short);
			$ID=$ansa[0];
			$cnt=$ansa[1];
			if ($ID!==false)
				$reason="Found $cnt metaphone 'sounds-like' word matches to $plink";
			else
				$reason="failed all searches";
		}
		if( $ID!==false)  {
			$r404[5]=get_permalink( $ID );
			$r404[6]=$reason;
			kpg_find_permalink_fixed_log($options,$f404,$r404,$stats);
			wp_redirect(get_permalink( $ID ).$query,(int)$kpg_pf_301); // let wp do it - more compatable.
			exit();
		}
	}
	// still here, it must be a real 404, we should log it
	$reason="Not found - slug:$flink, loose url:$plink";
	//echo "\r\n\r\n<!-- step 5 -->\r\n\r\n";
	$r404[6]=$reason;
	kpg_find_permalink_error_log($options,$e404,$r404,$stats);

	return; // end of permalink fixer
}

Hope this helps somebody else out there facing a similar problem.

I have to figure out how to add this to the next version. The place were you put this changed a little.

I was going to release a new version soon, but I’ll wait until I get your code working.

Thank you very much,

Keith

There is a new beta version available on https://www.blogseye.com for the permalink finder. It has a bunch of changes including this one.

I am copying it to my websites now and I’ll wait a week or so before I release it.

The program is getting quite large and out of hand, so I am not yet confident that everything I changed will work.

Keith

Hey Keith,
Is the one on blogseye the updated one with the changes? I’d be happy to help you test. Let me know.

Regards,
Sidd

Sidd,

The one on Blogseye has your changes in it. You can download it on the beta test page.

Keith