Hey guys,

I'm modifying a content management system written in PHP which uses quite unattractive URLs like this: http://domain.com/cms/website.php?id=/de/index.htm

Using the following replacement rules I managed to convert the addresses to something like this: http://domain.com/de/index.htm

$strBuffer = preg_replace( '#([\"\'])website\.php\?id=(/.+)(?:&|&)(.+)(\\1)#','\\1\\2?\\3\\4', $strBuffer );
$strBuffer = preg_replace( '#([\"\'])website\.php\?id=(/[^\\1]+)(\\1)#U','\\1\\2\\3', $strBuffer );
$strBuffer = preg_replace(' #([\"\'])\.\.(/.+\\1)#U','\\1\\2', $strBuffer );

So far everything works fine. However, some features of the CMS require adding one or more parameters to the URL, for instance:

http://domain.com/cms/website.php?id=/de/search.htm&action=search&searchterm=seo&tagsearch=1 

Since all additional values except id may vary in any order, I want to leave them untouched, only changing the first '&'-character to a question mark so the address remains valid:

http://domain.com/de/search.htm?action=search&searchterm=seo&tagsearch=1 

However, with the current expressions I get this broken URL: http://domain.com/de/search.htm&action=search&searchterm=roman?amp;tagsearch=1 
Can anyone tell me how to fix this?

Thank you in advance
Michael 

It looks like you use a greedy operator within your regular expression

.....       =(/.+)(?:&|&)(.+)(   .....................

(/.)+ does match the longest string, and only the rightmost & will be matched by your rule. You want the first one &  (leftmost).

The following lazy expressions are equivalent, the first one being likely the most performant one (less backtracking on most implementations)

(/[^&]+)(?:&|&)

(/.+?)(?:&|&)

Does it solve your problem ?