Question

I would like to strip the leading slash and the querystring from a URL but can’t work out how to do them both. I have this code which works perfectly for stripping the querystring, but it leaves the leading slash

preg_replace('/\?.*$/', '', $_SERVER['REQUEST_URI'])

If my URL is www.mysite.com/myPage?querystring=123, the above leaves me with /myPage. How can I tweak this so I can remove the leading slash too?

Also, can you point me at a resource to help me understand preg_replace pattern matching please?

Answer 1

I might favor PHP’s simple string functions over regex in such a simple case:

$addr  = ltrim($_SERVER['REQUEST_URI'], '/'); // zap a leading slash
$q_pos = strpos($addr, '?'); // get the string position of a '?'
$addr  = $q_pos !== FALSE ? substr($addr, 0, $q_pos) : $addr; // pull out URI

PHP’s strpos()^docs returns an integer value, so it’s possible that $q_pos === 0 … this is why we check $q_pos !== FALSE.

UPDATE

I suppose I should answer the question, though … so to actually use a regex in this situation …

$address = '/test/url/test.php?extra';
$pattern = '{^/?([^\?]+)\?.*$}';
$replace = '$1';
$address = preg_replace($pattern, $replace, $address);
echo "$address"; // outputs: test/url/test.php

How does this work? Well … our pattern specifies a capture group ([^\?]+) using the parentheses that grabs everything after an optional forward slash /? up to the first occurrence of an optional \? in the string. Note that we escape the actual question mark character with a backslash because it has meaning in the context of regex patterns. The final part of the regex pattern .* simply matches zero or more characters out to the end of the string.

Finally, our replacement simply specifies the $1 to reference the text we captured with our original parentheses grouping ([^\?]+).

One other thing to note that regex novices often fail to realize is that you aren’t required to use / as pattern delimiters. In a case like this where we’re matching actual forward slash characters I use something else (like the curly braces).

I usually point regex beginners to this link to help them get started.

UPDATE 2

The regex above assumes that there is always going to be a query string, so if you run up against a URI that doesn’t have one, (for example, /All-Products), that regex won’t work. To account for this, simply alter your pattern to make the query string optional:

$pattern = '{^/?([^\?]+)(?:\?.*)?$}'; // use an optional non-capturing group

-or-

$pattern = '{^/?([^\?]+)\??.*$}'; // make the escaped ? optional