Reasons

Two main reasons should drive you towards rewriting your URLs into friendly URLs. The most important one is so search engines can index your pages, because most search engines do not index dynamic URLs that contain question mark (?) or equal sign (=). The other main reason is usability. By creating friendly URLs for a site you make it usable. If you care to read more about the usability of this case, you can read Adaptive Path's User-Centered URL Design or Adam Baker's theory on How to make URLs user-friendly.

Implementation

Although the idea of rewriting URLs is now old, and a MUST have in a dynamic web site or web-based application, I still see sites using the index.php?page=X URL structure.

I think every web-based application or site should use a global loading and unloading script. By that I mean that every script that is called by the browser should call the loading script at the beginning and the unloading script at the end. This way you wrap your scripts with a customizable set of operations both at the start and at the end of them.

Using $_SERVER['PATH_INFO'] is the method I rely on when creating friendly URLs. When a request is made to a PHP-enabled server, PHP fills the $_SERVER global array with several variables regarding the server's enviroment and the request made. Among these variables the PATH_INFO one contains a certain part of the URL requested that is between the actual URL path and the query string of the URL. Let me explain with some examples. Let's take this URL for example:

For this request, the URL path would be /path/to/script.php and the query string ?var1=value1&var2=value2. Note that there is not any information that would go in the PATH_INFO variable. But if you take this next example:

For this request, the URL path and query string are the same as the above, but also the PATH_INFO variable contains /foo/bar/.

Planning our URLs

Now that you figured out what would come in your help. Now decide how are you going make your new URLs. Suppose you have this kind of URLs:

http://www.example.com/categories.php?cat_id=3
http://www.example.com/articles.php?art_id=15&page=2

You would want to try and convert them to something like this:

http://www.example.com/categories.php/cat_id/3/
http://www.example.com/articles.php/art_id/15/page/2/

But wait, that is not enough, because many search engines would still not index you properly. The problem is with the .php extension in the URL followed by /. So what you need to do is transform them into the following:

http://www.example.com/categories/cat_id/3/
http://www.example.com/articles/art_id/15/page/2/

To do this, you need to copy or move the categories.php and articles.php scripts to categories and articles. Now you have to tell the server that those 2 files need to be parsed by PHP. You can do this by creating a .htaccess file in your root directory (or in the same directory you want those kind of files to be parsed by PHP). Then you write in that file:

<Files ~ "categories|articles">
ForceType application/x-httpd-php
</Files>

Ok, now those 2 files are now parsed by PHP. Give it a try. Try to access http://www.example.com/categories?cat_id=3. It will have the same effect as trying to access categories.php?cat_id=3.

Parsing the URL

How you parse the URL (that is the data contained in $_SERVER['PATH_INFO']) is the most important part. For this you must put the method I am going to present below in the global loading script.

if (isset($_SERVER['PATH_INFO'])) {
	$url = substr($_SERVER['PATH_INFO'], 1);
	$urlParts = explode('/', $url);
	if ($urlParts[count($urlParts) - 1] == '') 
		array_pop($urlParts); 

	$urlPartsCount = count($urlParts);
	if ($urlPartsCount % 2 != 0) { 
		$urlPartsCount++; 
	}
	for ($i = 0; $i < $urlPartsCount; $i += 2) { 
		$_GET[$urlParts[$i]] = $urlParts[$i + 1];
	}
}

The next level

Now you have a functional PHP-based friendly URL interpreting engine. You have two choices now: either you hard code every URL in your links to the new pattern (example: /articles/art_id/15/page/2/), either you leave every link intact and build a PHP-based URL rewrite engine. Since this post is named as such, I will continue describing how to achieve this.

It is done by simply by using the preg_replace_callback function. You feed it an array of URL patterns, which you'll have to construct based on every URL in your application you wish to convert to friendly URL, and a replace callback function. First the URL patterns array:

$urlPatterns = array(
	'~'.preg_quote(BASE_DIR).'([^\.]+)\.php(\?([0-9a-zA-Z]+[^#"\']*))?~i',
);

The BASE_DIR constant I used in this example is the base dir of your application, it can be either / or the entire http://www.example.com/ depending on how you wrote your links in the application.

Next step is that you need to catch all the contents of one request before sending it to the browser. You can do this with the help of the output buffering functions builtin the PHP. For more on output buffering read Output Control Functions on PHP.net. First we insert in the global loading script:

ob_start();

And in the global unloading script:

$pageContents = ob_get_contents();
ob_end_clean();
echo preg_replace_callback($urlPatterns,'urlRewriteCallback',$pageContents);

The final task is creating the callback function to handle the replacement of the URLs.

function urlRewriteCallback($match) {
	$extra = '';
	if ($match[3]) {
		$params = explode('&', $match[3]);
		if ($params[0] == '') array_shift($params);
		foreach ($params as $param) {
			$paramEx = explode('=', $param);
			$extra .= $paramEx[0].'/'.$paramEx[1].'/';
		}
	}
	return BASE_DIR.$match[1].'/'.$extra;
}

Now you should have a ready-to-work PHP-based URL rewriting engine.

Thinking ahead

Of course you souldn't stop here. You can do a lot of more things to imporve the URLs.

For example you could drop the variables names at a first glance. But you should map all the variables you've got in a request to certain variables in your application. That way you get:

http://www.example.com/categories/3/
http://www.example.com/articles/15/2/

For this to work you must modify the parsing of the PATH_INFO in the loading script. Using a switch on the basename($_SERVER['SCRIPT_NAME']) you can map the URL data into different variables your application uses.

And another good thing to do is to lose the number ids and to replace them with words. For exmaple instead of 3 put the title of the category Web development. This way you will have even more friendly URLs:

http://www.example.com/categories/web-development/
http://www.example.com/articles/rewriting-dynamic-urls/2/

I hope this post helps everyone who reads it. And let's pray together for a web made of pretty URLs. Any comments on my method would be appreciated.

Comments

echo preg_replace_callback($urlOld, 'urlRewriteCallback', $pageContents);

Options +MultiViews
DirectoryIndex index index.php index.html

<Directory "path/to/localhost">

AllowOverride All