The methods/functions in the Fileinfo PHP extension "try to guess the content type and encoding of a file by looking for certain magic byte sequences at specific positions within the file. While this is not a bullet proof approach, the heuristics used do a very good job."¹ (Note that this module replaces the Mimetype module.)

The primary use I see for Fileinfo (and the reason I installed it last weekend!) is to test the MIME types of uploaded files against an array of allowed MIME types. I intend to use Fileinfo to supplement the standard check against the file type specified in the $_FILES array, since this is known to be unreliable (more on this in a future post!).

The Fileinfo extension is extremely simple to use. It can be implemented in either the object oriented or procedural style of programming, though my code examples will only demonstrate the former. As an object, we only need to worry about the constructor and two methods to test the MIME type; in the procedural style there are three functions to be aware of.

Here is a quick example from the PHP manual of how to use Fileinfo in the object oriented style:

<?php
$finfo = new finfo(FILEINFO_MIME, '/usr/share/misc/magic'); // return mime type ala mimetype extension
if (!$finfo) {
echo 'Opening fileinfo database failed';
exit();
}
/* get mime-type for a specific file */
$filename = '/usr/local/something.txt';
echo $finfo->file($filename);
/* close connection */
$finfo->close();
?>

The above example should output text/plain, however there are a few "gotchas" I'll mention after I briefly discuss the various methods.

The Fileinfo constructor takes two optional parameters: a Fileinfo constant and the path to the "magic" database file. While the manual states that these are optional, I haven't had any luck getting Fileinfo to work on Media Temple without specifying both! The code I use:

$finfo = new finfo(FILEINFO_MIME, '/usr/share/file/magic');

The first method to know about is file, which is used to get the MIME type of a file. It takes one required and two optional parameters. The required parameter is a string file name:

$finfo->file('path/to/file.ext');

The second method is close, which supposedly closes the resource created by the constructor.

As you can see, quite simple. However, as I mentioned above, not everything works as the documentation states! I'm not sure if this is a general problem with the module or simply the version I'm using (1.0.4 under PHP 5.2.6), but there are two errors:

1. The constructor is supposed to return "a magic database resource on success or false on failure."² In the procedural style it does, but in the object oriented style it always returns an object, so the if (!$finfo) { } condition in the example above always returns true. This is a problem if the path to the magic database file is wrong, because calling the file method without a magic database resource will throw an error.
2. In both the object oriented and procedural styles, calling the close method triggers a fatal error. It seems the method/function simply isn't defined, despite what the documentation claims!

To solve these issues and to add a little functionality, I wrote a Fileinfo wrapper class that duplicates (some of) the native Fileinfo functionality. Unfortunately, the class is currently only available for PHP 5 and the documentation is poor (I intend to remedy both this weekend!).

To use this class, include it in your PHP file. The constructor, file and close methods all work similar to Fileinfo, only I provide a property to test for failure and my close method doesn't throw a fatal error! I also add a new method, test, that allows one to test a given file against an array of allowable MIME types.

Here's a sample of how to use it:

<?php
include_once('fileinformation.php');
$finfo = new finformation(FILEINFO_MIME, '/usr/share/file/magic');
if ($finfo->ready) {
echo '<p>' . $finfo->file('themechanism-144x60.gif') . '</p>';
if ($finfo->test('themechanism-144x60.gif', array('image/gif'))) {
echo '<p>This is a valid file.</p>';
} else {
echo '<p>This is not a valid file.</p>';
}
}
?>

If you have a file named "themechanism-144×60.gif" on your server and if it's a gif, this should output <p>image/gif</p><p>This is a valid file.</p>, otherwise it will output <p>This is not a valid file</p>.

References

1. http://us.php.net/manual/en/intro.fileinfo.php
2. http://us.php.net/manual/en/function.finfo-open.php

As of PHP 5.3.0, the Fileinfo extension has been enabled by default and is no longer maintained at PECL. However, the Media Temple (dv) 3.5 service only includes PHP 5.2.6, so we'll need to manually install it.

Please note that prior to installing Fileinfo, you'll need to make sure that root access has been enabled on your (dv) and that the developer tools have been installed. Both of these actions can be completed from the Media Temple AccountCenter. For more information on root access, please review the first two articles in the references/more info section below. The third link is to an article on the developer tools.

Once you're sure that you have root access and that the developer tools are installed, SSH into your (dv). If Media Temple had the PECL binary installed, we could easily install Fileinfo by using the following statement from the command line: pecl install fileinfo.

Since we can't, we'll need to install Fileinfo from source. From the command line as root, enter the following statements:

mkdir /var/pecl-install
cd /var/pecl-install
wget http://pecl.php.net/get/Fileinfo-1.0.4.tgz
tar -zxf Fileinfo-1.0.4.tgz
cd Fileinfo-1.0.4
phpize
./configure
make
make install
cd /var
rm -f -r pecl-install

First we download and unpack the extension in our working directory, /var/pecl-install. Next we phpize the contents (ie prepare the build environment for a PHP extension). Once that's done, we perform a standard configure and make routine to install the module. The final two commands remove our working directory from the server.

The outcome is that our extension, fileinfo.so, should be created and exist in /usr/lib/php/modules. However, PHP still doesn't know about it. To activate the extension we need to add a line to our php.ini file.

From the command line, enter vi /etc/php.ini. Scroll down until you hit the extensions section of the file and then press i to enter vi's insert mode. Add extension=fileinfo.so and then press <esc> to exit insert mode and return to command mode. Type :wq<return> to exit vi and save your changes.

The final step to activate the Fileinfo extension is to restart the web server. From the command line enter:

/etc/init.d/httpd stop
/etc/init.d/httpd start

At this point, if you were to run phpinfo() on your server, you'd see that fileinfo is listed among the loaded modules, and if you tried to use any of the Fileinfo functions, you'd find they'd run without error.

However, if you actually try to use Fileinfo to determine the MIME type of a file, you'll discover it still doesn't work! The reason why is that Fileinfo depends on a "magic" file to determine the MIME type and, on a Media Temple (dv), it cannot access this file without our next (and final) step.

By default, Media Temple restricts PHP from accessing files outside of httpdocs using the open_basedir directive. However, to use Fileinfo, we have to let PHP access the directory that contains the "magic" file, in this case /usr/share/file/.

So, from the command line as root, enter (replacing your-domain.com with your actual domain!):

cd /var/www/vhosts/your-domain.com/conf/
vi vhost.conf

Enter vi's insert mode and add the following lines to the empty vhost.conf file:

<Directory "/var/www/vhosts/your-domain.com/httpdocs">
php_admin_value open_basedir "/var/www/vhosts/your-domain.com/httpdocs:/usr/share/file:/tmp"
</Directory>

Exit vi (saving your changes!) and reconfigure the server to look for the new vhost.conf file by entering:

/usr/local/psa/admin/sbin/websrvmng --reconfigure-vhost --vhost-name=your-domain.com

Restart the web server once again:

/etc/init.d/httpd stop
/etc/init.d/httpd start

And that's it! Fileinfo should be now be successfully installed and activated! For more info on changing the open_basedir configuration, please see the last link in the resources/more info section below.

References/More info

1. Media Temple KB. (dv) An introduction to the root user
2. Media Temple KB. Disabling SSH Login for root user
3. Media Temple KB. (dv) 3.5 Tech Specs – Developer’s Tools package listing.
4. Jelly and Custard. Installing PECL Modules
5. Media Temple KB. How can I edit php.ini on the (dv) & (dpv) Dedicated-Virtual Servers?
6. Media Temple KB. (dv) HOWTO: Enable PEAR/Set open_basedir and include_path.

To correctly serve XML, call the header function with the following arguments:

header('Content-Type: text/xml; charset=utf-8');

Correctly serving XHTML is a bit more complicated. The $_SERVER array contains the server variables, allowing us to interrogate the Accept HTTP header:

header('Vary: Accept');
if (stristr($_SERVER[HTTP_ACCEPT], 'application/xhtml+xml')) {
header('Content-Type: application/xhtml+xml; charset=utf-8');
} else {
header('Content-Type: text/html; charset=utf-8');
}

More information on correctly serving XHTML.