CODEWITHSUNDEEP
phpcommand-line
Mikhail
Mikhail

Reputation: 9007

Running web-intended PHP code from command line

I have PHP code that requires other php files by address $_SERVER['DOCUMENT_ROOT'].'/subdir/file.php';

First of all -- is this the best proper way to include things? Obviously I don't want to use path like '../../subdir/file.php'; because moving the file would break it.

But another interesting issue is that if I run this file through command line then $_SERVER is not created. I can fake it via $_SERVER['DOCUMENT_ROOT'] = '.'; but I'm curious to if this is the best practice. Seems like not.

Edit: Obviously there are many ways to skin this cat, although I think the best practice is to define a variable (or a constant) responsible for the include directory. Such as:

define('INC_DIR', $_SERVER['DOCUMENT_ROOT'].'/../includes');

or

if (PHP_SAPI == 'cli') {
    $_includes = '../includes';
} else {
    $_includes = $_SERVER['DOCUMENT_ROOT'].'/../includes/');
}

And use the aforementioned variable or constant throughout the code.

Upvotes: 0

Views: 605

Answers (3)

Hemaulo
Hemaulo

Reputation: 999

if(PHP_SAPI == 'cli')
{
    $_SERVER['DOCUMENT_ROOT'] = '/path/to/webroot';
}

Upvotes: 0

Brian
Brian

Reputation: 3023

I prefer to use a folder definition system in my architectures. Something like this:

define( 'DIR_ROOT',dirname(__FILE__) );

That works both in command line and web mode. Use that in your application entry point (index.php in most cases) and then load the rest of your framework from that file outward. All inbound calls to your application should be routed via .htaccess or other method so that they call index.php?foo=bar etc.

I also hate typing DIRECTORY_SEPARATOR all the time so I usually make the first definition:

define( 'DS' , DIRECTORY_SEPARATOR );

This enables you later to do something like this:

require_once( DIR_ROOT.DS.'myfolder'.DS.'myfile.class.php' );

Upvotes: 2

sholsinger
sholsinger

Reputation: 3088

Alternatively if you don't want or need to modify your php files and you just need a page to be executed normally you could use curl. Most Linux and Unix systems have it installed.

$ curl http://www.example.com/myscript.php &> /dev/null

The &> /dev/null part sends the output into a black hole in the system so you don't have to see the HTML which was returned by the request.

Upvotes: 0

Related Questions