Help with require and file hierarchies

sander

In know there is probably a simple solution to this that I am overlooking, but what's the easiest way to require() or include() various configuration and definition files into files that themselves are at varying levels of the file hierarchy? Many of the required files themselves call require() to insert files relative to their own file hierarchy position, but this gets screwed up since their position changes to that of the calling file. Do I have to fully specify the paths in all calls?
There must be a more elegant solution, and I'd rather not have to fully specify paths in case I need to change locations/servers down the road.

Surely there is an elegant approach to this!

Thanks,

Sander

merkinmuffley

you can use a built-in PHP global variable that references the root of your html documents, such as $DOCUMENT_ROOT. you'd write your include() calls like
include ($DOCUMENT_ROOT."/path/to/whatever.php");

now, i think that $DOCUMENT_ROOT is specific to Apache, so that might not help you. but, to see lists of all the global server variables like this one, and what their values are, just make a page that does nothing but call php_info().

sander

Merkin,

THanks for that. I was actually starting to play with $DOCUMENT_ROOT, but I am still stumbling on the fact that the browser is interpreting the resulting path as relative to the location of my website. That is, $DOCUMENT_ROOT returns the absolute path from the ROOT of the server on which my site resides, wheras the browser takes any URL as relative to my site's location, which is already partway down the path FROM the server's root. It wants to insert my website URL in front of the path, so it doesn't work!

(I hope I am making myself clear..)

I'm kinda scratching my head over this one, actually.

Ideally, I'd really like to be able to use relative pathnames anyway, except that the paths keep shifting according to where the calling ("requiring()") files themselves reside!

-- sander

merkinmuffley

well, you're not going to get to use relative paths, for the exact reason you described in your first post. when you include a file from an included file, the path is always relative to the highest-level script.

what i don't understand is the part about how something "wants to insert my website URL in front of the path"... what is this something? it certainly isnt due to any built-in functionality of php. i think you're confusing the actual file tree on the webserver with the file tree as your browser sees it. when you're working with include() or require(), forget about web browsers entirely, and concentrate on how the files are actually arranged on the webserver.

maybe your web scripts are rooted at /user/www (just an arbitrary example). let's say you have 2 files include.inc and functions.inc, both sitting in the /user/www folder. include.inc is going to be include()-ed by lots of different scripts in lots of different directories below /user/www, and include.inc needs to make an include() call to functions.inc. that include() call would look like:

// inside include.inc
include ($DOCUMENT_ROOT."/functions.inc");

The $DOCUMENT_ROOT variable gets replaced with "/user/www" each and every time, regardless of where the calling script is on the server. this has absolutely nothing to do with the fact that the web browser sees the directory /user/www as the root ("/") of the web site. hope this makes things a little clearer...

ednark

try something similar to this.... i named it acquire... it starts in the directory the script was called in.. then moves one directory up and looks for the file... etc etc...

at the moment it just prints out the file iinstead of including/requirign it... but whatever

usage:

acquire( "file_name.php", "start/path/", "/web/server/root" );

look for 'file.php' from here up to three directories down.

acquire( 'file.php', './', '../../../' );

look for 'file.php' from here ot the web root.

acquire( 'file.php', './', '/www/htodcs/' );

function acquire( $file_name, $current_dir=NULL, $basement=NULL )
{
    echo "\n\nCALL TO ACQUIRE() MADE\n";

if ( $current_dir === NULL ) {
    $current_dir = $SERVER['PATH_TRANSLATED'];
}

if ( $basement === NULL ) {
    $basement = $current_dir;
}

$current_dir  = realpath($current_dir);
$basement     = realpath($basement);

echo "    Current Dir  = '{$current_dir}'\n";
echo "    Basement Dir = '{$basement}'\n\n";

if ( file_exists("{$current_dir}/{$file_name}") )
{
    echo "    FOUND '{$file_name}' at '{$current_dir}/{$file_name}'\n";
    echo "\n\n-------------------------\n\n";
    readfile("{$current_dir}/{$file_name}");
    echo "\n\n-------------------------\n\n";
    return TRUE;
}

echo "    NO '{$file_name}' in '{$current_dir}'\n";

if ( $basement !== $current_dir ) {
    $next_dir = preg_replace( '/[\w\d_]+[\\\\\/]?$/', '', $current_dir );
    echo "    TRYING AGAIN WITH '$next_dir'";
    return acquire( $file_name, $next_dir, $basement);
}

echo "    We hit our basement and no file found, it doesn't exist\n";
return FALSE;

}

merkinmuffley

ed, if you already know where the file is, and it's not going to get up and walk away, why would you need to do all that extra work? sander's trouble isn't that he can't find the file to include... your function seems very extraneous to me in this case (but a useful utility for other purposes).

ednark

well if you just keep a lib/ or library directory or something like that then its always just best to point straight there ,as you said...

but this technique is what Zope uses (in concept) to do includes...

good for templates...

make a file /header.php. put it in your base root directory...
make a 2nd header.php and place it in a heigher up directory

/header.php
/special/header.php

now all files that use acquire('header.php')

will find the base directory one, or if they are in the /special/ tree they will hit the special one instead and include that. you can place files anywhere in your tree without haveing to rewrite a single line of code... and they will switch which header they use

allows you to have multiple files of the same ilk for various directories of your site... allows you to think directory oriented when developing your site...

however... not necessary at all... but is one solution to not knowing relative paths to files... just stick them in some directory above you

sander

Merkin, Ednark, thanks a lot for your interest here (and sorry I've been away from the computer to respond...)

Ednark, it seems to me that to use your acquire function, that would itself (presumably) be in some kind of included/required file, I would still have the same problem getting to IT'S definition, no? I think I agree with Merkin's assessment.

Merkin, what I was trying to descibe before (and which puzzles me) is that the html which is generated by php with $DOCUMENT_ROOT appears to be correct, e.g. /home/user/www/..whatever, but the browser isn't finding it and, when I look at the properties for the link (or image), the browser is displaying the url as http://user.com/home/user/www..! (which is obviously incorrect), despite the "correct" HTML!
That's what I was trying (ineffectually, I admit) to describe before.

In other words, the browser seems to be insisting upon the location of my website's "root" directory on its server as being the root directory of the server itself, so when I pass it the absolute path (from the server's root), it instead appends it onto my website's location root (I really hope I'm making sense here, even if the behavior I'm observing doesn't!)

So I'm rather flummoxed by this! What's going on here?
Either me or my computer must be on drugs...

Thanks,

Sander

bobthedog

I dunno if this is answering your question or not, but I make some general assumptions when performing something like this, one being that I know where an include file will be realtive to myself... i.e in a subdir named "inc" if i am a main script, or in the same subdir as me if i am in included file.

So - I normally use the following in "inc" files which may include other "inc" files (not knowing where the original file will be included from).

<?php
include_once(dirname(__FILE__) . '/dbconn.inc.php');

....

?>

or to include from another dir

<?php
include_once(dirname(__FILE__) . '/../somedir/class.mine.inc.php');

....

?>

Anyhow - i've probably just explained that really badly... but it may help ya 🙂

[EDIT]
If you didnt know FILE points to the actual file the statement is in, not the file including that file, so dirname(FILE) gives you the directory name where your included file resides, so you can do the include relative to the included file.
[/EDIT]

merkinmuffley

sander,
i'm confused. i thought that your problem was related to using the include() function, but now you're talking about browsers, images, and links (hrefs?), none of which should have anything at all to do with the include() function. if you can, make a small example that exhibits this weird behavior and post the example. or, you could use bob's suggestion, which sounds like it would work just fine...

bob,
that's a new one for me, is that documented anywhere or is it a sort of unannounced feature.

sander

Merkin,

YOU'RE CONFUSED! (Tell me about it!)
No, actually, you're right, I guess I've slightly changed the issue here, haven't I? I AM trying your $DOCUMENT_ROOT suggestion.
I think that I realized that the $DOCUMENT_ROOT isn't working because of the (additional?) problem I've enountered, which applies more generally (hence, the HREFs, etc. which I used with $DOCUMENT_ROOT just as an additional test). I really don't know why the browser sees the absolute path from the root of the server (whether for included files or just for linked files/images) as being appended to my websites location--as I've described). Is that totally weird, or what? Some bizzare Apache setting? Am I imagining the whole thing? Is this some kind of dream....

As for the original issue, and, yes your solution should really work here, I've discovered a plethora of possible solutions at the manual site (http://uk.php.net/manual/en/function.include.php) in the "user contributed notes", where there is quite the lively discussion on the topic (including your $DOCUMENT_ROOT, natch, as well as Bob's).

So, forgive me if I've wavered off topic (on my own thread, yet!), but I'm REALLY puzzled.

But, sure, have a look at http://artcourant.com/main/test.php and tell me whether either the image loads or the link works (Neither does for me), both of which use $DOCUMENT_ROOT.
(/home/artcoura/public_html is the location of my website).
(The website itself is temporarily, crippled, BTW, while I'm futzing with all this-- no big concern for me...)

In the end it will probably turn out to be some simple, dumb thing I've overlooked, like not being able to find the "any" key!

Cheers,

Sander

merkinmuffley

ok, right now that page just has the phpinfo() for your site. but, let's try a different tactic. i'm going to make up an example of some scripts, and maybe you can apply the example to solving your problem. this example entails 4 files: hello.php, main/index.php, include/myInclude.inc, include/functions.inc; all under your /home/artcoura/public_html directory

/home/artcoura/public_html/hello.php

<?
include_once ($DOCUMENT_ROOT."/include/myInclude.inc");
?>
<html>
<body>
Hello, user.  The current server time is <?= myCurrentTime() ?>.
<p>
Please <a href="/main/index.php">click here</a> to go to another page.
</body>
</html>

/home/artcoura/public_html/main/index.php

<?
include_once ($DOCUMENT_ROOT."/include/myInclude.inc");
?>
<html>
<body>
Hello again, user.  Thanks for clicking that link.  As a reward, 
upon your death bed you will receive total consciousness.
<p>
Also, the current server time is <?= myCurrentTime() ?>.

/home/artcoura/public_html/include/myInclude.inc

<?
include_once ($DOCUMENT_ROOT."/include/functions.inc");
?>

/home/artcoura/public_html/include/functions.inc

<?
function myCurrentTime()
{
    return date("h:i a");
}
?>

The moral of the story is this: $DOCUMENT_ROOT is only for creating paths to files so PHP can find them. When I made an href, I didn't use it; if I has included an image tag, I wouldn't have used it there; because those paths are for the web browser, not PHP. But I did use it for every call to include(). Did that help, or is it still clear as mud?

sander

WOw! Okay, THAT worked! Very nice.

(I am quite puzzled by your getting the phpinfo() func, since that has been commented out since yesterday...)!

But in any case, if I understand correctly, PHP will access directories on the server from the root (allowing for permissions, natch), wheras the browser (HTML code, that is) will only access from the virtual root of the website itself, yes? So I can still use constructs such as "../.." to navigate dirs in HTML, but only as far up (or down, depending upon your perspective) as my website's home (root) directory (at /home/artcoura/public_html in my case).

Funny that I hadn't quite grasped that, isn't it?

So, since I need to access directories of images (for example) from (again) various directory levels, rather than using DOCUMENT_ROOT for the HTML, I guess I will use DOCUMENT_ROOT for accessing a config file within PHP that will set various "html-friendly" (assuming "/home/artcoura/public_html" as the root of the path--a sort of "HTML_DOCUMENT_ROOT", if you will) variables which can be then php'ed to provide the path to my other directories.

Does that make sense? Have I got it?

(I certainly hope so, and will play around with it now....)

WOw, thanks, Merkin (and others too..)!

Sander

(geez....)

sander

Okay, and scratch that bit about creating a "HTML_DOCUMENT_ROOT" since, in the pure HTML code, that's what a beginning '/' is!

Things seem cool now...