Forum: NGINX How to remove the "IF" in this fcgi config

Hi, I'm trying to setup a php app using fpm (owncloud).

I am trying to match urls which can all over the filesystem and of the
form: something.php/some/path?params

So far I have something like this:


                 location / {
                         try_files $uri $uri/ index.php;
                 }

                 location ~ ^(?P<script_name>.+\.php)(/|$) {
                         fastcgi_split_path_info ^(.+\.php)(/.*)$;
                         if (!-f $script_name) {
                                 #return 404;
                                 break;
                         }
                         include fastcgi2.conf;
                         fastcgi_pass 127.0.0.1:9000;
                 }

where:
fastcgi2.conf is a copy of fastcgi.conf with one change:
fastcgi_param  REQUEST_URI        $uri$is_args$args;


How do I avoid using an IF here to check that the php file really
exists?  Also, why does uncommenting the return 404 cause some kind of
breakage (not even sure I understand exactly what happens? Seems like
the paths get broken?)

Thanks for any help?

Ed W

On Feb 21, 2013, at 21:48 , Ed W wrote:

>
> where:
> fastcgi2.conf is a copy of fastcgi.conf with one change:
> fastcgi_param  REQUEST_URI        $uri$is_args$args;
>
>
> How do I avoid using an IF here to check that the php file really exists?  Also, 
why does uncommenting the return 404 cause some kind of breakage (not even sure I 
understand exactly what happens? Seems like the paths get broken?)

               location ~ ^(?<script_name>.+\.php)(?<path_info>/|$) {
                       try_files  $script_name  =404;

                       include fastcgi2.conf;
                       fastcgi_param  PATH_INFO  $path_info;
                       fastcgi_pass   127.0.0.1:9000;
               }


--
Igor Sysoev
http://nginx.com/support.html

On 21/02/2013 17:54, Igor Sysoev wrote:

>                 location ~ ^(?<script_name>.+\.php)(?<path_info>/|$) {
>                         try_files  $script_name  =404;
>
>                         include fastcgi2.conf;
>                         fastcgi_param  PATH_INFO  $path_info;
>                         fastcgi_pass   127.0.0.1:9000;
>                 }

Thanks!!
Can I ask you to confirm the correction of a typo in your answer. Do I
want this:

     ....(?<path_info>.*) {

ie is this ammended version correct in the face of a URL such as:

     blah.php/some/path?param=2

Thanks

Ed W

On Thu, Feb 21, 2013 at 07:08:56PM +0000, Ed W wrote:
> On 21/02/2013 17:54, Igor Sysoev wrote:

Hi there,

> >                location ~ ^(?<script_name>.+\.php)(?<path_info>/|$) {

> Can I ask you to confirm the correction of a typo in your answer. Do I
> want this:
>
>     ....(?<path_info>.*) {

You probably want

               location ~ ^(?<script_name>.+\.php)(?<path_info>/.*|$) {

because you want to match /X.php or /X.php/X but not /X.phpX.

  f
--
Francis Daly        francis@daoine.org

On Saturday 23 February 2013 00:11:26 Francis Daly wrote:
>
> You probably want
>
>                location ~ ^(?<script_name>.+\.php)(?<path_info>/.*|$) {
>
> because you want to match /X.php or /X.php/X but not /X.phpX.
>

IMHO,

   location ~ ^(?<script_name>.+\.php)(?<path_info>/.*)?$ {

looks better.


% pcretest -b -d -m -s+
PCRE version 8.30 2012-02-04

  re> !^(?<script_name>.+\.php)(?<path_info>/.*|$)!
Memory allocation (code space): 42
Memory allocation (JIT code): 1739
------------------------------------------------------------------
  0  38 Bra
  3     ^
  4  15 CBra 1
  9     Any+
 11     .php
 19  15 Ket
 22   9 CBra 2
 27     /
 29     Any*
 31   4 Alt
 34     $
 35  13 Ket
 38  38 Ket
 41     End
------------------------------------------------------------------
Capturing subpattern count = 2
Named capturing subpatterns:
  path_info     2
  script_name   1
Options: anchored
No first char
Need char = 'p'
Subject length lower bound = 5
No set of starting bytes
data>


  re> !^(?<script_name>.+\.php)(?<path_info>/.*)?$!
Memory allocation (code space): 40
Memory allocation (JIT code): 1732
------------------------------------------------------------------
  0  36 Bra
  3     ^
  4  15 CBra 1
  9     Any+
 11     .php
 19  15 Ket
 22     Brazero
 23   9 CBra 2
 28     /
 30     Any*
 32   9 Ket
 35     $
 36  36 Ket
 39     End
------------------------------------------------------------------
Capturing subpattern count = 2
Named capturing subpatterns:
  path_info     2
  script_name   1
Options: anchored
No first char
Need char = 'p'
Subject length lower bound = 5
No set of starting bytes


  wbr, Valentin V. Bartenev

--
http://nginx.com/support.html
http://nginx.org/en/donation.html

On Sat, Feb 23, 2013 at 02:05:56AM +0400, Valentin V. Bartenev wrote:
> On Saturday 23 February 2013 00:11:26 Francis Daly wrote:

> >                location ~ ^(?<script_name>.+\.php)(?<path_info>/.*|$) {

>    location ~ ^(?<script_name>.+\.php)(?<path_info>/.*)?$ {
>
> looks better.

Agreed.

And it means I don't have to think about whether AB|C groups as (AB)|(C)
or (A)(B|C) in this regex implementation :-)

> % pcretest -b -d -m -s+

Using smaller Memory allocation numbers is an extra bonus.

Cheers,

  f
--
Francis Daly        francis@daoine.org

On 2013-02-22 23:05, Valentin V. Bartenev wrote:
> IMHO,
>
>    location ~ ^(?<script_name>.+\.php)(?<path_info>/.*)?$ {
>
> looks better.

In order to put yet another iteration into the game, I prefer the
script_name part to be non-greedy

     location ~ ^(?<script_name>.+?\.php)(?<path_info>/.*)?$ {

Of course it still depends on someone's preference what to match in
case of

     /foo.php/bar.php/quux

-cs

On 24/02/2013 19:40, Christoph Schug wrote:
>     location ~ ^(?<script_name>.+?\.php)(?<path_info>/.*)?$ {
>
> Of course it still depends on someone's preference what to match in
> case of
>
>     /foo.php/bar.php/quux
>
> -cs

You beat me - I was just about to post the same!

I think the non greedy is probably the more common case

Thanks!

Ed W