Endless Loop in mod rewrite

[day811]

Hi,

I read some posts about this problem in this forum, but I can't work out what is wrong in my settings.

On my server (Debian Etch-Apache2.2.3), here's my .htaccess

RewriteEngine On
RewriteBase /example/
ErrorDocument 404 /example/index.php?option=com_404
RewriteRule ^((fr¦de¦en)\/(.*\.html)?)$ index.php?$1 [L,NC]

It translates
/example/fr/shop/article1.html

into
/example/index.php?fr/shop/article1.html

The real parameters rewriting is in a php process

All is working fine except :
On my first access to the site, the only valid adresses are

1. /example/
2. /example/fr/ (ou ../en/ ou ../de)

All others adresses generates an 'error in redirection'

But once i use one of these valid addresses, i can navigate thru all site adresses (even those which bugged before),
After a while without browsing thru the site, the problem happens again

In rewrite_log, the following sequence goes in an endless loop with IE and in a 20 steps loop with FF

192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8cae38/initial] (3) [perdir /var/www/example/] add path info postfix: /var/www/example/fr -> /var/www/example/fr/shop/order_details-10.html
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8cae38/initial] (3) [perdir /var/www/example/] strip per-dir prefix: /var/www/example/fr/shop/order_details-10.html -> fr/shop/order_details-10.html
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8cae38/initial] (3) [perdir /var/www/example/] applying pattern '^((fr¦de¦en)\/(.*\.html)?)$' to uri 'fr/shop/order_details-10.html'
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8cae38/initial] (2) [perdir /var/www/example/] rewrite 'fr/shop/order_details-10.html' -> 'index.php?fr/shop/order_details-10.html'
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8cae38/initial] (3) split uri=index.php?fr/shop/order_details-10.html -> uri=index.php, args=fr/shop/order_details-10.html
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8cae38/initial] (3) [perdir /var/www/example/] add per-dir prefix: index.php -> /var/www/example/index.php
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8cae38/initial] (2) [perdir /var/www/example/] trying to replace prefix /var/www/example/ with /example/
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8cae38/initial] (1) [perdir /var/www/example/] internal redirect with /example/index.php [INTERNAL REDIRECT]
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8d5e18/initial/redir#1] (3) [perdir /var/www/example/] strip per-dir prefix: /var/www/example/index.php -> index.php
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8d5e18/initial/redir#1] (3) [perdir /var/www/example/] applying pattern '^((fr¦de¦en)\/(.*\.html)?)$' to uri 'index.php'
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8d5e18/initial/redir#1] (1) [perdir /var/www/example/] pass through /var/www/example/index.php

Where is my mistake ?

Thanks for your answer

[edited by: jdMorgan at 2:11 pm (utc) on Mar. 12, 2008]
[edit reason] use example.com, fix character-entities [/edit]

[jdMorgan]

I fixed what appear to be character-entities in the code, possibly because your character-encoding does not match that of this forum. I assume that your code looks like this:

RewriteRule ^((fr¦de¦en)\/(.*\.html)?)$ index.php?$1 [L,NC]

(Of course, if your character-encoding *is* different, then this may not look right to you!)

Note that forward slashes need not be escaped in mod_rewrite; This is only required in PERL, PHP, or other languages which use slahses to denote regular expressions, as in "/regex-here/".

So you can write your rule as:

RewriteRule ^((fr¦de¦en)/(.*\.html)?)$ index.php?$1 [L,NC]

or perhaps, for better efficiency:

RewriteRule ^((fr¦de¦en)/([^.]+\.html)?)$ /index.php?$1 [L,NC]

Looking at the code and examining your RewriteLog, I don't see a loop; The internal rewrite is invoked on the initial request, and on the next pass, the log indicates a pass-through (no match on the /index.php request).

It is likely that the looping is occurring because of a problem in your script. Further note that if there was a problem, your code would cause an *internal* loop, because it invokes an internal redirection, not an external redirect. Therefore, the fact that the looping is 'visible' to Firefox and IE indicates that the problem lies outside this code, or in an unexpected interaction between this code and your script.

Look at your access logs to see what happened after the events shown in the RewriteLog above. I suspect you may see something odd about the subsequent requests -- and the 301 or 302 server status codes will be easy to examine with a server header checker, such as the "Live HTTP Headers" add-on for Firefox/Mozilla browsers.

Jim

[day811]

you said : I don't see a loop;

Sure, i paste only one full sequence but after
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8d5e18/initial/redir#1] (1) [perdir /var/www/example/] pass through /var/www/example/index.php

it starts again in

192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8cae38/initial] (3) [perdir /var/www/example/] add path info postfix: /var/www/example/fr -> /var/www/example/fr/shop/order_details-10.html
192.168.1.1 - example [12/Mar/2008:10:12:46 +0100] [test.aude-line.com/sid#6718d8][rid#8cae38/initial] (3) [perdir /var/www/example/] strip per-dir prefix: /var/www/example/fr/shop/order_details-10.html -> fr/shop/order_details-10.html

In my script there's no redirection.
I only analyse the $_SERVER['QUERY_STRING'] to set the post and get parameters and rewrite the $_SERVER['QUERY_STRING']

But the problem never reaches my php process !
In LiveHTTP headers, i've got no information

In acces.log : (only 2 of many repeated lines)
192.168.1.1 - - [11/Mar/2008:11:29:50 +0100] "GET /example/fr/shop/details_order-13.html HTTP/1.1" 401 561 "-" "Mozilla/5.0 (Windows; U; Windows NT 5.1; fr; rv:1.8.1.12) Gecko/20080201 Firefox/2.0.0.12"
192.168.1.1 - example[11/Mar/2008:11:30:04 +0100] "GET /example/fr/shop/details_order-13.html HTTP/1.1" 301 26 "-" "Mozilla/5.0 (Windows; U; Windows NT 5.1; fr; rv:1.8.1.12) Gecko/20080201 Firefox/2.0.0.12"

Thanks for your help

PS : this time i paste UTF8 codes, hopping this will be better

[jdMorgan]

Well, the first time the rule was invoked, it rewrote the details_order-NN.html URL-path to index.php. as expected. Then mod_rewrite restarted, and as expected, DID NOT rewrite index.php to anything. So how did the index.php URL-path get changed back to .html again?

This must have been done by another agent: A different RewriteRule (perhaps in a subdirectory), a script, a mod_alias directive, etc.

Jim

[day811]

Ok, I found the problem. A script make a 301 redirection to test cookie abitlity but only on first connection.
I found the origin of this bug in another forum.

Thanks a lot for you help