Bug 23460 - mod_rewrite seems to use all lowercase URLs under Windows
Summary: mod_rewrite seems to use all lowercase URLs under Windows
Status: RESOLVED WONTFIX
Alias: None
Product: Apache httpd-1.3
Classification: Unclassified
Component: mod_rewrite (show other bugs)
Version: 1.3.27
Hardware: PC Windows XP
: P3 normal with 1 vote (vote)
Target Milestone: ---
Assignee: Apache HTTPD Bugs Mailing List
URL:
Keywords:
: 47393 (view as bug list)
Depends on:
Blocks:
 
Reported: 2003-09-28 04:53 UTC by Webmaster33
Modified: 2009-06-21 15:31 UTC (History)
1 user (show)



Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Webmaster33 2003-09-28 04:53:53 UTC
It seems, that mod_rewrite use all lowercase URLs under Windows.
The typed URL was:
http://www.site.com/Computers/Printers/page=2/order=d/sort=1

I used the following Rewrite rules in .htaccess:
RewriteEngine on
RewriteCond   %{HTTP_HOST}                 ^www\.site\.com$
RewriteRule ^(.*)/cat/(.*)page([0-9]+)[/]?(.*)$   /script.cgi?cat=$1&page=$3
RewriteRule ^(.*)/order(a|d)[/]?(.*)$     /script.cgi?%{QUERY_STRING}&order=$2
RewriteRule ^(.*)/sort([0-9]+)[/]?(.*)$   /script.cgi?%{QUERY_STRING}&sort=$2
RewriteRule ^/script.cgi(.*)$     /cgi-bin/script.cgi?%{QUERY_STRING} [L]

But in rewrite_log.txt appears:
(2) [per-dir c:/path/] rewrite computers/printers/page=2/order=d/sort=1 ->
/script.cgi?cat=computers/printers&page=2&order=d&sort=1


I searched bugzilla, docs, faqs, asked IRC channels, but I did not find any
issues about this problem/behaviour.

I have disabled mod_spelling module, so that can not cause the lowercase URL
issue in rewrite engine. I don't know other feature, what is URL case related, 
and is enabled or used in my server config.


It may happen that mod_rewrite is treating URLs are all lowercase because of
Windows FAT32 file system. 
Scripts should always get the correct case URL input, because if not, some
specific scripts will fail.

Best regards,
Webmaster33
Comment 1 André Malo 2003-09-29 06:58:32 UTC
mod_rewrite doesn't modify the case without advice (aka RewriteMap). My first
guess is, that this is a browser/proxy problem. Did you try another one or
telnetting to the server?
Comment 2 Webmaster33 2003-09-29 13:14:16 UTC
BTW: there is a mistake in the example, however that does not affect the problem:
The typed URL should be for the current rewrite rule:
http://www.site.com/Computers/Printers/page2/orderd/sort1

And in rewrite_log.txt appears:
(2) [per-dir c:/path/] rewrite computers/printers/page2/orderd/sort1 ->
/script.cgi?cat=computers/printers&page=2&order=d&sort=1

>My first guess is, that this is a browser/proxy problem.
It can not be a browser problem. I tried with both Mozilla and IE.

Also tried the telnetting now. The response is:
HTTP/1.1 200 OK
Date: Mon, 29 Sep 2003 12:54:22 GMT
Server: Apache/1.3.27 (Win32)
Connection: close
Content-Type: text/html; charset=iso-8859-2

But the "Computers/Printers" category parameter, appears to get passed as all
lowercase to the cgi script, as: "computers/printers"

I have only these modules loaded:
LoadModule rewrite_module modules/mod_rewrite.so
# module order
ClearModuleList
AddModule mod_env.c
AddModule mod_log_config.c
AddModule mod_mime.c
AddModule mod_negotiation.c
AddModule mod_include.c
AddModule mod_autoindex.c
AddModule mod_dir.c
AddModule mod_isapi.c
AddModule mod_cgi.c
AddModule mod_asis.c
AddModule mod_imap.c
AddModule mod_actions.c
AddModule mod_userdir.c
AddModule mod_alias.c
AddModule mod_rewrite.c
AddModule mod_access.c
AddModule mod_auth.c
AddModule mod_so.c
AddModule mod_setenvif.c

Should I try another Apache version?
I can not imagine what causes that mod_rewrite uses lowercase URLs. The [NC]
flag is not used.
The only possibility would be, that mod_rewrite converts URLs to lowercase under
Windows/FAT32. And that would be a bug.

Any idea?
Comment 3 Webmaster33 2003-10-12 21:20:54 UTC
Now I'm sure it's not a config problem.
The problem also appears on Apache v1.3.28.

I would try to make clear the problem.
Test URL:
http://www.site.com/Computers/Printers/Cables/index.html

Using following rewrite rule I would like to use some part of the URL as input
into a script:
RewriteEngine on
RewriteRule ^(/.*)/index.html$   /script.cgi?cat=$1

The problem is, that $1 will be not "Computers/Printers/Cables", but
"computers/Printers/Cables".
The script input needs the category name in correct case, doesn't matter if we
use Unix or Windows.


But in rewrite_log.txt appears:
(3) [per-dir z:/dir/] add path-info postfix: z:/dir/computers ->
z:/dir/computers/Printers/Cables/index.html
(3) [per-dir z:/dir/] strip per-dir prefix:
z:/dir/computers/Printers/Cables/index.html -> computers/Printers/Cables/index.html
(3) [per-dir z:/dir/] applying pattern '^(.*)/index.html$' to uri
'computers/Printers/Cables/index.html'
(2) [per-dir z:/dir/] rewrite computers/Printers/Cables/index.html ->
/script.cgi?&cat=computers/Printers/Cables
(3) split uri=/script.cgi?&cat=computers/Printers/Cables -> uri=/script.cgi,
args=&cat=computers/Printers/Cables
(1) [per-dir z:/dir/] internal redirect with /script.cgi [INTERNAL REDIRECT]
(3) [per-dir z:/dir/] strip per-dir prefix: z:/dir/script.cgi -> script.cgi
(3) [per-dir z:/dir/] applying pattern '^(.*)/index.html$' to uri 'script.cgi'
(1) [per-dir z:/dir/] pass through z:/dir/script.cgi

Is that normal?
"computers/Printers/Cables" is inserted with $1, and not the expected
"Computers/Printers/Cables".

I know, it may be related to the Windows's case insensitive filesystem, however
the rewrite feature is wrong that way. Script inputs are case sensitive on all
systems, so mod_rewrite should be also case sensitive. Windows understands both
lowercase and uppercase paths, so paths with correct case from mod_rewrite are
treated well at system level.

I think the bug is at "add path-info postfix" action, since the base dir is not
"z:/dir" as would expected, but "z:/dir/computers".

Best regards,
Webmaster33
(using Apache v1.3.28)
Comment 4 André Malo 2003-10-12 21:37:46 UTC
It must be somewhere else. The path-info adding is correct, if z:/dir exists.
Can you attach your whole configuration? Or better can you break down the 
problem to a minimal configuration?

If you don't want to publish your httpd.conf you mail it me privately as well.

Thanks.
Comment 5 Webmaster33 2003-10-12 21:40:37 UTC
Better sending privately. Thanks!
Comment 6 André Malo 2003-10-12 22:03:12 UTC
Received, thanks. I'm digging into it tomorrow (it's quite late here :)
Comment 7 William A. Rowe Jr. 2003-10-13 07:54:03 UTC
  This is resolved in Apache 2.0 - Apache 1.3 simply lowercased the filenames
  to 'canonicalize' them to their "one true name".  Yes this was very hackish,
  but that's the Apache 1.3 on Windows solution - an interesting and clever hack.

  Apache 2.0 was written from the ground up - and this was taken into account.
  The file's *true names* are used in Apache 2.0 (as stored on disk as their
  long file names.)  Also the "one true name", this takes a few more cpu/disk
  cycles but is more reliable and doesn't cause the sorts of problems you are
  encountering.

  Please upgrade to 2.0 to resolve your issue.
Comment 8 Webmaster33 2003-10-13 17:17:26 UTC
Yes, I upgraded to 2.0, and it works fine now. 
Thanks!


>Apache 1.3 simply lowercased the filenames
>to 'canonicalize' them to their "one true name"
However I think, that this should be still fixed in v1.3.x.

Unfortunately not the filename (or not just) was lowercased, but the path itself
in the "add path-info postfix" phase.
I think the base path is detected incorrectly: "z:/dir/computers" instead "z:/dir".
The "/computers" string should belong to post part.

Opinion?

BTW: Why does it need to lowercase anything, when the FAT16-32 is case
insensitive? If the URL part of the path is kept with original case, the Windows
will still accept it.
Comment 9 Webmaster33 2004-05-25 21:23:56 UTC
Was this bug really fixed in v1.3.x?
Comment 10 André Malo 2004-05-25 21:30:03 UTC
Nope.

(a) it's not a bug in mod_rewrite
(b) it's fixed 2.0. As said, 1.3 will stay as is.

Perhaps it should have better been resolved to wontfix...
Comment 11 Webmaster33 2004-05-25 21:46:09 UTC
>(a) it's not a bug in mod_rewrite

If not a bug in mod_rewrite, then bug in Apache. Doesn't change the fact, that
v1.3.x is not working correctly.


>(b) it's fixed 2.0. As said, 1.3 will stay as is.

I'm glad, that it's fixed in 2.0, but it still should be fixed in 1.3.x.


I found a fix in 1.3.13, which seems interesting:
"  *) Add case_preserved_filename field to the request_rec structure.
     On systems with case insensitive file systems (Windows, OS/2, etc.), 
     r->filename is case canonicalized (folded to either lower or upper 
     case, depending on the specific system) to accomodate file access
     checking. case_preserved_filename is the same as r->filename 
     except case is preserved. There is at least one instance where Apache 
     needs access to the case preserved filename: Java class files published 
     with WebDAV need to preserve filename case to make the Java compiler 
     happy. [Bill Stoddard]"
Isn't it something that may help current issue?

Anyway, until it's not fixed in 1.3.x, I'm reopening the bug. Sorry.
Comment 12 André Malo 2004-05-25 21:53:29 UTC
Patches welcome ;-)
Comment 13 William A. Rowe Jr. 2005-02-05 02:33:37 UTC
Whoops.  Ok, this has simply got to be marked won'tfix.

As I said 15 months ago - it is simply NOT possible to straighten this
out in Apache 1.3/Win32.  Nothing has changed, or will change, in that
answer.  If you don't like the behavior (certainly I don't :-) then
upgrade to 2.0.




Comment 14 Bob Ionescu 2009-06-21 15:31:54 UTC
*** Bug 47393 has been marked as a duplicate of this bug. ***