Received: (qmail 20998 invoked by uid 2012); 11 May 1999 20:44:43 -0000
Message-Id: <19990511204443.20997.qmail@hyperreal.org>
Date: 11 May 1999 20:44:43 -0000
From: Axel Beckert <axel@dagstuhl.de>
Reply-To: axel@dagstuhl.de
To: apbugs@hyperreal.org
Subject: RedirectMatch quotes "#"
X-Send-Pr-Version: 3.2

>Number:         4388
>Category:       mod_alias
>Synopsis:       RedirectMatch quotes "#"
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    apache
>State:          closed
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          duplicate
>Submitter-Id:   apache
>Arrival-Date:   Tue May 11 13:50:01 PDT 1999
>Closed-Date:    Sun May 06 13:11:16 PDT 2001
>Last-Modified:  Sun May 06 13:11:16 PDT 2001
>Originator:     axel@dagstuhl.de
>Release:        1.3.4
>Organization:
>Environment:
SunOS dagoffice 4.1.4 11 sun4m, gcc 2.6.3
>Description:
Many people link to our site and forget the "#" for jumping to the appropriate
entry on a page. E.g. instead of linking to /DATA/Seminars/99/#99421 they link
to /DATA/Seminars/99/99421. Due to having more than one path where this happens
(for every year since 1990 at least three occurences) I want and because of 
Redirect matches only complete paths and not parts of directory names (see 
beyond and RTFM to Redirect) I'm forced to use RedirectMatch instead of 
Redirect. So I wrote:

RedirectMatch permanent ^/DATA/Seminars/([0-9][0-9])/([0-9][0-9][0-9][0-9][0-9]?)(\.html|/.*)?$	http://www.dagstuhl.de:8042/DATA/Seminars/$1/#$2

If I now request http://www.dagstuhl.de:8042/DATA/Seminars/99/99421 I get 
redirected to http://www.dagstuhl.de:8042/DATA/Seminars/99/%2399421 instead of 
http://www.dagstuhl.de:8042/DATA/Seminars/99/#99421 because RedirectMatch
quoted the "#" and returned it as "%23".

In comparision

Redirect permanent /99/99 http://www.dagstuhl.de:8042/DATA/Seminars/99/#99

redirects http://www.dagstuhl.de:8042/99/99 to 
http://www.dagstuhl.de:8042/DATA/Seminars/99/#99 as it should. Unfortunately
redirect matches no path components, e.g. it matches /99/99 but not /99/99421,
so I am forced to use RedirectMatch...

BTW: This seems to be similar to #4379 and #3333, which both have the same 
problem with quoted question marks. Mainly I wrote this bug report, because I 
fear that, if that both bugs are fixed and only the "?" won't be quoted anymore,
the same problem with quoted hashes may resist.
>How-To-Repeat:
http://www.dagstuhl.de:8042/DATA/Seminars/99/99421
>Fix:
Don't let RedirectMatch quote hashes and question marks. AFAIK there shouldn't
be any other exceptions.
>Release-Note:
>Audit-Trail:
State-Changed-From-To: open-closed
State-Changed-By: rbb
State-Changed-When: Sun May  6 13:11:16 PDT 2001
State-Changed-Why:
This bug is being tracked in PR 7503.  The second of the 
two patches should solve the problem of the '#' being
escaped.

Please test this patch, and let us know if it solves your problem.
Class-Changed-From-To: sw-bug-duplicate
Class-Changed-By: rbb
Class-Changed-When: Sun May  6 13:11:16 PDT 2001
>Unformatted:
[In order for any reply to be added to the PR database, ]
[you need to include <apbugs@Apache.Org> in the Cc line ]
[and leave the subject line UNCHANGED.  This is not done]
[automatically because of the potential for mail loops. ]
[If you do not include this Cc, your reply may be ig-   ]
[nored unless you are responding to an explicit request ]
[from a developer.                                      ]
[Reply only with text; DO NOT SEND ATTACHMENTS!         ]