Legacy web pages of Axel Beckert

Since I'm no more active at the Fachschaftsrat Informatik of the University of Saarland anymore, I have transferred all my university time legacy web pages from http://fsinfo.cs.uni-sb.de/~abe/ to this interim host at http://fsinfo.noone.org/~abe/ with only minimal modifications, mainly e-mail addresses.

Most pages on this interim host won't be updated anymore until they are moved (and redirected) step by step to their future home somewhere under http://noone.org/.

Please also note that my former e-mail address abe@fsinfo.cs.uni-sb.de is no more valid. Use abe@deuxchevaux.org instead.

Axel Beckert, Zürich, 23rd of September 2007


Axel's Apache Page

Just another little Apache Tips & Tricks page



Powered by Apache

About the Apache HTTPd:

The Apache HTTP Daemon is, according to the Netcraft Web Server Survey, the most popular web server software on the world. It's free, modular and runs on a lot of operating systems.

Servers, I work on:

Needless to say, that they're all running Apache HTTPds... ;-)

Useful Tools:

I've written several PERL scripts for use with an Apache HTTPd. They can all be downloaded on my software page.

Patches:

Until now I've only written one patch for the Apache HTTPd. It also can be can found on my software page.

If you have mod_perl installed into your Apache, you can also use Apache::AutoIndex instead, which has much more features than my small patch, but sometimes has strange problems with some other mod_perl modules like Embperl.

Some of my Apache configuration files:

I always keep here an global copy of a well done dynamic directory listing (mod_autoindex) configuration, MIME types and magic file. The MIME types configuration is heavily expanded and only few original entries are modified. The magic file is mainly based on the KMimeMagic magic file, which can be seen as an more actual and heavily enlarged version of the Apache magic file, which seems to be based on that from file(1). It's updated according to new mime-types and has no more missing tabs in it.

My sources for the MIME types and the appropriate file suffixes can be found in my bookmarks under the subject MIME. For some single entries the information about correct MIME types comes from the documentation of the document format, e.g. like in the cases of PNG, MNG and JNG, see my bookmarks under the subject Graphics

Some little tips for the configuration:

Here you'll find some of my own tips & tricks for configuring and administrating the Apache HTTPd. Maybe this page will become bigger some day...

Avoiding 404s of Frequently Made URL Typos

For German users: I've got those rules actually swapped out in a file called dau.conf... ;-)

  1. Inexperienced people, reading handwritten or badly printed URLs often misinterpret the "~" character.

    You may think, it is just my fantasy that people may mix up those characters. It isn't. Here is what I discovered in our logs at Dagstuhl: People really tried to get user pages by typing "''", "``", "-", "`" or '"' instead of "~"! Except the "`", I guess, all were misinterpretations of the character "~". "`" was probably a typo, because at least US-English keyboard layouts, you'll get a "`", if you try to type a "~" without pressing a shift or caps lock key before.

    Typos are mostly regcognized by the people themselves, but IMHO it's nice not to let them notice, that they made a typo. And configuring this problem for a few more typos isn't a big hack if we just started for those people who can't read URLs correctly. I think the following typos may appear quite often:

    Due to Apache cannot do spelling correction on user names or directories, it's sometimes necessary to lend him a hand... I've got two solutions for this:

    Explanations on those configuration commands:

    Example:

  2. People often forget the "~" in URLs.

    This also happened a few times on our server. (Some also did add a "~" to URLs... :-).

    Here's a solution, using mod_rewrite without doing too much subrequests for testing:

    # Part a: URL does not end with "/"
    RewriteCond /var/htdocs/$1$2 !-d
    RewriteCond /var/htdocs/$1$2 !-f
    RewriteCond /home/$1/public_html$2 -f [OR]
    RewriteCond /home/$1/public_html$2 -d
    RewriteRule ^/([^-{|}`'"+*^\\/~]+)(/.*[^/])?$ http://www.domain.com/~$1$2 [R=permanent,L]
    
    # Part b: URL ends with "/"
    RewriteCond /var/htdocs/$1$2 !-d
    RewriteCond /home/$1/public_html$2 -f [OR]
    RewriteCond /home/$1/public_html$2 -d
    RewriteRule ^/([^-{|}`'"+*^\\/~]+)(/.+)?/$ http://www.domain.com/~$1$2 [R=permanent,L]
    
    # Part c: long Username aliases.
    RewriteCond $1 ^(olafsdottir|rolshoven)$
    RewriteCond /~$1$2 -U
    RewriteRule ^/([^-{|}`'"+*^\\/~]+)(/.*)?$ http://www.domain.com/~$1$2 [R=permanent,L]
    

    Explanations on those configuration commands:

Workaround for Apache Bugs #7503, #4388, #4379 and #3333

[This bug escapes "?" and "#" when using RedirectMatch.]

Since 6-May-2001 there exists a patch against Apache 2.0, which was submitted on 11-May-2001. See #7503.

This patch seems to work for Apache 1.3.x, if you replace any occurence of the string "apr" with "ap" in the patch. But I can't yet say if it's stable, because that modification just was a guess, which at least did not crash. :-)

If your problem are question marks and not hashes, you may pass data instead of via a query string via the path info. Therefore you just have to substitute the question mark in front of the query string with als slash: Use

RedirectMatch ^/test([0-9]+) http://www.dagstuhl.de/test.shtml/$1

instead of

RedirectMatch ^/test([0-9]+) http://www.dagstuhl.de/test.shtml?$1

because the last line doesn't work (yet).

If you use mod_rewrite, you can use a non-parsed-header (nph) cgi-script to do the redirect, because -- as the name says -- the header won't be parsed and therefore nothing will be escaped:

/var/cgi-bin/nph-xredirect.cgi:

#!/usr/bin/perl

$| = 1;

print <<EOF;
HTTP/1.0 302 Moved Temporarily
Server: $ENV{'SERVER_SOFTWARE'}
Location: $ENV{'PATH_INFO'}
Content-type: text/html

<HTML>
<HEAD>
<TITLE>302 Moved Temporarily (Extended)</TITLE>
</HEAD>
<BODY>
<H1>Moved Temporarily (Extended)</H1>
The document has moved <A HREF="$ENV{'PATH_INFO'}">here</A>.
</BODY>
</HTML>

EOF

You call the cgi-script somewhat like that:

RewriteRule ^/A/([0-9]{5})$ /var/cgi-bin/nph-xredirect.cgi/A/#$1 [T=application/x-httpd-cgi,L]

This line redirects e.g. request to "/A/12345" to "/A/#12345" and not to "/A/%2312345" as Apache 1.3.x would do if you use RedirectMatch for that.

I found this workaround at the Users Guide to URL Rewriting with the Apache Webserver by Ralf S. Engelschall and modified it a little bit for better readability and less redundancy.

Using mod_macro to ease configuration

mod_macro by Fabien Coelho provides a macro feature for the Apache configuration files without the overhead of e.g. m4 or mod_perl. This also reduces the number of onfiguration typos made during typing repetitive lines or cut and paste.

I use it most time to capsulate often repeated access configurations:

<Macro AllowDagstuhl>
	Allow from .dagstuhl.de 192.76.146.0/255.255.255.0 localhost 127.0.0.1
	AuthType Basic
	AuthName "IBFI Schloss Dagstuhl: Staff Only"
	AuthUserFile /www/etc/passwd
	AuthGroupFile /www/etc/group
	require group dagstuhl
	Satisfy any
</Macro>

<Macro AllowDagstuhlGuests>
	Allow from .dagstuhl.de 192.76.146.0/255.255.255.0 localhost 127.0.0.1
	AuthType Basic
	AuthName "IBFI Schloss Dagstuhl: Guests Only"
	AuthUserFile /www/etc/passwd
	AuthGroupFile /www/etc/group
	require group guest dagstuhl
	Satisfy any
</Macro>

<Macro AllowUniSB>
	Allow from .uni-sb.de 134.96.0.0/255.255.0.0
</Macro>

<Macro AllowMPII>
	Allow from .mpi-sb.mpg.de 139.19.0.0/255.255.0.0
</Macro>

<Macro AllowAllHosts>
    <Limit GET HEAD POST OPTIONS PROPFIND>
	Order allow,deny
	Allow from all
    </Limit>
    <LimitExcept GET HEAD POST OPTIONS PROPFIND>
	Order deny,allow
	Deny from all
    </LimitExcept>
</Macro>

<Macro AllowCampus>
    Use AllowDagstuhlGuests
    Use AllowUniSB
    Use AllowMPII
</Macro>

<Macro NoOtherMethods>
	# PUT DELETE PATCH PROPPATCH MKCOL COPY MOVE LOCK UNLOCK
	<LimitExcept GET HEAD POST OPTIONS PROPFIND>
		Order deny,allow
		Deny from all
	</LimitExcept>
</Macro>

So my <Limit> sections just look like this:

<Location /foobar>
	<Limit GET HEAD POST OPTIONS PROPFIND>
		Order deny,allow
	        Use AllowCampus
	</Limit>
	Use NoOtherMethods
</Location>

And not like this:

<Location /foobar>
	<Limit GET HEAD POST OPTIONS PROPFIND>
		Order deny,allow
		Allow from .uni-sb.de 134.96.0.0/255.255.0.0
		Allow from .mpi-sb.mpg.de 139.19.0.0/255.255.0.0
		Allow from .dagstuhl.de 192.76.146.0/255.255.255.0 localhost 127.0.0.1
		AuthType Basic
		AuthName "IBFI Schloss Dagstuhl: Guests Only"
		AuthUserFile /www/etc/passwd
		AuthGroupFile /www/etc/group
		require group guest dagstuhl
		Satisfy any
	</Limit>
	# PUT DELETE PATCH PROPPATCH MKCOL COPY MOVE LOCK UNLOCK
	<LimitExcept GET HEAD POST OPTIONS PROPFIND>
		Order deny,allow
		Deny from all
	</LimitExcept>
</Location>

There are a lot more possibilities to ease configuration with mod_macro and you also can have parameters to macros. See the mod_macro home page for details and download.


Letzte Änderung:
URL: http://fsinfo.noone.org:80/~abe/apache-tips/index.html
Handcoded and valid HTML 4.01 and CSS 2.
Copyright © 1999-2017 by Axel Beckert <abe at cs.uni-sb.de>
Server: Apache/2.2.22 (Debian) Embperl/2.5.0_3 DAV/2 SVN/1.6.17 PHP/5.4.45-0+deb7u9 mod_ssl/2.2.22 OpenSSL/1.0.1t mod_perl/2.0.7 Perl/v5.14.2 on SunOS 4.1.3 @ fsinfo.cs.uni-sb.de (Fachschaftsrat Informatik)
Browser: CCBot/2.0 (http://commoncrawl.org/faq/) @ (none)