Application

Application Notes: Apache httpd - Custom Indexes

Apache httpd - Custom Indexes

Inspired by the nice HTTP file server at [http://ftp.falsehope.com], I decided to see what it took to make my [http://ftp.nakedape.cc FTP/HTTP server] look nicer. First, I had to adjust the Apache configuration itself to look for my custom header and disable the generated HTML preamble. Settings like this were necessary:

AddType text/html .cgi
AddHandler cgi-script .cgi
<Directory /var/ftp>
    Options +Indexes +ExecCGI
    IndexOptions FancyIndexing NameWidth=* SuppressHTMLPreamble
    AllowOverride Indexes
    HeaderName /HEADER.cgi
    ReadmeName README.html
</Directory>

The AddType makes .cgi files type text/html, which is required by mod_autoindex. The AddHandler makes it a CGI, instead of just a plain file, and of course the ExecCGI option is required to run it as a CGI. SuppressHTMLPreamble makes it forego generating the HTML header. Setting the HeaderName to /HEADER.cgi means it always looks in the DocumentRoot for the header-generator, instead of whatever path it is current in. Setting the ReadmeName gives you the option of including HTML in your README files and I allow overriding the Indexes settings so I can use AddDescription in local .htaccess files.

Why do you need to use a CGI anyway? You can use a static HTML page, however, it won’t be able to tell the user what his current path is. In the CGI environment, this is stored in REQUEST_URI and is about the only thing you need from the CGI environment (although I suppose you could pass other stuff around). It’s rather too bad that mod_autoindex doesn’t define an escape sequence that will be replaced with the REQUEST_URI; you wouldn’t need to resort to CGI at all in that case.

I wrote two versions of the CGI, one in Perl and one in Python. I could probably have done it just with a shell script, but for some reason I have an aversion to doing anything like this in shell. Here’s the Python version:

#!/usr/bin/python

import os

template_html_header = """Content-Type: text/html

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2 Final//EN">
<html>
<head>
    <title>Index of %s</title>
</head>
<body style="background-color: white">
<h1>Index of %s</h1>
"""

request_uri = os.environ.get('REQUEST_URI')

print template_html_header % (request_uri, request_uri)

And here’s the version in Perl:

#!/usr/bin/perl

$path = $ENV{'REQUEST_URI'} ;

print <<EOF ;
Content-Type: text/html

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2 Final//EN">
<html>
<head>
    <title>Index of $path</title>
</head>
<body style="background-color: white">
<h1>Index of $path</h1>
EOF

And finally, the simplest way of all is to use server-side includes. SSI is generally too lame to be useful, but in this case it does enough and is probably slightly faster than running a CGI script. You’ll want to make sure SSI is enabled and change the HeaderName to /HEADER.shtml:

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2 Final//EN">
<html>
<head>
    <title>Index of <!--#echo var="DOCUMENT_URI" --></title>
</head>
<body style="background-color: white">
<h1>Index of <!--#echo var="DOCUMENT_URI" --></h1>

Pretty simple, eh?