freebsd-dev/usr.sbin/simple_httpd

Simple_httpd  -  A small and free Web server

"Simple_httpd is like /usr/bin/mail is to mail clients, no frills."

This HTTP server can be used in any FreeBSD/PicoBSD application.

It has been tested under FreeBSD 2.2.x, 3.x and 4.x. It might work 
on other OS systems, but it's for FreeBSD primarily.

The main advantage to Simple_httpd is that it is very small.
The 25K binary can satisfy most needs in a small or embedded
appplication.  If you want a full featured server see 
/usr/ports/www/apache* or http://www.apache.org 

Simple_httpd is released under a BSD style copyright that unlike
GPL is embedded developer friendly.

The server is designed to be run in one of two modes.  The standard
mode is a httpd server running in the background serving up a directory
of html,gif,cgi whatever.  Your traditional www server.

The "fetch" mode supports file transfer over httpd.  This 
is best thought of as mate for fetch(1).  This feature can be
useful to transfer a file from one host to another.

Simple_httpd has the ability to run CGI scripts.  All CGI
scripts must be located in ${DOCUMENT_ROOT}/cgi-bin.  The
server currently only sets 3 environment variables before calling
the script.

CGI Environment variables are below:

SERVER_SOFTWARE = FreeBSD/PicoBSD
REMOTE_HOST = client.canada_lower_taxes.com
REMOTE_ADDR = 200.122.13.108

In most target applications for this server the extra DNS traffic from
the remote_addr lookup will likely be on the local lan anyway and not
on the other side of the internet.  You can turn it off yourself in
the code if you want to speed the whole process up.  Be sure to turn
it off for the logfile also.

How to use it?
==============

Compile with make, run as follows

usage: simple_httpd 	[-vD]
			[-d directory]
			[-g grpid]
			[-l logfile]
			[-p port]
or
usage: simple_httpd [-p port] -f filename

-v
Run the server verbose.  Show the program options that will be used for this
process.  Will only show information during startup, no messages will
be displayed while serving requests.  In other words you can still 
daemonize without fear of output on stdout.

-D
Do not daemonize.  The server will not run in the background.  It will
stay attached to the tty.  This is useful for debugging.  In this
mode no log file is created.  Logging info is to stdout.

This option is automatically selected if fetch option is selected.

-d directory
The html document directory, if nothing is provided the default is 
/httphome if UID is root, otherwise document root is ${HOME}/public_html

-l logfile
Set the logfile to use. Log messages will be written to /var/log/jhttpd.log
if you are root and ${HOME}/jhttpd.log otherwise. If you don't want a 
log file try "-l /dev/null"

-p port
Set the port httpd server will listen to.  Default is port 80 if
you are root and 1080 if you are not. 

-f filename
This is the only option needed to use the "fetch" feature.  The file
specified will be the ONLY file served to ANY GET request from a browser
or fetch(1).

Example
=======

Standard Mode:
--------------
If you have the FreeBSD handbook installed on your machine and would 
like to serve it up over http for a quick look you could do this

simple_httpd -d /usr/share/doc/handbook -l /usr/tmp/jlog.txt -p 1088 -v

Any browser would be able to look at the handbook with
http://whatever_host/handbook.html:1088

I'm using 1088 as the port since I already have apache running on port 80
and port 1080 on my host.

Please note, the handbook is not installed by default in FreeBSD 3.x
It must be installed from the ports collection first if you want to
try this.

Another simple example is to browse your local ports collection:

cd /usr/ports
make readmes   #wait about 1 hour!
simple_httpd -p 1080 -v -d /usr/ports

Then point your browser at http://whatever_host/README.html

Fetch Mode:
--------------
This is designed to be used in conjunction with fetch(3).  It allows
for easy transfer of files from one host to another without messy
authentication or pathnames required with ftp.  The file to be
served up must be readable by the user running simple_httpd.
This is not a magic way to avoid permissions and read files.

The daemon will only serve up ONE file.  The file specified will 
be returned for every GET request regardless of what the browser
asks for.  This allows for on the fly naming.

sender# simple_httpd -f /usr/tmp/big_file.tgz
receiver# fetch http://sender.com/Industrial_Secrets.tgz

big_file.tgz was transferred from one machine to another and renamed
Industrial_Secrets.tgz at the same time.

Tunneling over other TCP ports.  Choose something that firewall
will probably pass. See /etc/services.

sender# simple_httpd -p 53 -f /usr/tmp/big_file.tgz
receiver# fetch http://sender.com:53/Industrial_Secrets.tgz

To Do
=====

Simple authentication would be very useful [understatment].
/etc/passwd or PAM would be nice.

I think a netmask option would be good. Most internet appliances
probably want to restrict traffic to local ethernet anyway.
ie: Allow anything from my class C.

The server always has 1 zombie process hanging around when it
runs as a daemon.  Should fix so that it doesn't happen.

Anything to make it faster!

Man page

If anyone has any improvements or ways to easily implement something
please let me <wlloyd@slap.net> know.  If you make some neat embedded
device with PicoBSD I want to know too!

Credits
=======

This program was originally contributed by Marc Nicholas <marc@netstor.com>

Major rewrite by William Lloyd <wlloyd@slap.net>

$FreeBSD$