Documentation Source Text

/*
** A small, simple HTTP server.
**
** Features:
**
**     * Launched from inetd/xinetd/stunnel4, or as a stand-alone server
**     * One process per request
**     * Deliver static content or run CGI
**     * Virtual sites based on the "Host:" property of the HTTP header
**     * Runs in a chroot jail
**     * Unified log file in a CSV format
**     * Small code base (this 1 file) to facilitate security auditing
**     * Simple setup - no configuration files to mess with.
** 
** This file implements a small and simple but secure and effective web
** server.  There are no frills.  Anything that could be reasonably
** omitted has been.
**
** Setup rules:

**        contain content.  The directory is chosen based on the HTTP_HOST
**        request header.  If there is no HTTP_HOST header or if the
**        corresponding host directory does not exist, then the
**        "default.website" is used.  If the HTTP_HOST header contains any
**        charaters other than [a-zA-Z0-9_.,*~/] then a 403 error is
**        generated.
**
**    (3) Any file or directory whose name begins with "." or "-" is ignored,
**        except for /.well-known/ at the top-level.  The exception is for
**        RFC-5785 to allow letsencrypt or certbot to generate a TSL cert
**        using webroot.
**
**    (4) Characters other than [0-9a-zA-Z,-./:_~] and any %HH characters
**        escapes in the filename are all translated into "_".  This is
**        a defense against cross-site scripting attacks and other mischief.
**
**    (5) Executable files are run as CGI.  All other files are delivered
**        as is.
**
**    (6) For SSL support use stunnel and add the -https 1 option on the
**        httpd command-line.
**
**    (7) If a file named "-auth" exists in the same directory as the file to
**        be run as CGI or to be delivered, then it contains information
**        for HTTP Basic authorization.  See file format details below.
**
**    (8) To run as a stand-alone server, simply add the "-port N" command-line
**        option to define which TCP port to listen on.
**
**    (9) For static content, the mimetype is determined by the file suffix
**        using a table built into the source code below.  If you have
**        unusual content files, you might need to extend this table.
**
** Command-line Options:
**
**  --root DIR       Defines the directory that contains the various
**                   $HOST.website subdirectories, each containing web content 
**                   for a single virtual host.  If launched as root and if
**                   "--user USER" also appears on the command-line and if

**    *  "user NAME LOGIN:PASSWORD" checks to see if LOGIN:PASSWORD 
**       authorization credentials are provided, and if so sets the
**       REMOTE_USER to NAME.
**    *  "realm TEXT" sets the realm to TEXT.
**
** There can be multiple "user" lines.  If no "user" line matches, the
** request fails with a 401 error.
**
** Because of security rule (7), there is no way for the content of the "-auth"
** file to leak out via HTTP request.
*/
#include <stdio.h>
#include <ctype.h>
#include <syslog.h>
#include <stdlib.h>
#include <sys/stat.h>
#include <unistd.h>