#/* Copyright 1999, IBM Corporation
# * All rights reserved
# *
# * Distribute freely, except: don't remove my name from the source or
# * documentation (don't take credit for my work), mark your changes
# * (don't get me blamed for your possible bugs), don't alter or
# * remove this notice.  No fee may be charged if you distribute the
# * package (except for such things as the price of a disk or tape,
# * postage, etc.).  No warranty of any kind, express or implied, is
# * included with this software; use at your own risk, responsibility
# * for damages (if any) to anyone resulting from the use of this
# * software rests entirely with the user.
# *
# * Send me bug reports, bug fixes, enhancements, requests, flames,
# * etc.  I can be reached as follows:
# *
# *          John Pfuntner      John_Pfuntner@tivoli.com
# */

Introduction
============

This is the readme file for websearch Version 1.0, a korn shell script
written by John Pfuntner.  The repository for the distributing the tool
is http://www.s390.ibm.com/ftp/os390/oe/toys/websearch.tar.Z.

I originally created this script as a "quick and dirty", "do it yourself"
way to perform search the OS/390 web site.  You can find the original 
version at work at http://www1.s390.ibm.com/cgi-bin/oefind.sh.  I based
it on the external behavior of search tools I had used on the WWW and 
my knowledge of OS/390 and the korn shell.  It is so ridiculously simple 
that you may not want to consider using it but if you don't need a lot of 
bells and whistles, it may suit your needs well.

I imagine you're familiar with other web search engines like AltaVista,
Yahoo, etc.  Here are some of the advantages and disadvantages I see of 
this script:

  Advantages
  ==========

  o no database to set up, the script reads from web pages - after
    some very simple customization (perhaps only one line), you can
    be performing searches!

  o source code is provided so changes are possible

  o output is highly customizable

  o hard to "break" - as long as OS/390 is behaving properly


  Disadvantages
  =============

  o rather slow

  o can find "nonsense" hits: the tool does not filter out
    HTML tags, comments, etc.

  o can't search for complete words easily: that is find "the"
    but not "there", "soothe", etc.

  o no logical expressions: find pages with "this" and/or "that"

Installation & Customization
============================

You need to place this script in an HFS directory that your web
server will recognize as a cgi-bin directory.  That means you'll 
have to have write permission to a directory specified by an EXEC
directive in your http configuration file.

You'll want to make sure the execute bit for the script is
turned on.  If it isn't, try something like:

	chmod +x websearch.sh

I would recommend permission settings of 755 (rwxr-xr-x).

The script looks for all *.html files in a hierarchical tree
which is rooted at a directory defined by the script variable
called SEARCH_BASE.  You should change this to the directory
you want to search.

You might also want to customize the output from the dynamic
pages the scripts generate.  Modify the "intro" and "outro"
functions in the script to suit your tastes.  Try it first with
the default beginning and ending HTML.  Then try modifying it 
to make it more "your own".  You can add text, pictures, links, 
etc.  You should know a little HTML if you want to modify these
functions and if you do, you'll have a lot of control.  You 
shouldn't need to know much about korn shell programming unless 
you want to do something really fancy.