
Description of the Example
==========================

Andrew Davison
The Department of Computer Science
The University of Melbourne
Parkville, Victoria 3052, Australia
Email: ad@cs.mu.oz.au
Tel: +61 3 9287 9172
Fax: +61 3 9348 1184
URL: http://www.cs.mu.oz.au/~ad

May 1996


1. Summary of curl (again)
==========================

curl is a tool for automatically generating the links between the pages
comprising a World Wide Web (WWW) document. curl examines a contents
file containing the file names of the individual pages, and modifies
each file to include the necessary links to its neighbours, parent,
initial page, and to pages listing the figures, tables, keywords and
new pages in the document. These figures, tables, keywords, and new
additions pages are generated automatically by curl. curl can also
produce two kinds of contents lists, which can be automatically
inserted into pages.

There is a search engine, available from every page, which can scan the
document for specified strings. The matching pages can be accessed
from a search results page via dynamically generated links.

If the author wishes to add extra pages, remove some, or to reorder
pages, this is achieved by modifying the contents file. When curl is
invoked on the new version of the contents file it will modify the
links necessary to reflect the new WWW document structure.

curl comes with a range of additional tools. For instance the frequency
of word occurrences in the document can be analysed in a number of
ways. This is useful when the document hierarchy is being developed, or
for deciding on keywords.

A set of tools are included for linking pages that cannot be connected
easily within a document hierarchy. An example would be a page in the
humour section of a newsletter and an article in the editorial section
which both discuss street maps. They already have well-defined
positions in the document (in the humour and editorial sections) but
could also be usefully inter-linked to capture the street maps theme.



2. The Example
=============

The files in this directory are intended to be linked together using
curl to form a WWW document with the following hierarchical form:

                          newsletter
                              |
                              |
            --------------------------------------
            |                 |                  |
         editorial          humour          related-info
            |                 |
            |                 |
            |            -----------
            |            |         |
            |         streets     nono
            |
            |
    -------------------
    |       |         |
  marks  pathology  whome


The tree represents the complete document, and each node corresponding
to a page of the document. 'newsletter' introduces the newsletter, and
contains links to the 'editorial' and 'humour' sections, and to the
'related-info' page.  The 'editorial' and 'humour' sections consist of
a series of related pages which are reached by links. Each of these
pages is represented by its own HTML file. For instance the 'marks'
page resides in the 'marks.html' file.

A quick look at one of the files will show that they are written in
HTML, but do not include any links which reflect the document format
defined by the contents file. There may be links, but these will be for
other purposes (e.g. see 'related-info.html').

There are also four GIF files (gball.gif, nono1.gif, nono2.gif and
street.gif), which are used as illustrations.

Last, but not least, there is a file called 'lpn-contents' which
holds the information curl needs to build the document. It contains:

   newsletter {
     editorial {
       marks
       pathology
       whome
     }
     humour {
       streets
       nono
     }
     related-info
   }

This is a textual version of the tree diagram above. 'lpn-contents' is
called a contents file.

For more discussion of the example, read 'guide.ps' or 'guide.txt'
in the 'docs' directory of the curl release.


3. How to Build the Document
===============================

1. *Copy* all the files in the 'example' directory to a new
   directory.

2. *Copy* everything in the 'Defaults' directory to the
   same new directory. The 'Defaults' directory contains
   files holding default information on link styles,
   and the keywords, figures, tables, new pages and search 
   HTML files. The 'arrows' subdirectory holds the icons 
   used to represent links graphically.

3. Execute curl on the contents file:

      curl lpn-contents

   This assumes that curl has been compiled and placed 
   in a suitable place. See the README file for details
   on how to do this.

4. To be able to search a document, search.html, must be configured.
   See the file 'srch-setup' in the 'docs' directory, or section 
   5.3.6 of the user guide ('guide.ps' or 'guide.txt' in 'docs').

5. Start a browser, such as netscape, on the 'top' page of the 
   WWW document:

      netscape newsletter.html

6. That's it!


