[greenstone-devel] [Fwd: Re: Hunting for files in GLI [Fwd: greenstone.org support]]

From jmt12@cs.waikato.ac.nz
DateTue, 22 Jul 2003 10:34:30 +1200 (NZST)
Subject [greenstone-devel] [Fwd: Re: Hunting for files in GLI [Fwd: greenstone.org support]]
-------- Original Message --------
Subject: Re: Hunting for files in GLI [Fwd: greenstone.org support]
From: <jmt12@cs.waikato.ac.nz>
Date: Fri, July 18, 2003 3:14 pm
To: <bethany@lanic.utexas.edu>

Hi Bethany,

To answer your first question the 'Hunt' view and its associated view
'Mirror' are currently disabled. They haven't been finished, nor tested
yet. I also don't think they'll do exactly what you need (please see my
comments below).

The second problem is little more fixable I suspect. There is a known
bug in the installer program that writes invalid path names into the
batch file that starts GLI. To fix this open the file 'Gatherer.bat' in
the GLI folder, and look for four 'set' commands located between two
rows of '*'. The commands probably contain paths like 'C:Program
Files...' however GLI currently doesn't like spaces in folder names so
you need to replace all of the folder names with spaces with their old
DOS16
shortened name. So 'Program Files' become 'Progra~1' (six characters
followed by '~' followed by 1).

Now for some other comments. I had better start by mentioning that the
GLI can't break down complex files yet, although Jar/Zip archive, Marc
and CDISIS support are planned (not html though, sorry). Furthermore it
generally expects the content to be available locally, in other words
mirrored into your collection. However it is entirely possible to use
GLI to create the library if you generate the records first.

I think you're on the right track by breaking each Lanic page down into
individual html 'records'. For each you might add several meta tags to
provide the information for the library. So a final record might look
like this:

<HTML>
<HEAD>
<TITLE>Cielo Sur</TITLE>
<META name="URL" content="http://www.cielosur.com/">
<META name="Subject" content="8.1">
</HEAD>
<BODY>
Astronom□□a desde el Hemisferio Sur para todos
</BODY>
</HTML>

(Note that the 8.1 is a value for a Hierarchy based browser. Please see
chapter 2 of the Greenstone Developers Guide).

Once this is done you can build a collection from these html records
using GLI. For an example of a Greenstone Collection which uses links to
external documents see:

http://www.nzdl.org/cgi-bin/library?a=p&p=about&c=aircraft

You can also view its collection configuration file here;

http://www.nzdl.org/gsdl/collect/aircraft/etc/collect.cfg

the most important parts to notice being;
a) the HTMLPlug plugin command containing the argument -metadata_fields
'Title,tagH1' which is how you extract metadata from the Meta tags in
html.
b) the CL2VList format command which creates a hyperlink based on the
URL metadata.

For further documentation of Greenstone please visit the 'docs' page
available from:

http://www.greenstone.org/

I would also recommend that the best (and fastest) way to get help on
Greenstone/GLI is to sign up to one or both of the mailing lists, also
available on the docs page.

Hope I've been of help,

John Thompson

Greenstone Digital Library, University of Waikato

>> I can't get the Librarian Interface to work at
>> all. The "Hunt view" never seems to appear, and I need to look
>> around the Internet. It starts out with a warning about specifying
>> the path to the locally running Greenstone library. From the DOS
>> prompt, I looked around c:program filesgsdlGLI but couldn't
>> figure out which file to edit in which way. Is there any
>> documentation beyond the help screens within the program? I'm by no
>> means computer illiterate but do haven't had to mess much with
>> anything even vaguely in the way of a backend other than hard-coding
>> Web pages except sometimes a config.sys file in, say, 11 years....
>> >
>> > I'm trying to create a library of links (the directory at
>> > http://lanic.utexas.edu/ is currently hard-coded with no engine
>> whatsoever behind it and that has to stop), but since the directory
>> is extremely idiosyncratic and we're thinking about adding
>> abstracts, for the demo/testbed, I wanted to create the records by
>> scratch rather than learning perl, attempting to break each LANIC
>> page down into hundreds of separate records that could be
>> reclassified, creating a mess, and still having to add the abstracts
>> and reclassify each record anyway. We do not intend to mirror, just
>> link, and we're using Greenstone rather than a tool specifically
>> designed for that job because this is also a class project. Creating
>> the records from scratch is part of the LI's functionality, isn't it?
>> Or am I so tired that I'm confused even about what it's supposed to
>> do?
>> >
>> > I have 2.40 installed, but that's not an option above, so I picked
>> "don't know."
>> >
>> > Thank you!!!