Re: Isite and ANZLIC metadata

[David Crossley <crossley@ozemail.com.au>]

Rob Atkinson asked some questions about getting the Isite search and
retrieve service to work with ANZLIC geospatial metadata.

Over the last few weeks I have been working on this as well - partly
so that I could give some feedback to the ANZLIC working group that
is developing the Document Type Definition (DTD) which defines the
SGML format.

>1. The ANZLIC doctype in the Isite 2.0.6 distribution is based upon
>the proposed SGML implementation no? And completely undocumented?

The developers of Isite seem to have set up a basic ANZLIC doctype
(based on the FGDC doctype) to get us started. However, it is not
ready to be used. I had to then edit the C++ code of that doctype to
reflect our current SGML field names and then recompile the application.

Isite 2.00.06 is completely undocumented because it is a pre-release.
Version 2.01 has not been released because there are still some bugs
and the documentation is still being enhanced. I guess "use at your
own risk and effort".

>2. The US "mp" metadata parser tool is required to make sensible HTML
>from a FGDC SGML metadata file, and this is required to make Isite
>retrieve FGDC documents properly. The ANZLIC doctype works the same way? 

The USGS has been developing and refining "mp" for some time now. It is
a strict parser that is meant to accept a text version of FGDC metadata
(in a special format) and generate an SGML file for indexing and transfer
and a text version and an HTML version for presentation. "mp" was
designed to work explicitly with FGDC metadata and to be very strict
about checking the syntax of the input file. I expect that we would have
a lot of trouble to get "mp" to do the same job on ANZLIC metadata
because our structure is too different to FGDC.

You only need a proper SGML file to get Isite to index and search. The
HTML file is only used for presentation. The HTML file needs the same
name as the SGML file but with a ".html" extension.

To get Isite to work with ANZLIC SGML metadata files I only had to alter
the date format of the ANZLIC SGML file. Isite and FGDC uses an ISO date
format of the form 19970602 (2 June 1997) or 199706 (June 1997). ANZLIC
has chosen a different date format which cannot be used.

>3. What is the syntax for numeric searches for Isearch?
> for zclient? 
>The syntax for Isearch bounding rectangle appears to be 
>-RECT{X1 X2 Y1 Y2} 
>but I cant persuade this to work with the ANZLIC or FGDC doctype using
>the ANZLIC SGML example.
>What is the Z39.50 syntax to make a FGDC bounding box search work?
>Is an ANZLIC doctype the same?

Mmmm. I am also having trouble here. Evidently the syntax is
-RECT(N S W E)  However, I can't yet get consistent results and am still
experimenting.

>4. The geogate program (referred to in documentation) has been merged
>into zgate?

Yes.

>5. The mp program wont parse ANZLIC metadata SGML. Are there mappings
>or configurations to make it or "cns" do this job?

See #2 above.

>6. I need to write a Perl module to extract spatial information from
>metadata to pass to geographically enabled data servers - I would
>prefer to use an ANZLIC standard that can be rendered in HTML. It
>looks like I might have to create the HTML manually, or use a generic
>SGML->HTML tool. Does anyone in Australia have a tool set for creating
>HTML from ANZLIC SGML? Or recommend an approach using other public
>domain tools (like SGMLS.pm). BTW This Perl module will be publically
>available from mid-June.

I have almost finished developing such a tool which will do the same
job as the "mp" parser. It will take an ANZLIC SGML file and produce
three files: a slightly different SGML file for indexing and searching,
an HTML file for presentation, and a text file for presentation.

I am using these publically available tools to do it:
SGMLS.pm ... http://www.uottawa.ca/~dmeggins/index.html
NSGMLS ..... part of the SP package http://www.jclark.com/sp.html

>7. Is there anything else obvious I have missed, or need to know but
>hasn't yet been comitted to documentation?

Make sure that you also set up the two other configuration files:
anzlic.fields ..... defines the field names and their type
(text, num, date, gpoly).
anzlic.localmap ... defines the mapping between the "bib1"
(bibliographic) attribute set and the
ANZLIC field names.

>Would much appreciate prompt responses if you have answers. 
>
>Thanks in advance
>-- 
>_______________________________________________________________
>Rob Atkinson              Email: rob@sydney.socialchange.net.au

I hope that this helps you Rob (and any others that may be
grappling with this).

regards



David Crossley
------------------------------------------------------------------
GeoGraphix Pty Ltd                -- Developing online services --
160 Cowper Street, Goulburn       -- for interactive access to  --
NSW 2580 Australia                -- geographic information     --
email: crossley@ozemail.com.au    --------------------------------
phone: +61 419 259705 fax: +61 48 225211
URL: http://purl.com/net/geographix
------------------------------------------------------------------