[Koha-zebra] zebra Multiple MARC management configuration

[Galen Charlton]

Hi,

On Tue, Apr 22, 2008 at 12:32 AM, Henri-Damien LAURENT
<laurenthdl at alinto.com> wrote:
>  At some point, zebra has been or was able to index multiple record
>  format in ***one*** database.
>  Is it still possible ? If yes, how can it be implemented ?
>  For instance, if I wanted to have MARC21 and UNIMARC and DC records be
>  indexed on one zebra database with DOM parsing. How should I do ?
>  Could it be possible to also parse data with different .chr depending on
>  the source ?

Yes, I think this can be done, and fairly easily at that, using the
DOM filter.  The XSLT used in the DOM extract and store pipelines just
needs to be able to determine the metadata format, then apply the
right transformation for indexing.

One way we could set up a hypothetical mutli-metaformat Koha database
is to define an XML wrapper, perhaps something like this:

<kohabib id="123">
  <metadata type="UNIMARC"> <!-- or MARC21, DC, MODS, etc. -->
    <!-- XML of MARC record using MARC21slim schema -->
  </metadata>
</kohabib>

The extract XSLT could then just check the metadata type, then refer
to the correct indexing transformation.

One thing to consider for this approach is moving the item data out of
the bib record for indexing purposes, and have something like this:

<kohabib id="456">
  <metadata type="DC" xmlns:dc="http://purl.org/dc/elements/1.1/">
      <dc:title>My cats</dc:title>
  </metadata>
  <kohaitems>
    <kohaitem>
      <homebranch>CPL</homebranch>
      <itemcallnumber>123.45 ABC</itemcallnumber>
   </kohaitem>
  </kohaitem>
</kohabib>

That way, we can separate figuring out how to store and display item
information from indexing the bib metadata, and not have to devise a
custom item storage mapping for each metadata type (and get the item
mapping out of the MARC frameworks except for perhaps MARC import and
export).

>  I am also MUCH interested in zebra agility in multiple record format
>  z3950 queries returns, that is one zebra setup could answer MARC21
>  UNIMARC, DC or else and send them on z3950 queries (depending on the
>  negociation with YAZ-CLIENT).
>  This would allow our Koha users to query one another database and get
>  the MARC data in their own format.
>  Is that possible ? If yes, how to achieve a good configuration of that ?

The DOM filter lets you specify multiple retrieval stylesheets based
on the desired record format.  For MARC format conversion, for which
stylesheets may not already exist, we can make use of the YAZ support
for USEMARCON.

Regards,

Galen
-- 
Galen Charlton
Koha Application Developer
LibLime
galen.charlton at liblime.com
p: 1-888-564-2457 x709