[greenstone-users] cds/isis importing - differences from 2.60 and upgrade

From ruben pandolfi
DateSun, 30 Jul 2006 10:54:20 +0200
Subject [greenstone-users] cds/isis importing - differences from 2.60 and upgrade
Hello,

(long message for CDS/ISIS users, switching to 2.70w )

we are working with CDS/ISIS importing, from time to time we need to
import new data from CDS/ISIS to greenstone. It is always the same
CDS/ISIS db structure, with incremental versions.

Our librarian prefer to use winisis to add bibliographical records, and
then we import a new CDS/ISIS db overwriting the previous.

Metadata set is dublin core + some new metadata records we have add, by
using 2.60 greenstone add/merge/ignore interface while exploding the
isis MST file.


In 2.60 this process was easy:

- import a first CDS/ISIS db
- exploding and assign metadata set (eg DCqualified)
- design and build collection

In 2.60, to add a more recent CDS/ISIS db, we just followed this process

- delete the current records
- importing a new CDS/ISIS db
- exploding
- creating

There was no need to reassign metadata, as they were correctly merged
already, because the explode process reads profile.xml


..............................................................

In 2.70w I can't not achieve that, everytime I import a CDS/ISIS db,
metadatas are extracted and appended to the metadata set, they do not
merge, I think because of the way now ISIS plug works and extract metadata.


..............................................................


I need some help because at the moment I have to redesign all the
collection, since metadata names change and therefore format, index,
classifier must be done again.

I still have the original 2.60 collection online, but I can't find a way
to use the existing metadata and avoid redesign the collection,
therefore I can not add the latest CDS/ISIS db.

You can view the collection here:

http://www02.unibg.it/~morin/gsdl/cgi-bin/library?e=d-000-00---0babel3--00-1-0--0prompt-10---4------0-1l--1-en-50---20-preferences---00131-001-1-0utfZz-8-00&a=q

metadata files are:

http://www02.unibg.it/~morin/gsdl/collect/babel3/metadata/dublin.mds
http://www02.unibg.it/~morin/gsdl/collect/babel3/metadata/ex.mds
http://www02.unibg.it/~morin/gsdl/collect/babel3/metadata/profile.xml

CDS/ISIS versions always have the same metadata set.

...................................................

After some (long :-) tests I can see that if I explode the new CDS/ISIS
file and remap to dublin core I have:

1)it appears that 2.70w creates the following file:

exp.mds

and does not consider and rewrites the original profile.xml

2) for some unknown reasons it capitalizes metadata in original .fdt
metadata files.

eg:

mods.extension.place (not included in dublin core) is added as:
Dc.ModsExtensionPlace

(in 2.60 we had dc.extension.place)

3) for some other unknown reasons ignore "." in metadata names,
therefore I now have a profile.xml file with double entries, although
the original metadata were the same.

eg:

<Action source="Dc.title" target="dc.Title"/> (from 2.60)
<Action source="DcTitle" target="dc.Title"/> (in 2.70w)

<Action source="Mods.extension.place" target="dc.extension.place"/>
(from 2.60)
<Action source="ModsExtensionPlace" target="dc.ModsExtensionPlace"/>
(in 2.70w)


Therefore the format/index in collect.cfg does not work anymore

...................................................

So my questions are:

1) is this correct and will be the future behaviour for next release
cds/isis plug? (in this case I will rewrite my collect.cfg accordingly)

2) can I edit profile.xml without GLI and expect to have the same
behaviour for net CDS/ISIS imports? (apparently not, is there a command
to make this on the shell )


Thank you for your help!

Ruben