[Dspace-general] Simple use case of DSpace -- can this work?

[Don Gourley]

Sergio,

I don't quite understand what kind of item you want to import into
DSpace...and if it is just one big item I'm not sure what value
DSpace would offer to managing it.  But in general for importing
items you would use the ItemImport program:

http://www.dspace.org/technology/system-docs/application.html#itemimporter

This program reads the DSpace "simple archive format" which is a
directory structure with folders for each item which contain a
very simple XML encoding of Dublin Core, the content files and
a file listing the content files.  I have written Perl scripts
to create this directory structure and it is pretty easy.

I think the content files can be structured in sub-folders but
I've never tried that and don't know how (or if) that structure
would be translated into DSpace's item structure.  The way I've
dealt with structural relationships between files in an item is
by including another file in each item which includes structural
metadata.  Another option in your case might be to use DSpace's
community/sub-community/collection/item hierarchy to map your
directories and files to multiple items instead of a single one.

In 1.4 you have another option which is to create a packager
plugin to ingest your item into DSpace.  However, the plugin
must be written in Java...I don't think there is any easy way
to use a scripting language.

-Don

On Wed, May 3, 2006 9:26 am, Sergio Trejo wrote:
> Hello All,
>
> I am about to install DSpace 1.4 alpha. I will gladly test it out and be
> happy to provide feedback to the maintainers. I had started to look at
> DSpace last year but was called to do work on a different project. Now I am
> returning to DSpace and I am looking forward to it!.
>
> I have a simple use case:
>
> * I have, on the file system of the server which I plan to install DSpace
> (Mac OS X Server), a top-level directory. This top-level directory contains
> files, sub-directories, and a few symbolic links (the links are to other
> directories within the top-level directory). The files contained in this
> directory structure on the file system are comprised mostly of web-related
> content (images in JPG and PNG), text, CSS, XHTML, etc. I also have one and
> only one RDF file for the entire top-level directory which contains Simple
> Dublin Core (15 elements maximum) that describe the entire directory of the
> content I just mentioned (DC: author, date, identifier, publisher, etc.).
>
> * I want to turn the above-described directory (and all of its content and
> RDF metdata file and sub-directories) as a DSpace "item" (a DSpace archival
> atom) as per the gorgeous diagram found at
> http://www.dspace.org/introduction/dspace-diagram.pdf
>
> * I would like to write a shell script that may be run on the Mac OS X
> Server machine that is also hosting the DSpace 1.4 alpha system, which
> script would be run by a designated Collection Curator and used to
> importthe above-mentioned DSpace item. I would thus like to avoid or
> highly
> minimize the requirement for a person (curator) to use the DSpace Web
> Interface and to avoid the need to fill out web forms for manually entering
> metadata about the "item". The motto I must take in my small and lean
> organization is borrowed from the Ruby on Rails community which espouses
> simplicity and agile approaches: DRY (Don't Repeat Yourself) <
> http://wiki.rubyonrails.com/rails/pages/DRY >
>
> Looking at the DSpace documentation, it is my understanding that in order to
> import an "item" into a DSpace repository, I will need to somehow create a
> SIP (Submission Information Package) file. A SIP apparently is "an XML
> metadata document with some content files" but I am having a hard time
> finding detailed documentation on how to create a SIP and just what goes
> into this "XML metadata document" as well as what "content files" are
> required.
>
> Could my proposed shell shell script, for example, parse the Simple Dublin
> Core contained in the RDF document that both describes the and is a part of
> the item, to generate a machine-meaningful SIP? How complex of a process
> might this be, to create a SIP? Will I need more than Simple Dublin Core to
> achieve all of this? Has anyone done something similar? My goal is to try
> and keep things as easy on people as possible. It is my job to make other
> people's lives as easy as possible ... I am fluent in scripting languages
> (python works great as does ruby) and am looking forward to creating SIPs
> for items.
>
> Thank you for any suggestions.
>
> -Sergio
> _______________________________________________
> Dspace-general mailing list
> Dspace-general at mit.edu
> http://mailman.mit.edu/mailman/listinfo/dspace-general
>