On Thu, Jun 10, 2010 at 3:38 PM, Samuele Kaplun <[email protected]> wrote:
> > I think that PyMARC is fully Python based and is using xml.sax parser > which is not the fastest. So probably for core usage the local Invenio > code is still better WRT performance. This is correct. > > On the other hand if PyMARC is fully adherent to MARC standard (thus it > > can work with the MARC21 binary format, including the leader etc.) it > might be an interesting option to exploit for users migrating from a > MARC based digital library or symmetrically for exporting to real > MARC21. I just spent a few days figuring out the best means of exporting some records from invenio for the purpose of loading into some other solr-based discovery systems (Blacklight, Kochief, VuFind). I found it necessary to write my own export script using pymarc combined with search_engine.perform_request_search and search_engine.get_record. The two main issues are a) that most utilities and systems out there for dealing with MARC expect MARC21, and b) they really, really expect a leader to indicate media type, encoding, etc. I think regardless of whether you decide to make use of pymarc in the future, the lack of a leader is the main compatibility blocker. There are some parsers/converters in the wild for aleph sequential format but they too expect a leader value. -- ****************************************************** Jay Luker Astrophysics Data System (ADS) [email protected] Center for Astrophysics 617-495-4588 60 Garden Street MS 67 617-495-7356 fax Cambridge, MA 02138 ******************************************************
