Support for UNIMARC

[stakats]

Either we need separate MARC translators, or the present MARC translator needs to be able to identify which kind of MARC it's reading. Presently we only really support USMARC (aka MARC-21), but many international library catalogs use UNIMARC. The LOC publishes conversion tables, and we need only a tiny subset of the full spec to have a functional translator. For examples of sites using UNIMARC, see ​http://catalogue.bnf.fr and ​http://opac.sbn.it/opacsbn/opac/iccu/informazioni.jsp.

[dstillman]

Just as a clarification, we currently support Latin-1 and UTF-8 flavors of MARC 21, not records using the MARC-8 character set, which would be difficult to support (since there's no native support in Mozilla). I don't know what character set is generally used for UNIMARC, but that first site above just outputs the data to the webpage and the latter doesn't appear to use any extended characters. Additional examples would probably be helpful.

I don't see a way to access the raw MARC file on catalogue.bnf.fr, though perhaps it's possible with a login. We currently only support binary MARC. Simon mentioned in the forums that ​reading human-readable MARC 21 records might be difficult. I don't know if the situation with UNIMARC is any different.

This might be a job for Simon, since he wrote the existing MARC support.

[ajlyon]

Unimarc is at least partly supported; I'll look into writing a translator for the BNF so that (1) we have a translator for France's national library and (2) to tease out remaining issues with UNIMARC, if there are any. Reading in the human readable UNIMARC should be no problem; we do it for plenty of other translators these days.

[dstillman]

Milestone 2.0 Beta 3 deleted