Tümer Garip wrote:
Adam wrote in reply to Paul:
cwd=/home/paul/koha.dev/head/misc/zebra/unimarc 15:40:34-03/02 zebrasrv(1) [log] record 0 type XML 15:40:34-03/02 zebrasrv(1) [log] 3129 bytes: <?xml version="1.0" encoding="UTF-8"?> <collection
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"; xsi:schemaLocation="http://www.loc.gov/MARC21/slim http://www.l ...
Your root element is collection. Not record. I don't
think melm will match that. Had you used record as root >> element - it should do it.
It's always a good idea to try things out with zebraidx -s update testrec.xml and see what gets matched.. (Look for the Idx: lines).
Ooops, this is getting more complicated for me. MARC21 slim defines the MARCXML as <collection><record></record></collection>. The perl MARC module produces XML according to this.
The MARC21 schema also allows <record>..</record>.. The collection element was only added as a container (XML allows exactly one root).
When you run zebraidx with a proper MARCXML you get an error saying collection.abs could not be found. So you are forced to rename marc21.abs as collection.abs by zebraidx or by ZOOM in that sense.
To make things more complicated everything seems to work until you try to use the recordID feature. If your XML file omits the MARC tags below 10 then RecordID works everytime. However if you have tags below 10 out of 50k records I got 30% with bad-match error.
So I am trying to eliminate whether its KOHA reproduced records doing or not. I'll post the outcome on this list. I hope to get a sample .. So that I can try this out myself.
/ Adam
Regards, Tumer Garip