The behavior of "any" index should be the same in DOM and GRS-1 indexing modes.
To do so, a change must be made to the xslt stylesheet used for transforming *-koha* files to *-zebra* files.
Created attachment 23114 [details] [review]
[PATCH] Bug 11291 : Make the any index more precise for DOM indexing
Any index is generated by a template in koha-indexdefs-to-zebra.xsl.
This template is applied to all fields of the record. This should be made more precise.
This patch removes this template, and adds "Any:w Any:p" to the $indexes variable generated in 4 templates of of koha-indexdefs-to-zebra.xsl
This patch also provides regenerated *-zebra-indexdefs.xsl files of biblio and auth of each marcflavor.
1. apply the patch
2. copy the modified authority-zebra-indexdefs.xsl and biblio-zebra-indexdefs.xsl into the configuration directory of your Zebra
3. rebuild Zebra for authorities (rebuild_zebra with -a -r -v )
4. rebuild Zebra for authorities (rebuild_zebra with -b -r -v )
5. make some searchs in staff interface and OPAC, with simple index and specific indexes (title, author...)
6. check no index is broken
7. put some very special value (like "qwerty000") in a record, in a field NOT indexed in any specific Zebra index (it depends of your marcflavor...)
8. search this special value in simple search => you should have 0 result
repeat these steps for MARC21 and UNIMARC
Maybe that's not a so good idea ;-)
Message sent to koha-devel.
I added a followup to bug 8962 to implement the anywhere index and cut down the any index for dom as per that bug. I think that mostly removes the need for this patch.
If I recall correctly, we're supposed to make changes to the XML then produce the requisite XSLT based on that using xsltproc.
This patch just uses the XSLT. Please provide a new patch that updates the XML and post the XSLT that is created from that XML.
Hello Martin and David
as I am no longer involved in Koha development, except for a few bugs I am still watching, I won't work anymore on this bug, or bug 8962. But I hope you will find a solution for improving Koha's indexing, which is not perfect ;-)