[Koha] Single character searches fail in koha 3.20.4 but work fine in v3.2.2

[Andreas Roussos]

Hello everyone,

We are running koha 3.20.4 in a development VM and have a problem when
making single character searches (i.e. searches that would normally return
a lot of results). Oddly enough, our production koha which is on v3.2.2
does not suffer from the same issue (please see the attached screenshots).

I followed the instructions in
http://wiki.koha-community.org/wiki/Troubleshooting_Zebra and confirmed
that the zebra daemon does indeed return a result set when making a single
character search ("f α" for example). So, I think the actual problem must
lie with one of the Perl scripts that process the data fed back from
zebra/MySQL.

How do we figure out which Perl script is choking on invalid UTF8 data?
Is there a way to find which biblios/authorities contain invalid characters
and to correct them?

Thanks in advance for your help.

Kind regards,
Andreas