[Koha-devel] Thoughts on Search Syntax (Zebra but maybe Elasticsearch)

[dcook at prosentient.com.au]

Hi all,

 

I ran into a problem today where the search query 'Emperor penguin
(Aptenodytes forsteri) foraging ecology' (without single quotes) was
creating a "CCL parsing error" in Zebra. I had QueryAutoTruncate enabled so
QueryWeightFields was automatically disabled.

 

Solutions include:

1.	Wrap the search query in double quotes

a.	"Emperor penguin (Aptenodytes forsteri) foraging ecology"

2.	Escape the parentheses with a backslash

a.	Emperor penguin \(Aptenodytes forsteri\) foraging ecology

3.	Turn off QueryAutoTruncate

a.	This lets the weighted query code thoroughly mangle the query which
includes wrapping it in double quotes.
b.	Note though that this option also makes it so that a search wrapped
in double quotes (like Solution 1) becomes a CCL parsing error

 

I think many of us know that Koha's Zebra-based query building is terrible,
but changing it obviously has consequences and we have so many different
search preference permutations that it makes it difficult to regression test
search changes.

 

We could start showing users when they have searches with syntax errors. but
then we all will probably start getting a lot of calls. Although I was
thinking that when showing that there was a syntax error with their query,
we could also provide a little "cheat sheet" with the basics of Zebra's CCL
syntax. (Of course, that's really only relevant when we're passing queries
straight through to Zebra and not butchering them first in Koha.)

 

We could try to fix search, but then who has the time/money for that? Why
fix something that is not (seemingly) broken?

 

I am wondering what other people think. 

 

Personally, I'm thinking that we should not allow complex queries in the
OPAC "Search" box. We treat that box (as we always should have) as the
operand to the index chosen in the "Library catalogue" drop down list.
However. that would be a breaking change for anyone that has bookmarked any
OPAC searches with complex search queries. I suppose that we could create a
new System Preference like "SimplifySearch" and just enable that for new
Koha installations though. Maybe one day we could then remove that system
preference and just make it "The Way". 

 

After all, I think that the *majority* of people are going to be performing
fairly straight forward searches in the OPAC "Search" box. They're not going
to be doing things like '(kw,phr,rtrn:"David writes verbose emails") OR (
(ti,phr,ext:"I think David's ideas have some merit") AND (su:Emails) )'. 

 

Relevant bugs:

https://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=27088

https://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=27139

 

David Cook

Software Engineer

Prosentient Systems

72/330 Wattle St

Ultimo, NSW 2007

Australia

 

Office: 02 9212 0899

Online: 02 8005 0595

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.koha-community.org/pipermail/koha-devel/attachments/20201203/84f89ec7/attachment.htm>