Re: solr field types and case sensitivity

[Dryganets Sergey <22...@gmail.com>]

ryantxu wrote:
> 
> yes, you would index a field twice - once with a LowerCaseFilter and 
> once without.  That is a good solution.
> 

I implement this and now there are some strange problems:

1. highlighting not always properly work for not case sensitive search

<fields>
        <field name="id" type="integer" indexed="true" stored="true"/>

        <field name="post_text" type="cs_text" indexed="false" stored="true"
multiValued="true"/>
        <field name="cs_post_text" type="cs_text" indexed="true"
stored="false" multiValued="true"/>
        <field name="ncs_post_text" type="ncs_text" indexed="true"
stored="false" multiValued="true"/>       
 </fields>

you can find full schema on:
https://issues.apache.org/jira/browse/SOLR-444

I know a bad solution for this problem - store data for the both cs and ncs
fields
(but I don't want store also 200mb for each search option)

As for me in solr now too much problems with highlighting :(

As I understand to highlight properly lucene-highlighter needs Analyzer, but
Analyzer linked to field type.
So I think it's root of this problem ...

maybe anyone know a good solution? 
ie solve this problem with just configuration changes

2.
add following document:
<add>
    <doc>
         <field name="id">3</field>
         <field name="post_text">Apples</field>
    </doc>
</add>



fl=*,score&q=ncs_post_text:apple&start=0&rows=10&hl=true&hl.fl=post_text
return document with id=3, but not return any highlighting info
(appl, apple and apples return same result)

fl=*,score&q=cs_post_text:Apple&start=0&rows=10&hl=true&hl.fl=post_text
return nothing, request for Appl return nothing too

and just Apples word return result

I don't understand it ... I just remove lower case factory









-- 
View this message in context: http://www.nabble.com/solr-field-types-and-case-sensitivity-tp14395912p14751690.html
Sent from the Solr - User mailing list archive at Nabble.com.