Citext data type

[budi]

Good day!

I have an email column that I want to index, and I'm using citext extension to make it easier for case-insensitivity cases (ha ha ha).

But zombodb always gives me elog:

WARNING:  Unrecognized data type citext, pretending it's of type 'text'

Sad. 😢

Any plan to put citext into the domain?

[eeeebbbbrrrr]

Sure, that's easy enough to do. I'll take care of this in the next version of ZomboDB.

Note that ZDB searches are case-insensitive by default, so there's no need to use citext for ZomboDB. If you're also querying that column directly from SQL, however, then there might be good reasons.

[budi]

Yup!
Suppressing logs when it is way too obvious is a huge deal :)

[budi]

Just to be clear, it just so happens that the citext (email) column is something I want to query with ES.. like so:

...
 USING zombodb(zdb('blah', blah.ctid), zdb(blah))
  WITH ( url='#{ENV['ELASTIC_HOST']}',
         shards=5,
         replicas=10,
         field_lists='search=[other_column, email]' );

so then I can:

SELECT * 
  FROM blah
 WHERE zdb('users', ctid) ==> 'search:"*combo*"';

its_so_fluffy.gif

[eeeebbbbrrrr]

Yep, we're on the same page. When I add support for the citext type, I'll do it exactly the way ZomboDB currently handles text. i.e., I'm going to add it to this block in zdbops.c:

        } else if (strcmp("text", typename) == 0 || strcmp("varchar", typename) == 0 ||
                   strcmp("character", typename) == 0 || strcmp("character varying", typename) == 0 ||
                   strcmp("text[]", typename) == 0 || strcmp("varchar[]", typename) == 0 ||
                   strcmp("character[]", typename) == 0 || strcmp("character varying[]", typename) == 0 ||
                   strcmp("uuid", typename) == 0 || strcmp("uuid[]", typename) == 0) {
            /* string field */
            appendStringInfo(result, "\"type\": \"string\",");
            appendStringInfo(result, "\"norms\": {\"enabled\":false},");
            appendStringInfo(result, "\"index_options\": \"docs\",");
            appendStringInfo(result, "\"ignore_above\":32000,");
            appendStringInfo(result, "\"analyzer\": \"exact\"");

        }

If you need tokenization/analysis then it's up to you to either use a predefined domain type like phrase or fulltext or to use zdb_define_mapping() to set custom mappings.

[budi]

I was just reading the SQL-API and got a eureka moment only to come back and read your comment :))

[eeeebbbbrrrr]

Should also read the TYPE-MAPPING.md document. :)

[eeeebbbbrrrr]

Change of plans... going to release in a v3.0.1

[eeeebbbbrrrr]

Released in v3.0.1