full text search index

On Thursday, May 26, 2016, Patrick Baker <patrickbakerbr(at)gmail(dot)com> wrote:

> Hi there,
>
> I've got the following query:
>
>>
>> SELECT COUNT(DISTINCT j0_.id) AS sclr10
>> FROM customers j0_
>> WHERE ((LOWER(j0_.name_first) LIKE '%some%'
>> OR LOWER(j0_.name_last) LIKE '%some%')
>> AND j0_.id = 5)
>> AND j0_.id = 5
>
>
> The query is taking ages to run.
>
> I read about wildcards and it seems I have to use a function with to_tsvector
> ?
>
>>
>> CREATE INDEX CONCURRENTLY ON public.customers USING gin ("clientid",
>> ("full_text_universal_cast"("name_first"::"text")),
>> ("full_text_universal_cast"("name_last"::"text")));
>
>
> full_text_universal_cast:
>
>> CREATE OR REPLACE FUNCTION public.full_text_universal_cast(doc_data
>> "text")
>> RETURNS "tsvector" AS
>> $BODY$
>> SELECT to_tsvector('english', COALESCE(TRIM(CAST(doc_data AS TEXT)), ''));
>> $BODY$
>> LANGUAGE sql IMMUTABLE
>> COST 1000;
>
>
>
>
> Would be something like above? Because it's not working...
>
> What am I missing guys?
> Thanks
>

Maybe Lucas Possamai can help. He seems to be a little bit further along
in this exercise.

I'm too tired to care at the moment. And I haven't had much practical work
here anyway.

David J.