Re: not quite a cross tab query... - Mailing list pgsql-general

From Steve Atkins
Subject Re: not quite a cross tab query...
Date
Msg-id A826A2D8-F33C-4B72-B5D2-FDD7C499CDF8@blighty.com
Whole thread Raw
In response to not quite a cross tab query...  (Richard Greenwood <richard.greenwood@gmail.com>)
Responses Re: not quite a cross tab query...  (Richard Greenwood <richard.greenwood@gmail.com>)
List pgsql-general
On Mar 5, 2009, at 11:17 AM, Richard Greenwood wrote:

> Hello pgsql listers,
>
> I've got a problem that is similar to, but I don't think identical to,
> a cross tab query. My data looks like:
>  ID | CAT
>  1  |   A
>  1  |   B
>  2  |   A
>  2  |   C
> So for each ID there may be many CAT (categories).
> The client wants it to look like:
>  ID | CATS
>  1  |  A,B
>  2  |  A,C
> Where each ID is unique in the results, and the CAT values are
> concatenated with a comma separator.
>
> There are about 100 unique CAT values. They only way I can see to do
> it is programatically with a for loop. But before I do that I wanted
> to bounce it off the fine minds that inhabit this list.

Sounds like a job for array_accum(), which you can find mentioned
at http://www.postgresql.org/docs/8.3/interactive/xaggr.html

select id, array_to_string(array_accum(cat), ',') from table group by
id;

There'll be a built-in array_agg() in 8.4, I think.

Cheers,
   Steve



pgsql-general by date:

Previous
From: "A. Kretschmer"
Date:
Subject: Re: Postgres Cookbook
Next
From: Sathish Duraiswamy
Date:
Subject: Re: converting older databases