Need help to dynamically access to colomns in function!

Started by Ivanmaraover 17 years ago6 messagesgeneral
Jump to latest
#1Ivanmara
aesthete2005@gmail.com

Hello. I have table classif with columns:
... , group1, group2, group3, ... , group48, ...

In function i do query and want run on every row and dynamically operate on
columns from group1 to group20. I do something like this:

OPEN curs FOR select * from classif;
loop
fetch curs into tmprec;
exit when not found;

for I in 1..20 loop
...
-- problem code is:
value := tmprec.group{I};
-- i cannot dynamically access to group1, group2, ... colomns according
to "I" variable.
...

end loop;
end loop;

I have to manually identify and handle each entry without a cycle do
something like this:
value := tmprec.group1;
...
value := tmprec.group2;
...
value := tmprec.group2;
...
value := tmprec.group20;

Please help me to do it dynamically with a loop, depending on the I?
something like this in a loop:
value := tmprec.group{I};

Thanks.

#2Ivan Pavlov
ivan.pavlov@gmail.com
In reply to: Ivanmara (#1)
Re: Need help to dynamically access to colomns in function!

If you need them one by one why fetch them into tmprec? Take a look at
the docs for FETCH:
http://www.postgresql.org/docs/8.3/static/sql-fetch.html

especially the FETCH ABSOLUTE...

regards,
Ivan Pavlov

Show quoted text

On Dec 16, 3:37 pm, aesthete2...@gmail.com ("Иван Марков") wrote:

Hello. I have table classif with columns:
... , group1, group2, group3, ... , group48, ...

In function i do query and want run on every row and dynamically operate on
columns from group1 to group20. I do something like this:

OPEN curs FOR select * from classif;
 loop
    fetch curs into tmprec;
    exit when not found;

    for I in 1..20 loop
        ...
    -- problem code is:
        value := tmprec.group{I};
    -- i cannot dynamically access to group1, group2, ... colomns according
to "I" variable.
    ...

    end loop;
end loop;

I have to manually identify and handle each entry without a cycle do
something like this:
value := tmprec.group1;
...
value := tmprec.group2;
...
value := tmprec.group2;
...
value := tmprec.group20;

Please help me to do it dynamically with a loop, depending on the I?
something like this in a loop:
value := tmprec.group{I};

Thanks.

#3Ivan Pavlov
ivan.pavlov@gmail.com
In reply to: Ivanmara (#1)
Re: Need help to dynamically access to colomns in function!

Please disregard my other message. I didn't get what you are trying to
do at first.
You can do this with dynamic SQL: Look at 38.5.4. Executing Dynamic
Commands (http://www.postgresql.org/docs/current/static/plpgsql-
statements.html).
I guess something like

EXECUTE 'SELECT tmprec.group' || i INTO value

will work in a loop. Didn't try it though.

regards,
Ivan Pavlov

#4Hoover, Jeffrey
jhoover@jcvi.org
In reply to: Ivan Pavlov (#2)
Re: Need help to dynamically access to colomns in function!

Create a plpgsql function that reformats your row as an array.

Example:
I have a table named task_parameter with three columns:

Table "camera.task_parameter"
Column | Type | Modifiers
--------------------------+---------------------------------+-----------
parameter_name | character varying(255) | not null
parameter_value | text |
task_id | bigint | not null

I create a plpgsql function that takes a task_parameter row and returns an array, one column per array entry:

create or replace function task_parameter_array(tp task_parameter) returns text[] as $$
declare
result text[];
begin
result[1] := tp.parameter_name;
result[2] := tp.parameter_value;
result[3] := tp.task_id;
return result;
end $$ language plpgsql;

select task_parameter_array(task_parameter) from task_parameter limit 1;
task_parameter_array
----------------------------------------------------
{"db alignments per query",25,1286428019358957945}
(1 row)

You can write a similar function for your table and then your code would look like:

OPEN curs FOR select classif_to_array(classif) as group_array from classif;
loop
fetch curs into tmprec;
exit when not found;

for I in 1..20 loop
...
value := tmprec.group_array{I};

of course, this begs the question, whjy not define you table to store an array...?

-----Original Message-----
From: pgsql-general-owner@postgresql.org [mailto:pgsql-general-owner@postgresql.org] On Behalf Of Ivan Pavlov
Sent: Tuesday, December 16, 2008 3:55 PM
To: pgsql-general@postgresql.org
Subject: Re: [GENERAL] Need help to dynamically access to colomns in function!

If you need them one by one why fetch them into tmprec? Take a look at
the docs for FETCH:
http://www.postgresql.org/docs/8.3/static/sql-fetch.html

especially the FETCH ABSOLUTE...

regards,
Ivan Pavlov

On Dec 16, 3:37 pm, aesthete2...@gmail.com ("Иван Марков") wrote:

Hello. I have table classif with columns:
... , group1, group2, group3, ... , group48, ...

In function i do query and want run on every row and dynamically operate on
columns from group1 to group20. I do something like this:

OPEN curs FOR select * from classif;
 loop
    fetch curs into tmprec;
    exit when not found;

    for I in 1..20 loop
        ...
    -- problem code is:
        value := tmprec.group{I};
    -- i cannot dynamically access to group1, group2, ... colomns according
to "I" variable.
    ...

    end loop;
end loop;

I have to manually identify and handle each entry without a cycle do
something like this:
value := tmprec.group1;
...
value := tmprec.group2;
...
value := tmprec.group2;
...
value := tmprec.group20;

Please help me to do it dynamically with a loop, depending on the I?
something like this in a loop:
value := tmprec.group{I};

Thanks.

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

#5Sam Mason
sam@samason.me.uk
In reply to: Ivanmara (#1)
Re: Need help to dynamically access to colomns in function!

On Tue, Dec 16, 2008 at 11:37:17PM +0300, IIIIIIII wrote:

Hello. I have table classif with columns:
... , group1, group2, group3, ... , group48, ...

In function i do query and want run on every row and dynamically operate on
columns from group1 to group20. I do something like this:

It sounds as though you should be using an ARRAY instead of having lots
of columns.

The best I could come up with would be doing something like:

OPEN curs FOR select * from classif;

OPEN curs FOR SELECT ARRAY[group1,group2,group3,group4] AS group FROM classif;

You'd obviously need to all the way up to "group20" here. If the syntax
gets a bit baroque you could create a view to do the same.

loop
fetch curs into tmprec;
exit when not found;

for I in 1..20 loop
...
-- problem code is:
value := tmprec.group{I};

value := tmprec.group[I];

-- i cannot dynamically access to group1, group2, ... colomns according
to "I" variable.
...

end loop;
end loop;

I have to manually identify and handle each entry without a cycle do
something like this:

You're using the wrong data type; RECORDs are for where you statically
know and care about the structure of the data, ARRAYs are when you care
at runtime.

Sam

#6David Fetter
david@fetter.org
In reply to: Ivanmara (#1)
Re: Need help to dynamically access to colomns in function!

On Tue, Dec 16, 2008 at 11:37:17PM +0300, Иван Марков wrote:

Hello. I have table classif with columns:
... , group1, group2, group3, ... , group48, ...

That's a very poor design because it's both denormalized and has very
poor naming. There are likely plenty of other things wrong with it,
too. Check http://thedailywtf.com/ for systems similar to yours. I
suspect it won't take long to find some.

In function i do query and want run on every row and dynamically
operate on columns from group1 to group20. I do something like this:

Nothing will really help until you fix your design, and dynamic
querying will only lead you further down this rat-hole.

The answer to, "how do I shoot myself in the foot?" is "Don't."

Cheers,
David.
--
David Fetter <david@fetter.org> http://fetter.org/
Phone: +1 415 235 3778 AIM: dfetter666 Yahoo!: dfetter
Skype: davidfetter XMPP: david.fetter@gmail.com

Remember to vote!
Consider donating to Postgres: http://www.postgresql.org/about/donate