PL/pgSQL syntax error - postgresql

I have a very simple PL/pgSQL script:
declare x varchar(100);
When I run it I get a message:
[WARNING ] declare x varchar(100)
ERROR: syntax error at or near "varchar"
LINE 1: declare x varchar(100)
^
I really don't understand what is wrong with this.

you can use procedural statements only inside function body in PostgreSQL.
CREATE OR REPLACE FUNCTION foo()
RETURNS int AS
$$ -- here start procedural part
DECLARE x int;
BEGIN
x := 10;
RETURN x;
END;
$$ -- here finish procedural part
LANGUAGE plpgsql; -- language specification
or in temporary function (anonymous block)
DO $$
DECLARE x int;
BEGIN
x := 10;
RAISE NOTICE '>>>%<<<', x;
END;
$$;
isn't possible to use procedural statements as SQL statements like T-SQL.

Use exemple
DO $$
Declare
test varchar;
begin
test := 'teste';
if (char_length(test) > 0) then
RAISE NOTICE '>>>%<<<', test;
end if;
end;
$$;

I have resolved this by the following piece of code:
do $$
declare TypeId int;
declare MasterId int;
begin
TypeId := null;
MasterId := null;
end;
$$;

Related

"Not found" condition being recognized as a type/variable?

CREATE OR REPLACE FUNCTION increaseSomeOfferPricesFunction(IN theOffererID INTEGER, IN numOfferIncreases INTEGER) RETURNS INT AS $$
BEGIN
DECLARE a INTEGER;
DECLARE b INTEGER;
DECLARE c DATE;
DECLARE d INTEGER = 0;
DECLARE NotFound CONDITION FOR SQLSTATE '02000';
When I try to create this stored function, I get the error CONTEXT: invalid type name "CONDITION FOR SQLSTATE '02000'"
I'm confused as to why this is happening, as I'm almost positive the syntax is correct.
This is SQL/PSM language (it is used by MySQL or DB2) - It is not PL/pgSQL. MySQL uses different language for stored procedures than Postgres does.
PL/pgSQL requires variable declaration before BEGIN keyword. Second, PL/pgSQL doesn't support conditions declaration:
CREATE OR REPLACE FUNCTION increaseSomeOfferPricesFunction(IN theOffererID INTEGER,
IN numOfferIncreases INTEGER)
RETURNS INT AS $$
DECLARE
a INTEGER;
b INTEGER;
c DATE;
d INTEGER = 0;
BEGIN
SELECT ...
IF NOT FOUND THEN ...
END IF;
END;
$$ LANGUAGE plpgsql;

postgreSQL : oracle sqlerrm equivalent of postgres

I'm new to PostgreSQL. I have experience in oracle. In oracle , to find the exact error, I use code 'dbms_output.put_line(sqlerrm)' . Here I have a postgresql function returning an integer value
CREATE OR REPLACE FUNCTION public.fn_sqltest(
p_id integer)
RETURNS integer
LANGUAGE 'plpgsql'
COST 100
VOLATILE
AS $BODY$
Declare
n integer;
begin
select off_id into n from office
where per_id=p_id;
return n ;
exception when others then
return -1;
end;
$BODY$;
ALTER FUNCTION public.fn_sqltest(character varying)
OWNER TO postgres;
I call this function as below
DO $$
DECLARE
ae integer;
BEGIN
ae:=fn_sqltest(10);
RAISE NOTICE 'exception: % % ', sqlstate , sqlerrm ;
RAISE NOTICE 'Return value is: % ', ae;
END $$;
and I get the error
ERROR: column "sqlstate" does not exist
How can I show the exact error message like sqlerrm in oracle.
I updated the function and wrote a code in function exception section
CREATE OR REPLACE FUNCTION public.fn_sqltest(
p_id integer)
RETURNS integer
LANGUAGE 'plpgsql'
COST 100
VOLATILE
AS $BODY$
Declare
n integer;
text_var1 text;
text_var2 text;
text_var3 text;
begin
select off_id into n from office
where per_id=p_id;
return n ;
exception when others then
GET STACKED DIAGNOSTICS text_var1 = MESSAGE_TEXT,
text_var2 = PG_EXCEPTION_DETAIL,
text_var3 = PG_EXCEPTION_HINT;
RAISE NOTICE 'Return value is: % % %',text_var1 , text_var2, text_var3;
return -1;
end;
$BODY$;
ALTER FUNCTION public.fn_sqltest(character varying)
OWNER TO postgres;
Another method is by changing the return type. I changed the return type to text and rewrite the exception section code as below
return sqlerrm;
Unlike PL/SQL, in PL/pgSQL SQLSTATE and SQLERRM are not defined outside an exception handler. See the documentation, section "Obtaining Information About An Error".
This also means that you can't get SQLSTATE and SQLERRM of a successful operation, unlike PL/SQL.
So, if you want to use these special variables outside an exception handler, you have to store them in variables.
I don't know how would you like to return them from the function. I can demonstrate this idea inside your function code:
CREATE OR REPLACE FUNCTION public.fn_sqltest(
p_id integer)
RETURNS integer
LANGUAGE 'plpgsql'
COST 100
VOLATILE
AS $BODY$
Declare
n integer;
v_sqlerrm text;
v_sqlstate text;
begin
begin
select off_id into n from office
where per_id=p_id;
return n ;
exception when others then
v_sqlerrm := sqlerrm;
v_sqlstate := sqlstate;
end;
RAISE NOTICE 'exception: % % ', v_sqlstate , v_sqlerrm ;
return -1;
end;
$BODY$;
I know this is old but just for the record: The easiest way to output the state and message of error is to raise them from within the exception clause. You don't need to declare extra variables.
CREATE OR REPLACE FUNCTION fn_sqltest(p_id integer)
RETURNS integer
AS $$
DECLARE
n integer;
BEGIN
SELECT off_id
INTO n
FROM office
WHERE per_id = p_id;
RETURN n;
EXCEPTION WHEN OTHERS THEN
RAISE NOTICE 'exception: % - %', SQLSTATE, SQLERRM;
RETURN -1;
END;
$$ LANGUAGE plpgsql;

How to initialize row datatype variables?

I would like to create a function that initialize and return the row datatype of a table as
CREATE FUNCTION get_default_table_row_object()
RETURNS mytable AS $$
DECLARE
row mytable;
BEGIN
row.field1 := 0;
row.field2 := -1;
row.record_reg_id := 1;
row.record_upd_id := 1;
row.record_reg_date := current_timestamp;
row.record_upd_date := current_timestamp;
RETURN row;
END;
$$ LANGUAGE plpgsql;
becuase my table has alot of columns and I need to create dozens of variables at several functions. I would like to use above function as
CREATE FUNCTION some_function() RETURNS VOID AS $$
DECLARE
i_obj1 mytable := get_default_table_row_object(); -- declare and initialize default values
BEGIN
-- function body
END;
$$ LANGUAGE plpgsql;
But this give me the error ERROR: default value for row or record variable is not supported. Has someway to figure it out ?
You can set it in the body instead, like so:
CREATE FUNCTION some_function() RETURNS VOID AS $$
DECLARE
i_obj1 mytable; -- declare only
BEGIN
i_obj1 := get_default_table_row_object(); -- set default values
END;
$$ LANGUAGE plpgsql;

How to pass table name to plpgsql function

I am trying to run the code:
CREATE OR REPLACE FUNCTION anly_work_tbls.testfncjh (tablename text) returns int
AS $$
DECLARE
counter int;
rec record;
tname text;
BEGIN
counter = 0;
tname := tablename;
FOR rec IN
select *
from tname
loop
counter = counter + 1;
end loop;
RETURN counter;
END;
$$
LANGUAGE 'plpgsql' IMMUTABLE
SECURITY DEFINER;
The goal of this code is to return the number of rows in the table you input. I know that this might not be the best way to accomplish this task, but the structure of this function would extend nicely to another question I am trying to tackle. Every time I run the code, I get the error:
ERROR: syntax error at or near "$1"
All online resources I have found tell me how to use the input variable within and EXECUTE block, but not in the above situation.
Currently running PostgreSQL 8.2.15.
CREATE OR REPLACE FUNCTION anly_work_tbls.testfncjh (tbl regclass, OUT row_ct int) AS
$func$
BEGIN
EXECUTE 'SELECT count(*) FROM '|| tbl
INTO row_ct;
END
$func$ LANGUAGE plpgsql IMMUTABLE SECURITY DEFINER;
Call:
SELECT anly_work_tbls.testfncjh('anly_work_tbls.text_tbl');
This should work for Postgres 8.2, but you consider upgrading to a current version anyway.
I pass the table name as object identifier type regclass, which takes care of quoting automatically and works with schema-qualified names. Details:
Table name as a PostgreSQL function parameter
Using an OUT parameter simplifies the function.
Don't quote the language name. It's an identifier.
If you actually need to loop through the result of a dynamic query:
CREATE OR REPLACE FUNCTION anly_work_tbls.testfncjh (tbl regclass)
RETURNS int AS
$func$
DECLARE
counter int := 0; -- init at declaration time
rec record;
BEGIN
FOR rec IN EXECUTE
'SELECT * FROM ' || tbl
LOOP
counter := counter + 1; -- placeholder for some serious action
END LOOP;
RETURN counter;
END
$func$ LANGUAGE plpgsql IMMUTABLE SECURITY DEFINER;
Read this chapter in the manual: Looping Through Query Results
The documented assignment operator in plpgsql is :=:
The forgotten assignment operator "=" and the commonplace ":="
Yes is really not the best way, but this would work:
CREATE OR REPLACE FUNCTION testfncjh (tablename text) returns int
AS $$
DECLARE
counter int;
rec record;
BEGIN
counter = 0;
FOR rec IN
EXECUTE 'select * from '||quote_ident(tablename) loop
counter = counter + 1;
end loop;
RETURN counter;
END;
$$
LANGUAGE 'plpgsql' IMMUTABLE
SECURITY DEFINER;
This would be nicer:
CREATE OR REPLACE FUNCTION testfncjh (tablename text) returns int
AS $$
DECLARE _count INT;
BEGIN
EXECUTE 'SELECT count(*) FROM '|| quote_ident(tablename) INTO _count;
RETURN _count;
END;
$$
LANGUAGE 'plpgsql' IMMUTABLE
SECURITY DEFINER;

How postgresql return a data set like the follows?

Just like the below function. I don't know how to return a set of inside parameter in postgresql?
create or replace function g_i(num int)
returns setof integer
as $$
declare
i int;
begin
while i < $1 loop
select i; -- How to write statements here?
end loop;
end;
$$ language plpgsql;
create or replace function g_i(num int)
returns setof integer
as $$
declare
i int;
begin
i := 0;
while i< $1 loop
i := i+1;
return query select i;
end loop;
end;
$$ language plpgsql;