I need to call a function dynamically (say handler() ) in PL/SQL function which returns a Nested Table.
Code:
BEGIN
...
v_function := 'handler'; //shown like this of simplicity.
EXECUTE IMMEDIATE 'BEGIN :result := ' || v_function || '(...); END;'
USING OUT v_error_msg;
... //process v_error_msg
END;
and the handler() specification:
TYPE t_error_msgs IS TABLE OF VARCHAR2(2000);
FUNCTION handle (...)
RETURN t_error_msgs;
Issue is I get PL-00457:expressions have to be of SQL types while compiling as execute immediate wont allow non-sql types to be binded.
Is there any way around ?
It depends what you mean by 'workaround' The type will have to be declared at SQL level, not within a PL/SQL block (presumably a package in this case). This would work, for example:
CREATE OR REPLACE TYPE t_error_msgs AS TABLE OF VARCHAR2(2000)
/
CREATE OR REPLACE PACKAGE p42 AS
FUNCTION handler RETURN t_error_msgs;
END p42;
/
CREATE OR REPLACE PACKAGE BODY p42 AS
FUNCTION handler RETURN t_error_msgs IS
BEGIN
RETURN null; -- put real data here, obviously...
END handler;
END p42;
/
DECLARE
v_error_msg t_error_msgs;
v_function varchar2(30);
BEGIN
v_function := 'p42.handler';
EXECUTE IMMEDIATE 'BEGIN :result := ' || v_function || '; END;'
USING OUT v_error_msg;
END;
/
Alternatively you can reconsider whether you really need this to be dynamic. Presumably you're passing or somehow determining the function to call on the fly and populating v_function. If there's a relatively short list of possible values it might be simpler to have a case with individual static function calls.
Related
i am trying to call a function from stored procedure in Oracle, but not getting any idea how to do.
my function has two IN parameter and one OUT parameter.
in my procedure i am using out sys refcursor . Any refrence or example will help me a lot.
Here is a simple example for calling function inside procedure. Also as mentioned by APC using OUT in function is a bad practice. Instead you can return your required output. And I'm not sure how you are using sys_refcursor, so modify your procedure accordingly
CREATE OR REPLACE FUNCTION SUM_OF_2(NUM1 IN NUMBER,NUM2 IN NUMBER) RETURN NUMBER
IS
RESULT_SUM NUMBER;
BEGIN
RESULT_SUM:=NUM1+NUM2;
RETURN RESULT_SUM;
END;
CREATE OR REPLACE PROCEDURE CALL_FUNCTON(NUM1 NUMBER,NUM2 NUMBER)
AS
V_FINAL_RESULT NUMBER;
BEGIN
V_FINAL_RESULT:=SUM_OF_2(NUM1,NUM2);
DBMS_OUTPUT.PUT_LINE(V_FINAL_RESULT);
END;
BEGIN
CALL_FUNCTON(5,10);
END;
/
CHECK DEMO HERE
Not sure on what your requirement is , maybe you are just trying the code for education purposes. Generally I have not seen much code which uses OUT parameter with functions, in case you want to return multiple values to the caller object then you could use a procedure with more then one OUT variables. There are some limitation on how an oracle function with OUT parameter would differ from a normal function.
CREATE OR REPLACE FUNCTION temp_demo_func(out_var1 OUT NUMBER)
RETURN VARCHAR2 IS
BEGIN
out_var1 := 1;
RETURN 'T';
EXCEPTION
WHEN OTHERS THEN
RETURN 'F';
END temp_demo_func;
/
CREATE OR REPLACE PROCEDURE temp_demo_proc
(
in_var1 NUMBER
,cur_refcur_out OUT SYS_REFCURSOR
) IS
res VARCHAR2(1);
out_var1 NUMBER;
BEGIN
res := temp_demo_func(out_var1 => out_var1);
dbms_output.put_line(out_var1);
OPEN cur_refcur_out FOR
SELECT in_var1
,out_var1
,res
FROM dual;
END;
/
set serveroutput on
declare
cur_refcur_out Sys_Refcursor;
in_var1 number := 22;
begin
temp_demo_proc(in_var1 => in_var1
,cur_refcur_out => cur_refcur_out);
end;
/
I have a scenario that I want to execute dynamically fetched methods from a cursor with different arguments. Those argument values are replaced (using Get_Parameter_Value___(head_rec_.objkey,parameter_); ) with values in a loop as you can see in the following example.
PROCEDURE Dynamic_exe(
keyvalue_ IN VARCHAR2)
IS
param_str_ VARCHAR2(2000);
temp_param_str_ VARCHAR2(2000);
method_stmt_ VARCHAR2(100);
CURSOR get_method IS
SELECT exe_method
FROM method_tab
BEGIN
param_str_ := Substr(rec_.exe_method,Instr(rec_.exe_method,'(')+1,(Instr(rec_.exe_method,')')-1)-Instr(rec_.exe_method,'('));
temp_param_str_ := param_str_;
method_stmt_ := rec_.exe_method;
WHILE temp_param_str_ IS NOT NULL LOOP
IF (Instr(temp_param_str_,',') > 0 )THEN
parameter_ := trim(Substr(temp_param_str_,1,Instr(temp_param_str_,',')-1));
temp_param_str_ := Substr(temp_param_str_,Instr(temp_param_str_,',')+1);
ELSE
parameter_ := trim(temp_param_str_);
temp_param_str_ := NULL;
END IF;
parameter_value_ := Get_Parameter_Value___(head_rec_.objkey,parameter_);
method_stmt_ := REPLACE(method_stmt_,parameter_,''''||parameter_value_||'''');
END LOOP;
FOR rec_ IN get_method LOOP
EXECUTE IMMEDIATE 'BEGIN '||method_stmt_||'; END;';
END LOOP;
END Dynamic_exe;
This is not safe, SQL injection can be done for this, I need a solution, associated with bind variables, Does anyone have a solution for this?
You can eliminate the possibility of SQL injection by using DBMS_ASSERT.SQL_OBJECT_NAME to protect the method name, and use DBMS_SQL and bind variables to protect the arguments.
DBMS_ASSERT.SQL_OBJECT_NAME throws an error if the value is not the same as an existing object. (Although for packages it only checks that the package name exists, not the procedure name. But the procedure name must still be a realistic name.)
For example, if the package name exists, the function will simply return the name:
SQL> select dbms_assert.SQL_OBJECT_NAME('test_package.test_procedure') name from dual;
NAME
--------------------------------------------------------------------------------
test_package.test_procedure
But any SQL injection shenanigans will raise an exception (which you can catch and handle if necessary):
SQL> select dbms_assert.sql_object_name('; delete from employees;') from dual;
select dbms_assert.sql_object_name('; delete from employees;') from dual
*
ERROR at line 1:
ORA-44002: invalid object name
ORA-06512: at "SYS.DBMS_ASSERT", line 401
Instead of building the entire statement as a string, add :bind_variable_n and DBMS_SQL to run it.
So the final string will look something like this (add the bind variable numbers in the loop):
method_stmt_ := 'begin '||method_name||'(:1, :2); end;';
Executing a dynamic number of bind variables requires DBMS_SQL.BIND_VARIABLE. Switching from native dynamic SQL to DBMS_SQL is going to be annoying, but it will let you pass in the bind variables without any injection concerns.
I have a procedure that has DML commands. the procedure accepts a variable of type out, and it returns a value.
i need call this procedure from a function.
the goal is that the function will return the value of the variable out returns the procedure.
(i need it for SSIS, but I believe that it is useful in other cases.)
During attempts, I got these errors:
ORA-14551: cannot perform a DML operation inside a query tips.
ORA-06519: active autonomous transaction detected and rolled back.
I'm looking for the right syntax to do it.
Example of a solution that works:
The procedure:
create or replace procedure MyProc(outRes OUT NUMBER)
is
begin
update some_table set some_description = 'abc';
commit;
if (some_condition) then
outRes := 666;
else
outRes := 999;
end if;
end MyProc;
Note: You must do commit; at the end of the DML command.
The function:
CREATE or replace FUNCTION MyFunc
RETURN int IS
PRAGMA AUTONOMOUS_TRANSACTION;
myVar number;
begin
MyProc(myVar);
return myVar;
END MyFunc;
Note that at the beginning of the function has: PRAGMA AUTONOMOUS_TRANSACTION;
And this function call:
select MyFunc() as res from dual;
Here is an example of what you need to do. Do know this is UNTESTED, but should give you a general idea of which way to go. This is known as Dynamic SQL and uses bind variables. There's a lot more I don't know such as the data type your procedure spits out and what not... so if it's not varchar2 then change it accordingly...
FUNCTION myFunc(procedure_call varchar2) RETURN VARCHAR2
IS
v_out1 varchar2(500);
BEGIN
EXECUTE IMMEDIATE 'begin '||procedure_call||'( :out1 ); end;' using v_out1;
RETURN v_out;
END;
I am new to PL/SQL. I have created a pipelined function inside a package which takes as its parameter input an array of numbers (nested table).
But I am having trouble trying to run it via an sql query. Please see below
my input array
CREATE OR REPLACE TYPE num_array is TABLE of number;
my function declaration
CREATE OR REPLACE PACKAGE "my_pack" as
TYPE myRecord is RECORD(column_a NUMBER);
TYPE myTable IS TABLE of myRecord;
FUNCTION My_Function(inp_param num_array) return myTable PIPELINED;
end my_pack;
my function definition
CREATE OR REPLACE PACKAGE BODY "my_pack" as
FUNCTION My_Function(inp_param num_array) return myTable PIPELINED as
rec myRecord;
BEGIN
FOR i in 1..inp_param.count LOOP
FOR e IN
(
SELECT column_a FROM table_a where id=inp_param(i)
)
LOOP
rec.column_a := e.column_a;
PIPE ROW (rec);
END LOOP;
END LOOP;
RETURN;
END;
end my_pack;
Here is the latest code I've tried running from toad. But it doesn't work
declare
myarray num_array;
qrySQL varchar2(4000);
begin
myarray := num_array(6341,6468);
qrySQL := 'select * from TABLE(my_pack.My_Function(:myarray))';
execute immediate qrySQL;
end;
So my question is how can I feed an array into this pipelined function from either TOAD or SQL Developer. An example would be really handy.
Thanks
The error is fairly clear, you have a bind variable that you haven't assigned anything to. You need to pass your actual array with:
qrySQL := 'select * from TABLE(my_pack.My_Function(:myarray))';
execute immediate qrySQL using myarray;
It's maybe more useful, if you want to call it from PL/SQL, to use static SQL as a cursor:
set serveroutput on
declare
myarray num_array;
begin
myarray := num_array(6341,6468);
for r in (select * from TABLE(my_pack.My_Function(myarray))) loop
dbms_output.put_line(r.column_a);
end loop;
end;
/
Or just query it statically as a test, for fixed values:
select * from TABLE(my_pack.My_Function(num_array(6341,6468)));
SQL Fiddle with some minor tweaks to the function to remove errors I think came from editing to post.
How to pass pl/sql record type to a procedure :
CREATE OR REPLACE PACKAGE BODY PKGDeleteNumber
AS
PROCEDURE deleteNumber (
list_of_numbers IN List_Numbers
)
IS
i_write VARCHAR2(5);
BEGIN
--do something
END deleteNumber;
END PKGDeleteNumber;
/
In this procedure deleteNumber I have used List_Numbers, which is a record type. The package declaration for the same is :
CREATE OR REPLACE PACKAGE PKGDeleteNumber
AS
TYPE List_Numbers IS RECORD (
IID NUMBER
);
TYPE list_of_numbers IS TABLE OF List_Numbers;
PROCEDURE deleteNumber (
list_of_numbers IN List_Numbers
);
END PKGDeleteNumber;
I have to execute the procedure deleteNumber passing a list of values. I inserted numbers in temp_test table, then using a cursor U fetched the data from it :
SELECT *
BULK COLLECT INTO test1
FROM temp_test;
Now, to call the procedure I am using
execute immediate 'begin PKGDELETENUMBER.DELETENUMBER(:1); end;'
using test1;
I have tried many other things as well(for loop, dbms_binding, etc). How do I pass a pl/sql record type as argument to the procedure?
EDIT:
Basically, I want to pass a list of numbers, using native dynamic sql only...
adding the table temp_test defn (no index or constraint):
create table test_temp (
IID number
);
and then inserted 1,2,3,4,5 using normal insert statements.
For this solution,
In a package testproc
CREATE TYPE num_tab_t IS TABLE OF NUMBER;
CREATE OR REPLACE PROCEDURE my_dyn_proc_test (p_num_array IN num_tab_t) AS
BEGIN
dbms_output.put_line(p_num_array.COUNT);
END;
/
this is called from sql prompt/toad
DECLARE
v_tab testproc.num_tab_t := testproc.num_tab_t(1, 10);
BEGIN
EXECUTE IMMEDIATE 'BEGIN testproc.my_dyn_proc_test(:1); END;' USING v_tab;
END;
this will not work.This shows error.I am not at my workstation so am not able to reproduce the issue now.
You can't use RECORD types in USING clause of EXECUTE IMMEDIATE statement. If you just want to pass a list of numbers, why don't you just use a variable of TABLE OF NUMBER type? Check below example:
CREATE TYPE num_tab_t IS TABLE OF NUMBER;
CREATE OR REPLACE PROCEDURE my_dyn_proc_test (p_num_array IN num_tab_t) AS
BEGIN
dbms_output.put_line(p_num_array.COUNT);
END;
/
DECLARE
v_tab num_tab_t := num_tab_t(1, 10);
BEGIN
EXECUTE IMMEDIATE 'BEGIN my_dyn_proc_test(:1); END;' USING v_tab;
END;
Output:
2
Edit
Try this:
CREATE TYPE num_tab_t IS TABLE OF NUMBER;
CREATE OR REPLACE PACKAGE testproc AS
PROCEDURE my_dyn_proc_test (p_num_array IN num_tab_t);
END;
/
CREATE OR REPLACE PACKAGE BODY testproc AS
PROCEDURE my_dyn_proc_test (p_num_array IN num_tab_t) AS
BEGIN
dbms_output.put_line(p_num_array.COUNT);
END;
END;
/
DECLARE
v_tab num_tab_t := num_tab_t(1, 10);
BEGIN
EXECUTE IMMEDIATE 'BEGIN testproc.my_dyn_proc_test(:1); END;' USING v_tab;
END;
Use an object type. object types are visible to all packages