This is a question about Oracle PL/SQL.
I have a procedure in which the exact WHERE clause is not known until the run time:
DECLARE
CURSOR my_cursor is
SELECT ...
FROM ...
WHERE terms in (
(SELECT future_term2 FROM term_table), -- whether this element should be included is conditional
(SELECT future_term1 FROM term_table),
(SELECT present_term FROM term_table)
);
BEGIN
(the processing)
END;
/
What the (SELECT ... FROM term_table) query returns is a 4-character string.
For a solution to this, I am thinking of using a parameterized cursor:
DECLARE
target_terms SOME_DATATYPE;
CURSOR my_cursor (pi_terms IN SOME_DATATYPE) IS
SELECT ...
FROM ...
WHERE terms in my_cursor.pi_terms;
BEGIN
target_terms := CASE term_digit
WHEN '2' THEN (
(SELECT future_term2 FROM term_table),
(SELECT future_term1 FROM term_table),
(SELECT present_term FROM term_table)
) ELSE (
(SELECT future_term1 FROM term_table),
(SELECT present_term FROM term_table)
)
END;
FOR my_record IN my_cursor (target_terms) LOOP
(the processing)
END LOOP;
END;
/
The problem is what the datatype for SOME_DATATYPE should be is not known to me, nor is it known whether Oracle supports such a cursor parameter at all. If supported, is the way shown above to fabricate the value for target_terms correct? If not, how?
Hope someone who know can advise. And thanks a lot for the help.
You can certainly pass a parameter to a cursor, just like you can to a function - but only IN parameters. However, PL/SQL is a strongly typed language, so the datatype must be specified at the time of compilation.
It looks to me like what you will need to do is construct the query dynamically and then use
OPEN cursor FOR l_query;
where l_query is the constructed string. This should give you a feel for what you can do:
CREATE OR REPLACE PACKAGE return_id_sal
AUTHID DEFINER
IS
TYPE employee_rt IS RECORD
(
employee_id employees.employee_id%TYPE,
salary employees.salary%TYPE
);
FUNCTION allrows_by (append_to_from_in IN VARCHAR2 DEFAULT NULL)
RETURN SYS_REFCURSOR;
END return_id_sal;
/
CREATE OR REPLACE PACKAGE BODY return_id_sal
IS
FUNCTION allrows_by (append_to_from_in IN VARCHAR2 DEFAULT NULL)
RETURN SYS_REFCURSOR
IS
l_return SYS_REFCURSOR;
BEGIN
OPEN l_return FOR
'SELECT employee_id, salary FROM employees ' || append_to_from_in;
RETURN l_return;
END allrows_by;
END return_id_sal;
/
DECLARE
l_cursor SYS_REFCURSOR;
l_row return_id_sal.employee_rt;
BEGIN
l_cursor := return_id_sal.allrows_by ('WHERE department_id = 10');
LOOP
FETCH l_cursor INTO l_row;
EXIT WHEN l_cursor%NOTFOUND;
END LOOP;
END;
/
You will need to take precautions against SQL injection with this sort of code. Certainly a user should never be able to pass SQL text directly to such a function!
You can use also some built-in VARRAY SQL types like SYS.ODCIVARCHAR2LIST or create your own :
CREATE OR REPLACE NONEDITIONABLE TYPE VARCHARLIST
AS VARRAY(32767) OF VARCHAR2(4000);
Then you can use it with SELECT COLUMN_VALUE FROM TABLE(COLLECTION) statement in your cursor:
DECLARE
l_terms SYS.ODCIVARCHAR2LIS; --or VARCHARLIST
CURSOR my_cursor (p_terms IN SYS.ODCIVARCHAR2LIS) IS
SELECT your_column
FROM your_table
WHERE terms in (select COLUMN_VALUE from table (p_terms));
BEGIN
select term
bulk collect into l_terms
from (
select 'term1' term from dual
union all
select 'term2' term from dual
);
FOR my_record IN my_cursor (l_terms) LOOP
--process data from your cursor...
END LOOP;
END;
Related
Lets say I have a table as follows--
create table employees
(
eno number(4) not null primary key,
ename varchar2(30),
zip number(5) references zipcodes,
hdate date
);
And I'm trying to create a trigger with--
CREATE OR REPLACE TRIGGER TWELVE_ONE
BEFORE INSERT OR UPDATE
ON EMPLOYEES
FOR EACH ROW
DECLARE
V_DATE VARCHAR2 (10);
BEGIN
SELECT TO_CHAR (SYSDATE, 'hh24:mi:ss') INTO V_DATE FROM DUAL;
IF (V_DATE >= '12:00:01' AND V_DATE < '13:00:00')
THEN
INSERT INTO TABLE ?????
ELSE
ROLLBACK? TERMINATE TRANSACTION?
END IF;
END;
Purpose of the trigger is to allow an insertion/update during 12:00-13:00 and prevent the insertion at any other time. The trigger construction (thanks to #Melkikun) is seems ok. However now I'm facing the following issues--
How is it possible to pass the values here? I mean lets say my create statement is:
Insert into employees Values (1, 'someone', 11111, '17-12-2015')
And lets say the time is 12:30:01 now. How would the trigger perform the insertion without knowing the values?
And lets say the time is now 13:00:1 now. How would the trigger stop/prevent the insertion?
I'm using Oracle SQL Developer 4.02.15
Many Thanks
You just have to do it the other way.
If the time is not correct,then you raise an exception, so the insert won't be done.
CREATE OR REPLACE TRIGGER TWELVE_ONE
BEFORE INSERT OR UPDATE
ON EMPLOYEES
FOR EACH ROW
DECLARE
V_DATE VARCHAR2 (10);
MyException exception;
BEGIN
SELECT TO_CHAR (SYSDATE, 'hh24:mi:ss') INTO V_DATE FROM DUAL;
IF (V_DATE < '12:00:01' OR V_DATE > '13:00:00')
THEN
raise MyException;
END IF;
EXCEPTION
When MyException then
ROLLBACK;
//output message ...
END;
How would the trigger perform the insertion without knowing the values?
The trigger knows the value thanks to :NEW and :OLD.
You normally use the terms in a trigger using :old to reference the old value and :new to reference the new value.So you will have :NEW.eno ,:NEW.ename ...
Here is an example from the Oracle documentation :
CREATE OR REPLACE TRIGGER Print_salary_changes
BEFORE DELETE OR INSERT OR UPDATE ON Emp_tab
FOR EACH ROW
WHEN (new.Empno > 0)
DECLARE
sal_diff number;
BEGIN
sal_diff := :new.sal - :old.sal;
dbms_output.put('Old salary: ' || :old.sal);
dbms_output.put(' New salary: ' || :new.sal);
dbms_output.put_line(' Difference ' || sal_diff);
END;
I would like to create a procedure which returns a list of the first five records. I must use record type and table type. What am I doing wrong?
CREATE OR REPLACE PROCEDURE procedure_example(v_table OUT v_rec) IS
CURSOR cur1 IS
SELECT * FROM (
SELECT
type_note,
note
FROM dd_note ORDER BY type_note)
WHERE rownum < 5;
TYPE v_rec IS RECORD ( v_type_note NUMBER(2)
, v_note VARCHAR(30));
TYPE v_table IS TABLE OF v_rec INDEX BY BINARY_INTEGER;
BEGIN
OPEN cur1;
LOOP
FETCH cur1 INTO v_type_note, v_note;
dbms_output.put_line(v_type_note || '. --- ' || v_note);
EXIT WHEN cur1%NOTFOUND;
--enter code here
END LOOP;
CLOSE cur1;
END procedure_example;
The place where you declared your types is wrong. You must declare the type "v_rec" and type "v_table" in the package where this procedure belongs, NOT inside this procedure. Here's an improved version of your code. First, declare these types in your package:
TYPE v_rec IS RECORD(
v_type_note number(2),
v_note varchar(30));
TYPE v_table is table of v_rec index by pls_integer;
Then here's your function:
create or replace PROCEDURE procedure_example (out_table OUT v_table) IS
BEGIN
select type_note, note BULK COLLECT into out_table from (select type_note, note from dd_note order by type_note) where rownum < 5
FOR i in 1..out_table.COUNT LOOP
dbms_output.put_line(out_table(i).v_type_note || '. --- ' || out_table(i).v_note);
end loop;
END procedure_example;
I have a stored procedure that has following pl/sql block. This block was using select query in for statement but i need to change that static variable to dynamic query. As I changed that it has error. Is there any way to use variable with FOR LOOP in implicit cursor.
declare
sql_query varchar2(32767) := 'select ctlchar ';
kpiNameQuery varchar2(600);
isWg boolean := true;
begin
IF isWG then
kpiNameQuery := 'select distinct KPI_NAME from weeklykpi where kpi_name in (select kpi_wg from auxillary.kpi_types) order by 1';
Else
kpiNameQuery := 'select distinct KPI_NAME from weeklykpi where kpi_name in (select kpi_wg1 from auxillary.kpi_types) order by 1';
End IF;
for KPI_NAME in kpiNameQuery
loop
sql_query := sql_query || ' , min(case when KPI_NAME = '''||x.KPI_NAME||''' then KPI_VALUE end) as '||x.KPI_NAME;
dbms_output.put_line(sql_query);
end loop;
end;
You can achieve similar functionality with the following using cursor
declare
type t_cursor is ref cursor;
c_cursor t_cursor;
l_sql varchar2(512);
l_var number;
begin
l_sql := 'select count(*) from emp'; -- do dynamic check before here for
-- correct sql
open c_cursor for l_sql;
loop
fetch c_cursor
into l_var;
exit When c_cursor%notfound;
DBMS_OUTPUT.put_line ('val '||l_var);
end loop;
close c_cursor;
end;
Unfotunately no, the doc states:
If the dynamic SQL statement is a SELECT statement that returns multiple rows, native dynamic SQL gives you these choices:
Use the EXECUTE IMMEDIATE statement with the BULK COLLECT INTO clause.
Use the OPEN FOR, FETCH, and CLOSE statements.
So you will have to use a REF cursor (or EXECUTE IMMEDIATE and loop over the results).
Incidentally, in your case you could go for static SQL and have comparable performance:
BEGIN
FOR cc IN (SELECT DISTINCT KPI_NAME
FROM weeklykpi
WHERE kpi_name IN (SELECT CASE WHEN l_variable = 1
THEN kpi_wg
ELSE kpi_wg1
END
FROM auxillary.kpi_types) LOOP
ORDER BY 1
-- do something
END LOOP;
END;
You'll have to use some other type than boolean though since it's unknown to SQL.
I am a beginner in Oracle. I created the table type as like as follows:
TYPE metertable IS TABLE OF pseb.metermaster.meterid%type;
I dnt know how to insert the value into that table type. I need to store the whole resultset of the following query into the table type.
select distinct(meterid)
from pseb.consumerfeedermetermapper
where feederid in (select distinct (fm.FeederID)
from pseb.feedermaster fm,pseb.consumerfeedermetermapper cfm
where fm.substationid=v_v_type
and cfm.feederid=fm.feederid
and cfm.FeederID>0)
and meterid >0
order by meterid;
Help me to do that.
Use BULK COLLECT to select the data into a variable of that type:
declare
mt metertable;
begin
select distinct(meterid)
bulk collect into mt
from pseb.consumerfeedermetermapper
where feederid in (select distinct (fm.FeederID)
from pseb.feedermaster fm,pseb.consumerfeedermetermapper cfm
where fm.substationid=v_v_type
and cfm.feederid=fm.feederid
and cfm.FeederID>0)
and meterid >0
order by meterid;
-- Now use mt...
end;
I had the same issue, and this code help me :
SET SERVEROUTPUT ON
DECLARE
TYPE t_bulk_collect_test_tab IS TABLE OF bulk_collect_test%ROWTYPE;
l_tab t_bulk_collect_test_tab;
l_cursor SYS_REFCURSOR;
BEGIN
-- Way 1
OPEN l_cursor FOR 'SELECT * FROM bulk_collect_test';
FETCH l_cursor
BULK COLLECT INTO l_tab;
CLOSE l_cursor;
DBMS_OUTPUT.put_line('Dynamic FETCH : ' || l_tab.count);
-- Way 2
EXECUTE IMMEDIATE 'SELECT * FROM bulk_collect_test' BULK COLLECT INTO l_tab;
DBMS_OUTPUT.put_line('Dynamic EXECUTE: ' || l_tab.count);
END;
/
http://www.dba-oracle.com/plsql/t_plsql_dynamic.htm
declare
type tab_type is table of consumerfeedermetermapper%rowtype;
tab_t tab_type;
begin
select distinct(meterid) bulk collect into tab_t
from pseb.consumerfeedermetermapper
where feederid in (select distinct (fm.FeederID)
from pseb.feedermaster fm,pseb.consumerfeedermetermapper cfm
where fm.substationid=v_v_type
and cfm.feederid=fm.feederid
and cfm.FeederID>0)
and meterid >0
order by meterid;
end;
You can use the above code
I have a procedure that performs some calculations on all records returned by a cursor. It looks a bit like this:
PROCEDURE do_calc(id table.id_column%TYPE)
IS
CURSOR c IS
SELECT col1, col2, col3
FROM table
WHERE ...;
BEGIN
FOR r IN c LOOP
-- do some complicated calculations using r.col1, r.col2, r.col3 etc.
END LOOP;
END;
Now I have the case where I need to perform the exact same calculation on a different set of records that come from a different table. However, these have the same "shape" as in the above in example.
Is it possible to write a procedure that looks like this:
PROCEDURE do_calc2(c some_cursor_type)
IS
BEGIN
FOR r IN c LOOP
-- do the calc, knowing we have r.col1, r.col2, r.col3, etc.
END LOOP;
END;
I know about SYS_REFCURSOR, but I was wondering if it was possible to use the much more convenient FOR ... LOOP syntax and implicit record type.
Create a package.
Declare your cursor as package variable.
Use %rowtype to set function parameter type.
create or replace package test is
cursor c is select 1 as one, 2 as two from dual;
procedure test1;
function test2(test_record c%ROWTYPE) return number;
end test;
create or replace package body test is
procedure test1 is
begin
for r in c loop
dbms_output.put_line(test2(r));
end loop;
end;
function test2(test_record c%ROWTYPE) return number is
l_summ number;
begin
l_summ := test_record.one + test_record.two;
return l_summ;
end;
end test;
I had a similar problem, where I had two cursors that needed to be processed the same way, so this is how I figured it out.
DECLARE
--Define our own rowType
TYPE employeeRowType IS RECORD (
f_name VARCHAR2(30),
l_name VARCHAR2(30));
--Define our ref cursor type
--If we didn't need our own rowType, we could have this: RETURN employees%ROWTYPE
TYPE empcurtyp IS REF CURSOR RETURN employeeRowType;
--Processes the cursors
PROCEDURE process_emp_cv (emp_cv IN empcurtyp) IS
person employeeRowType;
BEGIN
LOOP
FETCH emp_cv INTO person;
EXIT WHEN emp_cv%NOTFOUND;
DBMS_OUTPUT.PUT_LINE('Name = ' || person.f_name ||
' ' || person.l_name);
END LOOP;
END;
--Defines the cursors
PROCEDURE mainProcedure IS
emp empcurtyp;
BEGIN
OPEN emp FOR SELECT first_name, last_name FROM employees WHERE salary > 50000;
process_emp_cv(emp);
CLOSE emp;
OPEN emp FOR SELECT first_name, last_name FROM kuren WHERE first_name LIKE 'J%';
process_emp_cv(emp);
CLOSE emp;
END;
BEGIN
mainProcedure;
END;
/
You can also use this if you want to bulk collect your cursors. You just need to change your helper procedure process_emp_cv; the rest can stay the same.
Using BULK COLLECT
--Processes the cursors
PROCEDURE process_emp_cv (emp_cv IN empcurtyp) IS
TYPE t_employeeRowTable IS TABLE OF employeeRowType;
employeeTable t_employeeRowTable;
BEGIN
LOOP
FETCH emp_cv BULK COLLECT INTO employeeTable LIMIT 50;
FOR indx IN 1 .. employeeTable.Count
LOOP
DBMS_OUTPUT.PUT_LINE('Name = ' || employeeTable(indx).f_name ||
' ' || employeeTable(indx).l_name);
END LOOP;
EXIT WHEN emp_cv%NOTFOUND;
END LOOP;
END;
Try this one, Usong ref cursor.
declare
type c is ref cursor;
c2 c;
type rec is record(
id number,
name varchar(20)
);
r rec;
procedure p1(c1 in out c,r1 in out rec)is begin
loop
fetch c1 into r1;
exit when c1%notfound;
dbms_output.put_line(r1.id || ' ' ||r1.name);
end loop;
end;
begin
open c2 for select id, name from student;
p1(c2,r);
end;
Yes you can use Cursor explicitly into procedure and function,for that cursor need to declare into package as variable