PL/SQL compile conditionally on existence of database object - oracle

Is it possible to have conditional compilation in Oracle, where the condition is the existence of a database object (specifically, a table or view or synonym)? I'd like to be able to do something like this:
sp_some_procedure is
$IF /*check if A exists.*/ then
/* read from and write to A as well as other A-related non-DML stuff...*/
$ELSE /*A doesn't exist yet, so avoid compiler errors*/
dbms_output.put_line('Reminder: ask DBA to create A!')
$ENDIF
end;

Yes it is. Here a sample where the first stored procedure wants to select from XALL_TABLES, but if this table doesn't exist, select from dual. Finally, because I haven't got an XALL_TABLES object, the first stored procedure selects from dual. The second one, does the same thing on the ALL_TABLES object. Because the ALL_TABLES exists, the second stored procedure selects from all_tables but not from DUAL.
This kind of construction is useful where the package have to be deployed on all your database and use tables that are not deployed everywhere ... (ok, perhaps there is a conceptual problem, but it happens).
--conditionals compilation instructions accept only static condition (just with constants)
--passing sql bind variable doesn't work
--To pass a value to a conditional compilation instruction, I bypasses the use of input parameters of the script
--these 4 next lines affect a value to the first and the second input parameter of the script
--If your originally script use input script parameter, use the next free parameter ...
column param_1 new_value 1 noprint
select nvl(max(1), 0) param_1 from all_views where owner = 'SYS' and view_name = 'XALL_TABLES';
column param_2 new_value 2 noprint
select nvl(max(1), 0) param_2 from all_views where owner = 'SYS' and view_name = 'ALL_TABLES';
CREATE or replace PACKAGE my_pkg AS
function test_xall_tables return varchar2;
function test_all_tables return varchar2;
END my_pkg;
/
CREATE or replace PACKAGE BODY my_pkg AS
function test_xall_tables return varchar2 is
vch varchar2(50);
begin
$IF (&1 = 0) $THEN
select 'VIEW XALL_TABLES D''ONT EXISTS' into vch from dual;
$ELSE
select max('VIEW XALL_TABLES EXISTS') into vch from XALL_TABLES;
$END
return vch;
end test_xall_tables;
function test_all_tables return varchar2 is
vch varchar2(50);
begin
$IF (&2 = 0) $THEN
select 'VIEW ALL_TABLES D''ONT EXISTS' into vch from dual;
$ELSE
select max('VIEW ALL_TABLES EXISTS') into vch from ALL_TABLES;
$END
return vch;
end test_all_tables;
END my_pkg;
/
the test :
select my_pkg.test_xall_tables from dual;
give
VIEW XALL_TABLES D'ONT EXISTS
select my_pkg.test_all_tables from dual;
give
VIEW ALL_TABLES EXISTS

I would use 'EXECUTE IMMEDIATE' and a EXCEPTION clause.

Use dynamic SQL to create package constants to track which objects exist, and then use those constants in conditional compilation.
--E.g., say there are two possible tables, but only one of them exists.
--create table table1(a number);
create table table2(a number);
--Create a package with boolean constants to track the objects.
--(Another way to do this is to use ALTER SESSION SET PLSQL_CCFLAGS)
declare
table1_exists_string varchar2(10) := 'true';
table2_exists_string varchar2(10) := 'true';
temp number;
begin
begin
execute immediate 'select max(1) from table1 where rownum <= 1' into temp;
exception when others then
table1_exists_string := 'false';
end;
begin
execute immediate 'select max(1) from table2 where rownum <= 1' into temp;
exception when others then
table2_exists_string := 'false';
end;
execute immediate '
create or replace package objects is
table1_exists constant boolean := '||table1_exists_string||';
table2_exists constant boolean := '||table2_exists_string||';
end;
';
end;
/
--Look at the results in the source:
select * from user_source where name = 'OBJECTS';
--Create the object that refers to the tables.
create or replace function compile_test return varchar2 is
v_test number;
begin
$if objects.table1_exists $then
select max(1) into v_test from table1;
return 'table1 exists';
$elsif objects.table2_exists $then
select max(1) into v_test from table2;
return 'table 2 exists';
$else
return 'neither table exists';
$end
end;
/
--Check the dependencies - only TABLE2 is dependent.
select * from user_dependencies where name = 'COMPILE_TEST';
--Returns 'table 2 exists'.
select compile_test from dual;
Mixing dynamic SQL, dynamic PL/SQL, and conditional compilation is usually a very evil idea. But it will allow you to put all of your ugly dynamic SQL in one installation package, and maintain real dependency tracking.
This may work well in a semi-dynamic environment; for example a program that is installed with different sets of objects but does not frequently change between them.
(Also, if the whole point of this is just to replace scary error messages with friendly warnings, in my opinion that is a very bad idea. If your system is going to fail, the failure should be obvious so it can be immediately fixed. Most people ignore anything that starts with "Reminder...".)

No - that is not possible... but if you create a stored procedure referencing a non-existent DB object and try to compile it the compilation will show errors... the stored procedure will be there but "invalid"... and the compilation errors are accessible for the DBA whenever he looks at it... so I would just go ahead and create all needed stored procedures, if any compilation errors arise ask the DBA (sometimes the object exists but the stored procedure need permissions to access it...)... after the reason for the error(s) is fixed you can just recompile the stored procedure (via ALTER PROCEDURE MySchema.MyProcName COMPILE;) and all is fine...
IF you don't want code to be there you can just DROP the strored procedure and/or replace is via CREATE OR REPLACE... with dbms_output.put_line('Reminder: ask DBA to create A!') in the body.
The only other alternative is as kevin points out EXECUTE IMMEDIATE with proper EXCEPTION handling...

What I would do is check the existence via all_objects, something like:
declare
l_check_sql varchar2(4000);
l_cnt number;
begin
l_check_sql := q'{
select count(1)
from all_objects
where object_name = 'MY_OBJ'
and owner = 'MY_OWNER'
}';
execute immediate l_check_sql into l_cnt;
if (l_cnt > 0) then
-- do something referring to MY_OBJ
else
-- don't refer to MY_OBJ
end if;
end;

Related

Create record from refcursor

I would like to create record from refcursor. My code:
set serveroutput on
DECLARE
c_curs SYS_REFCURSOR;
v_id NUMBER;
BEGIN
pck_prov.get_value_type_list (1, c_curs); --> procedure called here
-- I guess this is the place where record can be created from cursor.
LOOP
FETCH c_curs
INTO v_id;--instead of fetching into variable I would like to fetch into row
EXIT WHEN c_curs%NOTFOUND;
DBMS_OUTPUT.PUT_LINE(v_id);--if fetching is done into row, only selected columns can be printed, like myrow.id
END LOOP;
CLOSE c_curs;
END;
Please note: I know how to create record from cursor which is defined with select statement as it is described here. What I don't know is how to use same technique for refcursors.
EDIT:
Code from here is just what I need, but it throws error:
set serveroutput on
VAR c_curs refcursor;
EXECUTE pck_prov.get_value_type_list(1, :c_curs);
BEGIN
FOR record_test IN :c_curs LOOP
dbms_output.put_line(record_test.id);
END LOOP;
END;
Error: error PLS-00456: item 'SQLDEVBIND1Z_1' is not a cursor.
Just to clarify question:
In my database there is around 200 packages.
Every package has several stored procedures inside - and usually each procedure is combined with columns from different tables. That is why it would be the best to have some dynamically created cursor, so I can make simple select just like in the example I've posted.
From Oracle 12.1 onward, you could use the DBMS_SQL.return_result procedure. SQL Plus displays the contents of implicit statement results automatically. So, rather than defining explicit ref cursor out parameters, the RETURN_RESULT procedure in the DBMS_SQL package allows you to pass them out implicitly.
DECLARE
c_curs SYS_REFCURSOR;
v_id NUMBER;
BEGIN
pck_prov.get_value_type_list (1, c_curs);
DBMS_SQL.return_result(c_curs); --> Passes ref cursor output implicitly
END;
/
In fact, no need of this separate PL/SQL block, you could add the DBMS_SQL.return_result(c_curs) statement in your original pck_prov.get_value_type_list procedure itself.
Just define a PL/SQL RECORD type that matches the cursor and FETCH into it.
DECLARE
c_curs SYS_REFCURSOR;
TYPE rec_t IS RECORD ( object_name VARCHAR2(30), object_type VARCHAR2(30) );
v_rec rec_t;
BEGIN
OPEN c_curs FOR
SELECT object_name,
object_type
FROM dba_objects
WHERE object_name like 'DBA%TAB%';
LOOP
FETCH c_curs
INTO v_rec;
EXIT WHEN c_curs%NOTFOUND;
DBMS_OUTPUT.PUT_LINE(v_rec.object_name || ' - ' || v_rec.object_type);
END LOOP;
CLOSE c_curs;
END;
DBA_ADVISOR_SQLA_TABLES - VIEW
DBA_ADVISOR_SQLA_TABVOL - VIEW
DBA_ADVISOR_SQLW_TABLES - VIEW
DBA_ADVISOR_SQLW_TABVOL - VIEW
DBA_ALL_TABLES - VIEW
etc...

Check Package exist or not before create it in Oracle

Need help
How do I check for a package if it exists or not and skip the creation if it already exists.
I have done the following and I got an error
DECLARE
l_cnt INTEGER;
own VARCHAR(200);
BEGIN
SELECT sys_context( 'userenv', 'current_schema' ) INTO own FROM dual;
SELECT count(*)
INTO l_cnt
FROM ALL_OBJECTS
WHERE object_type = 'PACKAGE'
and object_name = 'JUSTICE_LEAGUE'
AND OWNER = own;
IF( l_cnt <= 0) THEN
EXECUTE IMMEDIATE
'create or replace PACKAGE "JUSTICE_LEAGUE" AS
FUNCTION BATMAN(argPSN INT)
RETURN INT;
FUNCTION SUPERMAN(argSN int)
RETURN Info.infovalue%Type;
PROCEDURE AQUAMAN(argASN INT,argAssignedUser folderProcess.assignedUser%Type DEFAULT 'None');
END JUSTICE_LEAGUE';
DBMS_OUTPUT.PUT_LINE('Package created successfully');
ELSE
DBMS_OUTPUT.PUT_LINE('Package exists already');
END IF;
END;
/
Error report -
ORA-06550: line 23, column 70:
PLS-00103: Encountered the symbol "ALL" when expecting one of the following:
Is it a right way to put the create command for package within EXECUTE IMMEDIATE ?
First of all - no, it is wrong way to do that. The fact that dynamic SQL exists doesn't mean that you should use it, especially not for creating packages (or any other objects). There are really rare situations you'd want to do that.
PL/SQL procedures (functions, packages, triggers) offer create OR REPLACE option so - it is safe to run that statement as is (not as dynamic SQL). It means that:
if it doesn't exist, it'll be created
if it exists, it'll be overwritten by code you've just ran
If you insist on dynamic SQL, check its (package's) existence by querying user_objects:
SQL> select count(*)
2 from user_objects
3 where object_name = 'JUSTICE_LEAGUE'
4 and object_type = 'PACKAGE';
COUNT(*)
----------
0
SQL>
Depending on result, run execute immediate (or not).
There is only one issue with your code.
You have not handled the dynamic query properly. single-quote in the dynamic query must be escaped.
Two single quotes('') in the string are considered as one single quote(') in the final string.
Or you can use the quoted-string (q'{<your_string>}')
replace -- DEFAULT 'None'); with DEFAULT ''None'');

trying to add an agent (not allowing duplicate last names). Initial total sales should be zero

I am receiving the following error when trying to compile and execute. I am having issues how to figure this out.
14/7 PL/SQL: Statement ignored
14/10 PLS-00204: function or pseudo-column 'EXISTS' may be used inside a SQL statement only
Errors: check compiler log
CREATE OR REPLACE PROCEDURE AddAgent(
p_Agent_Fname IN Agent.Agent_Fname%TYPE,
p_Agent_Lname IN Agent.Agent_Lname%TYPE,
p_Agent_Address IN Agent.Agent_Address%TYPE,
p_Agent_Tsales IN Agent.Agent_Tsales%TYPE,
p_Agent_Salary IN Agent.Agent_Salary%TYPE)
IS
p_ErrorCode number; --USED FOR ERROR CHECKING
p_ErrorMsg Varchar2(200);
p_CurrentUser Varchar2(100);
BEGIN
IF EXISTS
(SELECT * FROM Agent WHERE Agent_Lname = p_Agent_Lname) THEN
dbms_output.put_line('Failure');
ELSE
INSERT INTO Agent (Agent_Fname, Agent_Lname, Agent_Address, Agent_Tsales, Agent_Salary)
SELECT p_Agent_Fname, p_Agent_Lname, p_Agent_Address, 0, p_Agent_Salary
from Dual;
COMMIT;
dbms_output.put_line('Success');
END IF;
END;
As you were told, you can't use EXISTS out of the SELECT statement. Therefore, you'll have to check for existence in the AGENT table elsewhere.
Here's one option you might consider. Note that I've also rewritten the INSERT INTO statement - there's no need to SELECT FROM DUAL, you already have all those values.
CREATE OR REPLACE PROCEDURE AddAgent(
p_Agent_Fname IN Agent.Agent_Fname%TYPE,
p_Agent_Lname IN Agent.Agent_Lname%TYPE,
p_Agent_Address IN Agent.Agent_Address%TYPE,
p_Agent_Tsales IN Agent.Agent_Tsales%TYPE,
p_Agent_Salary IN Agent.Agent_Salary%TYPE)
IS
p_ErrorCode number; --USED FOR ERROR CHECKING
p_ErrorMsg Varchar2(200);
p_CurrentUser Varchar2(100);
l_cnt number; --> newly added
BEGIN
-- check whether something exists in a table
select count(*)
into l_cnt
from dual
where exists (select null
from agent
where agent_lname = p_agent_lname
);
IF l_cnt > 0 then
dbms_output.put_line('Failure');
ELSE
INSERT INTO Agent
(Agent_Fname, Agent_Lname, Agent_Address, Agent_Tsales, Agent_Salary)
VALUES
(p_Agent_Fname, p_Agent_Lname, p_Agent_Address, 0, p_Agent_Salary);
COMMIT;
dbms_output.put_line('Success');
END IF;
END;
/

FORALL+ EXECUTE IMMEDIATE + INSERT Into tbl SELECT

I have got stuck in below and getting syntax error - Please help.
Basically I am using a collection to store few department ids and then would like to use these department ids as a filter condition while inserting data into emp table in FORALL statement.
Below is sample code:
while compiling this code i am getting error, my requirement is to use INSERT INTO table select * from table and cannot avoid it so please suggest.
create or replace Procedure abc(dblink VARCHAR2)
CURSOR dept_id is select dept_ids from dept;
TYPE nt_dept_detail IS TABLE OF VARCHAR2(25);
l_dept_array nt_dept_detail;
Begin
OPEN dept_id;
FETCH dept_id BULK COLLECT INTO l_dept_array;
IF l_dept_array.COUNT() > 0 THEN
FORALL i IN 1..l_dept_array.COUNT SAVE EXCEPTIONS
EXECUTE IMMEDIATE 'INSERT INTO stg_emp SELECT
Dept,''DEPT_10'' FROM dept_emp'||dblink||' WHERE
dept_id = '||l_dept_array(i)||'';
COMMIT;
END IF;
CLOSE dept_id;
end abc;
Why are you bothering to use cursors, arrays etc in the first place? Why can't you just do a simple insert as select?
Problems with your procedure as listed above:
You don't declare procedures like Procedure abc () - for a standalone procedure, you would do create or replace procedure abc as, or in a package: procedure abc is
You reference a variable called "dblink" that isn't declared anywhere.
You didn't put end abc; at the end of your procedure (I hope that was just a mis-c&p?)
You're effectively doing a simple insert as select, but you're way over-complicating it, plus you're making your code less performant.
You've not listed the column names that you're trying to insert into; if stg_emp has more than two columns or ends up having columns added, your code is going to fail.
Assuming your dblink name isn't known until runtime, then here's something that would do what you're after:
create Procedure abc (dblink in varchar2)
is
begin
execute immediate 'insert into stg_emp select dept, ''DEPT_10'' from dept_emp#'||dblink||
' where dept_id in (select dept_ids from dept)';
commit;
end abc;
/
If, however, you do know the dblink name, then you'd just get rid of the execute immediate and do:
create Procedure abc (dblink in varchar2)
is
begin
insert into stg_emp -- best to list the column names you're inserting into here
select dept, 'DEPT_10'
from dept_emp#dblink
where dept_id in (select dept_ids from dept);
commit;
end abc;
/
There appears te be a lot wrong with this code.
1) why the execute immediate? Is there any explicit requirement for that? No, than don't use it
2) where is the dblink variable declared?
3) as Boneist already stated, why not a simple subselect in the insert statement?
INSERT INTO stg_emp SELECT
Dept,'DEPT_10' FROM dept_emp#dblink WHERE
dept_id in (select dept_ids from dept );
For one, it would make the code actually readable ;)

PLSQL - Drop all database objects of a user

I'm trying to use a procedure (no parameters) to drop all of the user-created database objects located within the schema from where the procedure is launched, but I'm really not sure on how to go about this. Here's what I have so far, but I think I'm going about this the wrong way.
create or replace procedure CLEAN_SCHEMA is
cursor schema_cur is
select 'drop '||object_type||' '|| object_name|| DECODE(OBJECT_TYPE,'TABLE',' CASCADE CONSTRAINTS;',';')
from user_objects;
schema_rec schema_cur%rowtype;
begin
select 'drop '||object_type||' '|| object_name|| DECODE(OBJECT_TYPE,'TABLE',' CASCADE CONSTRAINTS;',';')
into schema_rec
from user_objects;
end;
/
create or replace
FUNCTION DROP_ALL_SCHEMA_OBJECTS RETURN NUMBER AS
PRAGMA AUTONOMOUS_TRANSACTION;
cursor c_get_objects is
select object_type,'"'||object_name||'"'||decode(object_type,'TABLE' ,' cascade constraints',null) obj_name
from user_objects
where object_type in ('TABLE','VIEW','PACKAGE','SEQUENCE','SYNONYM', 'MATERIALIZED VIEW')
order by object_type;
cursor c_get_objects_type is
select object_type, '"'||object_name||'"' obj_name
from user_objects
where object_type in ('TYPE');
BEGIN
begin
for object_rec in c_get_objects loop
execute immediate ('drop '||object_rec.object_type||' ' ||object_rec.obj_name);
end loop;
for object_rec in c_get_objects_type loop
begin
execute immediate ('drop '||object_rec.object_type||' ' ||object_rec.obj_name);
end;
end loop;
end;
RETURN 0;
END DROP_ALL_SCHEMA_OBJECTS;
Create the above function (autonomous so DDL can be called via a function)
then you can just:
select DROP_ALL_SCHEMA_OBJECTS from dual;
when you want to drop all your objects, make sure you dont try to drop the proc your running (i dont care about the procs thats why i dont have procs or functions in the object_type list)
if you want to drop everything you need an anonymous block
but i needed to be able to do this from a tool that only allowed ansi sql (not plsql) hence a stored proc.
Enjoy.
declare
cursor ix is
select *
from user_objects
where object_type in ('TABLE', 'VIEW', 'FUNCTION', 'SEQUENCE');
begin
for x in ix loop
execute immediate('drop '||x.object_type||' '||x.object_name);
end loop;
end;
Unless the user has hard to reapply permissions, its probably easier to just drop the user and recreate them.
Thanks Martin Brambley,
I feel we can simplify your answer in the following way.
CREATE OR REPLACE
procedure DROP_ALL_SCHEMA_OBJECTS AS
PRAGMA AUTONOMOUS_TRANSACTION;
cursor c_get_objects is
select object_type,'"'||object_name||'"'||decode(object_type,'TABLE' ,' cascade constraints',null) obj_name
FROM USER_OBJECTS
where object_type in ('TABLE','VIEW','PACKAGE','SEQUENCE','SYNONYM', 'MATERIALIZED VIEW', 'TYPE')
order by object_type;
BEGIN
begin
for object_rec in c_get_objects loop
execute immediate ('drop '||object_rec.object_type||' ' ||object_rec.obj_name);
end loop;
end;
END DROP_ALL_SCHEMA_OBJECTS;
/
execute DROP_ALL_SCHEMA_OBJECTS;
What you've got is a good start.
Here is the rest:
You have a cursor AND a select statement. You only need the cursor.
Your next step is to call the drop statement using dynamic PLSQL. I'd use the EXECUTE IMMEDIATE statement. Its more elegant and preformance friendly to just select the name of the thing you're dropping and submit it as a bind variable to EXECUTE IMMEDIATE.
In order to drop the objects of the schema calling the method and not the schema owning the method you have to use "AUTHID CURRENT_USER". See the Oracle documentation for more info.
Other things to drop: packages, functions, procedures (the system will likely hang then timeout if you try to drop this method while its running), Java classes, triggers, views, types
Lastly, this is obviously a very dangerous method so you may want to consider putting it in a script instead of a stored procedure so it isn't left in the database for anyone to run.
You're close - as someone else has noted you need an "EXECUTE IMMEDIATE" for the statement. You should consider:
Instead of creating a procedure to do this, run this as an anonymous PL/SQL block so you don't have the issue of trying to drop a procedure that is running.
Add a test for object type of TABLE and for that case modify the drop statement to include the cascade option to handle tables that are "parents" of other tables via foreign key constraints. Remember that you'll probably be generating the cursor list in an order that doesn't consider dependencies that will block the drop.
Also on the subject of dependencies, it is probably best to drop tables first (add a DECODE in your cursor that assigns a lower numeric value to this object type and order the cursor select by this value). If you have Oracle objects of type TYPE that are used as column types in a table definition the table must be dropped first.
If you use Oracle Advanced Queuing the objects related to this MUST be dropped with the AQ package API calls. Although you can drop the Oracle-generated tables for queue support with a regular DROP TABLE, you will find yourself in the catch-22 position of then not being able to drop the related queues nor add them back. Up to version 10g at least you couldn't even drop the containing schema without putting the database in a special mode when this situation existed
Thanks Martin Brambley and Vijayan Srinivasan!
But Vijayan Srinivasan's version is not correct, because dependent objects of type 'TYPE' sometime generates errors during drop them:
ORA-02303: cannot drop or replace a type with type or table dependents
My version drop ALL objects from Schema with additional:
drop procedures and functions (expect 'DROP_ALL_SCHEMA_OBJECTS')
drop all jobs and dbms_jobs
drop all db_links
do not drop nested tables, because DROPing of nested tables not supported
CREATE OR REPLACE
procedure DROP_ALL_SCHEMA_OBJECTS AS
PRAGMA AUTONOMOUS_TRANSACTION;
cursor c_get_objects is
select uo.object_type object_type_2,'"'||uo.object_name||'"'||decode(uo.object_type,'TABLE' ,' cascade constraints',null) obj_name2
FROM USER_OBJECTS uo
where uo.object_type in ('TABLE','VIEW','PACKAGE','SEQUENCE','SYNONYM', 'MATERIALIZED VIEW', 'FUNCTION', 'PROCEDURE')
and not (uo.object_type = 'TABLE' and exists (select 1 from user_nested_tables unt where uo.object_name = unt.table_name))
and not (uo.object_type = 'PROCEDURE' and uo.object_name = 'DROP_ALL_SCHEMA_OBJECTS')
order by uo.object_type;
cursor c_get_objects_type is
select object_type, '"'||object_name||'"' obj_name
from user_objects
where object_type in ('TYPE');
cursor c_get_dblinks is
select '"'||db_link||'"' obj_name
from user_db_links;
cursor c_get_jobs is
select '"'||object_name||'"' obj_name
from user_objects
where object_type = 'JOB';
cursor c_get_dbms_jobs is
select job obj_number_id
from user_jobs
where schema_user != 'SYSMAN';
BEGIN
begin
for object_rec in c_get_objects loop
execute immediate ('drop '||object_rec.object_type_2||' ' ||object_rec.obj_name2);
end loop;
for object_rec in c_get_objects_type loop
begin
execute immediate ('drop '||object_rec.object_type||' ' ||object_rec.obj_name);
end;
end loop;
for object_rec in c_get_dblinks loop
execute immediate ('drop database link '||object_rec.obj_name);
end loop;
for object_rec in c_get_jobs loop
DBMS_SCHEDULER.DROP_JOB(job_name => object_rec.obj_name);
end loop;
commit;
for object_rec in c_get_dbms_jobs loop
dbms_job.remove(object_rec.obj_number_id);
end loop;
commit;
end;
END DROP_ALL_SCHEMA_OBJECTS;
/
execute DROP_ALL_SCHEMA_OBJECTS;
drop procedure DROP_ALL_SCHEMA_OBJECTS;
exit;

Resources