Can't seem to create a trigger for insert -Oracle - sql

Question

I am trying to create a trigger on a table for an insert. Here is the creation script:
CREATE OR REPLACE TRIGGER INTEG_QRY_NEW
AFTER INSERT
ON INTEG_LOG
FOR EACH ROW
DECLARE
VAR2 NUMBER(10);
LOG_TEXT1 VARCHAR2(1000);
RESULT1 VARCHAR2(1000);
QUERY_NUM1 VARCHAR2(1000);
QUERY_TIME1 VARCHAR2(1000);
LOG_INDEX1 NUMBER(10);
BEGIN
SELECT COUNT(*) INTO VAR2 FROM USER_TAB_COLS WHERE (COLUMN_NAME = 'RESULT' OR COLUMN_NAME = 'QUERY_NUM' OR COLUMN_NAME = 'DATE_TIME' ) AND TABLE_NAME = 'INTEG_LOG';
IF VAR2=0 THEN
EXECUTE IMMEDIATE 'ALTER TABLE INTEG_LOG ADD RESULT VARCHAR2(50); ALTER TABLE INTEG_LOG ADD DATE_TIME DATE; ALTER TABLE INTEG_LOG ADD QUERY_NUM VARCHAR2(50);';
END IF;
LOG_TEXT1 := :NEW.LOG_TEXT;
QUERY_TIME1 := :NEW.QUERY_TIME;
LOG_INDEX1 := :NEW.LOG_INDEX;
IF QUERY_TIME1 = '' THEN
RESULT1 := '-1';
QUERY_NUM1 := '';
UPDATE INTEG_LOG
SET RESULT = RESULT1,
DATE_TIME = CURRENT_DATE,
QUERY_NUM = QUERY_NUM1
WHERE LOG_INDEX = LOG_INDEX1;
ELSE
RESULT1 := SUBSTR(LOG_TEXT1,(INSTR(LOG_TEXT1,'Result = ')+9),9);
QUERY_NUM1 := SUBSTR(SUBSTR(LOG_TEXT1,INSTR(LOG_TEXT1,'Q# ')+3,20),1,INSTR(SUBSTR(LOG_TEXT1,INSTR(LOG_TEXT1,'Q# ')+3,20),' '));
UPDATE INTEG_LOG
SET RESULT = RESULT1,
DATE_TIME = CURRENT_DATE,
QUERY_NUM = QUERY_NUM1
WHERE LOG_INDEX = LOG_INDEX1;
END IF;
END;
/
The trigger is taking care of a regular insert for the following columns (that are coming from an application):
LOG_DATE NUMBER(10),
LOG_TIME NUMBER(10),
LOG_TYPE NUMBER(10),
LOG_TEXT VARCHAR2(2000 BYTE),
LOG_INDEX NUMBER(10),
QUERY_TIME VARCHAR2(8 BYTE)
the trigger checks for the following column to whether they exist or not (if not, it adds them):
RESULT VARCHAR2(50 BYTE),
DATE_TIME DATE,
QUERY_NUM VARCHAR2(50 BYTE)
Now it is taking the string (LOG_TEXT column), and withdraws the result value and query number (what comes after '#Q ').
Next, it should update the same insert opperation, and just adds these values to the new 3 columns (result,date_time and query_num).
The trigger does get created successfully but the problem is, when i am trying to insert into the table (INTEG_LOG), I get the following error:
table string.string is mutating, trigger/function may not see it
Cause: A trigger (or a user defined plsql function that is referenced
in this statement) attempted to look at (or modify) a table that was
in the middle of being modified by the statement which fired it.
Action: Rewrite the trigger (or function) so it does not read that
table.
Might this be a syntax error ? Please help. Thank you.

Answer 1

The only reason the trigger compiles is because INTEG_LOG already has the columns you're dynamic SQL is trying to add. If the table didn't have those columns the trigger wouldn't add them because the trigger would be invalid as those UPDATE statements wouldn't compile.
One of the reasons why dynamic SQL should be avoid is that it turns compilation errors into runtime errors.
But trying to add columns to a table in a trigger built on that table is an astonishingly bad idea. Apart from anything else, DDL statement issue implicit commits and we cannot commit in triggers because that would play havoc with the transactions. (yes there is pragma autonomous_transaction but using that is usually a code smell).
The correct solution is:
Write a one-off DDL script to add those columns
If necessary make the script idempotent by checking for the prior existence of those columns before executing the ALTER TABLE statements
Run the script.
Populate the table with all the columns.
Of course this has nothing to do with the mutating table error which is due to those UPDATE statements. We cannot execute DML (including selects) on the table which hosts the trigger. I'm not sure what business rule you are trying to implement but you need a different way of doing it.