Create a DML trigger

This topic describes how to create a trigger.

Prerequisites

To create a trigger, you must have the following privileges:

• Privileges on the table associated with the trigger, such as SELECT, INSERT, UPDATE, and DELETE privileges

• CREATE TRIGGER privilege

• Privileges on statements to be executed after the trigger is activated

Syntax

You can use the CREATE statement to create a trigger.

The syntax is as follows:

CREATE
    TRIGGER trigger_name
    trigger_time trigger_event
    ON tbl_name FOR EACH ROW
    [trigger_order]
    trigger_body;

trigger_time: { BEFORE | AFTER }

trigger_event: { INSERT | UPDATE | DELETE }

trigger_order: { FOLLOWS | PRECEDES } other_trigger_name

where

• trigger_name specifies the name of the trigger. The name must be unique.

• tbl_name specifies the name of the table for which the trigger is to be created.

• BEFORE | AFTER specifies whether the trigger is activated before or after the triggering event. For example, whether the trigger is activated before or after each row is inserted into the associated table.

• INSERT | UPDATE | DELETE specifies the triggering event.

• FOR EACH ROW specifies that each time when the trigger is activated, the statements defined in the trigger are executed on each row affected by the event that causes the trigger to activate.

• FOLLOWS | PRECEDES specifies the order of triggers. OceanBase Database allows you to associate multiple triggers of the same triggering event and activating time to the same table. By default, triggers with the same triggering event and activating time are activated in the order of their creation time. You can use FOLLOWS and PRECEDES to specify the order of the triggers. If you specify FOLLOWS, a new trigger is activated after the current trigger. If you specify PRECEDES, a new trigger is activated before the current trigger.

OceanBase Database also supports NEW.columnName and OLD.columnName.

• For an INSERT trigger, NEW.columnName specifies new data that is inserted in a BEFORE scenario or was inserted in an AFTER scenario.

  Here, columnName indicates a column name in the corresponding data table.

• For an UPDATE trigger, OLD.columnName specifies existing data that is updated. NEW.columnName specifies new data after the update.

• For a DELETE trigger, OLD.columnName specifies existing data that is deleted.

• Values in OLD.columnName are read-only, whereas values in NEW.columnName can be specified by using SET statements.

In addition, if you want to create a trigger that contains multiple statements, you can use the BEGIN … END statement to specify the start and end of the entire code block.

The BEGIN … END syntax is as follows:

BEGIN
[statement_list]
END

In the preceding syntax, statement_list specifies a list of one or more statements to be executed. Each statement in the list must end with a semicolon (;). A semicolon (;) indicates the end of an SQL statement. When the system detects a semicolon (;), the system determines that a statement ends, and then starts to execute the statement. In this case, the interpreter fails to find the END keyword that matches the BEGIN keyword during SQL execution. Therefore, an error is reported. To prevent the error, you can use a DELIMITER command to change the delimiter of a statement.

Here is a sample DELIMITER command:

DELIMITER new_delimiter

Here, new_delimiter specifies the delimiter of a statement. A delimiter can be a sign of one or more bytes in length. The default delimiter is a semicolon (;). You can change the semicolon (;) to another delimiter, such as a number sign (#).

After the DELIMITER command is added to the syntax, statements with semicolons (;) following the command can be executed without an error reported. This is because the system does not consider that a statement ends until it detects the specified delimiter, such as a number sign (#).

Examples

• Create a trigger named test_trg, associate it with the test table, and activate the Insert trigger. The trigger serves as an accumulator to calculate the sum of the values of columns inserted into the table.

  obclient>CREATE TABLE test (user_id INT, user_num DECIMAL(10,2));
  Query OK, 0 rows affected
  
  obclient> CREATE TRIGGER test_trg BEFORE INSERT ON test
        FOR EACH ROW SET @sum = @sum + NEW.user_num;
  Query OK, 0 rows affected
  

• Create a trigger with multiple statements.

  obclient>CREATE TABLE test (user_id INT, user_num DECIMAL(10,2));
  Query OK, 0 rows affected
  
  obclient>DELIMITER #
  
  obclient>CREATE TRIGGER test_trg BEFORE UPDATE ON test
                FOR EACH ROW
                BEGIN
                IF NEW.user_num < 1 THEN
                SET NEW.user_num  = 1;
                ELSEIF NEW.user_num > 45 THEN
                SET NEW.user_num= 45;
              END IF;
              END;#
  Query OK, 0 rows affected
  
  obclient>DELIMITER;
  

Limitations

Triggers in MySQL mode have the following limitations:

• You can create a trigger only for a permanent table, but not for a temporary table.

• Triggers cannot use the CALL statement to return data to clients or use the stored procedures of dynamic SQL statements. However, stored procedures and functions can use OUT or INOUT to return data to triggers.

• Triggers cannot use clauses to start or end transactions, such as START TRANSACTION for starting a transaction, COMMIT for committing a transaction, and ROLLBACK for rolling back a transaction. However, triggers can roll back a transaction to a specific savepoint because this operation does not end the transaction.

• A foreign key action cannot activate a trigger.

• A trigger cannot return a value. Make sure that the trigger does not contain a statement that is used to return values. If you want to instantly stop a trigger, use a LEAVE statement.