A data change trigger is declared as a function with no arguments and a return type of trigger. Note that the function must be declared with no arguments even if it expects to receive some arguments specified in CREATE TRIGGER — such arguments are passed via TG_ARGV, as described below.

When a PL/pgSQL function is called as a trigger, several special variables are created automatically in the top-level block. They are:

A trigger function must return either NULL or a record/row value having exactly the structure of the table the trigger was fired for.

Row-level triggers fired BEFORE can return null to signal the trigger manager to skip the rest of the operation for this row (i.e., subsequent triggers are not fired, and the INSERT/UPDATE/DELETE does not occur for this row). If a nonnull value is returned then the operation proceeds with that row value. Returning a row value different from the original value of NEW alters the row that will be inserted or updated. Thus, if the trigger function wants the triggering action to succeed normally without altering the row value, NEW (or a value equal thereto) has to be returned. To alter the row to be stored, it is possible to replace single values directly in NEW and return the modified NEW, or to build a complete new record/row to return. In the case of a before-trigger on DELETE, the returned value has no direct effect, but it has to be nonnull to allow the trigger action to proceed. Note that NEW is null in DELETE triggers, so returning that is usually not sensible. The usual idiom in DELETE triggers is to return OLD.

INSTEAD OF triggers (which are always row-level triggers, and may only be used on views) can return null to signal that they did not perform any updates, and that the rest of the operation for this row should be skipped (i.e., subsequent triggers are not fired, and the row is not counted in the rows-affected status for the surrounding INSERT/UPDATE/DELETE). Otherwise a nonnull value should be returned, to signal that the trigger performed the requested operation. For INSERT and UPDATE operations, the return value should be NEW, which the trigger function may modify to support INSERT RETURNING and UPDATE RETURNING (this will also affect the row value passed to any subsequent triggers, or passed to a special EXCLUDED alias reference within an INSERT statement with an ON CONFLICT DO UPDATE clause). For DELETE operations, the return value should be OLD.

The return value of a row-level trigger fired AFTER or a statement-level trigger fired BEFORE or AFTER is always ignored; it might as well be null. However, any of these types of triggers might still abort the entire operation by raising an error.

Example 41.3 shows an example of a trigger function in PL/pgSQL.

CREATE TABLE emp (
    empname           text,
    salary            integer,
    last_date         timestamp,
    last_user         text
);

CREATE FUNCTION emp_stamp() RETURNS trigger AS $emp_stamp$
    BEGIN
        -- Check that empname and salary are given
        IF NEW.empname IS NULL THEN
            RAISE EXCEPTION 'empname cannot be null';
        END IF;
        IF NEW.salary IS NULL THEN
            RAISE EXCEPTION '% cannot have null salary', NEW.empname;
        END IF;

        -- Who works for us when they must pay for it?
        IF NEW.salary < 0 THEN
            RAISE EXCEPTION '% cannot have a negative salary', NEW.empname;
        END IF;

        -- Remember who changed the payroll when
        NEW.last_date := current_timestamp;
        NEW.last_user := current_user;
        RETURN NEW;
    END;
$emp_stamp$ LANGUAGE plpgsql;

CREATE TRIGGER emp_stamp BEFORE INSERT OR UPDATE ON emp
    FOR EACH ROW EXECUTE FUNCTION emp_stamp();

Another way to log changes to a table involves creating a new table that holds a row for each insert, update, or delete that occurs. This approach can be thought of as auditing changes to a table. Example 41.4 shows an example of an audit trigger function in PL/pgSQL.

CREATE TABLE emp (
    empname           text NOT NULL,
    salary            integer
);

CREATE TABLE emp_audit(
    operation         char(1)   NOT NULL,
    stamp             timestamp NOT NULL,
    userid            text      NOT NULL,
    empname           text      NOT NULL,
    salary            integer
);

CREATE OR REPLACE FUNCTION process_emp_audit() RETURNS TRIGGER AS $emp_audit$
    BEGIN
        --
        -- Create a row in emp_audit to reflect the operation performed on emp,
        -- making use of the special variable TG_OP to work out the operation.
        --
        IF (TG_OP = 'DELETE') THEN
            INSERT INTO emp_audit SELECT 'D', now(), current_user, OLD.*;
        ELSIF (TG_OP = 'UPDATE') THEN
            INSERT INTO emp_audit SELECT 'U', now(), current_user, NEW.*;
        ELSIF (TG_OP = 'INSERT') THEN
            INSERT INTO emp_audit SELECT 'I', now(), current_user, NEW.*;
        END IF;
        RETURN NULL; -- result is ignored since this is an AFTER trigger
    END;
$emp_audit$ LANGUAGE plpgsql;

CREATE TRIGGER emp_audit
AFTER INSERT OR UPDATE OR DELETE ON emp
    FOR EACH ROW EXECUTE FUNCTION process_emp_audit();

A variation of the previous example uses a view joining the main table to the audit table, to show when each entry was last modified. This approach still records the full audit trail of changes to the table, but also presents a simplified view of the audit trail, showing just the last modified timestamp derived from the audit trail for each entry.