Question

I’m new to PostgreSQL. I have tables like:

CREATE TABLE Person (
  ID SERIAL PRIMARY KEY,
  Name VARCHAR(32) NOT NULL DEFAULT '',
  Surname VARCHAR(32) NOT NULL DEFAULT '',
  Birthday DATE,
  Gender VARCHAR(8)
);

-- Student table inherits from person
CREATE TABLE Student (
  ID_Student SERIAL PRIMARY KEY,
  MajorDept VARCHAR(32),
) INHERITS(Person);

-- Student table inherits from person
CREATE TABLE Employee (
  ID_Employee SERIAL PRIMARY KEY,
  Position VARCHAR(32),
  Rank VARCHAR(32),
  Salary NUMERIC(12,2)
) INHERITS(Person);

-- Address table references person
CREATE TABLE Address (
  ID_Address SERIAL PRIMARY KEY,
  Person_id INTEGER REFERENCES Person(ID) NOT NULL,
  Email VARCHAR(32) UNIQUE,
  Country VARCHAR(32),
  CityCode INTEGER,
  City VARCHAR(32),
  AddressLine VARCHAR(60),
);

According to these tables, when I want to INSERT data into Adress table, Postgres gives that error:

ERROR: insert or update on table “address” violates foreign key
constraint “address_person_id_fkey” DETAIL: Key (person_id)=(1) is
not present in table “person”.

I’ve learned that in Postgres

indexes (including unique constraints) and foreign key constraints
only apply to single tables, not to their inheritance children.

My question is how can I fix this with using triggers? Sample code would be very useful.

After inserting a few rows to child tables, I can see the data with ‘SELECT * FROM Person;’ as well. It looks like:

Person Table

1;"Bill";"Smith";"1985-05-10";"male"
2;"Jenny";"Brown";"1986-08-12";"female"
3;"Bob";"Morgan";"1986-06-11";"male"
4;"Katniss";"Everdeen";"1970-08-12";"female"
5;"Peter";"Everdeen";"1968-08-12";"male"

Student Table

1;"Bill";"Smith";"1985-05-10";"male";1;"chemistry"
2;"Jenny";"Brown";"1986-08-12";"female";2;"physics"
3;"Bob";"Morgan";"1986-06-11";"male";3;"physics"

Employee Table

4;"Katniss";"Everdeen";"1970-08-12";"female";1;"Prof";"1";3500.00
5;"Peter";"Everdeen";"1968-08-12";"male";2;"Assist-Prof";"5";1800.00

Answer 1

First get rid of the FK with something like this:

alter table address drop constraint address_person_id_fkey

If that complains about there not being an address_person_id_fkey constraint then use \d address; in psql to find out what the FK is called.

Then a simple trigger like this should do the trick:

create or replace function pseudo_fk_for_address() returns trigger as $$
begin
    if not exists(select 1 from person where id = new.person_id) then
        raise exception 'No such person: %', new.person_id;
    end if;
    return new;
end;
$$ language plpgsql;

And attach it like this:

create trigger pseudo_fk_for_address_trigger before insert or update on address 
for each row execute procedure pseudo_fk_for_address();

Then you’ll get an error like this if you try to add an address for someone that doesn’t exist in person (including the tables that inherit from it):

playpen=> insert into address (person_id, email, country, citycode, city, addressline) values (3, 'ab', 'b', 2, 'c', 'd');
ERROR:  No such person: 3

You’d want to add a BEFORE DELETE trigger to person to avoid dangling references, that basic structure would be pretty much the same. You might want an index on address.person_id to help support the BEFORE DELETE trigger as well.

References:

• PL/pgSQL
• PL/pgSQL Trigger Procedures
• RAISE