Based on this example:
create table t (
id serial primary key,
a int not null,
b int default null
c int not null
);
Is there a way in PostgreSQL, without resorting to triggers, to incorporate a unique constraint that if there is already a record with values for a and c while b is null, no further records may be inserted in the same combination of a and c. The same should applied the other way round, if there is already a data set for a, b and c with concrete values, that a record with the same a and c may never be inserted where b is null, only further variants with different values for b.
examples:
(5, null, 1)
(4, 6, 9)
(4, 7, 9)
(5, null, 2)
(5, 3, 1) -- refused / error
(4, null, 9) -- also refused
—
Here is my specific use case to better illustrate the problem:
create table if not exists booking (
id serial4,
media_id int4 not null,
data_access int2 not null default session_user_id(),
product_line text not null,
product_variant text default null,
area_nr int2 not null,
net int2 not null,
period int2range not null,
-- If the start and end calendar weeks match, only one week is booked.
entry_time timestamp not null default now(),
constraint "internally used surrogate key to a booking" primary key (id),
constraint "no overlapping booking periods of a distributer" exclude using gist (
data_access with =,
product_line with =,
area_nr with =,
net with =,
period with &&
) where (product_variant is null),
constraint "no overlapping booking period of a sub-distributer" exclude using gist (
data_access with =,
product_line with =,
product_variant with =,
area_nr with =,
net with =,
period with &&
),
/*
constraint "known booked product"
foreign key (data_access, product_line, product_variant)
references product(data_access, line_abbr, variant_abbr),
constraint "known booked area"
foreign key (data_access, area_nr) references area(data_access, nr),
constraint "known booked net"
foreign key (data_access, product_line, net)
references net(data_access, product_line, nr),
constraint "known media of a booking"
foreign key (media_id) references media(id) on delete cascade,
*/
constraint "closed booking periods of areas" check (
upper_inc(period)
)
);
I commented out the foreign keys and only left them in for better understanding.
insert into booking
(product_line, product_variant, area_nr, net, period, media_id)
values
('CB', null, 1, 4, '[2239, 2245]', 5);
insert into booking
(product_line, product_variant, area_nr, net, period, media_id)
values
('CB', 'P', 1, 4, '[2239, 2245]', 5);
The last insert should result in an error since the entire distributor has already been booked for the given period.
insert into booking
(product_line, product_variant, area_nr, net, period, media_id)
values
('CB', 'F', 2, 6, '[2230, 2245]', 5);
insert into booking
(product_line, product_variant, area_nr, net, period, media_id)
values
('CB', 'P', 2, 6, '[2230, 2245]', 5);
insert into booking
(product_line, product_variant, area_nr, net, period, media_id)
values
('CB', null, 2, 6, '[2230, 2245]', 5);
The last one should give an error too since one or more sub-distributors have already been booked, so the entire distributor may no longer be available.
The problem: You first have to check whether a distributor or sub-distributor has already been booked in the same period / range for the same area, net and products, which I solve using a trigger. (due to the cool range feature and related constraints, I was already able to very elegantly exclude overlapping booking periods without a trigger):
create or replace function _avoid_incomplete_distributers() returns trigger as $$ begin
if exists (select from booking where
product_line = new.product_line and case
when new.product_variant is null then product_variant is not null
when new.product_variant is not null then product_variant is null
end and
area_nr = new.area_nr and
net = new.net and
period = new.period
) then
raise exception 'collision of distributer % with related sub-distributers for area % and net % in period %', new.product_line, new.area_nr, new.net, new.period;
else
return new;
end if;
end $$ language plpgsql;
create trigger before_insert_incomplete_distributer before insert on booking for each row
execute procedure _avoid_incomplete_distributers();
You can do this without triggers using write your own check function and adding to table check constraint.
Fox example:
This is function sample:
CREATE OR REPLACE FUNCTION check_booking(v_pline text, v_pvar text, v_area int2)
RETURNS bool
LANGUAGE plpgsql
AS $function$
BEGIN
if (v_pvar is null) then
if (exists(select 1 from booking where product_line = v_pline and area_nr = v_area)) then
return false;
end if;
end if;
if (v_pvar is not null) then
if (exists(select 1 from booking where product_line = v_pline and area_nr = v_area and product_variant is null)) then
return false;
end if;
end if;
return true;
END;
$function$
This is add constraint sample:
ALTER TABLE booking
ADD CONSTRAINT booking_check_constraint
CHECK (check_booking(product_line, product_variant, area_nr));
Related
Some background info: i have a table named defects which has column named status_id and another column named date_closed ,i want to set date_closed after status_id has been updated
i already try to do this using after update trigger with the following code:
after update on eba_bt_sw_defects
for each row
declare
l_status number(20) := null;
begin
select status_id into l_status from eba_bt_sw_defects D,eba_bt_status S where D.status_id = S.id;
if l_status in ( select id from eba_bt_status where is_open = 'N' and NVL(is_enhancement,'N')='N') then
:NEW.DATE_CLOSED := LOCALTIMESTAMP ;
end if;
end;
but an error occured ( subquery not allowed in this contextCompilation failed)
i want a help
A couple of things that need fixing in your code:
In a trigger do not select from the table the trigger you're on. This will probably raise a ORA-04091: table name is mutating, trigger/function may not see it error.
IF l_variable IN (SELECT ...) is not a valid oracle syntax. It raises PLS-00405: subquery not allowed in this context
I don't have your data so here is a similar example:
drop table todos;
drop table statuses;
-- create tables
create table statuses (
id number generated by default on null as identity
constraint statuses_id_pk primary key,
status varchar2(60 char),
is_open varchar2(1 char) constraint statuses_is_open_ck
check (is_open in ('Y','N'))
)
;
create table todos (
id number generated by default on null as identity
constraint todos_id_pk primary key,
name varchar2(255 char) not null,
close_date timestamp with local time zone,
status_id number
constraint todos_status_id_fk
references statuses on delete cascade
)
;
-- load data
insert into statuses (id, status, is_open ) values (1, 'OPEN', 'Y' );
insert into statuses (id, status, is_open ) values (2, 'COMPLETE', 'N' );
insert into statuses (id, status, is_open ) values (3, 'ON HOLD', 'Y' );
insert into statuses (id, status, is_open ) values (4, 'CANCELLED', 'N' );
commit;
insert into todos (name, close_date, status_id ) values ( 'Y2 Security Review', NULL, 1 );
-- triggers
CREATE OR REPLACE TRIGGER todos_biu BEFORE
INSERT OR UPDATE ON todos
FOR EACH ROW
DECLARE
l_dummy NUMBER;
BEGIN
SELECT
1
INTO l_dummy
FROM
statuses
WHERE
is_open = 'N' AND
id = :new.status_id;
:new.close_date := localtimestamp;
EXCEPTION
WHEN no_data_found THEN
-- I'm assuming you want close_date to NULL if todo is re-opened.
:new.close_date := NULL;
END todos_biu;
/
update todos set status_id = 2;
select * from todos;
id name close_date status_id
1 Y2 Security Review 11-MAY-22 05.27.04.987117000 PM 2
I have 2 tables like this
drop table if exists public.table_1;
drop table if exists public.table_2;
CREATE TABLE public.table_1 (
id serial NOT NULL,
user_id bigint not null,
status varchar(255) not null,
date_start date NOT NULL,
date_end date NULL
);
CREATE TABLE public.table_2 (
id serial NOT NULL,
user_id bigint not null,
status varchar(255) not null,
date_start date NOT NULL,
date_end date NULL
);
alter table public.table_1
add constraint my_constraint_1
EXCLUDE USING gist (user_id with =, daterange(date_start, date_end, '[]') WITH &&)
where (status != 'deleted');
alter table public.table_2
add constraint my_constraint_2
EXCLUDE USING gist (user_id with =, daterange(date_start, date_end, '[]') WITH &&)
where (status != 'deleted');
Every table contains rows which are related to a user, and all the rows of the same user cannot overlap in range. In addition, some rows may be logically deleted, so I added a where condition.
So far it's working w/o problems, but the 2 constraints work separately for each table.
I need to create a constraint which cover the 2 set of tables, so that a single daterange (of the same user and not deleted), may appaer only once across the 2 different tables.
Does the EXCLUDE notation be extended to work with different tables or do I need to check it with a trigger? If the trigger is the answer, which is the simplier way to do this? Create a temporary table with the union of the 2, add the constraint on it and check if fails?
Starting from #Laurenz Albe suggestion, this is what I made
-- #################### SETUP SAMPLE TABLES ####################
drop table if exists public.table_1;
drop table if exists public.table_2;
CREATE TABLE public.table_1 (
id serial NOT NULL,
user_id bigint not null,
status varchar(255) not null,
date_start date NOT NULL,
date_end date NULL
);
CREATE TABLE public.table_2 (
id serial NOT NULL,
user_id bigint not null,
status varchar(255) not null,
date_start date NOT NULL,
date_end date NULL
);
alter table public.table_1
add constraint my_constraint_1
EXCLUDE USING gist (user_id with =, daterange(date_start, date_end, '[]') WITH &&)
where (status != 'deleted');
alter table public.table_2
add constraint my_constraint_2
EXCLUDE USING gist (user_id with =, daterange(date_start, date_end, '[]') WITH &&)
where (status != 'deleted');
-- #################### SETUP TRIGGER ####################
create or REPLACE FUNCTION check_date_overlap_trigger_hook()
RETURNS trigger as
$body$
DECLARE
l_table text;
l_sql text;
l_row record;
begin
l_table := TG_ARGV[0];
l_sql := format('
select *
from public.%s as t
where
t.user_id = %s -- Include only records of the same user
and t.status != ''deleted'' -- Include only records that are active
', l_table, new.user_id);
for l_row in execute l_sql
loop
IF daterange(l_row.date_start, COALESCE(l_row.date_end, 'infinity'::date)) && daterange(new.date_start, COALESCE(new.date_end, 'infinity'::date))
THEN
RAISE EXCEPTION 'Date interval is overlapping with another one in table %', l_table
USING HINT = 'You can''t have the same interval across table1 AND table2';
END IF;
end loop;
RETURN NEW;
end
$body$
LANGUAGE plpgsql;
-- #################### INSTALL TRIGGER ####################
create trigger check_date_overlap
BEFORE insert or update
ON public.table_1
FOR EACH row
EXECUTE PROCEDURE check_date_overlap_trigger_hook('table_2');
create trigger check_date_overlap
BEFORE insert or update
ON public.table_2
FOR EACH row
EXECUTE PROCEDURE check_date_overlap_trigger_hook('table_1');
-- #################### INSERT DEMO ROWS ####################
insert into public.table_1 (user_id, status, date_start, date_end) values (1, 'active', '2020-12-10', '2020-12-20');
insert into public.table_1 (user_id, status, date_start, date_end) values (1, 'deleted', '2020-12-15', '2020-12-25');
insert into public.table_1 (user_id, status, date_start, date_end) values (2, 'active', '2020-12-10', '2020-12-20');
insert into public.table_1 (user_id, status, date_start, date_end) values (2, 'deleted', '2020-12-15', '2020-12-25');
-- This will fail for overlap on the same table
-- insert into public.table_1 (user_id, status, date_start, date_end) values (1, 'active', '2020-12-15', '2020-12-25');
-- This will fail as the user 1 already has an overlapping period on table 1
-- insert into public.table_2 (user_id, status, date_start, date_end) values (1, 'active', '2020-12-15', '2020-12-25');
-- This will fail as the user 1 already has an overlapping period on table 1
insert into public.table_2 (user_id, status, date_start, date_end) values (1, 'deleted', '2020-12-15', '2020-12-25');
update public.table_2 set status = 'active' where id = 1;
select 'table_1' as src_table, * from public.table_1
union
select 'table_2', * from public.table_2
You can probably use a trigger, but triggers are always vulnerable to race conditions (unless you are using SERIALIZABLE isolation).
If your tables really have the same columns, why don't you use a single table (and perhaps add a type column to disambiguate)?
I have table tariffs, with two columns: (tariff_id, reception)
I have table users, with two columns: (user_id, reception)
And I have table users_tariffs with two columns: (user_id, tariff_id).
I want to prevent situation when tariff from one reception is assigned to user from another reception. How can I do that?
E.G
Users:
user_id | reception
Putin | Russia
Trump | USA
Tariffs:
tariff_id | reception
cheap | USA
expensive | Russia
Wrong situation at users_tariffs, because Cheap tariff is for USA only:
user_id | tariff_id
Putin | Cheap
SOLUTION 1: FOREIGN KEY CONSTRAINTS
I am assuming the following table definitions.
In particular, the composite key in user_tariffs makes this a many-to-many relationship between users and tariffs.
CREATE TABLE tariffs (tariff_id int NOT NULL PRIMARY KEY,
reception text NOT NULL);
CREATE TABLE users (user_id int NOT NULL PRIMARY KEY,
reception text NOT NULL);
CREATE TABLE user_tariffs (tariff_id int NOT NULL REFERENCES tariffs (tariff_id),
user_id int NOT NULL REFERENCES users (user_id),
PRIMARY KEY (tariff_id, user_id));
You probably need a combination of all three columns somewhere, so let's create this:
ALTER TABLE user_tariffs ADD COLUMN reception text;
UPDATE user_tariffs a
SET reception = b.reception
FROM (SELECT * FROM tariffs) b
WHERE a.tariff_id = b.tariff_id;
ALTER TABLE user_tariffs ALTER COLUMN reception SET NOT NULL;
Now we can use FOREIGN KEY REFERENCES (user_id, reception) into users.
CREATE UNIQUE INDEX ON tariffs (tariff_id, reception);
ALTER TABLE user_tariffs ADD FOREIGN KEY (tariff_id, reception)
REFERENCES tariffs (tariff_id, reception);
In addition, we can use FK REFs (tariff_id, reception) into tariffs.
CREATE UNIQUE INDEX ON users (user_id, reception);
ALTER TABLE user_tariffs ADD FOREIGN KEY (user_id, reception)
REFERENCES users (user_id, reception);
Populate with data:
INSERT INTO users VALUES (1, 'cheap'), (2, 'expensive');
INSERT INTO tariffs VALUES (1, 'cheap'), (2, 'expensive');
Now assume we have the following data (user_id, tariff_id) to insert:
WITH data (user_id, tariff_id)
AS (VALUES (1, 2), (2, 1)), -- here is your application data
datas (user_id, tariff_id, reception)
AS (SELECT user_id,
tariff_id,
(SELECT u.reception -- reception calculated by user
FROM users u
WHERE u.user_id = d.user_id)
FROM data d)
INSERT INTO user_tariffs SELECT * FROM datas ;
Then you cannot insert the data, because you can only add (1, 1) or (2, 2) with the same reception, but not (1, 2) or (2, 1) with different reception's. The error message is:
ERROR: insert or update on table "user_tariffs" violates foreign key constraint "user_tariffs_user_id_fkey1"
DETAIL: Key (user_id, reception)=(2, cheap) is not present in table "users".
But you can insert with data AS VALUES (1, 1), (2, 2).
I think the FOREIGN KEY CONSTRAINT solution is to be preferred.
Please describe your functional dependencies, if you want better table designs.
SOLUTION 2: TRIGGER
-- DROP TABLE user_tariffs CASCADE;
-- DROP TABLE users CASCADE;
-- DROP TABLE tariffs CASCADE;
CREATE TABLE tariffs (tariff_id int NOT NULL PRIMARY KEY,
reception text NOT NULL);
CREATE TABLE users (user_id int NOT NULL PRIMARY KEY,
reception text NOT NULL);
CREATE TABLE user_tariffs (tariff_id int NOT NULL REFERENCES tariffs (tariff_id),
user_id int NOT NULL REFERENCES users (user_id),
PRIMARY KEY (tariff_id, user_id));
INSERT INTO users VALUES (1, 'cheap'), (2, 'expensive');
INSERT INTO tariffs VALUES (1, 'cheap'), (2, 'expensive');
-- table user_tariffs (user_id, tariff_id) only, without reception column.
Create a function with return type trigger:
CREATE OR REPLACE FUNCTION check_reception()
RETURNS trigger AS $$
DECLARE valid boolean := false;
BEGIN
SELECT (SELECT u.reception FROM users u WHERE u.user_id = NEW.user_id)
= (SELECT t.reception FROM tariffs t WHERE t.tariff_id = NEW.tariff_id)
INTO valid FROM user_tariffs ;
IF valid = false
THEN RAISE EXCEPTION '(user, tariff, reception) invalid.';
END IF;
RETURN NEW;
END; $$ LANGUAGE plpgsql ;
and register it:
CREATE TRIGGER reception_trigger
AFTER INSERT OR UPDATE ON user_tariffs
FOR EACH ROW EXECUTE PROCEDURE check_reception();
Now try to insert (1, 2), which would be (cheap, expensive) and is not allowed:
INSERT INTO user_tariffs VALUES (1, 2);
ERROR: (user, tariff, reception) invalid.
KONTEXT: PL/pgSQL function check_reception() line 7 at RAISE
But we can insert (1, 1), which is (cheap, cheap) without problem:
INSERT INTO user_tariffs VALUES (1, 1);
SELECT * FROM user_tariffs;
Remark
Triggers are not the best solution here, in my opinion. Try to avoid triggers, if possible. They can have side effects (transactions etc). Check StackOverflow for further details :)
I'm currently going through the growing pains of trying to learn about functions and triggers. I'm trying to do a problem from a book I'm reading , but i dont understand how to do certain parts.
using this table
create table movies (
id integer primary key,
title varchar(255) not null,
year integer
);
insert into movies values (1, 'The Croods', 2013);
insert into movies values (2, 'Now You See Me', 2013);
insert into movies values (3, 'Argo', 2012);
insert into movies values (4, 'Jurassic World', 2015);
create table discs (
id integer primary key,
movie_id integer not null references movies(id),
type_id integer references disc_types(id),
price decimal(10,2),
available boolean
);
insert into discs values (1, 1, 1, 1.59, 't');
insert into discs values (2, 1, 1, 1.59, 'f');
insert into discs values (3, 1, 2, 2.99, 'f');
insert into discs values (4, 2, 1, 1.29, 't');
insert into discs values (5, 2, 1, 1.29, 't');
insert into discs values (6, 2, 2, 2.99, 't');
insert into discs values (7, 3, 2, 2.59, 't');
insert into discs values (8, 3, 2, 2.59, 't');
create table customers (
id integer primary key,
name varchar(255),
email varchar(255)
);
insert into customers values (1, 'John', 'john#hotmail.com');
insert into customers values (2, 'Jane', 'jane#gmail.com');
create table rentals (
id integer primary key,
customer_id integer not null references customers(id),
disc_id integer not null references discs(id),
date_rented date,
date_returned date
);
insert into rentals values (1, 1, 7, '2013-10-01', '2013-10-03');
insert into rentals values (2, 2, 5, '2013-10-05', '2013-10-06');
insert into rentals values (3, 2, 2, '2013-11-02', null);
insert into rentals values (4, 2, 3, '2013-11-02', null);
create table ratings (
customer_id integer not null references customers(id),
movie_id integer not null references movies(id),
rating integer,
primary key (customer_id, movie_id)
);
insert into ratings values (1, 1, 1);
insert into ratings values (1, 2, 4);
insert into ratings values (1, 3, 5);
insert into ratings values (2, 1, 4);
my logic was that i would have the new values of the ratings table that were going to be inserted or updated and use those to compare to whats in the rentals table to see if that customer had rented that movie already, if they did then they could enter a rating. but i cant transfer that logic in this lol. unless there an easier way to do this.
The loop inside the function complicates matters a bit, let's see if we can get rid of it. Your ratings table has a reference to customer and movie so we need a join.
SELECT COUNT(*) INTO rented FROM rentals WHERE disc_id IN
(SELECT id from discs INNER JOIN
rentals ON disc_id = discs.id where movie_id = new.movie_id)
AND customer_id = new.customer_id
Right this should make the logic of your stored procedure a lot easier. I am now leaving you to finish it because this after all is a learning exercise.
You need this sort of a join because it's more efficient and simpler than the loop. The ratings table has a reference to the movie_id but the rentals table only has a disc_id thus to find out if the user has rented a particular movie, you need to join it through the disc table.
You will need to change the return values. ref: http://www.postgresql.org/docs/9.2/static/plpgsql-trigger.html
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
And also note that you do not do an INSERT inside your trigger function. You just return a non null value for the insert to proceed.
This is the EXISTS() version. (BTW: the definition for movies is missing)
CREATE OR REPLACE FUNCTION rate_only_rented()
RETURNS TRIGGER AS $func$
BEGIN
IF ( NOT EXISTS (
SELECT *
FROM rentals r
JOIN discs d ON r.disc_id = d.id
WHERE d.movie_id = NEW.movie_id
AND r.customer_id = NEW.customer_id
) ) THEN
RAISE EXCEPTION 'you(%) have not rented this movie(%) before'
, NEW.customer_id ,NEW.movie_id;
RETURN NULL;
ELSE
RETURN NEW;
END IF;
END;
$func$ language plpgsql;
And the trigger:
CREATE TRIGGER rate_only_rented
AFTER INSERT OR UPDATE
ON ratings
FOR EACH ROW
EXECUTE PROCEDURE rate_only_rented()
;
I have the following scenario in a Postgres 9.3 database:
Tables B and C reference Table A.
Table C has an optional field that references table B.
I would like to ensure that for each row of table C that references table B, c.b.a = c.a. That is, if C has a reference to B, both rows should point at the same row in table A.
I could refactor table C so that if c.b is specified, c.a is null but that would make queries joining tables A and C awkward.
I might also be able to make table B's primary key include its reference to table A and then make table C's foreign key to table B include table C's reference to table A but I think this adjustment would be too awkward to justify the benefit.
I think this can be done with a trigger that runs before insert/update on table C and rejects operations that violate the specified constraint.
Is there a better way to enforce data integrity in this situation?
There is a very simple, bullet-proof solution. Works for Postgres 9.3 - when the original question was asked. Works for the current Postgres 13 - when the question in the bounty was added:
Would like information on if this is possible to achieve without database triggers
FOREIGN KEY constraints can span multiple columns. Just include the ID of table A in the FK constraint from table C to table B. This enforces that linked rows in B and C always point to the same row in A. Like:
CREATE TABLE a (
a_id int PRIMARY KEY
);
CREATE TABLE b (
b_id int PRIMARY KEY
, a_id int NOT NULL REFERENCES a
, UNIQUE (a_id, b_id) -- redundant, but required for FK
);
CREATE TABLE c (
c_id int PRIMARY KEY
, a_id int NOT NULL REFERENCES a
, b_id int
, CONSTRAINT fk_simple_and_safe_solution
FOREIGN KEY (a_id, b_id) REFERENCES b(a_id, b_id) -- THIS !
);
Minimal sample data:
INSERT INTO a(a_id) VALUES
(1)
, (2);
INSERT INTO b(b_id, a_id) VALUES
(1, 1)
, (2, 2);
INSERT INTO c(c_id, a_id, b_id) VALUES
(1, 1, NULL) -- allowed
, (2, 2, 2); -- allowed
Disallowed as requested:
INSERT INTO c(c_id, a_id, b_id) VALUES (3,2,1);
ERROR: insert or update on table "c" violates foreign key constraint "fk_simple_and_safe_solution"
DETAIL: Key (a_id, b_id)=(2, 1) is not present in table "b".
db<>fiddle here
The default MATCH SIMPLE behavior of FK constraints works like this (quoting the manual):
MATCH SIMPLE allows any of the foreign key columns to be null; if any of them are null, the row is not required to have a match in the referenced table.
So NULL values in c(b_id) are still allowed (as requested: "optional field"). The FK constraint is "disabled" for this special case.
We need the logically redundant UNIQUE constraint on b(a_id, b_id) to allow the FK reference to it. But by making it out to be on (a_id, b_id) instead of (b_id, a_id), it is also useful in its own right, providing a useful index on b(a_id) to support the other FK constraint, among other things. See:
Is a composite index also good for queries on the first field?
(An additional index on c(a_id) is typically useful accordingly.)
Further reading:
Differences between MATCH FULL, MATCH SIMPLE, and MATCH PARTIAL?
Enforcing constraints “two tables away”
I ended up creating a trigger as follows:
create function "check C.A = C.B.A"()
returns trigger
as $$
begin
if NEW.b is not null then
if NEW.a != (select a from B where id = NEW.b) then
raise exception 'a != b.a';
end if;
end if;
return NEW;
end;
$$
language plpgsql;
create trigger "ensure C.A = C.B.A"
before insert or update on C
for each row
execute procedure "check C.A = C.B.A"();
Would like information on if this is possible to achieve without database triggers
Yes, it is possible. The mechanism is called ASSERTION and it is defined in SQL-92 Standard(though it is not implemented by any major RDBMS).
In short it allows to create multiple-row constraints or multi-table check constraints.
As for PostgreSQL it could be emulated by using view with WITH CHECK OPTION and performing operation on view instead of base table.
WITH CHECK OPTION
This option controls the behavior of automatically updatable views. When this option is specified, INSERT and UPDATE commands on the view will be checked to ensure that new rows satisfy the view-defining condition (that is, the new rows are checked to ensure that they are visible through the view). If they are not, the update will be rejected.
Example:
CREATE TABLE a(id INT PRIMARY KEY, cola VARCHAR(10));
CREATE TABLE b(id INT PRIMARY KEY, colb VARCHAR(10), a_id INT REFERENCES a(id) NOT NULL);
CREATE TABLE c(id INT PRIMARY KEY, colc VARCHAR(10),
a_id INT REFERENCES a(id) NOT NULL,
b_id INT REFERENCES b(id));
Sample inserts:
INSERT INTO a(id, cola) VALUES (1, 'A');
INSERT INTO a(id, cola) VALUES (2, 'A2');
INSERT INTO b(id, colb, a_id) VALUES (12, 'B', 1);
INSERT INTO c(id, colc, a_id) VALUES (15, 'C', 2);
Violating the condition(connecting C with B different a_id on both tables)
UPDATE c SET b_id = 12 WHERE id = 15;;
-- no issues whatsover
Creating view:
CREATE VIEW view_c
AS
SELECT *
FROM c
WHERE NOT EXISTS(SELECT 1
FROM b
WHERE c.b_id = b.id
AND c.a_id != b.a_id) -- here is the clue, we want a_id to be the same
WITH CHECK OPTION ;
Trying update second time(error):
UPDATE view_c SET b_id = 12 WHERE id = 15;
--ERROR: new row violates check option for view "view_c"
--DETAIL: Failing row contains (15, C, 2, 12).
Trying brand new inserts with incorrect data(also errors)
INSERT INTO b(id, colb, a_id) VALUES (20, 'B2', 2);
INSERT INTO view_c(id, colc, a_id, b_id) VALUES (30, 'C2', 1, 20);
--ERROR: new row violates check option for view "view_c"
--DETAIL: Failing row contains (30, C2, 1, 20)
db<>fiddle demo