Auto Increment FK how does it update? - sql

Question

Ok so i am currently developing in Oracle 11G express edition for a college assignment. I have ran into an issue on how to auto increment and update in the following parent table. So i have an address table and then a city table for instance. Here is the address SQL code
create table address(
addressid int primary key,
cityid int,
countyid int,
streetnameid int,
postcodeid int,
doornumid int,
natid int,
foreign key (doornumid) references doornum,
foreign key (postcodeid) references postcode,
foreign key (streetnameid) references streetname,
foreign key (countyid) references county,
foreign key (cityid) references city,
foreign key (natid) references nat);
So as you can see I am referencing the city table as a foreign key and this is the city SQL table code below:
create table city(
cityid int primary key,
city varchar(45));
The city code is using a sequence to auto increment when stuff is inserted it:
create SEQUENCE seq_cityID
MINVALUE 1
START WITH 1
INCREMENT BY 1
CACHE 11;
So simply I am auto incrementing all inputs by a cityid like this:
INSERT INTO city (cityid, city)
values(seq_cityID.nextval, Oxford);
INSERT INTO city (cityid, city)
values(seq_cityID.nextval, Oxford);
But my issue is that when I am referencing this in its parent table e.g. the address table how do i reference the ID for that row of data and make sure that the correct ID is pulled without having to manually type it in?
INSERT INTO address (addressid, cityid, countyid, streetnameid, postcodeid, doornumid, natid)
values(seq_addressID.nextval, seq_cityID.nextval, seq_countyID.nextval, seq_streetnameID.nextval, seq_postcodeID.nextval, seq_doornumID.nextval, seq_natID.nextval);
INSERT INTO address (addressid, cityid, countyid, streetnameid, postcodeid, doornumid, natid)
values(seq_addressID.nextval, seq_cityID.nextval, seq_countyID.nextval, seq_streetnameID.nextval, seq_postcodeID.nextval, seq_doornumID.nextval, seq_natID.nextval);
This is the simple insert into address which is only currently referencing nextval but i do not believe it will pull that row of data from that ID. How do i effectively pull through that ID from the child table and have it correctly into the parent automatically?

Answer 1

You have a misconception about tables and their relations.
Your datamodel has one table for all city names, one for all street names, one for all door numbers, etc. But why would you have a table that contains door numbers? What does it tell you? Door numbers are no entity by themselves; they belong to a street. And streets belong to cities. It would make no sense to find all people who live in some number 12 or in some Main Street.
One possible data model:
country (country_id, country_name, country_code)
pk country_id
county (county_id, county_name)
pk county_id
fk country_id -> country
city (city_id, city_name, postcode, county_id)
pk city_id
fk county_id -> county
street (street_id, street name, city_id)
pk street_id
fk city_id -> city
address (address_id, street_id, door_number
pk address_id
fk street_id -> street
With this data model we can check for consistency. If we want to enter the address 111 Millstreet, Oxford, Italy, the database will tell us there is no Oxford in Italy. We can also easily find addresses in the same street (and not like "there are five million addresses in Main Street", but "800 addresses in Main Street, Ohio").
If you want to insert a new address, you look up the country, then the county, etc. until you get to the street ID.

Answer 2

With your original tables:
declare
v_countyid integer;
v_cityid integer;
v_streetnameid integer;
begin
insert into city (city_name) values ('Saint Petersburg') returning cityid into v_city_d;
insert into streetname (street_name) values ('Park Drive') returning streetnameid into v_streetnameid;
...
insert into address (cityid, streetnameid, ...) values (v_cityid, v_streetnameid, ...);
commit;
end;
But then, there would be duplicate cities, street names, etc. in the tables. So we'd rather:
declare
v_countyid integer;
v_cityid integer;
v_streetnameid integer;
begin
select cityid into v_cityid from city where city_name = 'Saint Petersburg';
if cityid is null then
insert into city (city_name) values ('Saint Petersburg') returning cityid into v_city_d;
end if;
select streetnameid into v_streetnameid from streetname where street_name = 'Park Drive';
if v_streetnameid is null then
insert into streetname (street_name) values ('Park Drive') returning streetnameid into v_streetnameid;
end if;
...
insert into address (cityid, streetnameid, ...) values (v_cityid, v_streetnameid, ...);
commit;
end;
But as mentioned in my other answer. This data model doesn't make much sense anyway.