Postgres: change primary key in existing table - postgresql

I have a situation with two tables where one has a foreign key pointing to the other table (simplified) schema:
CREATE TABLE table1 (
name VARCHAR(64) NOT NULL,
PRIMARY KEY(name)
);
CREATE TABLE table2 (
id SERIAL PRIMARY KEY,
table1_name VARCHAR(64) NOT NULL REFERENCES table1(name)
);
Now I regret using the name column as primary key in table1 - and would like to add integer serial key instead. Since I already have data in the database I guess I need to do this carefully. My current plan is as follows:
Drop the foreign key constraint: table2(name) with ALTER TABLE table2 DROP CONSTRAINT table2_table1_name_fkey;
Drop the primary key constraint on table1(name) with ALTER TABLE table1 DROP CONSTRAINT name_pkey;.
Add a unique constraint on table1(name) with ALTER TABLE table1 ADD UNIQUE(name);
Add a automatic primary key to table1 with ALTER TABLE table1 ADD COLUMN ID SERIAL PRIMARY KEY;.
Add a new column table1_id to table2 with ALTER TABLE table2 ADD COLUMN table1_id INT;
Update all rows in table2 - so that the new column (which will be promoted to a foreign key) gets the correct value - as inferred by the previous (still present) foreign key table1_name.
I have completed steps up to an including step 5, but the UPDATE (with JOIN?) required to complete 6 is beyond my SQL paygrade. My current (google based ...) attempt looks like:
UPDATE
table2
SET
table2.table1_id = t1.id
FROM
table1 t1
LEFT JOIN table2 t2
ON t2.table1_name = t1.name;

You do not need JOIN in UPDATE.
UPDATE
table2 t2
SET
table1_id = t1.id
FROM
table1 t1
WHERE
t2.table1_name = t1.name;

Related

postgresql, deferred constraint not working as I expect

I'm struggling to understand the way DEFERRED constraints work in postgres (v13 if that matters).
I have the following schema and data:
CREATE TEMP TABLE t1 (
id int,
CONSTRAINT unique_id PRIMARY KEY (id)
);
CREATE TEMP TABLE t2 (
id int,
ref int,
CONSTRAINT fk FOREIGN KEY (ref) REFERENCES t1 (id) ON DELETE SET NULL DEFERRABLE INITIALLY IMMEDIATE
);
INSERT INTO t1 VALUES (1);
INSERT INTO t2 (id,ref) VALUES (1,1);
Then I execute the instructions below:
BEGIN;
SET CONSTRAINTS ALL DEFERRED;
DELETE FROM t1;
INSERT INTO t1 VALUES (1);
COMMIT;
Because constraints are deferred, I would expect the ON DELETE SET NULL to trigger at the end of the transaction and preserve the link from t1 to t2. Yet it does not.
SELECT * FROM t2;
id | ref
----+-----
1 |
(1 row)
What am I missing here ?
DEFERRABLE only means that the check if no row violates the constraint is postponed until the transaction is committed. It does not mean that the effect of the DML that you run, is postponed until the end of the transaction.
In your example at the end of the transaction no row in t2 violates the foreign key constraint. So the COMMIT is successful.
Additionally: the DELETE FROM t1 will set all ref columns to NULL in the table t2 that contained the deleted IDs. If you then insert a new row into t1, how should Postgres know which of the (possibly millions) of ref columns that are NULL to reset to the initial value?

PostgreSQL: Foreign key between composite type and independent columns

Minimal definitions:
CREATE TYPE GlobalId AS (
id1 BigInt,
id2 SmallInt
);
CREATE TABLE table1 (
id1 BigSerial NOT NULL,
id2 SmallInt NOT NULL,
PRIMARY KEY (id1, id2)
);
CREATE TABLE table2 (
global_id GlobalId NOT NULL,
FOREIGN KEY (global_id) REFERENCES table1 (id1, id2)
);
In short, I use a composite type for table2 (and many other tables), but for the primary table (table1), I don't directly use the composite type because composite types don't support the use of Serial.
The above produces the following error due to the ostensible mismatch between global_id and id1, id2: number of referencing and referenced columns for foreign key disagree.
Alternatively, if I define the foreign key as FOREIGN KEY (global_id.id1, global_id.id2) REFERENCES table1 (id1, id2), I get a syntax error on using an accessor on global_id.
Any ideas on how to define this foreign key relationship? Alternatively, if there's a way for table1 to use the GlobalId composite type while still getting serial/sequence behavior for id1, that works also.
You can define table1 using your composite type and fill the value using a BEFORE trigger:
CREATE TABLE table1 (id globalid PRIMARY KEY);
CREATE SEQUENCE s OWNED BY table1.id;
CREATE FUNCTION ins_trig() RETURNS trigger LANGUAGE plpgsql AS
$$BEGIN
NEW.id = (nextval('s'), (NEW.id).id2);
RETURN NEW;
END;$$;
CREATE TRIGGER ins_trig BEFORE INSERT ON table1 FOR EACH ROW
EXECUTE PROCEDURE ins_trig();
INSERT INTO table1 VALUES (ROW(NULL, 42));
SELECT * FROM table1;
id
--------
(1,42)
(1 row)

Use COPY FROM command in PostgreSQL to insert in multiple tables

I'm trying to use the performance of COPY FROM command in PostgreSQL to get all data of 1 table of a CSV file (CSV -> table1) and I need to insert other data, but, in a new table. I will need of a primary key of first table to put as a foreign key in second table.
Example:
I need to insert 1,000,000 of names in table1 and 500,000 of names in table2, but, all names in table2 reference to 1 tuple in table1.
CREATE TABLE table1 (
table1Id bigserial NOT NULL,
Name varchar(100) NULL,
CONSTRAINT table1Id PRIMARY KEY (table1Id)
);
CREATE TABLE table2 (
table2Id bigserial NOT NULL,
Other_name varchar(100) NOT NULL
table1_table1Id int8 NOT NULL,
CONSTRAINT table2_pk PRIMARY KEY (table2Id)
);
Command COPY does not allow table manipulations while copying data (such as look up to other table for fetching proper foreign keys to insert). To insert into table2 ids for corresponding rows from table1 you need to drop NOT NULL constraint for that field, COPY data and then UPDATE that fields separately.
Assuming table1 and table2 tables can be joined by table1.Name = table2.Other_name, the code is:
Before COPY:
ALTER TABLE table2 ALTER COLUMN table1_table1Id DROP NOT NULL;
After COPY:
UPDATE table2 SET table2.table1_table1Id = table1.table1Id
FROM table1
WHERE table1.Name = table2.Other_name;
ALTER TABLE table2 ALTER COLUMN table1_table1Id SET NOT NULL;

How to add a foreign key constraint to same table using ALTER TABLE in PostgreSQL

To create table I use:
CREATE TABLE category
(
cat_id serial NOT NULL,
cat_name character varying NOT NULL,
parent_id integer NOT NULL,
CONSTRAINT cat_id PRIMARY KEY (cat_id)
)
WITH (
OIDS=FALSE
);
ALTER TABLE category
OWNER TO pgsql;
parent_id is a id to another category. Now I have a problem: how to cascade delete record with its children? I need to set parent_id as foreign key to cat_id.
I try this:
ALTER TABLE category
ADD CONSTRAINT cat_cat_id_fkey FOREIGN KEY (parent_id)
REFERENCES category (cat_id) MATCH SIMPLE
ON UPDATE CASCADE ON DELETE CASCADE
But it falls with:
ERROR: insert or update on table "category" violates foreign key constraint "cat_cat_id_fkey"
DETAIL: Key (parent_id)=(0) is not present in table "category".
The problem you have - what would be the parent_id of a category at the top of the hierarchy?
If it will be null - it will break the NOT NULL constratint.
If it will be some arbitrary number like 0 - it will break the foreign key (like in your example).
The common solution - drop the NOT NULL constratint on the parent_id and set parent_id to null for top categories.
-- create some fake data for testing
--
DROP SCHEMA tmp CASCADE;
CREATE SCHEMA tmp ;
SET search_path=tmp;
CREATE TABLE category
(
cat_id serial NOT NULL,
cat_name character varying NOT NULL,
parent_id integer NOT NULL,
CONSTRAINT cat_id PRIMARY KEY (cat_id)
);
INSERT INTO category(cat_name,parent_id)
SELECT 'Name_' || gs::text
, gs % 3
FROM generate_series(0,9) gs
;
-- find the records with the non-existing parents
SELECT ca.parent_id , COUNT(*)
FROM category ca
WHERE NOT EXISTS (
SELECT *
FROM category nx
WHERE nx.cat_id = ca.parent_id
)
GROUP BY ca.parent_id
;
-- if all is well: proceed
-- make parent pointer nullable
ALTER TABLE category
ALTER COLUMN parent_id DROP NOT NULL
;
-- set non-existing parent pointers to NULL
UPDATE category ca
SET parent_id = NULL
WHERE NOT EXISTS (
SELECT *
FROM category nx
WHERE nx.cat_id = ca.parent_id
)
;
-- Finally, add the FK constraint
ALTER TABLE category
ADD CONSTRAINT cat_cat_id_fkey FOREIGN KEY (parent_id)
REFERENCES category (cat_id) MATCH SIMPLE
ON UPDATE CASCADE ON DELETE CASCADE
;
This is quite simple.
Here the foreign key parent_id refers to cat_id.
Here a record with parent_id=0 exists but not a record with cat_id=0.

PostgreSQL - properly change ID of table row

How to change id of some table's row?
Like:
UPDATE table SET id=10 WHERE id=5;
But in way that it would cascade changes to every other table that references this table with that id?
I want to do this, because I need to import data from another database which has most of the same tables, but ids are different. So if ids would match old database, it would be easier to import data correctly.
Suppose you have these two tables:
create table referenced (id integer primary key);
create table referencer (a integer references referenced (id));
Table referencer references table referenced:
=> \d referencer
Table "public.referencer"
Column | Type | Modifiers
--------+---------+-----------
a | integer |
Foreign-key constraints:
"referencer_a_fkey" FOREIGN KEY (a) REFERENCES referenced(id)
Then you insert a value in both:
insert into referenced values (1);
insert into referencer values (1);
select *
from
referenced rd
inner join
referencer rr on rd.id = rr.a
;
id | a
----+---
1 | 1
Now you want to change the reference to on update cascade:
alter table referencer
drop constraint referencer_a_fkey,
add foreign key (a) references referenced (id) on update cascade;
And update it:
update referenced set id = 2;
select *
from
referenced rd
inner join
referencer rr on rd.id = rr.a
;
id | a
----+---
2 | 2
Now you will have another problem in the referenced table primary key if the updated id already exists. But that would make another question.
UPDATE
This is dangerous so backup the db first. It must be done as superuser:
update pg_constraint
set confupdtype = 'c'
where conname in (
select
c.conname
from
pg_constraint c
inner join
pg_class referenced on referenced.oid = c.confrelid
where
referenced.relname = 'referenced'
and
c.contype = 'f'
);
It will change all the foreign key constraints on the referenced table to on update cascade
You will need to change your foreign key and set ON UPDATE action to CASCADE. When you change a value, all associated values will be changed too.
This is an example how to define it:
CREATE TABLE order_items (
product_no integer REFERENCES products ON UPDATE CASCADE,
order_id integer REFERENCES orders ON UPDATE CASCADE,
quantity integer,
PRIMARY KEY (product_no, order_id)
);
For more information see http://www.postgresql.org/docs/current/static/ddl-constraints.html