如何为postgres编写DELETE CASCADE？

Question

I'm manually constructing a DELETE CASCADE statement for postgres.

I have a 'transaction' and a 'slice' table, related as shown below:

    Table "public.slice"
  Column  | Type | Modifiers 
----------+------+-----------
 id       | text | not null
 name     | text | 
Referenced by:
    TABLE "transaction" CONSTRAINT "transaction_slice_id_fkey" FOREIGN KEY (slice_id) REFERENCES slice(id)
 Table "public.transaction"
  Column  | Type | Modifiers 
----------+------+-----------
 id       | text | not null
 slice_id | text | 
Referenced by:
    TABLE "classification_item" CONSTRAINT "classification_item_transaction_id_fkey" FOREIGN KEY (transaction_id) REFERENCES transaction(id)
Table "public.classification_item"
     Column     | Type | Modifiers 
----------------+------+-----------
 id             | text | not null
 transaction_id | text | 
Foreign-key constraints:
    "classification_item_transaction_id_fkey" FOREIGN KEY (transaction_id) REFERENCES transaction(id)

Say I want to delete all transactions and classification_items referenced by the slice whose name is 'my_slice'. What do I need to write?

=# delete from classification_item where transaction_id= #...? 
=# delete from transaction where slice_id= #...? 
=# delete from slice where name='my_slice';

Answer 1

Postgres foreign keys support the CASCADE deletes:

slice_id integer REFERENCES slice(id) ON DELETE CASCADE

etc

Answer 2

In case you can't do what others have suggested:

begin;
delete from classification_item where transaction_id in (select id from "transaction" where slice_id = (select id from slice where name = 'my_slice'));
delete from "transaction" where slice_id in (select id from slice where name='my_slice');
delete from slice where name='my_slice';
commit;

Answer 3

It's soemthing that is defined in the table rather than the DELETE Query. Example (look at order_id):

CREATE TABLE order_items (
    product_no integer REFERENCES products ON DELETE RESTRICT,
    order_id integer REFERENCES orders ON DELETE CASCADE,
    quantity integer,
    PRIMARY KEY (product_no, order_id)
);

Answer 4

You should use CASCADE deletes, and it should be possible to do so even if you inherited a database schema. You would just modify the constraints to add the CASCADE deletes to the schemas:

1. Drop and re-create the constraints to add CASCADE deletes:

    ALTER TABLE ONLY "transaction" DROP CONSTRAINT transaction_slice_id_fkey; ALTER TABLE ONLY "transaction" ADD CONSTRAINT transaction_slice_id_fkey FOREIGN KEY (slice_id) REFERENCES slice(id) ON DELETE CASCADE; ALTER TABLE ONLY "classification_item" DROP CONSTRAINT classification_item_transaction_id_fkey; ALTER TABLE ONLY "classification_item" ADD CONSTRAINT classification_item_transaction_id_fkey FOREIGN KEY (transaction_id) REFERENCES transaction(id) ON DELETE CASCADE; 

2. Now the following query will delete not just the my_slice record from table slice , but also all records from tables transaction and classification_item referencing it:

    DELETE FROM slice WHERE name='my_slice'; 

That procedure will work even if the original schema is created by an object-relational mapper like SQLAlchemy. However in such a case, take care to re-apply that "patch" whenever the schema changes or is re-created. Only if that can't be implemented automatically, it might not be a good idea after all …

Answer 5

It can be delegated to DBMS by set a constraint property 'On delete' = CASCADE. Please see an example .