Question

Having a bit of a problem. Firstly, here’s my table structure.

Table doc_perms

id    user_id    doc_id   can_view    can_edit    can_delete

ID being the PK and user_id and doc_id being FK’s of other tables.

Basically, what I need is to prevent an INSERT query from executing if the INSERT data for user_id and doc_id already exists in a row regardless of the rest of the columns.

For example

if you have the data

user_id  doc_id
   1        1
   1        2

And then you try do an INSERT with user_id = 1 and doc_id = 2, the query fails because there’s already a row with that data in it.

Hope this makes sense.

Appreciate your help in advance.

Answer 1

Assuming there are no duplicates in doc_perms, you can create a unique index on (user_id, doc_id):

CREATE UNIQUE INDEX doc_perms_index 
  ON doc_perms (user_id,doc_id)

If there are duplicates, you can use ALTER IGNORE … ADD UNIQUE INDEX will remove them:

ALTER IGNORE TABLE doc_perms 
  ADD UNIQUE INDEX doc_perms_index (user_id, doc_id)

(Among all rows that share the same (user_id, doc_id), all the rows except one will be dropped. The row that is kept is unspecified.)

After you create the unique index, INSERT INTO doc_perms ... will raise an error if (user_id, doc_id) is a duplicate.

If you wish to INSERT, but UPDATE other columns when (user_id, doc_id) is a duplicate, then (as @moopet has already mentioned) you can use INSERT … ON DUPLICATE KEY UPDATE. For example:

INSERT INTO doc_perms (user_id, doc_id, foo) 
  VALUES (1, 2, 3) ON DUPLICATE KEY UPDATE foo = 3'''