在一个查询中插入3个级别

Question

I have an Oracle data structure like this:

Clubs

Id
Name

A club can have many teams:

Teams

Id
Name
ClubId

A team can have many players:

Players

Id
Name
TeamId

How do I make an insert with a club with teams with players? If it is just a club with teams I can do:

Insert all 
into Clubs (Id, Name) Values (ClubIdSequence.nextval,'Manchester United')
into Teams (Name, ClubId) Values ('A-Team',ClubIdSequence.currval)
into Teams (Name, ClubId) Values ('B-Team',ClubIdSequence.currval)
Select * from dual

But how do I insert Players as well?

This doesn't work:

Insert all 
into Clubs (Id, Name) Values (ClubIdSequence.nextval,'Manchester United')
into Teams (Id,Name, ClubId) Values (TeamIdSequence.nextval,'A-Team',ClubIdSequence.currval)
into Players (Name,Teamid) Values ('Michael Laudrup',TeamIdSequence.currval
into Teams (Id,Name, ClubId) Values (TeamIdSequence.nextval,'B-Team',ClubIdSequence.currval)
into Players (Name,Teamid) Values ('Brian Laudrup',TeamIdSequence.currval)
Select * from dual

Because

An UPDATE or INSERT statement attempted to insert a duplicate key.
        For Trusted Oracle configured in DBMS MAC mode, you may see
        this message if a duplicate entry exists at a different level

Answer 1

This is a single statement needing the ClubIdSequence.nextval and TeamIdSequence.nextval. These values are being retrieved and then the statement gets executed. Thus you get two teams with the same id.

With your first statement you could even swap currval and nextval:

Insert all 
into Clubs (Id, Name) Values (ClubIdSequence.currval,'Manchester United')
into Teams (Name, ClubId) Values ('A-Team',ClubIdSequence.nextval)
into Teams (Name, ClubId) Values ('B-Team',ClubIdSequence.currval)
Select * from dual

It still does the same: excecute nextval, and then do the inserts with currval.

When working with ids got from sequences, I highly recommend writing before-insert triggers reading the nextval of the appropriate sequence and fill the id. Thus the writer of an insert statement doesn't have to find the according sequence and can hence not be mistaken in choosing one. Also someone might be tempted to write an insert statement with id = max(id) + 1 oblivious to the fact that there even is a sequence to be used.

Once you are using triggers, you would simply omit the id in your insert statements.

Still that doesn't solve your problem, because you would have to use currval, which gives the same value even when used more than once in one statement. (Strange enough it even gives you the current value after the first insert statement, ie neither the one before the two inserts nor the one after the two, as one would expect.)

So you end up with two scenarios: Either use sequences and not use insert all or use insert all and not use sequences. They just don't get along well.

Here is a solution without the use of sequences:

Insert all 
  into Clubs (id, Name) Values (max_id + 1, 'Manchester United')
  into Teams (Name, ClubId) Values ('A-Team',max_id + 1)
  into Clubs (id, Name) Values (max_id + 2, 'Manchester Apart')
  into Teams (Name, ClubId) Values ('B-Team',max_id + 2)
Select nvl(max(id),0) max_id from clubs

Insert all into Clubs (Name) Values ('Manchester United') into Teams (Name, ClubId) Values ('A-Team',max_id+1) into Clubs (Name) Values ('Manchester Apart') into Teams (Name, ClubId) Values ('B-Team',max_id+2) Select max(id) max_id from clubs