Question

Is there any way how can I use result for specifying table to join?

I’d like to do something like

SELECT id, some_number, ... FROM sometable NATURAL JOIN someothertable_$some_number;

I know that there’s nothing like this in relational algebra, so probably I’ll not succeed, I just wanted to ask to be sure.

I don’t want to use any SQL scripts.

Answer 1

Runnable Example Here: http://sqlfiddle.com/#!2/5e92c/36

Code to setup tables for this example:

create table if not exists someTable 
(
  someTableId bigint not null auto_increment
  , tableId int not null
  , someOtherTableId bigint not null

  , primary key (someTableId)
  , index (tableId, someOtherTableId)
);

create table if not exists someOtherTable_$1
(
  someOtherTableId bigint not null auto_increment
  , data varchar(128) character set utf8

  , primary key (someOtherTableId)
);

create table if not exists someOtherTable_$2
(
  someOtherTableId bigint not null auto_increment
  , data varchar(128) character set utf8

  , primary key (someOtherTableId)
);

insert sometable (tableId, someOtherTableId) values (1, 1);
insert sometable (tableId, someOtherTableId) values (1, 2);
insert sometable (tableId, someOtherTableId) values (2, 2);
insert sometable (tableId, someOtherTableId) values (2, 3);

insert someothertable_$1(data) values ('table 1 row 1');
insert someothertable_$1(data) values ('table 1 row 2');
insert someothertable_$1(data) values ('table 1 row 3');

insert someothertable_$2(data) values ('table 1 row 1');
insert someothertable_$2(data) values ('table 1 row 2');
insert someothertable_$2(data) values ('table 1 row 3');

STATIC SOLUTION

Here’s a solution if your tables are fixed (e.g. in the example you only have someOtherTable 1 and 2 / you don’t need the code to change automatically as new tables are added):

select st.someTableId
, coalesce(sot1.data, sot2.data)
from someTable st
left outer join someOtherTable_$1 sot1
on st.tableId = 1
and st.someOtherTableId = sot1.someOtherTableId
left outer join someOtherTable_$2 sot2
on st.tableId = 2
and st.someOtherTableId = sot2.someOtherTableId;

DYNAMIC SOLUTION

If the number of tables may change at runtime you’d need to write dynamic SQL. Beware: with every successive table you’re going to take a performance hit. I wouldn’t recommend this for a production system; but it’s a fun challenge. If you can describe your tool set & what you’re hoping to achieve we may be able to give you a few pointers on a more suitable way forward.

select group_concat(distinct ' sot' , cast(tableId as char) , '.data ') 
into @coalesceCols 
from someTable;

select group_concat(distinct ' left outer join someOtherTable_$', cast(tableId as char), ' sot', cast(tableId as char), ' on st.tableId = ', cast(tableId as char), ' and st.someOtherTableId = sot', cast(tableId as char), '.someOtherTableId ' separator '') 
into @tableJoins 
from someTable;

set @sql = concat('select someTableId, coalesce(', @coalesceCols ,') from someTable st', @tableJoins);

prepare stmt from @sql;
execute stmt;