Question

Considering the following table:

CREATE TABLE `foo`(
  `aaa` VARCHAR(45) NOT NULL,
  `bbb` VARCHAR(45) NOT NULL,
  `ccc` INT NOT NULL,
  `ddd` INT NOT NULL,
  PRIMARY KEY(`aaa`, `bbb`, `ccc`)
);

the following data:

INSERT INTO `foo` (`aaa`, `bbb`, `ccc`, `ddd`)
VALUES
('qwe', 'rty', 0, 123),
('asd', 'fgh', 0, 456),
('asd', 'fgh', 1, 678);

the following result is to be gotten:

'qwe', 'rty', 0, 123
'asd', 'fgh', 1, 678

The objective is to get

• Only one record for each (aaa, bbb) keypair
• Favour records with ccc = 1 if there are more than 1 record for a (aaa, bbb) keypair in the table

So I think I need to concatenate result sets of 2 queries:

SELECT * FROM `foo` WHERE `ccc` = 1

and

SELECT * FROM `foo` WHERE `ccc` = 0

discarding a second query result row when there is already a same (aaa, bbb) keypair row in the first query result.

How to code it in SQLite?

Answer 1

Try this one,

SELECT  a.*
FROM    foo a
        INNER JOIN
        (
            SELECT  aaa, bbb, MAX(ccc) maxC
            FROM foo
            GROUP BY aaa, bbb
        ) b ON a.aaa = b.aaa AND
                a.ccc = b.maxC AND
                a.bbb = b.bbb

SQLFiddle Demo