How can I rewrite a multi-column IN clause to work on SQLite?

Question

I\'ve got a query that looks something like this:

SELECT
  *
FROM table
WHERE
  (col1, col2) in (
    (\'col1_val1\', \'col2_val1\'),
    (\'col1_val2\', \'c         


        

Answer 1

In sqlite try to add the VALUES keyword:

SELECT *
FROM table
WHERE
  (col1, col2) in ( VALUES     --> add this keyword and remove the last ,
    ('col1_val1', 'col2_val1'),
    ('col1_val2', 'col2_val2'),
    ('col1_val3', 'col2_val3')
  )

Basically in sqLite executing the query:

VALUES
('col1_val1', 'col2_val1'),
('col1_val2', 'col2_val2');

is the same as:

SELECT 'col1_val1' AS column1, 'col2_val1' AS column2
UNION 
SELECT 'col1_val2' AS column1, 'col2_val2' AS column2;

or combined:

SELECT 'col1_val1' AS column1, 'col2_val1' AS column2
UNION  VALUES  ('col1_val2', 'col2_val2');

So you could even write it like:

SELECT * 
  FROM table
  WHERE (col1, col2) IN ( 
      SELECT 'col1_val1', 'col2_val1'
      UNION 
      SELECT 'col1_val2', 'col2_val2'
  );

which is a simple subquery and works in all/most databases.

Answer 2

Here's an easy solution that works, but it might not perform well on large data sets because it can't use any of your indexes.

SELECT * FROM table
WHERE col1 || '-' || col2 in (
  'col1_val1-col2_val1',
  'col1_val2-col2_val2',
  'col1_val3-col2_val3'
)

Try it in sqlfiddle

Enjoy!

Answer 3

Choose your favourite version:

http://sqlfiddle.com/#!5/6169b/9

using temporary table

CREATE TEMPORARY TABLE pair (a INTEGER, b INTEGER);
INSERT INTO pair (a, b) VALUES (1, 1);
INSERT INTO pair (a, b) VALUES (2, 2);
....

data IN pairs; if pair(a,b) is not unique

SELECT *
FROM data
WHERE EXISTS (
  SELECT NULL
  FROM  pair
  WHERE pair.a = data.a
    AND pair.b = data.b
);

data IN pairs; if pair(a,b) is unique

SELECT     data.*
FROM       data
INNER JOIN pair
        ON pair.a = data.a
       AND pair.b = data.b;

data NOT IN pairs; if pair(a,b) is unique

SELECT    data.*
FROM      data
LEFT JOIN pair
   ON pair.a = data.a
  AND pair.b = data.b
WHERE pair.a IS NULL
   OR pair.b IS NULL;

using inline table

data IN pairs; if pair(a,b) is not unique

SELECT *
FROM data
WHERE EXISTS (
  SELECT NULL
  FROM  (
    SELECT 1 AS a, 1 AS b
    UNION ALL
    SELECT 2 AS a, 2 AS b
    UNION ALL
    SELECT 3 AS a, 3 AS b
  ) AS pair
  WHERE pair.a = data.a
    AND pair.b = data.b
);

data IN pairs; if pair(a,b) is unique

SELECT     data.*
FROM       data
INNER JOIN (
  SELECT 1 AS a, 1 AS b
  UNION ALL
  SELECT 2 AS a, 2 AS b
  UNION ALL
  SELECT 3 AS a, 3 AS b
) AS pair
 ON  pair.a = data.a
AND  pair.b = data.b;

data NOT IN pairs; if pair(a,b) is unique

SELECT    data.*
FROM      data
LEFT JOIN (
  SELECT 1 AS a, 1 AS b
  UNION ALL
  SELECT 2 AS a, 2 AS b
  UNION ALL
  SELECT 3 AS a, 3 AS b
 ) AS pair
   ON pair.a = data.a
  AND pair.b = data.b
WHERE pair.a IS NULL
   OR pair.b IS NULL;