select from unknown number of tables with parameter - mysql

Question

I want to select (union) rows from multiple tables using Parameter
I have table w with two columns:
The column table_name is referring to other tables in my DB, and condition is the 'where' that should be added to the query.
table_name | condition
---------------------
x | y=2
x | r=3
t | y=2
the query should be something like:
select * from x where y=2
union
select * from x where r=3
union
select * from t where y=2
of course that the number of unions is unknown.
Should it be stored procedure? cursor?

Answer 1

One way to get this done. Initial answer was SQL Server syntax. This edit has the MySQL syntax. Make sure your temp table cannot be accessed at the same time. E.g. In MySQL temp tables are unique to the connection. Also add your error checking. In MySQL set the appropriate varchar size for your needs. I used 1024 across the board just for testing purposes.
MySQL syntax
CREATE table test (
id int,
table_name varchar(1024),
where_c varchar(1024)
);
INSERT into test(id, table_name, where_c) values
(1,'x','y=2'),
(2,'x','r=3'),
(3,'t','y=2');
DROP PROCEDURE IF EXISTS generate_sql;
DELIMITER //
CREATE PROCEDURE generate_sql()
BEGIN
DECLARE v_table_name VARCHAR(1024);
DECLARE v_where_c VARCHAR(1024);
DECLARE table_id INT;
DECLARE counter INT;
DECLARE v_SQL varchar(1024);
CREATE TEMPORARY table test_copy
SELECT * FROM test;
SET v_SQL = '';
SET counter = (SELECT COUNT(1) FROM test_copy);
WHILE counter > 0 DO
SELECT id, table_name, where_c
INTO table_id, v_table_name, v_where_c
FROM test_copy LIMIT 1;
SET v_SQL = CONCAT(v_SQL, 'SELECT * FROM ', v_table_name, ' WHERE ', v_where_c);
DELETE FROM test_copy WHERE id=table_id;
SET counter = (SELECT COUNT(1) FROM test_copy);
IF counter > 0 THEN
SET v_SQL = CONCAT(v_SQL,' UNION ');
ELSE
SET v_SQL = v_SQL;
END IF;
END WHILE;
DROP table test_copy;
SELECT v_SQL;
END //
DELIMITER ;
call generate_sql()
SQL Server syntax
CREATE table test (
id int,
table_name varchar(MAX),
condition varchar(MAX)
);
INSERT into test(id, table_name, condition) values
(1,'x','y=2'),
(2,'x','r=3'),
(3,'t','y=2');
SELECT * INTO #temp FROM test;
DECLARE #SQL varchar(MAX);
SET #SQL='';
while exists (select * from #temp)
begin
DECLARE #table_name varchar(MAX);
DECLARE #condition varchar(MAX);
DECLARE #table_id int;
SELECT top 1 #table_id=id, #table_name=table_name, #condition=condition FROM #temp;
SET #SQL += 'SELECT * FROM ' + #table_name + ' WHERE ' + #condition;
delete #temp where id = #table_id;
if exists (select * from #temp) SET #SQL += ' UNION ';
end
SELECT #SQL;
drop table #temp;

Answer 2

Assuming that tables x and t have the same definition and that you want to ignore duplicate results by using UNION rather than UNION ALL, the following should work:
SET #sql = '';
SELECT GROUP_CONCAT(
CONCAT('SELECT * FROM `', `table_name`, '` WHERE ', `condition`)
SEPARATOR ' UNION ') INTO #sql
FROM w;
SET #sql = CONCAT('SELECT * FROM ( ', #sql, ' ) a;');
PREPARE stmt FROM #sql;
EXECUTE stmt;
DEALLOCATE PREPARE stmt;
(I've edited your question slightly because you had two different definitions for table x)