generate an integer sequence in MySQL

Question

I need to do a join with a table/result-set/whatever that has the integers n to m inclusive. Is there a trivial way to get that without just building the table?

(BTW what would that type of construct be called, a "Meta query"?)

m-n is bounded to something reasonable ( < 1000's)

Answer 1

There is no sequence number generator (CREATE SEQUENCE) in MySQL. Closest thing is AUTO_INCREMENT, which can help you construct the table.

Answer 2

How big is m?

You could do something like:

create table two select null foo union all select null;
create temporary table seq ( foo int primary key auto_increment ) auto_increment=9 select a.foo from two a, two b, two c, two d;
select * from seq where foo <= 23;

where the auto_increment is set to n and the where clause compares to m and the number of times the two table is repeated is at least ceil(log(m-n+1)/log(2)).

(The non-temporary two table could be omitted by replacing two with (select null foo union all select null) in the create temporary table seq.)

Answer 3

You appear to be able to construct reasonably large sets with:

select 9 union all select 10 union all select 11 union all select 12 union all select 13 ...

I got a parser stack overflow in the 5300's, on 5.0.51a.

Answer 4

You could try something like this:

select *
from
(
  SELECT @rn:=@rn+1 as id from (select @rn:=2)t, `order` limit 4
) n inner join `order` using (id)

Answer 5

If you were using Oracle, 'pipelined functions' would be the way to go. Unfortunately, MySQL has no such construct.

Depending on the the scale of the numbers you want sets of, I see two simple ways to go : you either populate a temporary table with just the numbers you need (possibly using memory tables populated by a stored procedure) for a single query or, up front, you build a big table that counts from 1 to 1,000,000 and select bounded regions of it.

Answer 6

Warning: if you insert numbers one row at a time, you'll end up executing N commands where N is the number of rows you need to insert.

You can get this down to O(log N) by using a temporary table (see below for inserting numbers from 10000 to 10699):

mysql> CREATE TABLE `tmp_keys` (`k` INTEGER UNSIGNED, PRIMARY KEY (`k`));
Query OK, 0 rows affected (0.11 sec)

mysql> INSERT INTO `tmp_keys` VALUES (0),(1),(2),(3),(4),(5),(6),(7);
Query OK, 8 rows affected (0.03 sec)
Records: 8  Duplicates: 0  Warnings: 0

mysql> INSERT INTO `tmp_keys` SELECT k+8 from `tmp_keys`;
Query OK, 8 rows affected (0.02 sec)
Records: 8  Duplicates: 0  Warnings: 0

mysql> INSERT INTO `tmp_keys` SELECT k+16 from `tmp_keys`;
Query OK, 16 rows affected (0.03 sec)
Records: 16  Duplicates: 0  Warnings: 0

mysql> INSERT INTO `tmp_keys` SELECT k+32 from `tmp_keys`;
Query OK, 32 rows affected (0.03 sec)
Records: 32  Duplicates: 0  Warnings: 0

mysql> INSERT INTO `tmp_keys` SELECT k+64 from `tmp_keys`;
Query OK, 64 rows affected (0.03 sec)
Records: 64  Duplicates: 0  Warnings: 0

mysql> INSERT INTO `tmp_keys` SELECT k+128 from `tmp_keys`;
Query OK, 128 rows affected (0.05 sec)
Records: 128  Duplicates: 0  Warnings: 0

mysql> INSERT INTO `tmp_keys` SELECT k+256 from `tmp_keys`;
Query OK, 256 rows affected (0.03 sec)
Records: 256  Duplicates: 0  Warnings: 0

mysql> INSERT INTO `tmp_keys` SELECT k+512 from `tmp_keys`;
Query OK, 512 rows affected (0.11 sec)
Records: 512  Duplicates: 0  Warnings: 0

mysql> INSERT INTO inttable SELECT k+10000 FROM `tmp_keys` WHERE k<700;
Query OK, 700 rows affected (0.16 sec)
Records: 700  Duplicates: 0  Warnings: 0

edit: fyi, unfortunately this won't work with a true temporary table with MySQL 5.0 as it can't insert into itself (you could bounce back and forth between two temporary tables).

edit: You could use a MEMORY storage engine to prevent this from actually being a drain on the "real" database. I wonder if someone has developed a "NUMBERS" virtual storage engine to instantiate virtual storage to create sequences such as this. (alas, nonportable outside MySQL)

Answer 7

I found this solution on the web

SELECT @row := @row + 1 as row, t.* FROM some_table t, (SELECT @row := 0) r

Single query, fast, and does exactly what I wanted: now I can "number" the "selections" found from a complex query with unique numbers starting at 1 and incrementing once for each row in the result.

I think this will also work for the issue listed above: adjust the initial starting value for @row and add a limit clause to set the maximum.

BTW: I think that the "r" is not really needed.

ddsp

Answer 8

The following will return 1..10000 and is not so slow

SELECT @row := @row + 1 as row FROM 
(select 0 union all select 1 union all select 3 union all select 4 union all select 5 union all select 6 union all select 6 union all select 7 union all select 8 union all select 9) t,
(select 0 union all select 1 union all select 3 union all select 4 union all select 5 union all select 6 union all select 6 union all select 7 union all select 8 union all select 9) t2, 
(select 0 union all select 1 union all select 3 union all select 4 union all select 5 union all select 6 union all select 6 union all select 7 union all select 8 union all select 9) t3, 
(select 0 union all select 1 union all select 3 union all select 4 union all select 5 union all select 6 union all select 6 union all select 7 union all select 8 union all select 9) t4, 
(SELECT @row:=0)