DevOps, CI/CD and Automation

Hi,

When you say "CONNECT BY LEVEL <= x" , make sure at least one of the following is true:
(1) the table in the query has only one row, or
(2) you know what you're doing.

In other situations, create a Counter Table that has all the numbers you'll ever need, and join to the Counter Table where you otherwise would do a CONNECT BY.
In the definition for the Counter Table, the CONNECT BY clause will resemble your original CONNECT BY clause, but instead of "CONNECT BY LEVEL <= x" it will be "CONNECT LEVEL <= *(SELECT MAX (x) ...)* ".
When you use the counter table, the join condition will be similar to your original CONNECT BY condition.

WITH	cntr	AS
(
	SELECT	LEVEL	AS n
	FROM	dual
	CONNECT BY	LEVEL <= ( SELECT MAX (LENGTH (a_id))
			      	   FROM	  a
				 )
)
,	split_a_id	AS
(
	SELECT	  a.a_text,
--		  a.a_id,	-- For debugging only
--		  c.n,	-- For debugging only
		  SUM (TO_NUMBER ( SUBSTR ( a.a_id || '0'
	    	       	 	 	  , -c.n
					  , 1
					  )
				  )
	    	      ) OVER ( PARTITION BY  a.a_text
		      	       ORDER BY      c.n
			     )	AS cut_length
	FROM      a     a
	JOIN	  cntr	c	ON	c.n	<= LENGTH (a.a_id)
)
SELECT    SUBSTR ( a_text
	         , 1
	         , LENGTH (a_text) - cut_length
	         ) 	AS cut_text
FROM      split_a_id
ORDER BY  a_text
,     	  cut_text	DESC
;

Now that you're dealing with multiple rows, you need a PARTITION BY clause in the analytic SUM function.

Thanks for including the CREATE TABLE and INSERT statements.
Please format your code, and post it between

 tags so this site won't remove any spaces.
The link was a good idea, but it seems to be a link to the wrong thread.

Edited by: Frank Kulash on Nov 17, 2009 2:46 PM
Commented out 2 line I forgot to delete after testing.