CSV into rows

Question

Hi all,
From the example given in the following link I undersatand that we can convert comma seperated values into rows using a single SQL.
http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:73830657104020
But when I try this with multiple rows it gives me duplicate records.
i.e
CREATE TABLE tst1
(problem_no VARCHAR2(10),
cr_tracking_no VARCHAR2(256))
INSERT INTO tst1 VALUES('P1','106,211,243,250')
INSERT INTO tst1 VALUES('P1','1966,22023,38023,53914,56533')
SELECT
FROM           
(SELECT
problem_no,
SUBSTR (cr_tracking_no,
INSTR (cr_tracking_no, ',', 1, LEVEL  ) + 1,
INSTR (cr_tracking_no, ',', 1, LEVEL+1) - INSTR (cr_tracking_no, ',', 1, LEVEL) -1 )
AS token
FROM (
SELECT  problem_no, ','||trim(cr_tracking_no||',') cr_tracking_no
FROM tst1 )
CONNECT BY LEVEL <= LENGTH(cr_tracking_no)-LENGTH(REPLACE(cr_tracking_no,',',''))+1)
WHERE token IS NOT NULL
the desired output is
SQL> SELECT
  2  *
  3  FROM  
  4  (SELECT
  5  distinct problem_no,
  6  SUBSTR (cr_tracking_no,
  7  INSTR (cr_tracking_no, ',', 1, LEVEL  ) + 1,
  8  INSTR (cr_tracking_no, ',', 1, LEVEL+1) - INSTR (cr_tracking_no, ',', 1, LEVEL) -1 )
  9  AS token
10  FROM (
11  SELECT  problem_no, ','||trim(cr_tracking_no||',') cr_tracking_no
12  FROM tst1 )
13  CONNECT BY LEVEL <= LENGTH(cr_tracking_no)-LENGTH(REPLACE(cr_tracking_no,',',''))+1)
14  WHERE token IS NOT NULL
15  /
PROBLEM_NO TOKEN
P1         106
P1         1966
P1         211
P1         22023
P1         243
P1         250
P1         38023
P1         53914
P1         56533
[pre]
The above query return 46 row (Duplicates) , whereas the expected number of record is 9
Kindly tell me what is wrong and how to get the desired rows without using distinct.
Thanks in advance,
Jaggyam
null
null                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       

Answer 1

You are absolutely right about not wanting to use the distinct function here.
To make it work, you'll have to shortcut the hierarchy loop like this:
SQL> SELECT *
  2    FROM ( SELECT problem_no
  3                , SUBSTR
  4                  ( cr_tracking_no
  5                  , INSTR (cr_tracking_no, ',', 1, LEVEL  ) + 1
  6                  , INSTR (cr_tracking_no, ',', 1, LEVEL+1) - INSTR (cr_tracking_no, ',', 1, LEVEL) -1
  7                  ) AS token
  8             FROM ( SELECT problem_no
  9                         , ',' || trim(cr_tracking_no) || ',' cr_tracking_no
10                         , rownum rn
11                      FROM tst1
12                  )
13          CONNECT BY LEVEL <= LENGTH(cr_tracking_no)-LENGTH(REPLACE(cr_tracking_no,','))-1
14              AND connect_by_root rn = rn
15        )
16   WHERE token IS NOT NULL
17  /
PROBLEM_NO TOKEN
P1         106
P1         211
P1         243
P1         250
P1         1966
P1         22023
P1         38023
P1         53914
P1         56533
9 rows selected.More techniques for doing this can be found here:
http://rwijk.blogspot.com/2007/11/interval-based-row-generation.html
Regards,
Rob.