Multiple Entries in Single Field (seperate data...?)

Question

I have some mess of data and curious on if this can resolved simple from a query.
What I have, is multiple entries in a single cell. All the data is separated by pipes(|).
So, I have data similar to this in a table.(all made up of course :))..
Soda, coke|pepsi|dew|dtcoke,
Computer, dell|hp|gateway,
What I'm looking to do, is to turn these values into separate columns. So the final output would look
soda, coke
soda, pepsi
soda, dew
soda, dtcoke
computer, dell
computer, hp
computer, gateway
Any suggestions. And, sorry for the hefty examples. Figured more the better....
Edited by: Topher34 on Sep 5, 2008 9:20 AM

Answer 1

Hi,
Topher34 wrote:
With that, ...With what? There are a lot of queries in the thread to which C linked, not to mention links from there. If you need an explanation of any of them, post a copy.
I think this is what you want:
WITH     cntr     AS
     SELECT     LEVEL     AS n
     FROM     dual
     CONNECT BY     LEVEL <= 1 +
                    (     SELECT     MAX (REGEXP_COUNT (b_list, '[|]+'))
                         FROM     table_x
SELECT     a
,     b_list
,     REGEXP_SUBSTR     ( b_list
               , '[^|]+'
               , 1
               , n
               )     AS b
FROM     table_x
JOIN     cntr
ON     n     <= 1 + REGEXP_COUNT (b_list, '[|]+')
ORDER BY     a
,          n;In Oracle 11, this produces the following output from the sample data you posted (that is, the first two distinct columns):
A          B_LIST                         B
Computer   dell|hp|gateway                dell
Computer   dell|hp|gateway                hp
Computer   dell|hp|gateway                gateway
Soda       coke|pepsi|dew|dtcoke          coke
Soda       coke|pepsi|dew|dtcoke          pepsi
Soda       coke|pepsi|dew|dtcoke          dew
Soda       coke|pepsi|dew|dtcoke          dtcokeREGEXP_COUNT is new in Oracle 11, but here's how to adapt it for Oracle 10:
The solution above finds the number of entries in b_list (entries delimited by one or more consecutive | characters) like this:
1 + REGEXP_COUNT (b_list, '[|]+')In Oracle 10, you can get the number of entries delimited by a single | character like this:
1 + LENGTH (b_list) - LENGTH (REPLACE (b_list, '|', NULL))The results will be the same if b_list does not contain consecutive | characters. Otherwise, you may want to test for NULL values in column b.
Edited by: Frank Kulash on Sep 30, 2008 3:32 PM
Originally posted with error.
In the Oracle 10 work-around, when this expression
1 + LENGTH (b_list) - LENGTH (REPLACE (b_list, '|', NULL))is used in sub-query cntr, it is to find the worst case across the whole table, so it should be the argument of MAX. Say
MAX (1 + LENGTH (b_list) - LENGTH (REPLACE (b_list, '|', NULL)))or
1 + MAX (LENGTH (b_list) - LENGTH (REPLACE (b_list, '|', NULL)))In the main query, it applies just to the single value in the current row, so you should not use MAX there.