Regular Expression & LEVEL - how to split attribut value

Question

Hi Folks;
I have to transform the value of an attibut Attr_A (of table A) in multiple attibut's values in a another table B like that :
Table A
Attr_A = '[only one letter from A to Z][only 5 numerics from 0 to 9][space][Operator][space][only one letter from A to Z][only 5 numerics from 0 to 9][space][Operator][space][only one letter from A to Z][only 5 numerics from 0 to 9][space][Operator][space]etc...
with Operator = 'AND' or '(+),' or '(-),'
exemple Attr_A='L12345 AND T23456 (+), U12345 (-)'
In the result table B, I would have :
- first column egal to 'L12345'
- secund column egal to '1' (position of the first column value in the Attr_A)
- third column egal to 'AND'(the operator between secund column value of table B)
Next record :
- first column egal to 'T23456'
- secund to '2'
- third to '(+),'
etc 'U12345' '3' '(-),'
Thanks for your help ^^
Edited by: Moostiq on 27 avr. 2011 10:54
Edited by: Moostiq on 27 avr. 2011 10:54

Answer 1

Hi,
Whenever you post code or data on this this, please format it and type hese 6 characters:
\(small letters only, inside curly brackets) before and after each section of formatted text.  This will keep strings such as (+)
from looking like
(+)
You need to divide attr_a into parts, where each part consists of a 6-character word followed by a space, then followed by an operator.  Since the opertors are at least 3-characters long, that means a string of n characters will have (at most) n/10 parts.  There may be other text in attr_a (in your example, there was a ',') that will be ignored.
The first sub-query below, cntr, generates the numbers 1, 2, 3, ... up to the greatest possible numbr of parts in any attr_a.
The second sub-query, got_part, extracts each part from attr_a.
The main query parses each part into the columns you want.WITH cntr     AS
     SELECT     LEVEL     AS n
     FROM     (
               SELECT MAX (LENGTH (attr_a))
                    / 10     AS max_parts
               FROM     a
     CONNECT BY     LEVEL     <= max_parts
,     got_part     AS
     SELECT     REGEXP_SUBSTR ( a.attr_a
               , '[A-Z][0-9]{5} +(AND|\(\+\)|\(-\))'
               , 1
               , c.n
               )          AS part
     FROM     a
     JOIN     cntr     c ON c.n <= LENGTH (a.attr_a) / 10
SELECT     REGEXP_SUBSTR (part, '[A-Z][0-9]{5}')     AS column_1
,     REGEXP_SUBSTR (part, '[0-9]')          AS column_2
,     REGEXP_SUBSTR (part, '[^ ]+$')     AS column_3
FROM     got_part
WHERE     part     IS NOT NULL