Large string into the VARCHAR2 field parsing problem

Question

Guys,
SQL> select * from v$version;
BANNER
Oracle Database 10g Enterprise Edition Release 10.2.0.4.0 - 64bi
PL/SQL Release 10.2.0.4.0 - Production
CORE 10.2.0.4.0 Production
TNS for Solaris: Version 10.2.0.4.0 - Production
NLSRTL Version 10.2.0.4.0 - Production
My problem is: The frontend of the application sends a large string into the VARCHAR2 (4000 BYTES) field.
The string it sends is somewhat like the samples given below. I need to parse and put them in the tables
as shown below. Please let me know, if you have any questions. -
Crit: Some text1.
Proficient (points 2): Some more text1.
Comment1...........................................
..Comment1
Crit: Some text2.
Non-performance (points 0): Some more text2.
Crit: Some text3.
Distinguished (points 3): Some more text3.
Crit: Some text4.
Distinguished (points 3): Some more text4.
Table 1
Crit          points          comment        morecomment
Some text1.     2           Some more text1. Comment1.............................Comment1
Some text2.     0           Some more text2.
Some text3.     3           Some more text3.
Some text4.     3           Some more text4.
Table 2
Overall Comment
Crit: Some text1.
Proficient (points 2): Some more text1.
Crit: Some text2.
Non-performance (points 0): Some more text2.
Crit: Some text3.
Distinguished (points 3): Some more text3.
Crit: Some text4.
Distinguished (points 3): Some more text4.
Overall comments:
Blah2 Blah2 Blah2!
Some more blah2 blah2 blah2.
Table 1
Crit          points          comment        morecomment
Some text1.     2           Some more text1.
Some text2.     0           Some more text2.
Some text3.     3           Some more text3.
Some text4.     3           Some more text4.
Table 2
Overall Comment
Blah2 Blah2 Blah2!
Some more blah2 blah2 blah2.
Crit: Some text1.
Proficient (points 2): Some more text1.
Comment1...........................................
..Comment1.
Crit: Some text2.
Non-performance (points 0): Some more text2.
Comment2...............................Comment2.
Crit: Some text3.
Distinguished (points 3): Some more text3.
Crit: Some text4.
Distinguished (points 3): Some more text4.
Overall comments:
Blah1 Blah1 Blah1!Some more blah1 blah1 blah1.
Table 1
Crit          points          comment        morecomment
Some text1.     2           Some more text1. Comment1.............................Comment1
Some text2.     0           Some more text2. Comment2...............................Comment2.
Some text3.     3           Some more text3.
Some text4.     3           Some more text4.
Table 2
Overall Comment
Blah1 Blah1 Blah1!Some more blah1 blah1 blah1.--------------------------------------------------------------------
My code which doesn't work
WITH t AS
(SELECT 'Crit: Some text1.
Proficient (points 2): Some more text1.
Crit: Some text2.
Non-performance (points 0): Some more text2.
Crit: Some text3.
Distinguished (points 3): Some more text3.
Crit: Some text4.
Distinguished (points 3): Some more text4.
Overall comments:
Blah2 Blah2 Blah2!
Some more blah2 blah2 blah2.' string
FROM   dual)
SELECT SUBSTR(REGEXP_SUBSTR(string,'Crit: +$', 1, level), 12) "Crit",
       SUBSTR(REGEXP_SUBSTR(string,'points [^\)]+', 1, level), 8) "points",
       SUBSTR(REGEXP_SUBSTR(string,'\):^[[:alnum:]]+$', 1, level), 3) "comment",
       SUBSTR(REGEXP_SUBSTR(string,'^[[:alnum:]]+$', 1, level), 1) "morecomment"
FROM   t
CONNECT BY LEVEL <= (LENGTH(string) - LENGTH(REPLACE(string, 'Crit:')))/5
WITH t AS
(SELECT 'Crit: Some text1.
Proficient (points 2): Some more text1.
Comment1...........................................
..Comment1
Crit: Some text2.
Non-performance (points 0): Some more text2.
Crit: Some text3.
Distinguished (points 3): Some more text3.
Crit: Some text4.
Distinguished (points 3): Some more text4.' string
FROM   dual)
SELECT SUBSTR(string, INSTR(string, 'Overall comments:', -1) + 18) "overall_comment"
FROM   t
CONNECT BY LEVEL <= (LENGTH(string) - LENGTH(REPLACE(string, 'Crit:')))/5Edited by: abyss on Sep 17, 2010 12:26 PM

Answer 1

Hi,
Except for the morecomment column, you almost had it.
Oracle documentation often refers to ^ and $ as meaning beginnig- and end-of-line, but they seem to mean beginning- and end-of-string.
Aside from that , I just changed a few details (like ' +' to '.+').
SELECT SUBSTR (REGEXP_SUBSTR (string, 'Crit:.+',          1, level), 6, 18)     AS Crit,
       SUBSTR (REGEXP_SUBSTR (string, 'points [^\)]+',           1, level), 8    )     AS points,
       SUBSTR (REGEXP_SUBSTR (string, '\):[ [:alnum:]]+',      1, level), 3     )     AS cmnt,
       REGEXP_REPLACE ( SUBSTR ( REGEXP_SUBSTR ( string || CHR(10) ||'Crit:'
                                                     , '.*?Crit:'
                                      , 1
                                      , LEVEL + 1
                                      , 'n'          -- . matches newline
                          , LENGTH ( REGEXP_SUBSTR ( string
                                                            , 'Crit:.*'
                                           , 1
                                           , LEVEL
                                  ) +
                          LENGTH ( REGEXP_SUBSTR ( string
                                            , 1
                                            , LEVEL
                                  ) - 2
                , 'Crit:$'
                   ) more_cmnt
FROM   t
CONNECT BY LEVEL <= (LENGTH(string) - LENGTH(REPLACE(string, 'Crit:')))/5
; This gets the results you requested from the data you posted. I don't know if it gets those results for the right reasons or not, since you didn't post any reasons for why you want those results. Always explain why you want the results you do, giving specific examples from the sample data.
The morecomment (called more_cmnt above) is much more complicated, since it involves scanning across multiple lines. I wouldn't be surprised if there is a simpler way than what I posted above.
I do not recommend doing this in pure SQL. I suggest a PL/SQL function, because debugging and maintenance would be so much easier.
To populate table2:
WITH     got_pos         AS
     SELECT     string
     ,     INSTR (string, 'Overall comments:', -1)     AS pos
     FROM     t
SELECT  SUBSTR ( string
            , pos + 18
            )     AS overall_comment
FROM    got_pos
WHERE     pos     > 0
;