Replacing characters in a string when between double quotes - sql

Question

I would like to replace commas in my strings, between dynamic positions (ie. between double quotes). Note that I will not have more than 2 occurrences of double quotes in my strings if that matters.
My example:
'randomtext,123,"JEAN SEBASTIEN, GUY, DANIEL",sun'
Desired output:
'randomtext,123,"JEAN SEBASTIEN GUY DANIEL",sun'
So far I've tried things with REGEXP_REPLACE() mixed with INSTR() but could not get anything done.
Cheers

Answer 1

Short & clean.
with t(str) as (select 'randomtext,123,"JEAN SEBASTIEN, GUY, DANIEL",sun' from dual)
select regexp_replace(str,'(^[^"]*|[^"]*$)|,','\1') as result
from t
-
+------------------------------------------------+
| RESULT |
+------------------------------------------------+
| randomtext,123,"JEAN SEBASTIEN GUY DANIEL",sun |
+------------------------------------------------+
SQL Fiddle
In addition -
Short and clean generic version
with t(str) as
(
select 'Well,you,went,uptown,riding,in,your,limousine' from dual
union all select 'With,your,fine,"Park, Avenue, clothes"' from dual
union all select 'You,had,the,"Dom, Perignon",in,your,hand,"And, the, spoon",up,your,nose' from dual
union all select '"And, when, you",wake,"up, in, the, morning"' from dual
union all select '"With, your, head, on, fire"' from dual
union all select '"And",your,"eyes, too, bloody","to, see",Go,"on, and, cry, in",your,coffee,"But","don''t","come, bitchin''","to, me"' from dual
)
select regexp_replace(str, '((^|").*?("|$))|,', '\1') as result
from t
--
+------------------------------------------------------------------------------------------------------------+
| RESULT |
+------------------------------------------------------------------------------------------------------------+
| Well,you,went,uptown,riding,in,your,limousine |
| With,your,fine,"Park Avenue clothes" |
| You,had,the,"Dom Perignon",in,your,hand,"And the spoon",up,your,nose |
| "And when you",wake,"up in the morning" |
| "With your head on fire" |
| "And",your,"eyes too bloody","to see",Go,"on and cry in",your,coffee,"But","don't","come bitchin'","to me" |
+------------------------------------------------------------------------------------------------------------+
SQL Fiddle

Answer 2

Assuming you are working on CSV, then it is possible that you will also have nested double quotes as per this sample data:
CREATE TABLE test_data ( value ) AS
SELECT 'randomtext,123,"JEAN SEBASTIEN, GUY, DANIEL",sun' FROM DUAL UNION ALL
SELECT 'randomtext,123,"A, ""BC"", D",sun' FROM DUAL;
You can use the regular expression ^(.*?)("([^\"]|\\")+")(.*)$ to match the terms before, inside the quotes and after and then replace commas in only the middle parts:
SELECT value,
REGEXP_SUBSTR( value, '^(.*?)("([^\"]|"")+")(.*)$', 1, 1, NULL, 1 )
|| REPLACE(
REGEXP_SUBSTR( value, '^(.*?)("([^\"]|"")+")(.*)$', 1, 1, NULL, 2 ),
','
)
|| REGEXP_SUBSTR( value, '^(.*?)("([^\"]|"")+")(.*)$', 1, 1, NULL, 4 ) replaced_value
FROM test_data
Which outputs:
VALUE | REPLACED_VALUE
:----------------------------------------------- | :---------------------------------------------
randomtext,123,"JEAN SEBASTIEN, GUY, DANIEL",sun | randomtext,123,"JEAN SEBASTIEN GUY DANIEL",sun
randomtext,123,"A, ""BC"", D",sun | randomtext,123,"A ""BC"" D",sun
db<>fiddle here
Update
If you need to handle multiple quoted terms in a string (with nested quotes):
CREATE TABLE test_data ( value ) AS
SELECT 'randomtext,123,"JEAN SEBASTIEN, GUY, DANIEL",sun' FROM DUAL UNION ALL
SELECT 'randomtext,123,"A, ""BC"", D",sun' FROM DUAL UNION ALL
SELECT 'E,"F, G",H,"I, ""J""", K' FROM DUAL UNION ALL
SELECT 'L,M,N' FROM DUAL;
Then you can use a recursive sub-query factoring clause:
WITH replacements( value, prefix, suffix ) AS (
SELECT value,
REGEXP_SUBSTR( value, '^(.*?)("([^\"]|"")+"|$)(.*)$', 1, 1, NULL, 1 )
|| REPLACE(
REGEXP_SUBSTR( value, '^(.*?)("([^\"]|"")+"|$)(.*)$', 1, 1, NULL, 2 ),
','
),
REGEXP_SUBSTR( value, '^(.*?)("([^\"]|"")+"|$)(.*)$', 1, 1, NULL, 4 )
FROM test_data
UNION ALL
SELECT value,
prefix
|| REGEXP_SUBSTR( suffix, '^(.*?)("([^\"]|"")+"|$)(.*)$', 1, 1, NULL, 1 )
|| REPLACE(
REGEXP_SUBSTR( suffix, '^(.*?)("([^\"]|"")+"|$)(.*)$', 1, 1, NULL, 2 ),
','
),
REGEXP_SUBSTR( suffix, '^(.*?)("([^\"]|"")+"|$)(.*)$', 1, 1, NULL, 4 )
FROM replacements
WHERE suffix IS NOT NULL
)
SELECT value,
prefix AS replaced_value
FROM replacements
WHERE suffix IS NULL;
Which outputs:
VALUE | REPLACED_VALUE
:----------------------------------------------- | :---------------------------------------------
L,M,N | L,M,N
randomtext,123,"JEAN SEBASTIEN, GUY, DANIEL",sun | randomtext,123,"JEAN SEBASTIEN GUY DANIEL",sun
randomtext,123,"A, ""BC"", D",sun | randomtext,123,"A ""BC"" D",sun
E,"F, G",H,"I, ""J""", K | E,"F G",H,"I ""J""", K
db<>fiddle here