How to remove more than one space in Oracle - sql

Question

I have an Oracle table which contains data like 'Shiv------Shukla' (consider '-' as space).
Now I need to write a program which leaves just one space and removes all other spaces.
Here is the program which I've made but it is not giving me expected result.
DECLARE
MAX_LIMIT VARCHAR2(50):=NULL;
REQ VARCHAR2(20):=NULL;
CURSOR C1 IS
SELECT *
FROM ASSET_Y;
BEGIN
FOR REC IN C1
LOOP
MAX_LIMIT:=LENGTH(REC.NAME)-LENGTH(REPLACE(REC.NAME,'-'));
FOR I IN 1..MAX_LIMIT
LOOP
UPDATE ASSET_Y
SET NAME=REPLACE(REC.NAME,'--','-')
WHERE REC.SNO=ASSET_Y.SNO;
COMMIT;
SELECT ASSET_Y.NAME INTO REQ FROM ASSET_Y WHERE ASSET_Y.SNO=REC.SNO;
DBMS_OUTPUT.PUT_LINE(REQ);
END LOOP;
END LOOP;
COMMIT;
END;
/
My table is
SQL> select * from asset_y;
SNO NAME FL
---------- -------------------- --
1 Shiv------Shukla y
2 Jinesh y
after running the procedure i m getting the following output.
Shiv---Shukla
Shiv---Shukla
Shiv---Shukla
Shiv---Shukla
Shiv---Shukla
Shiv---Shukla
PL/SQL procedure successfully completed.

Answer 1

Since regexp_replace is not available in Oracle 9i maybe you can use owa_pattern routines for simple regex replaces:
owa_pattern.change(fStr, '\s+', ' ', 'g');
More info about owa_pattern package can be found here
Bear in mind, that "\s" will match tabs and newlines as well.

Answer 2

With Oracle 9 you could write your own function:
CREATE FUNCTION remove_multi_spaces( in_value IN VARCHAR2 )
RETURN VARCHAR2
AS
v_result VARCHAR2(32767);
BEGIN
IF( in_value IS NOT NULL ) THEN
FOR i IN 1 .. ( LENGTH(in_value) - 1 ) LOOP
IF( SUBSTR( in_value, i, 2 ) <> ' ' ) THEN
v_result := v_result || SUBSTR( in_value, i, 1 );
END IF;
END LOOP;
v_result := v_result || SUBSTR( in_value, -1 );
END IF;
RETURN v_result;
END;
and call it in a single update-statement:
UPDATE asset_y
SET name = replace_multi_spaces( name );
BTW: With Oracle 10 you could use REGEXP_REPLACE.

Answer 3

Your problem is this part:
SET NAME=REPLACE(REC.NAME,'--','-')
However many times you do that within the inner loop it starts with the same value of REC.NAME as before. Changing it to this would fix it:
SET NAME=REPLACE(NAME,'--','-')
However, it is a pretty inefficient way to do this job if the table is large. You could instead do this:
BEGIN
LOOP
UPDATE ASSET_Y
SET NAME=REPLACE(NAME,'--','-')
WHERE NAME LIKE '%--%';
EXIT WHEN SQL%ROWCOUNT = 0;
END LOOP;
END;
/

Answer 4

Another way:
CREATE OR REPLACE
FUNCTION remove_multi_spaces( in_value IN VARCHAR2 )
RETURN VARCHAR2 IS
v_result VARCHAR2(32767) := in_value;
BEGIN
LOOP
EXIT WHEN INSTR(v_result,' ') = 0;
v_result := REPLACE(v_result, ' ', ' ');
END LOOP;
RETURN v_result;
END remove_multi_spaces;

Answer 5

Ack loops! No need to loop this
This will work in T-SQL...unfortunately I have no pl/sql environment to write this in. PL/SQL will have equivlents to everything used here (I think substr instead of substring and | instead of +)
declare #name varchar(200)
set #name = 'firstword secondword'
select left(#name,(patindex('% %',#name)-1)) + ' ' + ltrim(substring(#name,(patindex('% %',#name)+1),len(#name)))
You'll have to retool it to work for oracle and you'll need to replace any reference to #name to asset_y.name
select left(asset_y.name,(patindex('% %',asset_y.name)-1)) || ' ' || ltrim(substring(asset_y.name,(patindex('% %',asset_y.name)+1),len(asset_y.name)))
Sorry if it won't run as is, as I mentioned I lack an oracle install here to confirm...
Just to add...I normally turn that query above into a function named formatname and call it as select formatname(array_y.name) from... This allows me to include some form of error handling. The query will fail if patindex('% %',array_v.name) returns a null...meaning there is no space. You could do the same in a select statement using cases I guess:
select case when patindex('% %',array_v.name) > 0 then
left(asset_y.name,(patindex('% %',asset_y.name)-1)) || ' ' || ltrim(substring(asset_y.name,(patindex('% %',asset_y.name)+1),len(asset_y.name)))
else asset_y.name
from...