T-SQL: Splitting character values between non-consistent instances of delimiters - sql-server

Question

There is a string accompanying a value that I need to extract from a column. I can extract the value from most of the rows, but there are a few cases where the value has different properties. This is a simplified example of the problem;
IF OBJECT_ID('TEMPDB..#TABLE') IS NOT NULL
DROP TABLE #TABLE
CREATE TABLE #TABLE(
colSTRING NVARCHAR(MAX) NULL
);
INSERT INTO #TABLE (colSTRING)
VALUES (',SHOULD NOT BE STORED THIS WAY:22.67')
,(',SHOULD NOT BE STORED THIS WAY:46.32')
,(',SHOULD NOT BE STORED THIS WAY:23.45')
,(',SHOULD NOT BE STORED THIS WAY:66.67')
,(',SHOULD NOT BE STORED THIS WAY:22.35,ANOTHER BAD THING:OK')
;
SELECT * FROM #TABLE
Output:
Notice that there is a number at the end of the string to the right of the ':'. This is the number I need to extract.
The bottom row however shows that there is a second string entry in the same cell. I need to extract 22.35 from this cell while omitting the rest of the string.
This is what I have so far;
SELECT
(RIGHT(colSTRING,CHARINDEX(':',REVERSE(colSTRING))-1)) [STRING NUMBER]
FROM #TABLE
output:
This works for the other values in the table, but the bottom row does not extract the correct value. It takes the string to the right of the ':' of the second string entry.
Is there some way to use this logic on only the first occurrence of the ':'?

Answer 1

So this is how I solved the problem, thanks to #MartinSmith 's idea. I adjusted the example a bit to show how this interacts with a number with more than 2 digits (>=100.00).
IF OBJECT_ID('TEMPDB..#TABLE') IS NOT NULL
DROP TABLE #TABLE
CREATE TABLE #TABLE(
colSTRING NVARCHAR(MAX) NULL
);
INSERT INTO #TABLE (colSTRING)
VALUES (',SHOULD NOT BE STORED THIS WAY:22.67')
,(',SHOULD NOT BE STORED THIS WAY:46.32')
,(',SHOULD NOT BE STORED THIS WAY:23.45')
,(',SHOULD NOT BE STORED THIS WAY:766.67')
,(',SHOULD NOT BE STORED THIS WAY:22.35,ANOTHER BAD THING:OK')
;
SELECT * FROM #TABLE
Solution: In this case, every string entry always starts with a comma. I can use that information in a CASE statement. I make a column with populated entries for each case when there are numbers <100.00 or >=100.00
SELECT ISNULL(CASE WHEN [2DIGITS] LIKE ',%' THEN NULL ELSE [2DIGITS] END,[3DIGITS]) [FIXED]
FROM(
SELECT
(RIGHT(colSTRING,CHARINDEX(':',REVERSE(colSTRING))-1)) [STRING NUMBER]
,SUBSTRING(colSTRING,1 + PATINDEX('%:[0-9][0-9].[0-9][0-9]%', colSTRING),5) [2DIGITS]
,SUBSTRING(colSTRING,1 + PATINDEX('%:[0-9][0-9][0-9].[0-9][0-9]%', colSTRING),6) [3DIGITS]
FROM #TABLE
)A