Replace strings between two characters using T-SQL - sql

Question

I need to update a string to amend any aliases - which can be 'H1.', 'H2.', 'H3.'... etc - to all be 'S.' and am struggling to work out the logic.
For example I have this:
'H1.HUB_CUST_ID, H2.HUB_SALE_ID, H3.HUB_LOC_ID'
But I want this:
'S.HUB_CUST_ID, S.HUB_SALE_ID, S.HUB_LOC_ID'
If you could use wildcards in REPLACE, I'd do something like this REPLACE(#string, 'H%.H', 'S.H').
Theoretically, there is no limit to how many H# aliases there could be. In practice there will almost definitely be less than 10.
Is there a better way than a nested replace of H1 - H10 separately, which both looks messy in a script and carries a small risk if more tables are joined in future?

Answer 1

SQL Server doesn't support pattern replacement. You are better off using a different language, that does support pattern/REGEX replacement or implementing a CLR function.
That said, however, considering you said that the value would always be below 10 you could brute force it, but it's not "pretty".
SELECT REPLACE(REPLACE(REPLACE(REPLACE(REPLACE(REPLACE(REPLACE(REPLACE(REPLACE(YourString,'H1.','S.'),'H2.','S.'),'H3.','S.'),'H4.','S.'),'H5.','S.'),'H6.','S.'),'H7.','S.'),'H8.','S.'),'H9.','S.')
FROM YourTable ...

Answer 2

You can convert your string to XML and then convert it into simple table:
DECLARE #txt nvarchar(max) = N'H1.HUB_CUST_ID, H2.HUB_SALE_ID, H3.HUB_LOC_ID',
#x xml
SELECT #x = '<a al="' + REPLACE(REPLACE(#txt,', ','</a><a al="'),'.','">')+ '</a>'
SELECT t.c.value('#al', 'nvarchar(max)') as alias_name,
t.c.value('.','nvarchar(max)') as col_name
FROM #x.nodes('/a') t(c)
Output:
alias_name col_name
H1 HUB_CUST_ID
H2 HUB_SALE_ID
H3 HUB_LOC_ID
You can put results into temp table, amend them using LIKE 'some basic pattern' and then build new string.

Answer 3

If you don't care about the result order, you can unaggregate and reaggregate:
select t.*, v.new_val
from t cross apply
(select string_agg(concat('S1', stuff(s.value, 1, charindex('.'), '') - 1, ',') within group (order by (select null) as newval
from string_split(t.col, ',') s
) s;
Note: This assumes that all values start with the prefix you want to replace -- as your sample data suggests. A case expression can be used if there are exceptions.
You can actually get the original ordering -- assuming no duplicates -- using charindex():
select t.*, v.new_val
from t cross apply
(select string_agg(concat('S1', stuff(s.value, 1, charindex('.'), '') - 1, ',')
within group (order by charindex(s.value, t.col)
) as newval
from string_split(t.col, ',') s
) s;