Encoding character type in sql server column

Question

Hello,
I am having troubles during REPLACE character in sql server column due to fact that, it seems to me, SQL Server does not recognize two "the same" characters i.e., some of them it translates into another encoding system correctly while, some of
them, keeps untouched, as entered. Obviously, according to my understanding, they "look the same shape" but, obviously, belong to different encoding systems. How to change encoding system for some of them?
Namely, this UPDATE statement:
UPDATE TOP (4000000) tblName
SET       FirstName = N'Đ' + SUBSTRING(FirstName, 2, 8000)
WHERE (FirstName LIKE N'Ð%')
works fine for my case but, as you see, it replaces just the first occurrence. Could it be possible to expand  this pattern to all occurrences?

Answer 1

I'm not sure that I have fully understood what you want to achieve, but it seems that you want to replace occurrances of Latin Captial Letter Edh with Latin Capital Letter D with Stroke, which looks virtually identical in uppercase (but are entirely different
in lowercase). Or vice versa.
Using the glyphs is certainly bound to cause confusion, and I think the best for clarity is to use the nchar function. Or possibly upper in combination with the lowercase glyphs. As for entering the characters, the Character Map utility, which you find in
the  folder Accessories/System Tools folder in the Start Menu is useful.
Below is a script that demonstrates the techniques above. You can also see that the characters are considered equal in a accent-insensitive collation. (Well, at least Latin1_General. They are of course unequal in Icelandic and Croatian AI collations.)
CREATE TABLE tblName(FirstName nvarchar(20) NOT NULL)
INSERT tblName(FirstName)
VALUES(upper('eðh'))
SELECT FirstName, lower(FirstName)
FROM tblName
UPDATE dbo.tblName
SET FirstName = REPLACE(FirstName, nchar(208), nchar(272))
WHERE FirstName LIKE N'%' + nchar(208) + '%';
SELECT FirstName, lower(FirstName)
FROM tblName
SELECT CASE WHEN nchar(208) = nchar(272) COLLATE Latin1_General_CI_AI THEN 1 ELSE 0 END AS Equal
go
DROP TABLE tblName
Erland Sommarskog, SQL Server MVP, [email protected]