Replace unicode number between to different delimiters with NCHAR function - sql-server

Question

I've created a MSSQL Server function which encodes special chars (example: हिन्दीabcde fG#) to the unicode number with "#" and ";" as delimiter. Only very simple chars like "abc" will not be encoded:
declare #position int, #txt nvarchar(max), #output as varchar(max);
set #position=1;
set #txt = N'हिन्दीabcde fG#';
set #output = '';
while #position <= len(#txt)
begin
declare #t int;
select #t=unicode(substring(#txt,#position,1))
--print '&#'+ CONVERT(nvarchar(5),#t)+';'
if ( (#t between 48 and 57) OR (#t between 65 and 90) or (#t between 97 and 122) )
BEGIN
SET #output = #output + CONVERT(nvarchar(5), substring(#txt,#position,1) );
END
else
BEGIN
SET #output = #output + '#'+ CONVERT(nvarchar(5),#t)+';'
END
set #position = #position+1
end
Print #output
The result is:
2361;#2367;#2344;#2381;#2342;#2368;abcde#32;fG#35;
I need it for working with ODBC drivers and to avoid problems with special chars.
But now I need the way back - to decode the encoded chars. Is there any smart solution or will I need at least two loops, the "NCHAR" function ...?
I'll try to build such a function - if it's successfull, I'll post it here :)

Answer 1

You might find this approach a little more appealing. First, create a split function that maintains order:
CREATE FUNCTION dbo.SplitStringsOrdered
(
#List NVARCHAR(MAX),
#delim NVARCHAR(10)
)
RETURNS TABLE
AS
RETURN
(
SELECT rn, v = LTRIM(RTRIM(SUBSTRING(#List, rn,
CHARINDEX(#delim, #List + #delim, rn) - rn)))
FROM
(
SELECT TOP (8000) rn = ROW_NUMBER() OVER (ORDER BY s1.[object_id])
FROM sys.all_objects AS s1 CROSS JOIN sys.all_objects AS s2
) AS n
WHERE rn <= LEN(#List)
AND SUBSTRING(#delim + #List, rn, LEN(#delim)) = #delim
);
GO
Usage:
DECLARE #x NVARCHAR(MAX) = N'#2361;#2367;#2344;#2381;'
+ N'#2342;#2368;abcde#32;fG#35;';
-- need one small adjustment to make the string more split-friendly:
SET #x = REPLACE(#x, '#', ';#');
DECLARE #output NVARCHAR(MAX);
SELECT #output = (SELECT
CASE WHEN v LIKE '#%' THEN NCHAR(REPLACE(v, '#', '')) ELSE v END
FROM dbo.SplitStringsOrdered(#x, ';') AS x
ORDER BY rn FOR XML PATH(''),
TYPE).value('./text()[1]','nvarchar(max)');
SELECT #output;
Output:
हिन्दीabcde fG#

Answer 2

I've solved the problem with this query:
declare #position int, #txt nvarchar(max), #output as nvarchar(max), #buffer as varchar(max);
set #position=1;
set #txt = '#2361;#2367;#2344;#2381;#2342;#2368;abcde#32;fG#35;';
set #output = '';
set #buffer = '';
while #position <= len(#txt)
begin
declare #t varchar(max);
select #t=(substring(#txt,#position,1))
if ( len(#buffer) = 0 and #t <> '#' and #t <> ';')
BEGIN
-- Append simple chars, which were not encoded
Print 'Hänge den String ganz normal an den Output: ' + #t
SET #output = #output + #t;
END
ELSE
BEGIN
if ( #t = '#' )
BEGIN
Print 'Raute wurde erkannt: #';
SET #buffer = '#';
END
else if ( #t = ';' )
BEGIN
SET #buffer = REPLACE( #buffer, '#' , '' );
Print 'Umwandeln: ' + #buffer
SET #output = #output + isnull( NCHAR(#buffer) , '');
SET #buffer = '';
END
else
BEGIN
Print 'Ganzzahl an den Buffer anhängen: ' + #t;
SET #buffer = #buffer + #t;
END
END
set #position = #position+1
end
Print #output