Question

For a SQL Server 2005 Instance what is the best way to take find/replace a column with multiple emails like

<[e-mail]>; [e-mail]; [e-mail] .....

and replace it with

<[e-mail]>; [e-mail]; [e-mail] .....

For testing purposes, I could think of some ways to do this in C# but I was wondering if there was a think way to do it in SQL Server, maybe with REGEX ? I want to keep the random weirdness as much as possible (some emails have brackets, some have semicolons at the end etc …)

Thanks

Answer 1

Here you go, you can accomplish it with a cte in a function.

create function dbo.FixupEmails(@s varchar(8000))
returns table
as
return (
      WITH splitter_cte AS (
      SELECT CHARINDEX(';', @s) as pos, 0 as lastPos, 1 as cte_level
      UNION ALL
      SELECT CHARINDEX(';', @s, pos + 1), pos, cte_level + 1 as cte_level
      FROM splitter_cte
      WHERE pos > 0
      ), each_email_cte AS(
      select replace(replace(replace(OneEmail, '>', ''), '<', ''), ' ', '') as OneEmail, cte_level
        from (select SUBSTRING(@s, lastPos + 1,
                         case when pos = 0 then 80000 else pos - lastPos -1 end) as OneEmail,
                         cte_level
                from splitter_cte) as t
      ), each_half_cte AS (
        select OneEmail, CHARINDEX('@', OneEmail) as atPos, cte_level
        from each_email_cte
        where len(OneEmail) > 6  -- 6 from [e-mail] (I think that 6 would be the minimum valid email length)
      ), new_email_cte as
      (
        select cte1.OneEmail, Replace(@s, cte1.OneEmail, '[e-mail]') as New, cte1.cte_level --, 1 as level
        from each_half_cte cte1
        where cte1.cte_level = 1

        UNION ALL

        select cte2.OneEmail, Replace(necte.New, cte2.OneEmail, '[e-mail]') as New, cte2.cte_level--, 1 as level
        from new_email_cte as necte
        inner join each_half_cte as cte2 on cte2.cte_level = necte.cte_level + 1


      )
      select New
      from new_email_cte
      where cte_level = (select max(cte_level) from new_email_cte)
)
go

set nocount on;

declare @emailString varchar(2048)
set @emailString = '<[e-mail]>; [e-mail]; [e-mail] ';
select @emailString as Original;
SELECT *
  FROM dbo.FixupEmails(@emailString);




set @emailString = '<[e-mail]>; [e-mail];';
select @emailString as Original;
SELECT *
  FROM dbo.FixupEmails(@emailString);


set @emailString = '<[e-mail]>';
select @emailString as Original;
SELECT *
  FROM dbo.FixupEmails(@emailString)
OPTION(MAXRECURSION 0);
-- include MAXRECURSION as shown above if you have more than 100 email addresses in the field.



set @emailString = '<[e-mail]>; [e-mail];';
select @emailString as Original;
SELECT *
  FROM dbo.FixupEmails(@emailString)

It is kind of long, but here is the output.

Original
----------------------------------------------------------------
<[e-mail]>; [e-mail]; [e-mail] 

New
-----------------------------------------------------------------
<[e-mail]>; [e-mail]; [e-mail] 




Original
----------------------------------------------------------------
<[e-mail]>; [e-mail];

New
----------------------------------------------------------------
<[e-mail]>; [e-mail];





Original
----------------------------------------------------------------
<[e-mail]>

New
----------------------------------------------------------------
<[e-mail]>





Original
----------------------------------------------------------------
<[e-mail]>; [e-mail];

New
----------------------------------------------------------------
<[e-mail]>; [e-mail];

This was lots of fun. I think that the function provided will do what you are looking for.