SQL Email Verification Function using Regex - sql

Question

I am trying to create a function in SQL Server 2005 to check to see if an email is in a valid format with regular expressions.
Here is what I have so far:
CREATE FUNCTION isValidEmailFormat
(
#Email varchar(100)
)
RETURNS bit
AS
BEGIN
DECLARE #Result bit
SET #Result = (SELECT CASE
WHEN #Email LIKE '%[a-zA-Z0-9_\-]+#([a-zA-Z0-9_\-]+\.)+ (com|org|edu|nz|au])%'
THEN 1
ELSE 0
END AS Valid)
RETURN #Result
END
Am I doing something wrong with my regular expression? Or do I have to do something more to compare a varchar to a regular expression?
-Edit-
Right now, whatever string I put in returns a 0, even if the email format itself is correct.

Answer 1

The short answer is that no, it cannot be done. The syntax for LIKE is not the same as (and way less powerful than) regular expressions. See also SQL Server Regular expressions in T-SQL
But you can make the hop to .Net and do the matching there. You can instantiate a VBScript.RegExp inside T-SQL using sp_OACreate and use that.
CREATE FUNCTION dbo.isValidEmailFormat
(
#Email varchar(100)
)
RETURNS bit
AS
BEGIN
DECLARE #pattern varchar(4000)
SET #pattern = '[a-zA-Z0-9_\-]+#([a-zA-Z0-9_\-]+\.)+(com|org|edu|nz|au)'
DECLARE #Result bit
DECLARE #objRegexExp INT
EXEC sp_OACreate 'VBScript.RegExp', #objRegexExp OUT
EXEC sp_OASetProperty #objRegexExp, 'Pattern', #pattern
EXEC sp_OASetProperty #objRegexExp, 'IgnoreCase', 1
EXEC sp_OASetProperty #objRegexExp, 'MultiLine', 0
EXEC sp_OASetProperty #objRegexExp, 'Global', false
EXEC sp_OASetProperty #objRegexExp, 'CultureInvariant', true
EXEC sp_OAMethod #objRegexExp, 'Test', #Result OUT, #Email
EXEC sp_OADestroy #objRegexExp
RETURN #Result
END
Do take a peek at Regex email verification error - using JavaScript to see if you want to be a bit less restrictive on what characters are allowed.

Answer 2

I used #flup 's answer and updated the RegExp pattern as follows:
dots are allowed in the portion before #, but not as first or last character there
the expression is anchored to start and end of string with ^ and $
specific TLDs and country codes were added for the data I was using
The TLDs are whitelisted explicitly to exclude many typos where .com was entered as .co or .cm -- valid country codes for Colombia or Cameroon but not for my data.
The updated line:
SET #pattern = '^([a-zA-Z0-9_\-]+\.)*[a-zA-Z0-9_\-]+#([a-zA-Z0-9_\-]+\.)+(com|org|edu|net|ca|au|coop|de|ee|es|fm|fr|gr|ie|in|it|jp|me|nl|nu|ru|uk|us|za)$'
I hope this adds value to an already-excellent answer.