Using T-SQL XML input element values for an IN Clause - sql-server

Question

I'm trying to write a procedure to sync users from Active directory into my local application database. From my code, I'm passing XML in the following format to the stored procedure:
<AdUsers>
<AdUser AccountSid="S-1-5-21-111111111-111111111-111111111-1111" DisplayName="Test User" EmailAddress="tuser#mail.local" ExchangeServerFk="4" ExchangeServer="https://mail.local" Department="" StatusFK="1" UserName="TUSER">
<AccountSids>
<Sid>S-1-5-21-111111111-111111111-111111111-1111</Sid>
</AccountSids>
</AdUser>
</AdUsers>
I'd like to do a sync between the XML and the rows in my tb_Mailboxes table with the following Stored Procedure:
#adUsers XML, #lastSyncBy VARCHAR (50), #lastSyncOn DATETIME, #defaultProfileId INT, #adDomainId INT
AS
begin try
BEGIN TRANSACTION
--First delete all the mailboxes exist in the database but not in the xml.
delete tb_Mailboxes
where AccountSid not in (
select
rtrim(element.value('text()[1]', 'varchar(100)')) as AccountSid
from
#adUsers.nodes('/AdUsers/AdUser/AccountSids/Sid') t(element)
) AND #adDomainId = AdDomainFk
--Then insert or update existing accounts
MERGE tb_Mailboxes as [target]
USING
(
select
rtrim(element.value('data(#AccountSid)', 'varchar(100)')) as AccountSid
,rtrim(element.value('data(#DisplayName)', 'varchar(100)')) as DisplayName
,rtrim(element.value('data(#EmailAddress)', 'varchar(500)')) as EmailAddress
,rtrim(element.value('data(#ExchangeServerFk)', 'varchar(100)')) as ExchangeServerFk
,rtrim(element.value('data(#ExchangeServer)', 'varchar(150)')) as ExchangeServer
,rtrim(element.value('data(#Department)', 'varchar(100)')) as Department
,rtrim(element.value('data(#StatusFK)', 'varchar(100)')) as StatusFK
,rtrim(element.value('data(#UserName)', 'varchar(100)')) as UserName
,element.query('AccountSids') as SidList
from
#adUsers.nodes('/AdUsers/AdUser') t(element)
) as [source]
on [target].AccountSid IN
(
SELECT rtrim(A.value('text()[1]', 'varchar(100)')) as CurSid
FROM [source].SidList.nodes('Sid') AS FN(A)
)
WHEN MATCHED THEN UPDATE SET
DisplayName = [source].DisplayName
,EmailAddress = [source].EmailAddress
,ExchangeServerFk = [source].ExchangeServerFk
,ExchangeServer = [source].ExchangeServer
,Department = [source].Department
,UserName = [source].UserName
/*,StatusFK = [source].StatusFK*/
,LastSyncOn = #lastSyncOn
,LastSyncBy = #lastSyncBy
WHEN NOT MATCHED THEN INSERT
(
AdDomainFk,
UserName,
DisplayName,
Department,
EmailAddress,
ExchangeServerFk,
ExchangeServer,
AccountSid,
IsAutoDeleteEnabled,
ProfileFk,
Settings,
QueueLastPickedUp,
QueueLastProcessed,
QueueLastFinished,
LastSyncOn,
LastSyncBy,
StatusFK
)
VALUES
(
#adDomainId
,[source].UserName
,[source].DisplayName
,[source].Department
,[source].EmailAddress
,[source].ExchangeServerFk
,[source].ExchangeServer
,[source].AccountSid
,0
,#defaultProfileId
,NULL
,NULL
,NULL
,NULL
,#lastSyncOn
,#lastSyncBy
,[source].StatusFK
);
COMMIT TRANSACTION
END TRY
BEGIN CATCH
ROLLBACK TRANSACTION
END CATCH
However, the "NOT IN" in the delete section AND "IN" in the match section don't seem to work. Is this type of IN clause using multiple values in the XML even feasible? Is there a better approach to this problem that I'm missing?

Answer 1

The issue with your MERGE query is the join between the [source] and the [target] tables. Rather than joining the target and source tables using
ON [target].AccountSid IN
(
SELECT rtrim(A.value('text()[1]', 'varchar(100)')) as CurSid
FROM [source].SidList.nodes('Sid') AS FN(A)
)
use this instead:
ON [target].AccountSid = [source].AccountSid
[source] will be materialised as a table and you join to it like you would any other table. Your IN statement doesn't make much sense as it is a completely different entity so will equivalent to a kind of Cartesian join (FULL OUTER).
Another comment I would make is why the separate DELETE statement to remove mailboxes that no longer exist in the XML? Why not simply put the DELETE within the MERGE statement by using the following statement?
WHEN NOT MATCHED BY SOURCE
THEN DELETE
Applying all this, your MERGE statement becomes:
MERGE tb_Mailboxes AS [target]
USING
(SELECT RTRIM(element.value('data(#AccountSid)', 'varchar(100)')) AS AccountSid
, RTRIM(element.value('data(#DisplayName)', 'varchar(100)')) AS DisplayName
, RTRIM(element.value('data(#EmailAddress)', 'varchar(500)')) AS EmailAddress
, RTRIM(element.value('data(#ExchangeServerFk)', 'varchar(100)')) AS ExchangeServerFk
, RTRIM(element.value('data(#ExchangeServer)', 'varchar(150)')) AS ExchangeServer
, RTRIM(element.value('data(#Department)', 'varchar(100)')) AS Department
, RTRIM(element.value('data(#StatusFK)', 'varchar(100)')) AS StatusFK
, RTRIM(element.value('data(#UserName)', 'varchar(100)')) AS UserName
FROM #adUsers.nodes('/AdUsers/AdUser') t (element)) AS [source]
ON [target].AccountSid = [source].AccountSid
WHEN MATCHED
THEN UPDATE
SET DisplayName = [source].DisplayName
, EmailAddress = [source].EmailAddress
, ExchangeServerFk = [source].ExchangeServerFk
, ExchangeServer = [source].ExchangeServer
, Department = [source].Department
, UserName = [source].UserName
/*,StatusFK = [source].StatusFK*/
, LastSyncOn = #lastSyncOn
, LastSyncBy = #lastSyncBy
WHEN NOT MATCHED BY TARGET
THEN INSERT (AdDomainFk
, UserName
, DisplayName
, Department
, EmailAddress
, ExchangeServerFk
, ExchangeServer
, AccountSid
, IsAutoDeleteEnabled
, ProfileFk
, Settings
, QueueLastPickedUp
, QueueLastProcessed
, QueueLastFinished
, LastSyncOn
, LastSyncBy
, StatusFK)
VALUES (#adDomainId
, [source].UserName
, [source].DisplayName
, [source].Department
, [source].EmailAddress
, [source].ExchangeServerFk
, [source].ExchangeServer
, [source].AccountSid
, 0
, #defaultProfileId
, NULL
, NULL
, NULL
, NULL
, #lastSyncOn
, #lastSyncBy
, [source].StatusFK)
WHEN NOT MATCHED BY SOURCE
THEN DELETE;