嗯，这和之前的完全不一样了...

Well, this is quite different to the one before...

看起来，您希望将 <D> 节点放在任何位置.这有帮助吗?

As it looks, you want to take <D> nodes whereever they are placed. Does this help?

WITH Tally(Nmbr) AS
(
    SELECT TOP (SELECT @xmlVar.value(N'count(//D)','int'))
           ROW_NUMBER() OVER(ORDER BY (SELECT NULL)) 
    FROM master..spt_values --just a pre-filled table with many rows
)
SELECT Nmbr
      ,e.value(N'@actioncode[1]','nvarchar(max)') AS Employee
FROM Tally
OUTER APPLY(SELECT  @xmlVar.query(N'//D').query(N'/D[sql:column("Nmbr")]')) AS B(ds)
OUTER APPLY ds.nodes(N'D/E') AS C(e);

第一个应用程序将使用深度搜索(使用//D)来获取仅包含这些节点的派生 XML.其余的和以前一样.

The first apply will use the deep search (with //D) to get a derived XML including these nodes only. The rest is rather the same as before.