Use a calculated column value to look up a specific value from the same source table - sql

Question

I have a table in SQL Server called ShippingDocSummary which contains all historical records for shipping documents. For a given document number, there can be any number of historical records/transactions. What I am trying to do is create a view that summarizes that table and calculates various column values. My issue comes when trying to get a specific value from the source table based on a calculated column value from the same source table.
Here is a simplified example of what the source table looks like:
DocNum DocHistID StatusCode StatusQty StatusDate
12345 10000001 AS1 2 2/16/2020
12345 10000002 D6T 2 4/20/2020
12345 10000003 COR 2 4/20/2020
12345 10000004 AS1 2 5/5/2020
My code to create my summary so far is as follows:
SELECT
DocNum,
SUM(CASE WHEN ShippingDocSummary.StatusCode LIKE 'AS_' THEN StatusQty ELSE 0 END) AS AS_QTY,
SUM(CASE WHEN ShippingDocSummary.StatusCode LIKE 'D6T' THEN StatusQty ELSE 0 END) AS D6T_QTY,
SUM(CASE WHEN ShippingDocSummary.StatusCode LIKE 'COR' THEN StatusQty ELSE 0 END) AS COR_QTY,
MAX(CASE WHEN ShippingDocSummary.StatusCode LIKE 'AS_' THEN DocHistID ELSE null END) AS LastAS_DHID
FROM ShippingDocSummary
GROUP BY DocNum
This gives me the following, which is correct:
DocNum AS_QTY D6T_QTY COR_QTY LastAS_DHID
12345 4 2 2 10000004
What I am trying to get next, as a column, is the StatusDate that correlates to the LastAS_DHID value that was calculated from the select (i.e. I need the next column to show '5/5/2020' since that date ties to the DocHistID '10000004').
In Power Bi, I'd use a lookupvalue function, but as I transfer my logic to the SQL server, I think I need to use a derived CASE WHEN type of query, but unsure of how to structure the syntax. Any help is greatly appreciated!

Answer 1

Maybe you have to create your on table and then do a left join. This being said, Try this:
SELECT A.DocNum,
A.AS_QTY,
A.D6T_QTY,
A.COR_QTY,
A.LastAS_DHID,
B.StatusDate
FROM (
SELECT DocNum,
SUM(CASE WHEN ShippingDocSummary.StatusCode LIKE 'AS_' THEN StatusQty ELSE 0 END) AS AS_QTY,
SUM(CASE WHEN ShippingDocSummary.StatusCode LIKE 'D6T' THEN StatusQty ELSE 0 END) AS D6T_QTY,
SUM(CASE WHEN ShippingDocSummary.StatusCode LIKE 'COR' THEN StatusQty ELSE 0 END) AS COR_QTY,
MAX(CASE WHEN ShippingDocSummary.StatusCode LIKE 'AS_' THEN DocHistID ELSE null END) AS LastAS_DHID
FROM ShippingDocSummary
GROUP BY DocNum) A
LEFT JOIN TABLE_WITH_DOCHIST_10000004 B
ON A.LastAS_DHID = B.DocHistID

Answer 2

SELECT DISTINCT
DocNum,
SUM(CASE WHEN StatusCode LIKE 'AS_' THEN StatusQty ELSE 0 END) OVER (PARTITION BY DocNum) AS AS_QTY,
SUM(CASE WHEN StatusCode LIKE 'D6T' THEN StatusQty ELSE 0 END) OVER (PARTITION BY DocNum) AS D6T_QTY,
SUM(CASE WHEN StatusCode LIKE 'COR' THEN StatusQty ELSE 0 END) OVER (PARTITION BY DocNum) AS COR_QTY,
FIRST_VALUE(StatusDate) OVER (PARTITION BY DocNum ORDER BY CASE WHEN StatusCode LIKE 'AS_' THEN DocHistID ELSE null END DESC) AS LastAS_DHID
FROM ShippingDocSummary
I've converted your aggregate functions to windowed versions with partitioning on the original grouping column. The select distinct is then necessary to actually collapse the rows to a single aggregated row. If first_value() isn't available on your platform there are other alternatives.
Here's one that worked for me as a scalar subquery:
(
SELECT StatusDate FROM ShippingDocSummary sds2
WHERE sds2.DocHistId =
MAX(
CASE WHEN ShippingDocSummary.StatusCode LIKE 'AS_'
THEN ShippingDocSummary.DocHistID ELSE null END
)
) AS LastAS_DHID