T-SQL: Ugly SQL - sql-server

Question

EDIT: Just so folks understand, I'm not worried about the formatting, I'm worried about the usage of the GROUP By and the usage of the aggregate fields when it doesn't make a whole lot of sense.
I've been tasked with making some SQL more readable. While I generally know what to do, this particular query escapes me. The gist of the query involves the writer grouping by a whole bunch of fields, and adding those fields to the query results. For fields that he/she doesn't GROUP BY, they use a MIN aggregate function I guess to "make the error go away"
MIN(ISNULL(dbo.V_CONNECT_ContactPartnerDetail_0010.SAPNr, N'')) AS sapkunr,
My difficulty comes from the fact that I can stuff the GROUP BY into a CTE, and then branch out from there, but I've never gotten the row counts to match up between the query I've created and the original one. Any help on making this SQL more readable and making its intent more clear (no functions to make the error go away) would be greatly appreciated.
SELECT dbo.V_CONNECT_ContactPartnerDetail_0010.ID_FI AS firmencode,
dbo.V_CONNECT_ContactPartnerDetail_0010.ID_KP AS partnercode,
dbo.V_CONNECT_ContactPartnerDetail_0010.Nachname,
Min(Isnull(dbo.V_CONNECT_ContactPartnerDetail_0010.Vorname, '')) AS vname,
Min(CASE V_CONNECT_ContactPartnerDetail_0010.Anrede
WHEN 'Frau' THEN 2
ELSE 1
END) AS anrede,
Min(Isnull(dbo.V_CONNECT_ContactPartnerDetail_0010.EMail, N'')) AS mail,
Min(Isnull(dbo.V_CONNECT_ContactPartnerDetail_0010.SAPNr, N'')) AS sapkunr,
Isnull(dbo.V_CONNECT_ContactPartnerDetail_0010.Titel, N'') AS titel
FROM dbo.V_CONNECT_ContactPartnerDetail_0010
INNER JOIN dbo.V_CONNECT_ContactPartnerPivot
ON dbo.V_CONNECT_ContactPartnerDetail_0010.ID_C005 = dbo.V_CONNECT_ContactPartnerPivot.ID_C005
LEFT OUTER JOIN dbo.V_CONNECT_Firmen_PZ_Download
ON dbo.V_CONNECT_ContactPartnerDetail_0010.ID_VF = dbo.V_CONNECT_Firmen_PZ_Download.ID_VF
WHERE ( dbo.V_CONNECT_ContactPartnerDetail_0010.VKO = '0010' )
GROUP BY dbo.V_CONNECT_ContactPartnerDetail_0010.ID_FI,
dbo.V_CONNECT_ContactPartnerDetail_0010.ID_KP,
dbo.V_CONNECT_ContactPartnerDetail_0010.Nachname,
dbo.V_CONNECT_ContactPartnerDetail_0010.Ort,
dbo.V_CONNECT_ContactPartnerPivot.flg_spl,
dbo.V_CONNECT_ContactPartnerPivot.flg_ha,
dbo.V_CONNECT_ContactPartnerPivot.flg_fu,
dbo.V_CONNECT_ContactPartnerPivot.flg_ma,
dbo.V_CONNECT_ContactPartnerPivot.flg_ph,
Isnull(dbo.V_CONNECT_ContactPartnerDetail_0010.Titel, N'')

Answer 1

This is much more "readable" to me
SELECT cpd.ID_FI AS firmencode
, cpd.ID_KP AS partnercode
, cpd.Nachname AS Nachname
, MIN(ISNULL( cpd.Vorname ,'')) AS vname
, MIN(CASE cpd.Anrede WHEN 'Frau' THEN 2 ELSE 1 END) AS anrede
, MIN(ISNULL( cpd.EMail ,N'')) AS mail
, MIN(ISNULL( cpd.SAPNr ,N'')) AS sapkunr
, ISNULL( cpd.Titel ,N'') AS titel
FROM dbo.V_CONNECT_ContactPartnerDetail_0010 cpd
JOIN dbo.V_CONNECT_ContactPartnerPivot cpp
ON cpd.ID_C005 = cpp.ID_C005
LEFT
JOIN dbo.V_CONNECT_Firmen_PZ_Download fpd
ON fpd.ID_VF = cpd.ID_VF
WHERE cpd.VKO = '0010'
GROUP
BY cpd.ID_FI
, cpd.ID_KP
, cpd.Nachname
, cpd.Ort
, cpp.flg_spl
, cpp.flg_ha
, cpp.flg_fu
, cpp.flg_ma
, cpp.flg_ph
, ISNULL(cpd.Titel ,N'')
EDIT
If I was "tasked with making some SQL more readable", I'd start with the changes above.
Beyond that, it's not clear why the GROUP BY clause includes expression that aren't in the SELECT list. It's valid to do that. But what's curious is that uf there are multiple rows from "cpd" that have different values of "Ort", then there's a pontential to get multiple rows returned, with the same values of "ID_FI", "ID_KP", "Nachname".
What really sticks out though is the outer join to "fpd", and apart from the reference to the "ID_VF" column in the join condition, there aren't any references to columns from "fpd" anywhere else in the query. It seems like if that outer join were removed, we'd get the same result.
The first structural change I would propose would be the removal of the join to "fpd".
SELECT cpd.ID_FI AS firmencode
, cpd.ID_KP AS partnercode
, cpd.Nachname AS Nachname
, MIN(ISNULL( cpd.Vorname ,'')) AS vname
, MIN(CASE cpd.Anrede WHEN 'Frau' THEN 2 ELSE 1 END) AS anrede
, MIN(ISNULL( cpd.EMail ,N'')) AS mail
, MIN(ISNULL( cpd.SAPNr ,N'')) AS sapkunr
, ISNULL( cpd.Titel ,N'') AS titel
FROM dbo.V_CONNECT_ContactPartnerDetail_0010 cpd
JOIN dbo.V_CONNECT_ContactPartnerPivot cpp
ON cpp.ID_C005 = cpd.ID_C005
WHERE cpd.VKO = '0010'
GROUP
BY cpd.ID_FI
, cpd.ID_KP
, cpd.Nachname
, cpd.Ort
, ISNULL( cpd.Titel ,N'')
, cpp.flg_spl
, cpp.flg_ha
, cpp.flg_fu
, cpp.flg_ma
, cpp.flg_ph
We can rearrange the expression in the GROUP BY clause, to move "Title" up with the other columns from "cpd". Without an ORDER BY clause, there's no guarantee what order the rows will be returned in.
We can't tell (from the query, and from the information provided) whether the "ID_C005" column is the PRIMARY KEY or a UNIQUE KEY in either "cpd" or "cpp".
And not knowing that, we can't really make other change to the query without potentially changing the result. If "ID_C005" is unique in "cpp", then we could eliminate all of the "cpp" column references from the GROUP BY.
If the purpose of the inner join (to "cpp") is to filter out rows from "cpd" that don't have a matching row in "cpp", we could make some other changes to the query. And that might make it more "readable".