如何创建动态变量以加载SQL Server中的表中的字段

Question

I'm stuck with a scenario where i have to create a dynamic variable for a select query which hold the data of a particular filed. Here is the scenario

Table 'Customer' has below data

Column0 Column1 Column2     Column3     Column4 Column5
--------------------------------------------------------
H       123X                11/27/2017  C       10.23
D       123X    78462F103   11/28/2017  A       112.35
D       123X    55024U109   11/28/2017  A       25.30
H       456X                11/27/2017  B       5263.44
D       456X    78462F103   11/28/2017  A       23.00
D       456X    55024U109   11/28/2017  A       12123.00
D       456X    78462F103   11/28/2017  A       56.08
D       456X    55024U109   11/28/2017  C       45.07

Now i have to create variable when Column0='H' using the Sum(column5) and load it into Column5 when Column0='D' until the Column0='H' value changes(Table has multiple H values).

Example: I have to calculate the variable when Column0='H' lets say If Column0='H' then Variable=Sum(Column5) which is '5.36'. Now When Coulmn1='D' then Column5's value shlould be loaded as 5.63 when the When the new Column0='H' comes the value of variable changes to '6.33' and it should load the updated value into 'D' records.

Any suggestions on this?

Answer 1

Assuming that I am correct you want the total of all rows for each Column1 to be the new value in Column5 when Column0 = H you can do something like this.

Notice I had to create sample data in a table before I could do anything.

declare @Something table 
(
    Column0 char(1)
    , Column1 char(4)
    , Column2 varchar(20)
    , Column3 date
    , Column4 char(1)
    , Column5 decimal (9,2)
)

insert @Something values
('H', '123X', '', '11/27/2017', 'C', 10.23)
, ('D', '123X', '78462F103', '11/28/2017', 'A', 112.35)
, ('D', '123X', '55024U109', '11/28/2017', 'A', 25.30)
, ('H', '456X', '', '11/27/2017', 'B', 5263.44)
, ('D', '456X', '78462F103', '11/28/2017', 'A', 23.00)
, ('D', '456X', '55024U109', '11/28/2017', 'A', 12123.00)
, ('D', '456X', '78462F103', '11/28/2017', 'A', 56.08)
, ('D', '456X', '55024U109', '11/28/2017', 'C', 45.07)
;

with SummaryData as
(
    select *
        , NewCol5 = case when Column0 = 'H' then sum(Column5) over(partition by Column1) else Column5 end
    from @Something
)

update SummaryData
set Column5 = NewCol5

select *
from @Something

I would very strongly urge to move away from this design immediately if not sooner. It is fraught with error. And even worse is this looks like sales information. With this design you can't track things like SalesTax, Shipping, discounts, etc....all the kinds of things you want to track about orders.

Answer 2

I was under the impression that the source was a flat file and am providing a SSIS solution.

read data -> Conditional Split ("H" and "D")

Dpath
Multicast -> Write out detail records
  -> Aggregate on Col1 --> Merge Join to H path

H path:
Merge Join on Col1 to Aggregated Data

Finally either AddH Col5 to Aggregate or replace col 5 with aggregate (it's not clear in question.)

Write out your header records.