I'm trying to extract number of adults, children and infants from the below example XML:
- Adults = AgeType 3
- Children = AgeType 2
- Infants = AgeType 1
and then count specifies how many of that age type. Correct result is:
- 35 Adults
- 5 Children
- 2 Infants
The below code returns the correct result, however I would like to avoid the cross apply
entering to the <Customer>
tags (as there are hundreds of millions per day).
declare @a table(a xml)
insert into @a
select '<Customers>
<Customer AgeType="3" Count="10" />
<Customer AgeType="3" Count="20" />
<Customer AgeType="3" Count="5" />
<Customer AgeType="2" Count="5" />
<Customer AgeType="1" Count="2" />
</Customers>'
select
sum(case when b.cust = 3 then b.cust*b.count_cust end)/3 as adt
,sum(case when b.cust = 2 then b.cust*b.count_cust end)/2 as chd
,sum(case when b.cust = 1 then b.cust*b.count_cust end)/1 as inf
from (
select c.value('(@AgeType)','int') As cust
,c.value('(@Count)','int') As count_cust
from @a cross apply a.nodes('Customers/Customer') as t(c)
) as b
Can anyone find any other logic that could be more effiecient? I was thinking using count
or sum
over the <Customer>
tags like below, however I can't get the correct reults.
a.value('count(Customers/Customer/@AgeType[.=3])','int') As adt
,a.value('count(Customers/Customer/@AgeType[.=2])','int') As chd
,a.value('count(Customers/Customer/@AgeType[.=1])','int') As inf