SQL添加每周和每月总计

Question

I am trying to follow a similar example of getting weekly and monthly totals from: StackOverFlow: SQL Add Sum Row for Week and At the End Add the Grand Total . This level of SQL is a stretch for me, so please be a clear as possible.

I have a table with over 200K records of ID, Store_ID, Sales_Date, Amount.

ID  Store_ID    Amount  Sales_Date
1   215            7    1/29/2012
2   215            7    1/30/2012
3   215            7    1/31/2012
4   215            7    2/1/2012
5   215            7    2/2/2012
6   215            7    2/3/2012
7   215            7    2/4/2012
8   215            8    2/5/2012
9   215            8    2/6/2012
10  215            8    2/7/2012
    ***More and More Data***        
162 218            4    10/30/2011
163 218            4    10/31/2011
164 218            4    11/1/2011
165 218            4    11/2/2011
166 218            4    11/3/2011
167 218            4    11/4/2011
168 218            4    11/5/2011
169 218            8    11/6/2011
170 218            8    11/7/2011
171 218            8    11/8/2011
           ******LOTS MORE DATA*****

I need to generate a view that will show the Weekly and Monthly Totals for each Store_ID and the associated date of the total. The issue that I am having is that the example is providing me with Week totals(with no dates associated), Month totals (with no dates associated) and the daily amounts( which is a side benefit).

I need to know how I can add in an additional column to the results that will show me the week ending date and the month ending date.

This is what I have so far (it is almost exactly like the example):

set datefirst 7

select top 100
    case
        when grouping(cast(datepart(week, [Sales_Date]) as varchar(255)))=1 then '<MonthEnd>'
        when grouping(cast([Sales_Date] as date))=1 then '<weektotal>'
        else cast(cast([Sales_Date] as date) as varchar(255))
    end as Period
    , WkSales = sum(Amount)
    , Store = Store_ID
From KF_Store_Sales_Daily

group by 
    grouping sets(  
    (cast(datepart(month, [Sales_Date]) as varchar(255)), cast(datepart(week, [Sales_Date]) as varchar(255)),cast([Sales_Date] as date)),
    (cast(datepart(month, [Sales_Date]) as varchar(255)), cast(datepart(week, [Sales_Date]) as varchar(255))),
    (cast(datepart(month, [Sales_Date]) as varchar(255)))
    )
    , Store_ID
ORDER BY Store_ID, Sales_Date    

Answer 1

The following query can be used to show daily, weekly, monthly and yearly totals:

select
    case
        when grouping(d.m)=1 then 'Year ' + cast(max(d.y) as varchar(10))
        when grouping(d.w)=1 then datename(m, max(Sales_Date)) + ' ' + cast(max(d.y) as varchar(10))
        when grouping(Sales_Date)=1 then 'Week ' + datename(m, max(ws)) + ' ' + cast(datepart(d, max(ws)) as varchar(20)) + ' - '
            + datename(m, max(we)) + ' ' + cast(datepart(d, max(we)) as varchar(20))
        else cast(cast([Sales_Date] as date) as varchar(255))
    end as Period
    , Sales = sum(Amount)
    , Store = Store_ID
from KF_Store_Sales_Daily
    cross apply (
        select -- aux. expressions for dates
            datepart(yy, [Sales_Date]), -- year
            datepart(m, [Sales_Date]), -- month
            datepart(wk, [Sales_Date]), -- week
            dateadd(d, 1-datepart(w, Sales_date), Sales_date), -- week start
            dateadd(d, 7-datepart(w, Sales_date), Sales_date) -- week end
    ) d(y, m, w, ws, we)
group by Store_ID, d.y, rollup (d.m, d.w, Sales_Date)
order by d.y desc,
    grouping(d.m), d.m,
    grouping(d.w), d.w,
    grouping(Sales_Date), Sales_Date

I'm not sure how convenient it is to have montly and weekly totals together (since week may belong to two months). Following two queries for the case if you will need them separartely.

For daily, monthly and yearly totals:

select
    case
        when grouping(d.m)=1 then 'Year ' + cast(max(d.y) as varchar(10))
        when grouping(Sales_Date)=1 then datename(m, max(Sales_Date)) + ' ' + cast(max(d.y) as varchar(10))
        else cast(cast([Sales_Date] as date) as varchar(255))
    end as Period
    , Sales = sum(Amount)
    , Store = Store_ID
from KF_Store_Sales_Daily
    cross apply (
        select
            datepart(yy, [Sales_Date]),
            datepart(m, [Sales_Date])
    ) d(y, m)
group by Store_ID, d.y, rollup (d.m, Sales_Date)
order by d.y desc,
    grouping(d.m), d.m,
    grouping(Sales_Date), Sales_Date

For daily, weekly and yearly totals (in this case a week may belong to two years):

select
    case
        when grouping(d.w)=1 then 'Year ' + cast(max(d.y) as varchar(10))
        when grouping(Sales_Date)=1 then 'Week ' + datename(m, max(ws)) + ' ' + cast(datepart(d, max(ws)) as varchar(20)) + ' - '
            + datename(m, max(we)) + ' ' + cast(datepart(d, max(we)) as varchar(20))
        else cast(cast([Sales_Date] as date) as varchar(255))
    end as Period
    , Sales = sum(Amount)
    , Store = Store_ID
from KF_Store_Sales_Daily
    cross apply (
        select
            datepart(yy, [Sales_Date]),
            datepart(wk, [Sales_Date]),
            dateadd(d, 1-datepart(w, Sales_date), Sales_date),
            dateadd(d, 7-datepart(w, Sales_date), Sales_date)
    ) d(y, w, ws, we)
group by Store_ID, d.y, rollup (d.w, Sales_Date)
order by d.y desc,
    grouping(d.w), d.w,
    grouping(Sales_Date), Sales_Date