SQL 服务器：运行 CTE 中两个表的总数

Question

我在为 CTE 中的两个表创建运行总数时遇到了一些麻烦。由于野兽的性质（我必须联合所有的两个表）我得到双日期行。

将它们相加的最佳方法是什么？

这是一个示例代码。 Date 是 date。 Cost 和 Total 是 decimal(20, 4)。

with CTE_Blah
As
(
    SELECT
        [Date]
        ,SUM([Cost]) as [Cost_Sum]
        ,SUM([Total]) as [Invoiced_Sum]
    FROM TSL_Q40_DATED
    WHERE   ([Date] BETWEEN '2016-01-01' AND '2016-01-31') AND [Rego] <> ''
    GROUP BY [Date]
    UNION ALL
    SELECT
        [Date]
        ,SUM([Cost]) as [Cost_Sum]
        ,SUM([Total]) as [Invoiced_Sum]
    FROM TSL_Q41_DATED
    WHERE   ([Date] BETWEEN '2016-01-01' AND '2016-01-31') AND [Rego] <> ''
    GROUP BY [Date]
)
select
    [Date]
    ,SUM([Cost_Sum]) over(order by [Date] rows unbounded preceding) as [Cost_Sum_Total]
    ,SUM([Invoiced_Sum]) over(order by [Date] rows unbounded preceding) as [Invoiced_Sum_Total]
from CTE_Blah

数据 headers:

Date    Cost_Sum_Total  Invoiced_Sum_Total
2016-01-03  106.7500    2976.5300
2016-01-03  2814.2184   6188.3800
2016-01-04  2814.2184   6626.3800
2016-01-05  25182.9184  69719.7260
2016-01-05  62910.4085  121623.9760
2016-01-06  76959.7885  169941.0178
2016-01-06  111582.4872 219210.3828
2016-01-07  130267.1772 276630.7298
2016-01-07  157257.1861 319858.9298
2016-01-08  181760.3061 374464.6628
2016-01-08  216602.5447 422211.4528
2016-01-09  216602.8647 422202.4528
2016-01-11  232240.6247 475385.5239
2016-01-11  292532.0367 555308.8015
2016-01-12  308909.0367 610309.8151
2016-01-12  352102.4805 667868.5984
2016-01-13  379601.3205 735715.6454
2016-01-13  443095.2996 819906.0354
2016-01-14  485450.7396 924794.9284
2016-01-14  570151.1295 1036974.5164
2016-01-15  599628.5995 1133408.4675
2016-01-15  682348.2252 1245186.5388
2016-01-16  682570.0052 1245727.4688
2016-01-16  683832.8852 1247475.2988
2016-01-18  710836.7252 1326957.7818
2016-01-18  773762.3701 1413407.1843
2016-01-19  798178.4701 1487877.5098
2016-01-19  855247.8828 1566829.2628
2016-01-20  875797.9428 1625227.4776
2016-01-20  919321.1908 1684962.4396
2016-01-21  945941.7808 1764925.5599
2016-01-21  991368.0220 1829901.9009
2016-01-22  1026347.2920    1934196.0364
2016-01-22  1125390.4969    2069489.9374
2016-01-23  1143017.4887    2089926.9674
2016-01-25  1167810.0887    2158779.0054
2016-01-25  1218388.1937    2229148.9254
2016-01-26  1265004.9037    2365723.2409
2016-01-26  1331139.5559    2463935.7624
2016-01-27  1374683.8859    2586818.1040
2016-01-27  1538084.3589    2802247.8524
2016-01-28  1602411.9689    2980781.7534
2016-01-28  1743563.6906    3170026.1664
2016-01-29  1836842.7906    3411791.7531
2016-01-29  2052226.0474    3693064.0627
2016-01-30  2060326.0474    3710266.4677
2016-01-30  2068572.2762    3721389.5477
2016-01-31  2068572.2762    3721422.8677
2016-01-31  2068943.5162    3722678.2677

Answer 1

您可以将 window 函数与普通聚合结合起来

SELECT
    [Date]
    ,SUM(SUM([Cost_Sum])) OVER (ORDER BY [Date] ROWS UNBOUNDED PRECEDING) AS [Cost_Sum_Total]
    ,SUM(SUM([Invoiced_Sum])) OVER (ORDER BY [Date] ROWS UNBOUNDED PRECEDING) AS [Invoiced_Sum_Total]
FROM CTE_Blah
GROUP BY [Date]

它是如何工作的？

GROUP BY 将同一日期的所有记录分组并将它们的值汇总为每日总和
然后 window 函数汇总所有每日总和

当然，您也可以将预聚合移出 CTE，然后根本运行不进入此问题。

SQL 服务器：运行 CTE 中两个表的总数

SQL Server : running totals of two tables in CTE

sql

sql-server

common-table-expression

cumulative-sum