Question

我现在对MySQL和高效查询有相当的经验，但仍遇到问题......

我有两张表，“发票”和“salesRecords”发票表包含发票的基本概览，销售记录逐行分解有关购买物品的特定信息。

为简单起见，我们假设表格如下：

发票

id    |   repairCharge     |    shippingCharge      |     total
 1    |       28.95        |         0              |     30.45   
 2    |       10.00        |         8.50           |     29.50

//注意;总列包括销售记录中显示的项目

SalesRecords

invNo  |    itemNo   |  price    |   quantity   | discount  |   
  1    |     123     |  1.50     |      1       |     0     |  
  2    |     121     |  6.50     |      1       |     1.5   |  
  2    |     128     |  5.50     |      1       |     0     |

我想在一个日期期间获得所售商品，人工费和运费的总价值，因此我需要：

汇总发票表中的所有行
JOIN（左？）SalesRecords表ON invNo和SUM（（价格折扣）*数量）
对日期使用WHERE子句

我马上写下这个SQL：

SELECT SUM((sr.price-sr.discount)*sr.quantity) as income,   
   SUM(i.repairCharge) as labour,
   SUM(i.shippingCharge) as carriage 
FROM invoices i 
    LEFT JOIN salesRecords sr ON sr.invNo=i.id
WHERE i.dateTime BETWEEN 1431647999 AND 1434360348`

错误

因为，对于包含多个项目行的任何发票，我获得了两个表的多个实例，因此repairCharge和shippingCharge值加倍或增加三倍，具体取决于产品线的数量在该发票上购买。

所以，我搞砸了并提出了一个解决方案...但它非常丑陋并且可能效率低下：

SELECT SUM(income) as income,
       SUM(labour) as labour,
       SUM(carriage) as carriage 
FROM (SELECT (SELECT SUM((price-discount)*quantity)
              FROM salesRecords
              where invno=i.id
              GROUP BY invNo) as income, 
             SUM(i.repairCharge) as labour,
             SUM(i.shippingCharge) as carriage 
      FROM invoices i 
      WHERE i.dateTime BETWEEN 1434326474 AND 1434361694
      GROUP BY id
     ) totals

任何人都可以建议最简化这种方法并提高效率吗？

Answer 1

就大多数其他DBMS的易读性和性能而言，我个人只是在子查询中总结销售记录：

SalesRecords

MySQL在子查询上使用中间物化（据我所知），如果这样做，可能会对上述查询产生负面影响，因为它会首先汇总SELECT SUM(COALESCE(i.Income, 0)) AS Income, SUM(i.repairCharge) AS Labour, SUM(i.shippingCharge) AS Carriage FROM ( SELECT i.ID, i.repairCharge, i.shippingCharge, SUM((sr.price - sr.discount) * sr.quantity) AS Income FROM Invoices AS i LEFT JOIN SalesRecords AS sr ON sr.InvNo = i.ID WHERE i.dateTime BETWEEN 1434326474 AND 1434361694 GROUP BY i.ID, i.repairCharge, i.ShippingCharge ) AS i;中的所有记录并将结果存储在在发票日期应用过滤器之前的哈希表，所以您的工作版本，但使用JOIN而不是相关子查询可能会表现更好：

Sources

Answer 2

试试这个：

select sum(labour) as labour,
       sum(carriage) as carriage,
       sum(income) as income
from( select sum(i.repairCharge) as labour,
             sum(i.shippingCharge) as carriage,
             coalesce(sum((sr.price - sr.discount) * sr.quantity), 0) as income
      from Invoices i
      left join SalesRecords sr on i.id = sr.invNo
      group by i.id)t

在多行上使用LEFT JOIN简化我的SQL

2 个答案: