INNER JOIN 查询使用三个表返回匹配条件的行的计数和总和答案

【问题标题】：INNER JOIN query using three tables to return Count and Sum of rows that match criteriaINNER JOIN 查询使用三个表返回匹配条件的行的计数和总和
【发布时间】：2013-02-24 15:25:40
【问题描述】：

我正在尝试连接三个 SQL 表中的数据。

表格格式如下：

客户

╔════════╗
║ CLIENT ║
╠════════╣
║ A      ║
║ B      ║
║ C      ║
║ D      ║
╚════════╝

工作时间

╔════════╦══════════╦════════╦════════════╗
║ Client ║   Work   ║ Amount ║    Date    ║
╠════════╬══════════╬════════╬════════════╣
║ A      ║ Web Work ║     10 ║ 2013-01-12 ║
║ B      ║ Research ║     20 ║ 2013-01-20 ║
║ A      ║ Web Work ║     15 ║ 2013-01-21 ║
║ C      ║ Research ║     10 ║ 2013-01-28 ║
╚════════╩══════════╩════════╩════════════╝

费用

╔════════╦══════════╦════════╦════════════╗
║ Client ║   Item   ║ Amount ║    Date    ║
╠════════╬══════════╬════════╬════════════╣
║ A      ║ Software ║     10 ║ 2013-01-12 ║
║ B      ║ Software ║     20 ║ 2013-01-20 ║
╚════════╩══════════╩════════╩════════════╝

我想要一个返回每个客户的工作和费用的计数和总和的查询，即：

╔════════╦═══════════╦═══════════╦══════════════╦══════════════╗
║ CLIENT ║ COUNTWORK ║ WORKTOTAL ║ COUNTEXPENSE ║ EXPENSETOTAL ║
╠════════╬═══════════╬═══════════╬══════════════╬══════════════╣
║ A      ║         2 ║        25 ║            1 ║           10 ║
║ B      ║         1 ║        20 ║            1 ║           20 ║
║ C      ║         1 ║        10 ║            0 ║            0 ║
╚════════╩═══════════╩═══════════╩══════════════╩══════════════╝

到目前为止，我有以下内容：

SELECT clients.Client,
 COUNT(distinct work_times.id) AS num_work,
 COUNT(expenses.id) AS num_expenses
FROM
 clients
 INNER JOIN work_times ON work_times.Client = clients.Client
   INNER JOIN expenses ON expenses.Client = work_times.Client
GROUP BY
  clients.Client

这似乎是正确的，但它跳过了没有费用的客户，并且似乎将 num_expenses 乘以 num_work。我还想添加一个 WHERE 子句来指定只返回两个日期之间的工作时间和费用。我需要对查询进行哪些更改才能获得所需的输出？

【问题讨论】：

标签： sql inner-join multiple-tables

【解决方案1】：

您需要单独计算子查询中的值。最外层查询上的WHERE 子句的目的是过滤掉在一张表上至少有记录的记录。所以在这种情况下，Client D 将不会显示在结果列表中。

SELECT  a.*,
        COALESCE(b.totalCount, 0) AS CountWork,
        COALESCE(b.totalAmount, 0) AS WorkTotal,
        COALESCE(c.totalCount, 0) AS CountExpense,
        COALESCE(c.totalAmount, 0) AS ExpenseTotal
FROM    clients A
        LEFT JOIN
        (
            SELECT  Client, 
                    COUNT(*) totalCount,
                    SUM(Amount) totalAmount
            FROM    work_times
            WHERE   DATE BETWEEN '2013-01-01' AND '2013-02-01'
            GROUP   BY Client
        ) b ON a.Client = b.Client
        LEFT JOIN
        (
            SELECT  Client, 
                    COUNT(*) totalCount,
                    SUM(Amount) totalAmount
            FROM    expenses
            WHERE   DATE BETWEEN '2013-01-01' AND '2013-02-01'
            GROUP   BY Client
        ) c ON a.Client = c.Client
WHERE   b.Client IS NOT NULL OR
        c.Client IS NOT NULL

SQLFiddle Demo

更新

╔════════╦═══════════╦═══════════╦══════════════╦══════════════╗
║ CLIENT ║ COUNTWORK ║ WORKTOTAL ║ COUNTEXPENSE ║ EXPENSETOTAL ║
╠════════╬═══════════╬═══════════╬══════════════╬══════════════╣
║ A      ║         2 ║        25 ║            1 ║           10 ║
║ B      ║         1 ║        20 ║            1 ║           20 ║
║ C      ║         1 ║        10 ║            0 ║            0 ║
╚════════╩═══════════╩═══════════╩══════════════╩══════════════╝

【讨论】：

我必须对您的查询进行轻微更改，以按月分组，然后按客户分组，然后按月排序。我试了一下，你可以看到here，但是输出是由客户而不是月份订购的。你有什么建议吗？
小提琴上的输出是否正确，唯一的问题是排序？
输出几乎是正确的。由于我对日期所做的更改，客户 B 的工作时间为 2013 年 1 月，客户 B 的费用为 2013 年 2 月，但它们却在同一行返回。除此之外，订购是一个问题。有些月份，客户可能需要支付费用，但没有工作时间，反之亦然。
好的，我现在分析问题。我可以给你的一个提示是为此打开另一个问题，这样很多其他人都可以看到问题:D
我又问了一个问题here

【解决方案2】：

您可以将group by 移动到子查询中，因此您不必为每个expense 重复每个work_time。一旦有了子查询，就可以轻松地为两者添加日期过滤器：

SELECT  clients.Client
,       work_times.cnt AS num_work
,       work_times.total AS total_work
,       expenses.cnt AS num_expenses
,       expenses.total AS total_expenses
FROM    clients
LEFT JOIN
        (
        SELECT  Client
        ,       COUNT(DISTINCT id) as cnt
        ,       SUM(Amount) as total
        FROM    work_times
        WHERE   Date between '2013-01-01' and '2013-02-01'
        GROUP BY
                Client
        ) work_times
ON      work_times.Client = clients.Client
LEFT JOIN
        (
        SELECT  Client
        ,       COUNT(DISTINCT id) as cnt
        ,       SUM(Amount) as total
        FROM    expenses
        WHERE   Date between '2013-01-01' and '2013-02-01'
        GROUP BY
                Client
        ) expenses
ON      expenses.Client = clients.Client

【讨论】：

【解决方案3】：

我没有合适的实例来测试，但我可能会先这样，然后检查是否可以进一步改进查询...

select 
  T1.client, 
  ce AS 'Count Work', 
  am AS 'Work Total', 
  ci AS 'Count Expense', 
  am2 AS 'Expense Total' 
from (
  select 
    client, 
    count (work) as ce, 
    sum(amount) as am 
  FROM 
    clients 
      left join work_times 
      on fk_client=client 
  group by 
    fk_client
) T1 
left join (
  select 
    client, 
    count(item) as ci, 
    sum(amount) as am2 
  from 
    clients 
      left join expenses 
      on fk_client=client 
  group by fk_client
) T2 
where T1.client=T2.client;

也许这看起来很复杂，但它可以确保每个客户只有一排。也许以后会更易读……

【讨论】：