获取日期差异的MYSQL查询

我有一个具有以下列和数据的表事务

id  transaction_date trans_type account_id agents_id transaction_date price miles
1   2012-02-08       Buy        1          1         2010-02-08       0.016 12000
2   2012-03-01       Sell       2          2         2012-03-10       0.256 -2000
3   2012-03-27       Buy        3          3         2012-03-27       0.256 10000
4   2012-03-28       Sell       4          4         2012-03-28       0.589 -11000
5   2012-03-29       Buy        5          5         2012-03-29       0.87  25000
6   2012-03-29       Sell       6          6         2012-02-29       0.879 -12000
7   2012-04-01       Sell       7          7         2012-04-01       0.058 -15000

  Account Table
  id    Program_id
  1     1
  2     1
  3     2

  Program table
  id      Abbreviation
  1       AA
  2       AC

  Agents table
  id      Name
  1       Bob
  2       Ben

我想得到第一次销售日期和第一次购买日期，以获得交易售出前的平均天数，以获得交易在库存中的天数，所以应该是

  (Sell date)2012-03-01 - (Buy date)2012-02-08

我在试这个

SELECT 
    case when t.trans_type ='Sell' then transaction_date end as SellDate
   ,case when t.trans_type ='Buy' then transaction_date end as BuyDate
   ,DATEDIFF(case when t.trans_type ='Sell' then transaction_date end
            ,case when t.trans_type ='Buy' then transaction_date end) as Date
   ,transaction_date
FROM transactions t
order by transaction_date

但在日期上总是空的

这是完整的查询

SELECT p.abbreviation,ag.name
  ,sum(-1.00 * t.miles * t.price - coalesce(t.fees,0) - coalesce(c.cost,0)) as profit
  ,sum(t.miles) 'Totakl Miles'
  ,avg(price / miles) 'Average'
  ,transaction_date
FROM transactions t
inner join accounts a on t.account_id = a.id
inner join programs p on a.program_id = p.id
inner join agents ag on t.agent_id = ag.id
LEFT JOIN (
           SELECT rp.sell_id, sum(rp.miles * t.price) as cost
           from report_profit rp
           inner join transactions t on rp.buy_id = t.id
           where t.miles > 50000
           group by rp.sell_id
           order by rp.sell_id
          ) c on t.id = c.sell_id
where t.transaction_date BETWEEN '2012-03-14' AND '2012-04-14'
Group by p.id , ag.id

编辑

我尝试了LiqorVicar的答案，但它给出了错误“子查询返回多个记录”，因为我添加了组

有人能在这方面指导我吗

---

提前感谢…

尝试这样的子查询

SELECT 
    DATEDIFF(
      (
      SELECT MIN(date)
      FROM Transaction
      WHERE trans_type='Sell'
      ) AS first_sell_date
   ,
      (
      SELECT MIN(date)
      FROM Transaction
      WHERE trans_type='Buy'
      ) AS first_buy_date
   )

编辑：跟随OP评论并使用完整查询更新问题

你能不能把DATEDIFF包装成一个分钟通话

DATEDIFF(
    MIN(case when t.trans_type ='Sell' then transaction_date end),
    MIN(case when t.trans_type ='Buy' then transaction_date end)
) as Date

这个怎么样-

SELECT *, DATEDIFF(sale.transaction_data, purchase.transaction_date)
FROM transactions purchase
INNER JOIN (
    SELECT *
    FROM transactions
    WHERE trans_type = 'Sell'
    ORDER BY transaction_date ASC
) sale
    ON purchase.transaction_date < sale.transaction_date
WHERE purchase.trans_type = 'Buy'
GROUP BY purchase.transaction_date

---

首先感谢大家的帮助

下面是返回精确结果的查询

select p_id,ag_id,
     p_abb,ag_name
    ,sum(-1.00 * miles * price - coalesce(fees,0) - coalesce(cost,0)) as profit
    ,sum(miles) 'Total Miles',avg(price / miles) 'Average'
    ,DATEDIFF(min(buy_dt),min(sell_dt)) as 'Days'
     From
     (
         SELECT p.id 'p_id',ag.id 'ag_id',p.abbreviation 'p_abb',ag.name 'ag_name'
         ,miles
         ,price
         ,fees
         ,c.cost
         ,case when t.trans_type ='Sell' then transaction_date end 'sell_dt'
         ,case when t.trans_type ='Buy' then transaction_date end 'buy_dt'
         ,transaction_date
       FROM transactions t
       inner join accounts a on t.account_id = a.id
       inner join programs p on a.program_id = p.id
       inner join agents ag on t.agent_id = ag.id
       LEFT JOIN (
            SELECT rp.sell_id, sum(rp.miles * t.price) as cost
           from report_profit rp
           inner join transactions t on rp.buy_id = t.id
           where t.miles > 50000
           group by rp.sell_id
           order by rp.sell_id
          ) c on t.id = c.sell_id

  ) t1
  group by p_id, ag_id

---

再次感谢大家。

获得了示例查询和结果吗？您需要向DATEDIFF传递2个参数，从我看来，您只向DATEDIFF传递了1个参数，因为您的case语句将只生成一个日期。这是因为您一次只对一行进行操作，这意味着您的

DATEDIFF

对于其中一个日期总是得到

null

。您需要在

交易

上进行自联接，以便将买入记录与卖出记录进行匹配。您似乎没有任何与买入和卖出交易相关的密钥。您如何知道哪个卖出交易与哪个买入交易相关？是的，您都是对的…为所有涉及的表添加结构将非常有用。更好的方法是为每个表和预期输出包含一些示例数据。感谢您的回复，但是还有其他更好的方法吗？事实上，我有很多数据需要通过多个join获取如果您的查询要复杂得多，您需要在问题中包含详细信息。我们不是读心术的人，至少我不是：）@lixorvicar，正如您在编辑中提到的那样使用DATEDIFF，为date@AnilD是否每行都有正确的日期值（即不是0000-00-00的日期值）。我在一个数据集上运行了一个类似的查询，其中一些日期是“0000-00-00”，它返回NULL。排除这些值后，效果很好。所有日期都必须有一些值，但案例与我第一次查询中的情况相同，当案例为Sell时，Buy记录将有空日期，反之亦然。

select p_id,ag_id,
     p_abb,ag_name
    ,sum(-1.00 * miles * price - coalesce(fees,0) - coalesce(cost,0)) as profit
    ,sum(miles) 'Total Miles',avg(price / miles) 'Average'
    ,DATEDIFF(min(buy_dt),min(sell_dt)) as 'Days'
     From
     (
         SELECT p.id 'p_id',ag.id 'ag_id',p.abbreviation 'p_abb',ag.name 'ag_name'
         ,miles
         ,price
         ,fees
         ,c.cost
         ,case when t.trans_type ='Sell' then transaction_date end 'sell_dt'
         ,case when t.trans_type ='Buy' then transaction_date end 'buy_dt'
         ,transaction_date
       FROM transactions t
       inner join accounts a on t.account_id = a.id
       inner join programs p on a.program_id = p.id
       inner join agents ag on t.agent_id = ag.id
       LEFT JOIN (
            SELECT rp.sell_id, sum(rp.miles * t.price) as cost
           from report_profit rp
           inner join transactions t on rp.buy_id = t.id
           where t.miles > 50000
           group by rp.sell_id
           order by rp.sell_id
          ) c on t.id = c.sell_id

  ) t1
  group by p_id, ag_id