Question

我不确定是否确实需要递归查询，但这就是我要做的事情（一个A查询，一个B查询，一个B查询，一个A ...递归查询）。

这是我最小的完整可验证代码：

我有一个这样的表（MySQL v5.7，InnoDB）：

CREATE TABLE transactions 
(
    id INT PRIMARY KEY AUTO_INCREMENT,
    code VARCHAR(10),
    date DATETIME,
    mode ENUM('Buy', 'Sell', 'Count', 'Return'),
    quantity INT,
    price DECIMAL(10,2),
    price_currency ENUM('ARS', 'USD'),
    usd_to_ars DECIMAL(10,2),
    return_id INT NULL DEFAULT NULL
)

然后我用一些物品填充它：

INSERT INTO transactions (code, date, mode, quantity, price, price_currency, usd_to_ars) 
VALUES 
("a", "20180101", 'Buy', 4, 10, 'ARS', 3.7),
("a", "20180102", 'Buy', 9, 8, 'ARS', 5.8),
("a", "20180103", 'Sell', -3, 0, 'USD', 0),
("b", "20180104", 'Buy', 5, 5, 'USD', 8.9),
("a", "20180105", 'Buy', 2, 7, 'USD', 3.4),
("b", "20180106", 'Buy', 1, 8, 'ARS', 9),
("a", "20180107", 'Sell', -8, 0, 'USD', 4.4),
("a", "20180108", 'Buy', 9, 9, 'ARS', 3.2);

INSERT INTO transactions (code, date, mode, quantity, price, price_currency, usd_to_ars, return_id) 
VALUES ("a", "20180109", 'Return', 6, 2, 'ARS', 2, 2);

最后我执行以下代码：

SELECT * 
FROM  
    (SELECT
         id, date, code, mode, quantity, price, price_currency, usd_to_ars, return_id,
       @acm := @acm + quantity as stock,
       @avr := (@avr * (@acm - quantity) +
               if(quantity > 0, quantity *
                   if(mode = "Return", @avr,
                        if(price_currency = 'USD', price, price / usd_to_ars)
                     ),
                   quantity * @avr)
                 ) / @acm as average_price_usd
FROM
    transactions t1,
    (SELECT @acm := 0) x,
    (SELECT @avr := 0) y) t2
ORDER BY id DESC

dbfiddle.uk

如您所见，它不会引发错误，它返回一个表，它可以工作……但是不像我想要的那样。

在第四行：

@avr := (@avr * (@acm - quantity) + if(quantity > 0, quantity * if(mode = "Return", @avr, if(price_currency = 'USD', price, price / usd_to_ars)), quantity * @avr)) / @acm as average_price_usd

我想更改以下参数的@avr：

If(mode = "Return", @avr, [...])

目前，如果mode = "Return"为true，则使用当前的@avr值，但是我想使用记录{{1} 1}}。由于@avr是一个计算值，我必须执行一个子查询才能再次计算它……我认为。问题是我不知道该怎么做。

因此，id = return_id的平均价格不应为@avr，而应为id = 9。

那么，如何在3.02...时在另一个查询中执行此查询，以便获取记录2.72...的{{1}}值？

如果可能的话，我是SQL知识的新手，所以我仍然不确定哪些事情是可行的，哪些是不可能的。对我来说，是否需要在表，视图或函数中创建列都没关系，我认为我可以使用它们中的任何一个。

Answer 1

简介：您的数据库小提琴表明您似乎正在使用MySQL 8.0。

由于您已经有一个有效的查询，并且只需要处理特定的mode = 'Return'情况，一种解决方案可以是将查询转换为CTE，并对其进行自我联接以将Return记录与其相关联原始购买。

这在this db fiddle中有效：

WITH cte AS (
    SELECT * FROM (SELECT
           id, date, code, mode, quantity, price, price_currency, usd_to_ars, return_id,
           @acm := @acm + quantity as stock,
           @avr := (@avr * (@acm - quantity) +
                   if(quantity > 0, quantity *
                       if(mode = "Return", @avr,
                            if(price_currency = 'USD', price, price / usd_to_ars)
                         ),
                       quantity * @avr)
                     ) / @acm as average_price_usd
    FROM
        transactions t1,
        (SELECT @acm := 0) x,
        (SELECT @avr := 0) y) t2
)
SELECT
    cte.id, 
    cte.date,
    cte.code,
    cte.mode,
    cte.quantity,
    cte.price,
    cte.price_currency,
    cte.usd_to_ars,
    cte.return_id,
    cte.stock,
    COALESCE(cte2.average_price_usd, cte.average_price_usd) average_price_usd
FROM cte LEFT JOIN cte cte2 ON cte.mode = 'Return' AND cte2.id = cte.return_id
ORDER BY cte.id DESC

此外，您似乎还可以使用 MySQL 8.0窗口函数实现您的目标。对我来说，计算average_price_usd的逻辑尚不清楚，但是这里有一个查询使用这种技术来计算库存。您可能可以对其进行修改，以添加平均价格的逻辑（应该进入上述查询的CTE部分）：

SELECT 
    id,
    code,
    date,
    mode,
    quantity,
    price,
    price_currency,
    usd_to_ars,
    return_id,
    SUM(quantity) OVER (ORDER BY date) stock,
    NULL average_price_usd
FROM transactions
ORDER BY date desc

编辑：这是MySQL <8.0的一种（丑陋的）解决方法，它使用两个子查询（请参见this db fiddle）：

SELECT
    cte.id, 
    cte.date,
    cte.code,
    cte.mode,
    cte.quantity,
    cte.price,
    cte.price_currency,
    cte.usd_to_ars,
    cte.return_id,
    cte.stock,
    COALESCE(cte2.average_price_usd, cte.average_price_usd) average_price_usd
FROM 
    (
        SELECT
            id, date, code, mode, quantity, price, price_currency, usd_to_ars, return_id,
            @acm := @acm + quantity as stock,
            @avr := (@avr * (@acm - quantity) +
                   if(quantity > 0, quantity *
                       if(mode = "Return", @avr,
                            if(price_currency = 'USD', price, price / usd_to_ars)
                         ),
                       quantity * @avr)
                     ) / @acm as average_price_usd
        FROM
            transactions t1,
            (SELECT @acm := 0) x,
            (SELECT @avr := 0) y
    ) cte
    LEFT JOIN (
        SELECT
            id, date, code, mode, quantity, price, price_currency, usd_to_ars, return_id,
            @acm2 := @acm2 + quantity as stock,
            @avr2 := (@avr2 * (@acm2 - quantity) +
                   if(quantity > 0, quantity *
                       if(mode = "Return", @avr2,
                            if(price_currency = 'USD', price, price / usd_to_ars)
                         ),
                       quantity * @avr2)
                     ) / @acm2 as average_price_usd
        FROM
            transactions t1,
            (SELECT @acm2 := 0) x,
            (SELECT @avr2 := 0) y
    ) cte2 ON cte.mode = 'Return' AND cte2.id = cte.return_id
ORDER BY cte.id DESC

如何在查询中进行递归查询或子查询？

1 个答案: