Question

我无法正确选择语法。请帮助
我有两个表：第一个包含工作数据

  CREATE TABLE ProdStatus (
  Id    INT(6) UNSIGNED ZEROFILL,
  ProdId INT,
  StatusId INT,
  Date  DATE,
  Total  INT
);

此表中的数据：

INSERT INTO ProdStatus
VALUES
  (000001, 1, 1, '2014-02-01', 5),
  (000002, 1, 2, '2014-02-01', 3),
  (000003, 1, 2, '2014-02-05', 4),
  (000004, 2, 1, '2014-02-07', 2),
  (000005, 2, 1, '2014-02-08', 7),
  (000006, 2, 2, '2014-02-08', 9);

第二个是日历表

    CREATE TABLE calendar (
      id int(11),
      Date date
);

日历表的数据：

INSERT INTO calendar (id, Date)
VALUES
(1, '2014-02-01'),
(2, '2014-02-02'),
(3, '2014-02-03'),
(4, '2014-02-04'),
(5, '2014-02-05'),
(6, '2014-02-06'),
(7, '2014-02-07'),
(8, '2014-02-08'),
(9, '2014-02-09'),
(10, '2014-02-10');

我需要能够从ProdStatus表中选择所有值的总和，按ProdId和Date分组（总和，因为我可以有多个状态）但是与日历表一起使用，因为我需要每天的值选定的范围。如果我在ProdStatus表中没有一天的信息，那么应该返回前一天的结果。

以下是我所做的查询：

SELECT p.ProdId, c.Date, sum(p.Total) as Result
FROM ProdStatus p
right outer join calendar c on c.Date = p.Date
WHERE p.ProdId in (1, 2)
and c.date between '2014-02-01' and '2014-02-10'
group by c.date, p.ProdId

此查询返回此结果：

ProdId    Date    Result
1    2014-02-01    8
1    2014-02-05    4
2    2014-02-07    2
2    2014-02-08    16

预期结果：

ProdId    Date    Result
1    2014-02-01    8
2    2014-02-01    0
1    2014-02-02    8
2    2014-02-02    0
1    2014-02-03    8
2    2014-02-03    0
1    2014-02-04    8
2    2014-02-04    0
1    2014-02-05    4
2    2014-02-05    0
1    2014-02-06    4
2    2014-02-06    0
1    2014-02-07    4
2    2014-02-07    2
1    2014-02-08    4
2    2014-02-08    16
1    2014-02-09    4
2    2014-02-09    16
1    2014-02-10    4
2    2014-02-10    16

有什么建议吗？谢谢

Answer 1

这似乎是编程语言的任务。你是如何消费结果的？你能用这种语言实现这个目标吗？

如果您需要这是一个100％的mysql解决方案，您可能需要编写一个存储过程来执行此操作，而不仅仅是一个简单的查询。

Answer 2

这是我的sqlfiddle。有一个派生表，表示列表产品ID，以确保每个日历日期的每个产品都有一个条目。

SELECT foo.p AS ProdId, calendar.Date, SUM(COALESCE(ProdStatus.Total,0)) as Result
FROM calendar
JOIN (SELECT 1 AS p UNION SELECT 2) AS foo
LEFT JOIN ProdStatus
  ON calendar.Date >= ProdStatus.Date
    AND foo.p = ProdStatus.ProdId
LEFT JOIN ProdStatus AS t1
  ON calendar.Date >= t1.Date
    AND ProdStatus.Date < t1.Date
    AND foo.p = t1.ProdId
WHERE calendar.date BETWEEN '2014-02-01' AND '2014-02-10'
  AND t1.Id IS NULL
GROUP BY calendar.date, foo.p
ORDER BY calendar.date, foo.p;

以下是使用您的Product表（假设密钥称为ID）：

SELECT product.id AS ProdId, calendar.Date, SUM(COALESCE(ProdStatus.Total,0)) as Result
FROM calendar
JOIN Product
LEFT JOIN ProdStatus
  ON calendar.Date >= ProdStatus.Date
    AND product.id = ProdStatus.ProdId
LEFT JOIN ProdStatus AS t1
  ON calendar.Date >= t1.Date
    AND ProdStatus.Date < t1.Date
    AND product.id = t1.ProdId
WHERE calendar.date BETWEEN '2014-02-01' AND '2014-02-10'
  AND t1.Id IS NULL
  AND product.id IN (1, 2, 253, 54, 78)
GROUP BY calendar.date, product.id
ORDER BY calendar.date, product.id;

Answer 3

我将您的查询切换为left outer join，只是因为我更好地按照这种方式查询查询（“在进行连接时保留第一个表中的所有行”）。

您只需将prodid上的条件从where子句移到on子句：

SELECT p.ProdId, c.Date, sum(p.Total) as Result
FROM calendar c left outer join
     ProdStatus p
     on c.Date = p.Date and p.ProdId in (1, 2)
WHERE c.date between '2014-02-01' and '2014-02-10'
group by c.date, p.ProdId
order by c.date, p.ProdId;

如果没有匹配项，则prodiId为NULL，因此where子句中的条件失败。

编辑：

我明白了。要解决此类查询，您需要生成两个prodids和日期的所有可能组合，然后加入prod状态：

SELECT prods.ProdId, c.date, coalesce(sum(p.Total), 0) as Result
FROM (select 1 as ProdId union all select 2) as prods cross join
     calendar c left outer join
     ProdStatus p
     on c.Date = p.Date and prods.ProdId = p.ProdId
WHERE c.date between '2014-02-01' and '2014-02-10'
GROUP BY prods.ProdId, c.date
OPRDER BY prods.ProdId, c.date;

如果没有数据，MySQL返回前一行

3 个答案: