Sql 基于同一表上的值在同一行上返回多个值

我已经有一段时间被难住了，一直在尝试根据其他结果从我们的一个表中获取结果，并且每个值都不返回多行，这很难为我解释，所以我将给出一个示例

我们有一个价值观，例如苹果。每个苹果都有自己的序列号，serial。对于某些苹果，有多个（最多6个）序列值

ID      APPLE       SERIAL
2052    5035        1
2055    5035        4
2058    5035        8
2070    5040        5

在我的结果中，我想返回每个苹果一次，以及分配给它的所有序列号，例如：

APPLE   SERIAL_1    SERIAL_2    SERIAL_3    SERIAL_4
5035    1           4           8
5040    5

我试着做了一个查询，它基本上连接了同一个表4次，说serial_x not in（其他serial）。这样做的问题是，它会产生如下多种结果：

APPLE   SERIAL_1    SERIAL_2    SERIAL_3    SERIAL_4
5035    1           4           8
5035    1           8           4
5035    4           1           8
5035    4           8           1
5035    8           1           4
5035    4           4           1
5040    5

with data as (
    select 2052 id, 5035 apple, 1 serial from dual union all
    select 2055 id, 5035 apple, 4 serial from dual union all
    select 2058 id, 5035 apple, 8 serial from dual union all
    select 2070 id, 5040 apple, 5 serial from dual
)
select
    apple,
    serial_1,
    serial_2,
    serial_3,
    serial_4,
    serial_5,
    serial_6
from (
    select
        apple,
        serial as serial_1,
        lead(serial,1) over (partition by apple order by serial) as serial_2,
        lead(serial,2) over (partition by apple order by serial) as serial_3,
        lead(serial,3) over (partition by apple order by serial) as serial_4,
        lead(serial,4) over (partition by apple order by serial) as serial_5,
        lead(serial,5) over (partition by apple order by serial) as serial_6,
        row_number() over (partition by apple order by serial) rn
    from data
)
where rn = 1;

---

这显然是因为具有不同序列的apple 5035在表中多次出现，并且它获得分配给它的三个序列的所有可能组合。我相信有一个简单的方法可以做到这一点，但我已经尝试了很长时间，并没有能够产生一个好的结果。有人能帮忙吗？

你可以试试这个，它不雅致，但很管用：

SELECT DISTINCT t1.apple, 
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 0,1) serial_1,
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 1,1) serial_2,
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 2,1) serial_3,
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 3,1) serial_4,
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 4,1) serial_5,
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 5,1) serial_6
FROM your_table t1

---

你可以试试这个，它不雅致，但很管用：

SELECT DISTINCT t1.apple, 
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 0,1) serial_1,
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 1,1) serial_2,
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 2,1) serial_3,
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 3,1) serial_4,
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 4,1) serial_5,
    (SELECT serial FROM your_table
     WHERE apple = t1.apple
     ORDER BY serial LIMIT 5,1) serial_6
FROM your_table t1

---

您可以尝试使用GROUP_CONCAT聚合函数和GROUP BY APPLE

SELECT 
    a.APPLE,
    GROUP_CONCAT(DISTINCT s.SERIAL) AS serials
FROM 
    apples a
    LEFT JOIN apples s ON s.APPLE = a.APPLE
GROUP BY a.APPLE

在MySQL中测试：

mysql> select * from apples;
+------+-------+--------+
| ID   | APPLE | SERIAL |
+------+-------+--------+
| 2052 |  5035 |      1 |
| 2055 |  5035 |      4 |
| 2058 |  5035 |      8 |
| 2070 |  5040 |      5 |
+------+-------+--------+
4 rows in set (0.00 sec)

mysql> SELECT 
    ->     a.APPLE,
    ->     GROUP_CONCAT(DISTINCT s.SERIAL) AS serials
    -> FROM 
    ->     apples a
    ->     LEFT JOIN apples s ON s.APPLE = a.APPLE
    -> GROUP BY a.APPLE;
+-------+---------+
| APPLE | serials |
+-------+---------+
|  5035 | 1,4,8   |
|  5040 | 5       |
+-------+---------+
2 rows in set (0.00 sec)

mysql> 

---

您可以尝试使用GROUP_CONCAT聚合函数和GROUP BY APPLE

SELECT 
    a.APPLE,
    GROUP_CONCAT(DISTINCT s.SERIAL) AS serials
FROM 
    apples a
    LEFT JOIN apples s ON s.APPLE = a.APPLE
GROUP BY a.APPLE

在MySQL中测试：

mysql> select * from apples;
+------+-------+--------+
| ID   | APPLE | SERIAL |
+------+-------+--------+
| 2052 |  5035 |      1 |
| 2055 |  5035 |      4 |
| 2058 |  5035 |      8 |
| 2070 |  5040 |      5 |
+------+-------+--------+
4 rows in set (0.00 sec)

mysql> SELECT 
    ->     a.APPLE,
    ->     GROUP_CONCAT(DISTINCT s.SERIAL) AS serials
    -> FROM 
    ->     apples a
    ->     LEFT JOIN apples s ON s.APPLE = a.APPLE
    -> GROUP BY a.APPLE;
+-------+---------+
| APPLE | serials |
+-------+---------+
|  5035 | 1,4,8   |
|  5040 | 5       |
+-------+---------+
2 rows in set (0.00 sec)

mysql> 

甲骨文11g有一个功能，似乎可以满足您的需求。我这里没有11g，但下面应该很接近：

SELECT apple, listagg(serial, ',') WITHIN GROUP (ORDER BY serial) "Serial Numbers"
  FROM tbl
 GROUP BY apple;

您无法为每个序列项获得良好的列标题，但它应该可以工作。

Oracle11g具有的功能似乎可以满足您的需要。我这里没有11g，但下面应该很接近：

SELECT apple, listagg(serial, ',') WITHIN GROUP (ORDER BY serial) "Serial Numbers"
  FROM tbl
 GROUP BY apple;

---

您无法为每个序列项获得良好的列标题，但它应该可以工作。

我会这样做：

APPLE   SERIAL_1    SERIAL_2    SERIAL_3    SERIAL_4
5035    1           4           8
5035    1           8           4
5035    4           1           8
5035    4           8           1
5035    8           1           4
5035    4           4           1
5040    5

with data as (
    select 2052 id, 5035 apple, 1 serial from dual union all
    select 2055 id, 5035 apple, 4 serial from dual union all
    select 2058 id, 5035 apple, 8 serial from dual union all
    select 2070 id, 5040 apple, 5 serial from dual
)
select
    apple,
    serial_1,
    serial_2,
    serial_3,
    serial_4,
    serial_5,
    serial_6
from (
    select
        apple,
        serial as serial_1,
        lead(serial,1) over (partition by apple order by serial) as serial_2,
        lead(serial,2) over (partition by apple order by serial) as serial_3,
        lead(serial,3) over (partition by apple order by serial) as serial_4,
        lead(serial,4) over (partition by apple order by serial) as serial_5,
        lead(serial,5) over (partition by apple order by serial) as serial_6,
        row_number() over (partition by apple order by serial) rn
    from data
)
where rn = 1;

显然，您不需要WITH块，因为您可以使用实际表，所以您的查询将从SELECT开始

这将产生以下输出：

---

---

我会这样做：

APPLE   SERIAL_1    SERIAL_2    SERIAL_3    SERIAL_4
5035    1           4           8
5035    1           8           4
5035    4           1           8
5035    4           8           1
5035    8           1           4
5035    4           4           1
5040    5

with data as (
    select 2052 id, 5035 apple, 1 serial from dual union all
    select 2055 id, 5035 apple, 4 serial from dual union all
    select 2058 id, 5035 apple, 8 serial from dual union all
    select 2070 id, 5040 apple, 5 serial from dual
)
select
    apple,
    serial_1,
    serial_2,
    serial_3,
    serial_4,
    serial_5,
    serial_6
from (
    select
        apple,
        serial as serial_1,
        lead(serial,1) over (partition by apple order by serial) as serial_2,
        lead(serial,2) over (partition by apple order by serial) as serial_3,
        lead(serial,3) over (partition by apple order by serial) as serial_4,
        lead(serial,4) over (partition by apple order by serial) as serial_5,
        lead(serial,5) over (partition by apple order by serial) as serial_6,
        row_number() over (partition by apple order by serial) rn
    from data
)
where rn = 1;

显然，您不需要WITH块，因为您可以使用实际表，所以您的查询将从SELECT开始

这将产生以下输出：

---

---

游戏进行得有点晚，但由于在Oracle中没有使用PIVOT子句的答案，因此可能会很有趣

SELECT *
FROM (
  SELECT apple, serial
  FROM fruits
) t
PIVOT ( 
  max(serial) for serial in (1,2,3,4,5,6)
) 

---

SQLFiddle示例：

游戏进行得有点晚，但由于在Oracle中没有使用PIVOT子句的答案，因此可能会很有趣

SELECT *
FROM (
  SELECT apple, serial
  FROM fruits
) t
PIVOT ( 
  max(serial) for serial in (1,2,3,4,5,6)
) 

---

---

SQLFiddle示例：

在SQL Server上，您可以使用PIVOT函数。我不确定MySql和Oracle中是否存在PIVOT。Oracle中有PIVOT函数。在SQL Server上，您可以使用PIVOT函数。我不确定MySql和Oracle中是否存在PIVOT。Oracle中有PIVOT函数。@Marco我没有，但由于担心它不起作用，我只是尝试了一下，结果成功了。看看我在答案后面附加了什么。嘿，就像上面的解决方案一样，GROUP_CONCAT函数不适用于Oracle，同样，我应该在最初的帖子中说我正在使用Oracle。无论如何，应该有一个甲骨文的等价物，我会研究一下。谢谢：）@Jamie Joe是的[mysql]标签有误导性。我试图从问题中确定您使用的是MySQL还是Oracle，但无法确定。关于Oracle中的GROUP_CONCAT，也许可以提供帮助？它不起作用：OP希望每个序列都在不同的列中，而不是一列序列都已连接@Marco它正在工作，但是它没有将序列分成单独的列。我建议OP可以尝试在查询之外进行分离。。但我的答案似乎对OP没有多大用处，因为他需要Oracle的解决方案。@Marco我没有，但因为担心它不起作用，我只是尝试了一下，它起了作用。看看我在答案后面附加了什么。嘿，就像上面的解决方案一样，GROUP_CONCAT函数不适用于Oracle，同样，我应该在最初的帖子中说我正在使用Oracle。无论如何，应该有一个甲骨文的等价物，我会研究一下。谢谢：）@Jamie Joe是的[mysql]标签有误导性。我试图从问题中确定您使用的是MySQL还是Oracle，但无法确定。关于Oracle中的GROUP_CONCAT，也许可以提供帮助？它不起作用：OP希望每个序列都在不同的列中，而不是一列序列都已连接@Marco它正在工作，但是它没有将序列分成单独的列。我建议OP可以尝试在查询之外进行分离。。但我的答案对OP来说似乎不是很有用，因为他需要一个针对Oracle的解决方案。嘿，极限函数不适用于Oracle，也许我应该在我使用Oracle的帖子中提到。无论如何，我会用谷歌搜索到oracle的LIMIT。感谢您的快速回复@JamieJoe：你应该检查你的标签：如果你在使用Oracle，为什么要放MySql？嘿，LIMIT函数不适用于Oracle，也许我应该在帖子中提到我在使用Oracle。无论如何，我会用谷歌搜索到oracle的LIMIT。感谢您的快速回复@JamieJoe：你应该检查一下你的标签：如果你在使用Oracle，为什么要放MySql？太好了，正是我需要的。非常感谢你！现在是学习代码的时候了：）太好了，正是我所需要的。非常感谢你！现在是学习代码的时候了：）