连接表并选择第一个表中所有行都满足条件的外键

Question

我有两张桌子

帐户

ID  | Deleted? |  Type
1   |     0    |   Father
2   |     0    |   Son
3   |     1    |   Son
4   |     1    |   Son
5   |     0    |   Father
6   |     0    |   Father
7   |     1    |   Son
8   |     0    |   Son
9   |     0    |   Father
10  |     1    |   Son

Rel_Accounts

ID  | SON | FATHER 
 1  |  4  |   6
 2  |  3  |   6
 3  |  2  |   5
 4  |  4  |   1
 5  |  7  |   1
 6  |  8  |   9
 7  |  10 |   9

我只想选择活跃（已删除 =0）的父亲 ID，他们确实有 SON 已删除 = 1：

FATHERS
6
1

当 FATHER = 0 但他的所有 SONs Deleted = 1 时，您如何获得这些记录？

我尝试了以下方法，但没有成功：

SELECT A.ID,
    case when A.DELETED = 0
    THEN (SELECT AH.SONS FROM ACCOUNTS_REL AH WHERE AH.FATHER = A.ID AND A.DELETED = 1)
    END
FROM ACCOUNTS A 
WHERE A.TYPE = 'Father'

预期结果是 1 和 6，因为他们是活跃的父亲，并且 所有他们的儿子都被删除。

Answer 1

如果你想要所有活跃的父亲，他的所有儿子都删除了这等于说所有活跃的父亲有儿子但没有活跃的儿子，那么试试下面：

select distinct rc.father
    from accounts a 
    join rel_accounts rc on a.id=rc.father
where a.deleted=0 
      and rc.father not in 
     (select qrc.father
             from accounts qa 
             join rel_accounts qrc on qa.id=qrc.father
      where qa.deleted=0 
            and qrc.son in 
           (select qracc.son
                   from rel_accounts qracc 
                   join accounts qacc on qracc.son=qacc.id
                   where qacc.deleted=0))
order by father desc

见SQLFIDDLE DEMO

Answer 2

在这种情况下，您需要使用多个连接来创建所需的数据集。如果我对您的理解正确，您希望过滤掉从结果集中删除的帐户记录，然后只返回代表有儿子的父亲的行。这样的事情就足够了：

Select distinct F.id from accounts F
join rel_accounts R on R.father=F.id
join accounts S on S.id=R.son
where F.deleted=0 and S.deleted=0;

联接本身会过滤掉您不想要的结果，然后您可以简单地从结果集中排除已删除的行。

其他人也许可以为你拼凑出一个稍微干净一点的版本。

Answer 3

正如我一直建议的那样，把它分解成几块，然后重新组合起来。

首先获取活跃父亲的列表：

SELECT id
FROM accounts
WHERE deleted = 0 AND type = 'Father';

了解不活跃的儿子也很重要：

SELECT id
FROM accounts
WHERE deleted = 1 AND type = 'Son'

现在，只需从 rel_accounts 表中提取，其中父亲在第一个查询中，儿子在第二个查询中：

SELECT *
FROM rel_accounts
WHERE father IN(
  SELECT id
  FROM accounts
  WHERE deleted = 0 AND type = 'Father')
 AND son IN(
   SELECT id
   FROM accounts
   WHERE deleted = 1 AND type = 'Son');

编辑我认为这是解决方案，但经过测试，我意识到这并不能检查 每个 儿子是否已被删除。为此，我首先得到了每个父亲的儿子数量：

SELECT father, COUNT(*) AS numSons
FROM rel_accounts
GROUP BY father;

我使用相同的方法来获取每个活跃父亲的被删除儿子的数量：

SELECT father, COUNT(*) AS numDeletedSons
FROM rel_accounts
WHERE father IN(
  SELECT id
  FROM accounts
  WHERE deleted = 0 AND type = 'Father')
 AND son IN(
   SELECT id
   FROM accounts
   WHERE deleted = 1 AND type = 'Son')
GROUP BY father;

然后我使用了一个连接到我的最后一个子查询。如果父亲删除的儿子数量与儿子总数相同，则应包括在内：

SELECT a.father
FROM(
  SELECT father, COUNT(*) AS numDeletedSons
  FROM rel_accounts
  WHERE father IN(
    SELECT id
    FROM accounts
    WHERE deleted = 0 AND type = 'Father')
  AND son IN(
    SELECT id
    FROM accounts
    WHERE deleted = 1 AND type = 'Son')
  GROUP BY father) a
JOIN(
  SELECT father, COUNT(*) AS numSons
  FROM rel_accounts
  GROUP BY father) b ON b.father = a.father AND b.numSons = a.numDeletedSons;

我在SQL Fiddle 中得到了您的预期结果。