如何从多对多表中选择一对一关系

Question

我有一个数据库，用于存放书籍及其作者。 在我的模型中，一位作者可以有很多书，而一本书可以由很多作者写。 例：

AUTHOR_ID|BOOK_ID
1|100
1|200
1|300
2|300
3|300
4|400

我试图让作者只写一本书，而那本书只能由该作者自己写。 在上面的示例中，只有有效作者的AUTHOR_ID = 4。
我需要编写一个select来获取满足上述要求的作者ID，我该如何编写快速有效的select来做到这一点？

Answer 1

select *
from BookAuthors t1
where not exists (select * from BookAuthors t2
                  where t2.BOOK_ID = t1.BOOK_ID
                    and t2.Author_ID <> t1.Author_ID)
  and not exists (select * from BookAuthors t3
                  where t3.Author_ID  = t1.Author_ID
                    and t3.BOOK_ID <> t1.BOOK_ID)

第一个NOT EXISTS用来确保相同的bookid没有第二个作者。

第二个NOT EXISTS用来确保同一Author_ID没有写另一本书。

组合版本：

select *
from BookAuthors t1
where not exists (select * from BookAuthors t2
                  where (t2.BOOK_ID = t1.BOOK_ID
                         and t2.Author_ID <> t1.Author_ID)
                     or (t2.Author_ID  = t1.Author_ID
                         and t2.BOOK_ID <> t1.BOOK_ID))

Answer 2

这是一个单独的答案，因为第一个错误。

如果您的数据库支持窗口函数，则一种方法是：

select Author_ID
from (select ba.*, count(*) over (partition by Author_ID) as numBooks,
             count(*) over (partition by Book_ID) as numAuthors
      from BookAuthors ba
     ) ba
where numBooks = 1 and numAuthors = 1;

通过将条件移至where子句或join s，也可以通过多种方式进行：

select ba.*
from BookAuthors ba
where Author_ID in (select Author_Id from BookAuthors group by Author_Id having count(*) = 1) and
      Book_ID in (select Author_Id from BookAuthors group by Book_ID having count(*) = 1);