使用SQL更新多对多表

我有一张放动物的桌子

CREATE TABLE Animals
(
AnimalId int NOT NULL,
Color int NOT NULL,
Breed int NOT NULL,
Genre int NOT NULL,
);

和一个完全相同的表，但所有内容都是可选的（除了键）

最后是多对多表：

CREATE TABLE AnimalsExpenses
(
ExpenseId int NOT NULL FOREIGN KEY REFERENCES Expenses(ExpenseId),
AnimalId int NOT NULL FOREIGN KEY REFERENCES Animals(AnimalId),
);

记录（a，e）应放在表上，如果

((SELECT Color FROM Animals WHERE AnimalId = a) = (SELECT Color FROM Expenses WHERE ExpenseId = e)
OR (NULL) = (SELECT Color FROM Expenses WHERE ExpenseId = e))
AND
((SELECT Breed FROM Animals WHERE AnimalId = a) = (SELECT Breed FROM Expenses WHERE ExpenseId = e)
OR (NULL) = (SELECT Breed FROM Expenses WHERE ExpenseId = e))
AND
((SELECT Genre FROM Animals WHERE AnimalId = a) = (SELECT Genre FROM Expenses WHERE ExpenseId = e)
OR (NULL) = (SELECT Genre FROM Expenses WHERE ExpenseId = e))

…怎么会是一个更新动物费用的查询？也就是说：它删除不应该在n-m表中的记录，并添加需要在那里的记录

例如：

------ Animals -------------------
 AnimalId  Color  Breed  Genre
----------------------------------
     1      1       1      1
     2      1       1      2
     3      1       2      2

----- Expenses -------------------
 ExpenseId  Color  Breed  Genre
----------------------------------
     1       NULL   NULL   NULL      (applies to every animal)
     2       NULL    2     NULL      (applies to animals of breed 2)
     3         1     2      2        (applies exactly to animal 3)

----- AnimalsExpenses -------------------------------------------
  AnimalId   ExpenseId   Is it ok?
-----------------------------------------------------------------
      1        1         yes, because "expense 1" applies to all
      2        1         yes, because "expense 1" applies to all
      3        1         yes, because "expense 1" applies to all

      1        2         no, breed doesnt match
      2        2         no, breed doesnt match
      3        2         yes, because "expense 2" matches breed with "animal 3"

      1        3         no, breed and genre doesnt match
      2        3         no, breed doesnt match
      3        3         yes, everything matches

---

为什么不直接截断表，然后运行

Insert into AnimalsExpenses
select a.AnimalId
  , e.ExpenseId
  from Animals a
  inner join Expenses e on a.Breed = ISNULL(e.Breed, a.Breed) 
          AND a.Color = ISNULL(e.Color, a.Color) 
          AND a.Genre = ISNULL(e.Genre, a.Genre) 

---

我建议您以这种方式重写

WHERE

标准：

ISNULL(Expenses.Color, Animals.Color) = Animals.Color
AND ISNULL(Expenses.Breed, Animals.Breed) = Animals.Breed
AND ISNULL(Expenses.Genre, Animals.Genre) = Animals.Genre

---

再看看

COALESCE

您使用的是哪家SQL供应商？我可能会在应用程序层而不是在database.SQL服务器中强制执行此操作。我使用LINQ/Entity Framework进行了此更新，但速度太慢了。您是否可以包含一些示例数据，并向我们展示一个允许和禁止的条目是什么样子？我已经给出了一个示例。您的逻辑是，

Animals

表的所有列都具有相同的值，还是品种匹配？否则会丢弃记录吗？解释“应用于所有”的含义。Truncate表示删除所有行？我会试试。。。听起来是个好答案。它在1000只动物和20个开支（aprox）情况下的性能如何？@sports Truncate意味着删除所有行，这是一个命令。应该不会太糟糕。顺便说一下，我更新了答案，ORs需要更新

ISNULL(Expenses.Color, Animals.Color) = Animals.Color
AND ISNULL(Expenses.Breed, Animals.Breed) = Animals.Breed
AND ISNULL(Expenses.Genre, Animals.Genre) = Animals.Genre