Question

我有一个遗留系统，它可以同时与多个数据库配合使用。假设他们是MASTER和SLAVE。

有用户组（层次结构）。每个组都有一个父组，父组可以拥有自己的父组等。

table MASTER.dbo.GROUPS: 
GROUP_ID
PARENTGROUP_ID
NAME

可以将用户分配到不同的组。

table MASTER.dbo.LK_GROUPS:
GROUP_ID
USER_ID

有实体，每个实体都有所有者（一对多）。

table SLAVE.dbo.ENTITY_OWNERSHIP
ENTITY_ID
USER_ID (column has a bad name because it is GROUP_ID)

任务：编写查询/函数/存储过程以仅选择用户有权访问的实体。应该可以使用解决方案作为sql查询的一部分，我无法修改。此外，应该可以使用2种模式：

假设实体的所有者是S1集并且用户组是S2集，如果S1-S2 = 0，则可以访问实体。
假设实体的所有者是S1集并且用户组是S2集，如果S1与S2相交，则可以访问实体。 0

所以我写了一个函数fn_checkUserRights：

create function fn_checkUserRights (@userId bigint, @entityId bigint, @compareMode int)
returns int
as
begin
    declare @result int;
    declare @count int;

    if @compareMode = 1
        begin

            with groups_with_ancestors
            as (
                select GROUP_ID, PARENTGROUP_ID, NAME
                from MASTER.dbo.GROUPS as GROUPS
                where 
                    GROUP_ID in (
                        select 
                            GROUP_ID
                        from 
                            MASTER.dbo.LK_GROUPS
                        where 
                            USER_ID = @userId
                    )

                union all

                select 
                    b.GROUP_ID, b.PARENTGROUP_ID, b.NAME
                from 
                    groups_with_ancestors a
                    inner join MASTER.dbo.GROUPS b on b.GROUP_ID = a.PARENTGROUP_ID
            ) 
            select @count = count(*) from (
                select
                    ENTITY_OWNERSHIP.USER_ID as OWNER_ID 
                from 
                    SLAVE.dbo.ENTITY_OWNERSHIP as ENTITY_OWNERSHIP
                where 
                    ENTITY_OWNERSHIP.ENTITY_ID = @entityId

                except

                select distinct GROUP_ID from groups_with_ancestors

            ) a

            if @count > 0
                set @result = 0
            else
                set @result = 1
        end

    else if @compareMode = 2
        begin

            with groups_with_ancestors
            as (
                select GROUP_ID, PARENTGROUP_ID, NAME
                from MASTER.dbo.GROUPS as GROUPS
                where GROUP_ID in (
                        select 
                            GROUP_ID
                        from 
                            MASTER.dbo.LK_GROUPS
                        where 
                            USER_ID = @userId
                    )

                union all

                select 
                    b.GROUP_ID, b.PARENTGROUP_ID, b.NAME
                from 
                    groups_with_ancestors a
                    inner join MASTER.dbo.GROUPS b on b.GROUP_ID = a.PARENTGROUP_ID
            ) 
            select @count = count(*) from (
                select
                    ENTITY_OWNERSHIP.USER_ID as OWNER_ID 
                from 
                    SLAVE.dbo.ENTITY_OWNERSHIP as ENTITY_OWNERSHIP
                where 
                    ENTITY_OWNERSHIP.ENTITY_ID = @entityId

                intersect

                select distinct GROUP_ID from groups_with_ancestors

            ) a

            if @count > 0
                set @result = 1
            else
                set @result = 0

        end


    return @result;

end;

使用示例：

select * 
from 
    SLAVE.dbo.ENTITY_OWNERSHIP as owners
where 
    MASTER.dbo.fn_checkUserRights(10010, owners.ENTITY_ID, 1) = 1

但是有一些问题：

函数使用2个DB中的表：MASTER和SLAVE。他们的名字事先不知道。我不能将名称作为参数传递，因为函数不允许动态查询执行。如何解决这个问题？
@ compareMode = 1和@ compareMode = 2的代码块几乎相同，将公共代码提取到变量或函数会很棒。你能推荐什么？

按用户访问权限过滤记录

0 个答案: