AppBox升级进行时 - Any与All的用法（Entity Framework）

AppBox 是基于 FineUI 的通用权限管理框架，包括用户管理、职称管理、部门管理、角色管理、角色权限管理等模块。

属于某个角色的用户列表（Any的用法）

使用Subsonic，我们有两种方法获取属于某个角色的用户列表，分别是表关联和子查询。

Subsonic的表关联实现：

// 查询 X_User 表
SqlQuery q = new Select().From<XUser>().InnerJoin(XRoleUser.UserIdColumn, XUser.IdColumn);
q.Where("1").IsEqualTo("1");

// 在用户名称中搜索
string searchText = ttbSearchUser.Text.Trim();
if (!String.IsNullOrEmpty(searchText))
{
	q.And(XUser.NameColumn).ContainsString(searchText);
}

// 过滤选中角色下的所有用户
object[] values = Grid1.DataKeys[Grid1.SelectedRowIndexArray[0]];
int roleId = Convert.ToInt32(values[0]);
q.And(XRoleUser.RoleIdColumn).IsEqualTo(roleId);

// 在查询添加之后，排序和分页之前获取总记录数
// Grid1总共有多少条记录
Grid2.RecordCount = q.GetRecordCount();

// 排列
q.OrderBys.Add(GetSortExpression(Grid2, XUser.Schema));

// 数据库分页
q.Paged(Grid2.PageIndex + 1, Grid2.PageSize);
XUserCollection items = q.ExecuteAsCollection<XUserCollection>();

Grid2.DataSource = items;
Grid2.DataBind();

Subsonic的子查询实现：

// 查询 X_User 表
SqlQuery q = new Select().From<XUser>();
q.Where("1").IsEqualTo("1");

// 在用户名称中搜索
string searchText = ttbSearchUser.Text.Trim();
if (!String.IsNullOrEmpty(searchText))
{
	q.And(XUser.NameColumn).ContainsString(searchText);
}

// 过滤选中角色下的所有用户
object[] values = Grid1.DataKeys[Grid1.SelectedRowIndexArray[0]];
int roleId = Convert.ToInt32(values[0]);
SqlQuery subQ = new Select(XRoleUser.UserIdColumn).From<XRoleUser>().Where(XRoleUser.RoleIdColumn).IsEqualTo(roleId);

q.And(XUser.IdColumn).In(subQ);

// 在查询添加之后，排序和分页之前获取总记录数
// Grid1总共有多少条记录
Grid2.RecordCount = q.GetRecordCount();

// 排列
q.OrderBys.Add(GetSortExpression(Grid2, XUser.Schema));

// 数据库分页
q.Paged(Grid2.PageIndex + 1, Grid2.PageSize);
XUserCollection items = q.ExecuteAsCollection<XUserCollection>();

Grid2.DataSource = items;
Grid2.DataBind();

　　

使用Entity Framework就不能从数据库的角度思考问题，而应该从实体类之间的关系考虑问题，具体的实现：

IQueryable<User> q = DB.Users;

// 在用户名称中搜索
string searchText = ttbSearchUser.Text.Trim();
if (!String.IsNullOrEmpty(searchText))
{
	q = q.Where(u => u.Name.Contains(searchText));
}

// 过滤选中角色下的所有用户
object[] values = Grid1.DataKeys[Grid1.SelectedRowIndexArray[0]];
int roleId = Convert.ToInt32(values[0]);
q = q.Where(u => u.Roles.Any(r => r.ID == roleId));

// 在查询添加之后，排序和分页之前获取总记录数
Grid2.RecordCount = q.Count();

// 排列和分页
q = SortAndPage<User>(q, Grid2);

Grid2.DataSource = q;
Grid2.DataBind();

　　

这里用到了 Any 方法，可以这么理解：检索一些用户，只要用户的任意一个角色是roleId就可以。

来看下生成的SQL语句：

exec sp_executesql N'SELECT TOP (20)
[Project2].[ID] AS [ID],
[Project2].[Name] AS [Name],
[Project2].[Email] AS [Email],
[Project2].[Password] AS [Password],
[Project2].[Enabled] AS [Enabled],
[Project2].[Gender] AS [Gender],
[Project2].[ChineseName] AS [ChineseName],
[Project2].[EnglishName] AS [EnglishName],
[Project2].[Photo] AS [Photo],
[Project2].[QQ] AS [QQ],
[Project2].[CompanyEmail] AS [CompanyEmail],
[Project2].[OfficePhone] AS [OfficePhone],
[Project2].[OfficePhoneExt] AS [OfficePhoneExt],
[Project2].[HomePhone] AS [HomePhone],
[Project2].[CellPhone] AS [CellPhone],
[Project2].[Address] AS [Address],
[Project2].[Remark] AS [Remark],
[Project2].[IdentityCard] AS [IdentityCard],
[Project2].[Birthday] AS [Birthday],
[Project2].[TakeOfficeTime] AS [TakeOfficeTime],
[Project2].[LastLoginTime] AS [LastLoginTime],
[Project2].[CreateTime] AS [CreateTime],
[Project2].[DeptID] AS [DeptID]
FROM ( SELECT [Project2].[ID] AS [ID], [Project2].[Name] AS [Name], [Project2].[Email] AS [Email], [Project2].[Password] AS [Password], [Project2].[Enabled] AS [Enabled], [Project2].[Gender] AS [Gender], [Project2].[ChineseName] AS [ChineseName], [Project2].[EnglishName] AS [EnglishName], [Project2].[Photo] AS [Photo], [Project2].[QQ] AS [QQ], [Project2].[CompanyEmail] AS [CompanyEmail], [Project2].[OfficePhone] AS [OfficePhone], [Project2].[OfficePhoneExt] AS [OfficePhoneExt], [Project2].[HomePhone] AS [HomePhone], [Project2].[CellPhone] AS [CellPhone], [Project2].[Address] AS [Address], [Project2].[Remark] AS [Remark], [Project2].[IdentityCard] AS [IdentityCard], [Project2].[Birthday] AS [Birthday], [Project2].[TakeOfficeTime] AS [TakeOfficeTime], [Project2].[LastLoginTime] AS [LastLoginTime], [Project2].[CreateTime] AS [CreateTime], [Project2].[DeptID] AS [DeptID], row_number() OVER (ORDER BY [Project2].[Name] DESC) AS [row_number]
	FROM ( SELECT
		[Extent1].[ID] AS [ID],
		[Extent1].[Name] AS [Name],
		[Extent1].[Email] AS [Email],
		[Extent1].[Password] AS [Password],
		[Extent1].[Enabled] AS [Enabled],
		[Extent1].[Gender] AS [Gender],
		[Extent1].[ChineseName] AS [ChineseName],
		[Extent1].[EnglishName] AS [EnglishName],
		[Extent1].[Photo] AS [Photo],
		[Extent1].[QQ] AS [QQ],
		[Extent1].[CompanyEmail] AS [CompanyEmail],
		[Extent1].[OfficePhone] AS [OfficePhone],
		[Extent1].[OfficePhoneExt] AS [OfficePhoneExt],
		[Extent1].[HomePhone] AS [HomePhone],
		[Extent1].[CellPhone] AS [CellPhone],
		[Extent1].[Address] AS [Address],
		[Extent1].[Remark] AS [Remark],
		[Extent1].[IdentityCard] AS [IdentityCard],
		[Extent1].[Birthday] AS [Birthday],
		[Extent1].[TakeOfficeTime] AS [TakeOfficeTime],
		[Extent1].[LastLoginTime] AS [LastLoginTime],
		[Extent1].[CreateTime] AS [CreateTime],
		[Extent1].[DeptID] AS [DeptID]
		FROM [dbo].[Users] AS [Extent1]
		WHERE (N''admin'' <> [Extent1].[Name]) AND ( EXISTS (SELECT
			1 AS [C1]
			FROM [dbo].[RoleUsers] AS [Extent2]
			WHERE ([Extent1].[ID] = [Extent2].[UserID]) AND ([Extent2].[RoleID] = @p__linq__0)
		))
	)  AS [Project2]
)  AS [Project2]
WHERE [Project2].[row_number] > 0
ORDER BY [Project2].[Name] DESC',N'@p__linq__0 int',@p__linq__0=1
go

　　

EF生成的SQL语句是很复杂，我们来稍微简化一下（为了看清本质，去掉了排序，过滤以及返回字段的个数等）：

exec sp_executesql N'SELECT [Project2].[ID] AS [ID], [Project2].[Name] AS [Name], [Project2].[Email] AS [Email], [Project2].[Password] AS [Password]
	FROM ( SELECT
		[Extent1].[ID] AS [ID],
		[Extent1].[Name] AS [Name],
		[Extent1].[Email] AS [Email],
		[Extent1].[Password] AS [Password]
		FROM [dbo].[Users] AS [Extent1]
		WHERE (N''admin'' <> [Extent1].[Name]) AND ( EXISTS (SELECT
			1 AS [C1]
			FROM [dbo].[RoleUsers] AS [Extent2]
			WHERE ([Extent1].[ID] = [Extent2].[UserID]) AND ([Extent2].[RoleID] = @p__linq__0)
		))
	)  AS [Project2]',N'@p__linq__0 int',@p__linq__0=1
go

进一步简化：

exec sp_executesql N' SELECT
		[Extent1].[ID] AS [ID],
		[Extent1].[Name] AS [Name],
		[Extent1].[Email] AS [Email],
		[Extent1].[Password] AS [Password]
		FROM [dbo].[Users] AS [Extent1]
		WHERE (N''admin'' <> [Extent1].[Name]) AND ( EXISTS (SELECT
			1 AS [C1]
			FROM [dbo].[RoleUsers] AS [Extent2]
			WHERE ([Extent1].[ID] = [Extent2].[UserID]) AND ([Extent2].[RoleID] = @p__linq__0)
		))
	',N'@p__linq__0 int',@p__linq__0=1
go

进一步简化：

SELECT
		[Users].[ID] AS [ID],
		[Users].[Name] AS [Name],
		[Users].[Email] AS [Email],
		[Users].[Password] AS [Password]
		FROM [dbo].[Users] AS [Users]
		WHERE EXISTS (SELECT
			1 AS [C1]
			FROM [dbo].[RoleUsers] AS [RoleUsers]
			WHERE ([Users].[ID] = [RoleUsers].[UserID]) AND ([RoleUsers].[RoleID] = 1)
		)

进一步简化：

SELECT
		ID,
		Name,
		Email,
		Password
		FROM Users
		WHERE EXISTS (SELECT
			1
			FROM RoleUsers
			WHERE (Users.ID = RoleUsers.UserID) AND (RoleUsers.RoleID = 1)
		)

　　

最终，我们看到了Entity Framework使用子查询和 EXISTS 关键字来完成 Any 的操作。

当然，如果是我们自己写SQL，可以使用 IN 关键字来达到相同的效果：

SELECT
		ID,
		Name,
		Email,
		Password
		FROM Users
		WHERE ID IN (SELECT
			UserID
			FROM RoleUsers
			WHERE (Users.ID = RoleUsers.UserID) AND (RoleUsers.RoleID = 1)
		)

也可以使用关联查询达到相同的效果：

SELECT
		ID,
		Name,
		Email,
		Password
		FROM Users
		INNER JOIN RoleUsers
		ON (Users.ID = RoleUsers.UserID) AND (RoleUsers.RoleID = 1)

　　

幸运的是，我么只需要一个 Any 关键字就完成了这个稍微复杂的查询。　　

不属于某个角色的用户列表（All的用法）

类似的，查询不属于某个角色的用户列表（用来添加用户到某个角色的UI界面中），使用Subsonic也有两种方法，我们只看下子查询的方式：

SqlQuery q = new Select().From<XUser>();
q.Where("1").IsEqualTo("1");

// 在职务名称中搜索
string searchText = ttbSearchMessage.Text.Trim();
if (!String.IsNullOrEmpty(searchText))
{
    q.And(XUser.NameColumn).ContainsString(searchText);
}

// 排除已经属于本角色的用户
int currentRoleId = GetQueryIntValue("id");
SqlQuery subQ = new Select(XRoleUser.UserIdColumn).From<XRoleUser>().Where(XRoleUser.RoleIdColumn).IsEqualTo(currentRoleId);

q.And(XUser.IdColumn).NotIn(subQ);

// 在查询添加之后，排序和分页之前获取总记录数
Grid1.RecordCount = q.GetRecordCount();

// 排列
q.OrderBys.Add(GetSortExpression(Grid1, XUser.Schema));

// 数据库分页
q.Paged(Grid1.PageIndex + 1, Grid1.PageSize);
XUserCollection items = q.ExecuteAsCollection<XUserCollection>();

Grid1.DataSource = items;
Grid1.DataBind();

　　

使用Entity Framework，我们只需借助 All 关键字就能简单实现：

IQueryable<User> q = DB.Users;

// 在职务名称中搜索
string searchText = ttbSearchMessage.Text.Trim();
if (!String.IsNullOrEmpty(searchText))
{
    q = q.Where(u => u.Name.Contains(searchText));
}

// 排除已经属于本角色的用户
int currentRoleId = GetQueryIntValue("id");
q = q.Where(u => u.Roles.All(r => r.ID != currentRoleId));

// 在查询添加之后，排序和分页之前获取总记录数
Grid1.RecordCount = q.Count();

// 排列和分页
q = SortAndPage<User>(q, Grid1);

Grid1.DataSource = q;
Grid1.DataBind();

　　

可以简单的理解：检索一些用户，要保证这些用户的所有角色没有一个是currentRoleId。

　　

去除分页和排序后，生成的SQL语句为：

exec sp_executesql N'SELECT
[Extent1].[ID] AS [ID],
[Extent1].[Name] AS [Name],
[Extent1].[Email] AS [Email],
[Extent1].[Password] AS [Password],
[Extent1].[Enabled] AS [Enabled],
[Extent1].[Gender] AS [Gender],
[Extent1].[ChineseName] AS [ChineseName],
[Extent1].[EnglishName] AS [EnglishName],
[Extent1].[Photo] AS [Photo],
[Extent1].[QQ] AS [QQ],
[Extent1].[CompanyEmail] AS [CompanyEmail],
[Extent1].[OfficePhone] AS [OfficePhone],
[Extent1].[OfficePhoneExt] AS [OfficePhoneExt],
[Extent1].[HomePhone] AS [HomePhone],
[Extent1].[CellPhone] AS [CellPhone],
[Extent1].[Address] AS [Address],
[Extent1].[Remark] AS [Remark],
[Extent1].[IdentityCard] AS [IdentityCard],
[Extent1].[Birthday] AS [Birthday],
[Extent1].[TakeOfficeTime] AS [TakeOfficeTime],
[Extent1].[LastLoginTime] AS [LastLoginTime],
[Extent1].[CreateTime] AS [CreateTime],
[Extent1].[DeptID] AS [DeptID]
FROM [dbo].[Users] AS [Extent1]
WHERE  NOT EXISTS (SELECT
	1 AS [C1]
	FROM [dbo].[RoleUsers] AS [Extent2]
	WHERE ([Extent1].[ID] = [Extent2].[UserID]) AND (([Extent2].[RoleID] = @p__linq__0) OR (CASE WHEN ([Extent2].[RoleID] <> @p__linq__0) THEN cast(1 as bit) WHEN ([Extent2].[RoleID] = @p__linq__0) THEN cast(0 as bit) END IS NULL))
)',N'@p__linq__0 int',@p__linq__0=1
go

　　

最终简化为：

SELECT
ID,
Name,
Email,
Password
FROM Users
WHERE  NOT EXISTS (SELECT
	1
	FROM RoleUsers
	WHERE (ID = RoleUsers.UserID) AND ((RoleUsers.RoleID = 1) OR (CASE WHEN (RoleUsers.RoleID <> 1) THEN cast(1 as bit) WHEN (RoleUsers.RoleID = 1) THEN cast(0 as bit) END IS NULL))
)

　　

按照我的理解，其中：

CASE WHEN (RoleUsers.RoleID <> 1) THEN cast(1 as bit) WHEN (RoleUsers.RoleID = 1) THEN cast(0 as bit) END IS NULL

类似于下面的判断：

RoleUsers.RoleID  IS NULL

只是不知道为啥会生成这么令人费解的代码。因为如果 RoleUsers.RoleID为NULL的话，既不会走进第一个WHEN，也不会走进第二个WHEN，自然就是NULL IS NULL为true了。

下面简单写个SELECT来验证我的想法：

select CASE WHEN (null <> 1)
					THEN cast(1 as bit)
					WHEN (null = 1)
					THEN cast(0 as bit)
			END 

　　

注意，这个NULL<>1的结果是FALSE，NULL=1的结果也是FALSE，所以最终的结果才是NULL。

再来看一个简单的SELECT查询：

　　

下载或捐赠AppBox

1. AppBox v2.1 是免费软件，免费提供下载：http://fineui.com/bbs/forum.php?mod=viewthread&tid=3788

2. AppBox v3.0 是捐赠软件，你可以通过捐赠作者来获取AppBox v3.0的全部源代码（http://fineui.com/donate/）。

日寇忘我之心不死，同志尚需警惕！纪念九一八。