Rafy 中的 Linq 查询支持(根据聚合子条件查询聚合父)
为了提高开发者的易用性,Rafy 领域实体框架在很早开始就已经支持使用 Linq 语法来查询实体了。但是只支持了一些简单的、常用的条件查询,支持的力度很有限。特别是遇到对聚合对象的查询时,就不能再使用 Linq,而只能通过构造底层查询树的接口来完成了。由于开发者的聚合查询的需求越来越多,所以本周我们将这部分进行了增强。
接下来,本文将说明 Rafy 框架原来支持的 Linq 语法,以及最新加入的聚合查询支持及用法。
使用 Linq 查询的代码示例
public WarehouseList GetByCode(string warehouseCode, string nameKeywords, PagingInfo pagingInfo)
{
return this.FetchList(r => r.DA_GetByCode(warehouseCode, nameKeywords, pagingInfo));
}
private EntityList DA_GetByCode(string warehouseCode, string nameKeywords, PagingInfo pagingInfo)
{
var q = this.CreateLinqQuery();
//条件对比
q = q.Where(e => e.Code == warehouseCode);
if (!string.IsNullOrEmpty(nameKeywords))
{
q = q.Where(e => e.Name.Contains(nameKeywords));
}
//排序
q = q.OrderByDescending(w => w.Name);
return this.QueryList(q, pagingInfo);//以指定的分页信息 pagingInfo 分页
}
支持的一般查询
使用 CreateLinqQuery 方法创建出一个 IQueryable<Warehouse> 对象,针对该对象,我们可以以下的标准 Linq 方法:Where、OrderBy、OrderByDescending、ThenBy、ThenByDescending、Count。
对于其中最重要的 Where 方法,Rafy 也支持许多操作,包括:
- 属性的各种对比操作(=,!=,>,>=,<,<=,!,Contains,StartsWith,EndsWith等)。
- 支持两个属性条件间的连接条件:&&、||。
- 支持引用查询。即间接使用引用实体的属性来进行查询,在生成 Sql 语句时,将会生成 INNER JOIN 语句,连接上这些被使用的引用实体对应的表。例如:
q = q.Where(warehouse => warehouse.Administrator.Name == "admin");
这部分的内容,之前的版本已经支持了,各位可参见 Rafy 框架的用户手册。
聚合查询
聚合查询的功能是,开发者可以通过定义聚合子的属性的条件,来查询聚合父。这是本次升级的重点。
例如,书籍管理系统中,Book (书)为聚合根,它拥有 Chapter (章)作为它的聚合子实体,而 Chapter 下则还有 Section(节)。那么,我们可以通过这个功能,来查询类似以下需求的数据:
- 查询拥有某个章的名字的所有书籍。
要实现这种场景的查询,我们可以在仓库的数据层,使用下面的 Linq 语法:
public BookList LinqGetIfChildrenExists(string chapterName)
{
return this.FetchList(r => r.DA_LinqGetIfChildrenExists(chapterName));
}
private EntityList DA_LinqGetIfChildrenExists(string chapterName)
{
var q = this.CreateLinqQuery();
q = q.Where(book => book.ChapterList.Concrete().Any(c => c.Name == chapterName));
q = q.OrderBy(b => b.Name);
return this.QueryList(q);
}
其生成的 Sql 如下:
SELECT [T0].[Id], [T0].[Author], [T0].[BookCategoryId], [T0].[BookLocId], [T0].[Code], [T0].[Name], [T0].[Price], [T0].[Publisher] FROM [Book] AS [T0] WHERE EXISTS ( SELECT 1 FROM [Chapter] AS [T1] WHERE [T1].[BookId] = [T0].[Id] AND [T1].[Name] = @p0 ) ORDER BY [T0].[Name] ASC
- 查询每个章的名字必须满足某条件的所有书籍。
我们可以在仓库的数据层,使用下面的 Linq 语法:
public BookList LinqGetIfChildrenAll(string chapterName)
{
return this.FetchList(r => r.DA_LinqGetIfChildrenAll(chapterName));
}
private EntityList DA_LinqGetIfChildrenAll(string chapterName)
{
var q = this.CreateLinqQuery();
q = q.Where(e => e.ChapterList.Cast<Chapter>().All(c => c.Name == chapterName));
q = q.OrderBy(e => e.Name);
return this.QueryList(q);
}
生成的 SQL 是:
SELECT [T0].[Id], [T0].[Author], [T0].[BookCategoryId], [T0].[BookLocId], [T0].[Code], [T0].[Name], [T0].[Price], [T0].[Publisher] FROM [Book] AS [T0] WHERE NOT (EXISTS ( SELECT 1 FROM [Chapter] AS [T1] WHERE [T1].[BookId] = [T0].[Id] AND [T1].[Name] != @p0 )) ORDER BY [T0].[Name] ASC
- 查询某个章中所有节必须满足某条件的所有书籍。
我们可以在仓库的数据层,使用下面的 Linq 语法:
public BookList LinqGetIfChildrenExistsSectionName(string sectionName)
{
return this.FetchList(r => r.DA_LinqGetIfChildrenExistsSectionName(sectionName));
}
private EntityList DA_LinqGetIfChildrenExistsSectionName(string sectionName)
{
var q = this.CreateLinqQuery();
q = q.Where(book => book.ChapterList.Concrete().Any(c => c.SectionList.Cast<Section>().Any(s => s.Name.Contains(sectionName))));
q = q.OrderBy(b => b.Name);
return this.QueryList(q);
}
将会生成如下 SQL:
SELECT [T0].[Id], [T0].[Author], [T0].[BookCategoryId], [T0].[BookLocId], [T0].[Code], [T0].[Name], [T0].[Price], [T0].[Publisher] FROM [Book] AS [T0] WHERE EXISTS ( SELECT 1 FROM [Chapter] AS [T1] WHERE [T1].[BookId] = [T0].[Id] AND EXISTS ( SELECT 1 FROM [Section] AS [T2] WHERE [T2].[ChapterId] = [T1].[Id] AND [T2].[Name] LIKE @p0 ) ) ORDER BY [T0].[Name] ASC
- 同时,这些查询也可以支持分页。例如,我们在上面的查询添加一个分页条件,代码如下:
public BookList LinqGetIfChildrenExistsSectionName(string sectionName)
{
return this.FetchList(r => r.DA_LinqGetIfChildrenExistsSectionName(sectionName));
}
private EntityList DA_LinqGetIfChildrenExistsSectionName(string sectionName)
{
var q = this.CreateLinqQuery();
q = q.Where(book => book.ChapterList.Concrete().Any(c => c.SectionList.Cast<Section>().Any(s => s.Name.Contains(sectionName))));
q = q.OrderBy(b => b.Name);
return this.QueryList(q, new PagingInfo(2, 1));//分页
}
分成的 SQL 如下:
SELECT TOP 1 [T0].[Id], [T0].[Author], [T0].[BookCategoryId], [T0].[BookLocId], [T0].[Code], [T0].[Name], [T0].[Price], [T0].[Publisher] FROM [Book] AS [T0] WHERE EXISTS ( SELECT 1 FROM [Chapter] AS [T1] WHERE [T1].[BookId] = [T0].[Id] AND EXISTS ( SELECT 1 FROM [Section] AS [T2] WHERE [T2].[ChapterId] = [T1].[Id] AND [T2].[Name] LIKE @p0 ) ) AND [T0].[Id] NOT IN ( SELECT TOP 1 [T0].[Id] FROM [Book] AS [T0] WHERE EXISTS ( SELECT 1 FROM [Chapter] AS [T1] WHERE [T1].[BookId] = [T0].[Id] AND EXISTS ( SELECT 1 FROM [Section] AS [T2] WHERE [T2].[ChapterId] = [T1].[Id] AND [T2].[Name] LIKE @p1 ) ) ORDER BY [T0].[Name] ASC ) ORDER BY [T0].[Name] ASC
头晕,越来越复杂……不过经过测试,上面都没有什么问题。
下面是一个单元测试生成的分页、复杂聚合查询的 SQL,贴上来观赏下:
SELECT TOP 2 [T0].[Id], [T0].[Author], [T0].[BookCategoryId], [T0].[BookLocId], [T0].[Code], [T0].[Name], [T0].[Price], [T0].[Publisher] FROM [Book] AS [T0] LEFT OUTER JOIN [BookCategory] AS [T1] ON [T0].[BookCategoryId] = [T1].[Id] WHERE [T0].[Name] != @p0 AND [T1].[Name] = @p1 AND EXISTS ( SELECT 1 FROM [Chapter] AS [T2] WHERE [T2].[BookId] = [T0].[Id] AND [T2].[Name] = @p2 ) AND EXISTS ( SELECT 1 FROM [Chapter] AS [T3] WHERE [T3].[BookId] = [T0].[Id] AND [T3].[Name] = @p3 AND NOT (EXISTS ( SELECT 1 FROM [Section] AS [T4] LEFT OUTER JOIN [SectionOwner] AS [T5] ON [T4].[SectionOwnerId] = [T5].[Id] WHERE [T4].[ChapterId] = [T3].[Id] AND ([T4].[Name] NOT LIKE @p4 OR [T4].[SectionOwnerId] IS NULL OR [T5].[Name] != @p5) )) ) AND [T0].[Id] NOT IN ( SELECT TOP 4 [T0].[Id] FROM [Book] AS [T0] LEFT OUTER JOIN [BookCategory] AS [T1] ON [T0].[BookCategoryId] = [T1].[Id] WHERE [T0].[Name] != @p6 AND [T1].[Name] = @p7 AND EXISTS ( SELECT 1 FROM [Chapter] AS [T2] WHERE [T2].[BookId] = [T0].[Id] AND [T2].[Name] = @p8 ) AND EXISTS ( SELECT 1 FROM [Chapter] AS [T3] WHERE [T3].[BookId] = [T0].[Id] AND [T3].[Name] = @p9 AND NOT (EXISTS ( SELECT 1 FROM [Section] AS [T4] LEFT OUTER JOIN [SectionOwner] AS [T5] ON [T4].[SectionOwnerId] = [T5].[Id] WHERE [T4].[ChapterId] = [T3].[Id] AND ([T4].[Name] NOT LIKE @p10 OR [T4].[SectionOwnerId] IS NULL OR [T5].[Name] != @p11) )) ) ORDER BY [T0].[Name] ASC ) ORDER BY [T0].[Name] ASC
刚开始支持 Linq 查询的时候,就已经把聚合查询的单元测试给写了。鉴于比较复杂,所以一直没有实现。这周总算完成了这部分代码,心中一块石头落了地。