Question

我正在尝试从旧的经典ASP项目中复制任何现有查询的结果。查询非常简单。

SELECT DISTINCT TOP 10 MAX(H.HistID) as MaxHID, C.CompanyID,C.CompanyName, P.ProsID, P.ProsName, P.Online 
FROM HISTORY H, PROSPECTUS P, COMPANY C 
WHERE H.ProsID = P.ProsID 
and P.CompanyID = C.CompanyID 
and H.UserID = 2712
GROUP BY C.CompanyID, C.CompanyName, P.ProsID, P.ProsName, P.Online 
ORDER BY MaxHID DESC

我（至少试图）使用以下查询使用Entity Framework复制它：

MyContext.HistoryItems.Where(h => h.UserId == userId)
    .GroupBy(h => new { h.Prospectus.Family.Id, h.ProsId })
    .Select(h => new { Max = h.Max(i => i.Id), Item = h.FirstOrDefault() })
    .OrderByDescending(h => h.Max)
    .Take(10)
    .Select(h => h.Item);

（GroupBy()旨在复制原始查询中的DISTINCT行为。）

此查询会生成完全相同的数据，但需要大约10秒才能执行。我查看了EF生成的查询，这是一个怪物。

SELECT TOP (10) 
    [Project6].[HistId] AS [HistId], 
    [Project6].[UserID] AS [UserID], 
    [Project6].[ProsID] AS [ProsID], 
    [Project6].[HDate] AS [HDate], 
    [Project6].[ProsDocId] AS [ProsDocId]
    FROM ( SELECT 
        [Project5].[HistId] AS [HistId], 
        [Project5].[UserID] AS [UserID], 
        [Project5].[ProsID] AS [ProsID], 
        [Project5].[HDate] AS [HDate], 
        [Project5].[ProsDocId] AS [ProsDocId], 
        [Project5].[C1] AS [C1]
        FROM ( SELECT 
            [Project4].[HistId] AS [HistId], 
            [Project4].[UserID] AS [UserID], 
            [Project4].[ProsID1] AS [ProsID], 
            [Project4].[HDate] AS [HDate], 
            [Project4].[ProsDocId] AS [ProsDocId], 
            (SELECT 
                MAX([Extent5].[HistId]) AS [A1]
                FROM  [dbo].[History] AS [Extent5]
                INNER JOIN [dbo].[Prospectus] AS [Extent6] ON [Extent5].[ProsID] = [Extent6].[ProsId]
                WHERE ([Extent5].[UserID] = @p__linq__0) AND (([Project4].[CompanyId] = [Extent6].[CompanyId]) OR (1 = 0)) AND ([Project4].[ProsID] = [Extent5].[ProsID])) AS [C1]
            FROM ( SELECT 
                [Project2].[ProsID] AS [ProsID], 
                [Project2].[CompanyId] AS [CompanyId], 
                [Limit1].[HistId] AS [HistId], 
                [Limit1].[UserID] AS [UserID], 
                [Limit1].[ProsID] AS [ProsID1], 
                [Limit1].[HDate] AS [HDate], 
                [Limit1].[ProsDocId] AS [ProsDocId]
                FROM   (SELECT 
                    @p__linq__0 AS [p__linq__0], 
                    [Distinct1].[ProsID] AS [ProsID], 
                    [Distinct1].[CompanyId] AS [CompanyId]
                    FROM ( SELECT DISTINCT 
                        [Extent1].[ProsID] AS [ProsID], 
                        [Extent2].[CompanyId] AS [CompanyId]
                        FROM  [dbo].[History] AS [Extent1]
                        INNER JOIN [dbo].[Prospectus] AS [Extent2] ON [Extent1].[ProsID] = [Extent2].[ProsId]
                        WHERE [Extent1].[UserID] = @p__linq__0
                    )  AS [Distinct1] ) AS [Project2]
                OUTER APPLY  (SELECT TOP (1) 
                    [Extent3].[HistId] AS [HistId], 
                    [Extent3].[UserID] AS [UserID], 
                    [Extent3].[ProsID] AS [ProsID], 
                    [Extent3].[HDate] AS [HDate], 
                    [Extent3].[ProsDocId] AS [ProsDocId]
                    FROM  [dbo].[History] AS [Extent3]
                    INNER JOIN [dbo].[Prospectus] AS [Extent4] ON [Extent3].[ProsID] = [Extent4].[ProsId]
                    WHERE ([Extent3].[UserID] = @p__linq__0) AND (([Project2].[CompanyId] = [Extent4].[CompanyId]) OR (1 = 0)) AND ([Project2].[ProsID] = [Extent3].[ProsID]) ) AS [Limit1]
            )  AS [Project4]
        )  AS [Project5]
    )  AS [Project6]
    ORDER BY [Project6].[C1] DESC

我的Linq查询显然有问题会导致EF产生如此混乱。我创建的EF查询是否效率不高？

Answer 1

每当分组发挥作用时，我更喜欢高于语法的查询语法。

当你这样做......

(
    from h in MyContext.HistoryItems
    where h.UserId == userId)
    group h by new { 
                        h.CompanyId, 
                        Comapany = h.Company.Name, 
                        Prospectus = h.Prospectus.Name, 
                        h.Prospectus.Online, 
                        h.ProsId 
                   } into grp
    select new { 
                    Max = grp.Max(i => i.Id),
                    grp.Key.CompanyId,
                    grp.Key.Comapany,
                    grp.Key.Prospectus,
                    grp.Key.Online,
                    grp.Key.ProsId,
               } into proj
    orderby proj.Max descending
    select proj
).Distinct().Take(10)

...查询应该更清晰，因为现在你只选择将出现在最终投影中的属性，并且FirstOrDefault()没有被混淆。连接甚至不存在，因为我让EF使用h.Company等导航属性来计算联接。

请注意，我尝试重现SQL查询。 LINQ查询对我来说看起来不一样。

Answer 2

试试这个，我不确定它是100％正常工作

    MyContext.HistoryItems
        .Join(MyContext.ProspectusItems, a => a.ProsId, b => b.ProsId, (a,b) => new {a, b})
        .Join(MyContext.CompanyItems, a => a.b.CompanyId, b => b.CompanyId, (a,b) => new {a, b})
        .Where(a => a.a.a.UserId == 2712)
        .GroupBy(x => new {x.b.CompanyID, x.bCompanyName, x.a.b.ProsID, x.a.b.ProsName, x.a.b.Online)
        .Select(x => new {maxHistID = x.Max(x => x.a.a.HistID), x.Key.x.b.CompanyID, x.Key.x.bCompanyName, x.Key.x.a.b.ProsID, x.Key.x.a.b.ProsName, x.Key.x.a.b.Online).
        .OrderByDesc(x => x.MaxHID)
        .Take(10)

Answer 3

试试这个，这是一个更直接可比的实现： MyContext.HistoryItems.Where(h => h.UserId == userId) GroupBy(h => new {h.Prospectus.Family.Id, h.ProsId},h=>h.HistId,(key,vals)=>new {key.Id, key.ProsId,Max=vals.Max()}) .OrderByDescending(h => h.Max) .Take(10)

实体框架 - Distinct和Max生成怪异（和慢）查询

3 个答案: