实体框架在有序列表中选择项目位置

Question

我想在有序列表中选择用户的位置。

我只能通过将所有项目加载到内存中然后使用IndexAt() 方法来查找位置来使其工作。但是，当数据库中有很多行时，这并不能很好地工作。

        public static async Task<int> GetUsersRank(DbEntities db, string userId)
    {           
      var items = await db.UserIqAnswers.Where(x => x.IqQuestion.CorrectAnswer == x.Answer).GroupBy(x => x.UserId).Select(x => new { userId = x.Key, points = x.Sum(y => y.IqQuestion.Points) })
         .OrderBy(x => x.points)
         .ToListAsync();
      return items.FindIndex(x => x.userId == userId) + 1;         
    }

我怎样才能更有效地做到这一点？

Answer 1

这样的？ 这将运行两个查询。一个加载当前用户的信息，另一个加载列表中的位置。 不过，查询应该相当有效。第二个应该作为 SQL 服务器端的 COUNT() 来完成。

var sumByUser = db.UserIqAnswers.Where(x => x.IqQuestion.CorrectAnswer == x.Answer)
    .GroupBy(x => x.UserId)
    .Select(x => new { userId = x.Key, points = x.Sum(y => y.IqQuestion.Points) });
var currentUser = sumByUser.Where(x => x.userId == userId).Single();
var rank = sumByUser.Where(x => x.points > currentUser.points).Count();

请注意，“sumByUser”查询永远不会执行，它只是用作接下来两个查询的基础。

您可以使用 LINQ 语法将其重写为作为一个查询运行：

(from currentUser in sumByUser.Where(x => x.userId == userId)
 from rank in sumByUser.Where(x => x.points > currentUser.points
 select new { currentUser, rank = rank.Count()}).Single()

但我会检查那个生成的 SQL 以确定。

Answer 2

answer by gnud 提出了正确的想法，但不幸的是，提供的两种解决方案都不能很好地转换为 SQL。这是因为 EF6 查询翻译仍然对您编写 LINQ 查询的方式很敏感（很遗憾）。

这是一个实现（通过实验找到），它产生了良好的转换（将 decimal? 强制转换为 Points 属性类型的可空类型 - 如果集合为空，需要强制转换以避免 NRE） ：

public static async Task<int> GetUsersRank(DbEntities db, string userId)
{
    var userPoints = await db.UserIqAnswers
        .Where(x => x.UserId == userId && x.IqQuestion.CorrectAnswer == x.Answer)
        .SumAsync(x => (decimal?)x.IqQuestion.Points) ?? 0;

    var rank = await db.UserIqAnswers
        .Select(x => new { x.UserId, x.Answer, x.IqQuestion })
        .Where(x => x.UserId != userId && x.IqQuestion.CorrectAnswer == x.Answer)
        .GroupBy(x => x.UserId)
        .Select(x => new { userId = x.Key, points = x.Sum(y => y.IqQuestion.Points) })
        .CountAsync(x => x.points < userPoints || (x.points == userPoints && string.Compare(x.userId, userId) < 0));

    return rank;
}

生成 2 个这样的 SQL：

SELECT
    [GroupBy1].[A1] AS [C1]
    FROM ( SELECT
        SUM([Extent2].[Points]) AS [A1]
        FROM  [dbo].[UserIqAnswers] AS [Extent1]
        INNER JOIN [dbo].[IqQuestions] AS [Extent2] ON ([Extent1].[Answer] = [Extent2].[CorrectAnswer]) AND ([Extent1].[IqQuestion_Id] = [Extent2].[Id])
        WHERE [Extent1].[UserId] = @p__linq__0
    )  AS [GroupBy1]

和

SELECT
    [GroupBy2].[A1] AS [C1]
    FROM ( SELECT
        COUNT(1) AS [A1]
        FROM ( SELECT
            [Extent1].[UserId] AS [K1],
            SUM([Extent2].[Points]) AS [A1]
            FROM  [dbo].[UserIqAnswers] AS [Extent1]
            INNER JOIN [dbo].[IqQuestions] AS [Extent2] ON ([Extent1].[IqQuestion_Id] = [Extent2].[Id]) AND ([Extent2].[CorrectAnswer] = [Extent1].[Answer])
            WHERE [Extent1].[UserId] <> @p__linq__0
            GROUP BY [Extent1].[UserId]
        )  AS [GroupBy1]
        WHERE ([GroupBy1].[A1] < @p__linq__1) OR (([GroupBy1].[A1] = @p__linq__2) AND ([GroupBy1].[K1] < @p__linq__3))
    )  AS [GroupBy2]

Answer 3

使用Skip：获取元素501 ..

queryable.Skip(500).Take(1).ToList();