在实体框架中将 Sum(case) SQL 语句转换为 LINQ 查询答案

【问题标题】：Convert Sum(case) SQL statement to LINQ query in Entity Framework在实体框架中将 Sum(case) SQL 语句转换为 LINQ 查询
【发布时间】：2016-09-20 21:35:51
【问题描述】：

下面是我希望转换为 LINQ 查询的 SQL 查询

SELECT 
    ProcessName,
    SUM(CASE WHEN Status = 'In-Progress' THEN 1 ELSE 0 END) As 'In-Progress',
    SUM(CASE WHEN Status = 'Success' THEN 1 ELSE 0 END) As 'Completed',
    Count(CASE WHEN status in ('In-Progress','Success') then 1 end) as Total
FROM
    TableName
GROUP BY 
    ProcessName

【问题讨论】：

标签： sql-server entity-framework linq

【解决方案1】：

第一部分很简单。 SQL 表达式SUM(CASE WHEN condition THEN 1 ELSE 0 END) 直接映射到LINQ Sum(condition ? 1 : 0)。

更有趣的是 SQL 表达式COUNT(CASE WHEN condition THEN 1 END。这是COUNT(CASE WHEN condition THEN 1 ELSE NULL END 的快捷方式。现在，考虑到 SQL COUNT 函数会跳过 NULL 值，LINQ 映射可以是 Count(condition) 或 Sum(condition ? 1 : 0)。根据我使用 EF 的经验，后者转换为更好的 SQL（前者生成从源表读取的附加子查询），所以我总是更喜欢在 LINQ to Entities 查询中使用它。

话虽如此，LINQ 查询可能是这样的：

// constants to avoid typos
const string StatusInProgress = "In-Progress";
const string StatusCompleted = "Success";
// needed for IN clause 
var statuses = new[] { StatusInProgress, StatusCompleted };
// the query
var query =
    from e in db.TableA
    group e by e.ProcessName into g
    select new
    {
        InProgress = g.Sum(e => e.Status == StatusInProgress ? 1 : 0),
        Completed = g.Sum(e => e.Status == StatusCompleted ? 1 : 0),
        C = g.Sum(e => statuses.Contains(e.Status) ? 1 : 0),
    };

生成的 SQL 如下所示：

SELECT
    1 AS [C1],
    [GroupBy1].[K1] AS [ProcessName],
    [GroupBy1].[A1] AS [C2],
    [GroupBy1].[A2] AS [C3],
    [GroupBy1].[A3] AS [C4]
    FROM ( SELECT
        [Extent1].[K1] AS [K1],
        SUM([Extent1].[A1]) AS [A1],
        SUM([Extent1].[A2]) AS [A2],
        SUM([Extent1].[A3]) AS [A3]
        FROM ( SELECT
            [Extent1].[ProcessName] AS [K1],
            CASE WHEN (N'In-Progress' = [Extent1].[Status]) THEN 1 ELSE 0 END AS [A1],
            CASE WHEN (N'Success' = [Extent1].[Status]) THEN 1 ELSE 0 END AS [A2],
            CASE WHEN ([Extent1].[Status] IN (N'In-Progress', N'Success')) THEN 1 ELSE 0 END AS [A3]
            FROM [dbo].[TableName] AS [Extent1]
        )  AS [Extent1]
        GROUP BY [K1]
    )  AS [GroupBy1]

如您所见，除了丑陋的外部部分什么都不做外，最里面的子查询中包含的主要部分几乎与所讨论的 SQL 查询相同。

【讨论】：

【解决方案2】：

你可以这样做：

   var statuses = new {"In-Progress","Success"};


var res = yourData
          .GroupBy(m => m.ProcessName)
          .Select(g => new {
             ProcessName = g.Key,

             //for InProgress, 2 (or 3) ways
             InProgress = g.Select(x => x.Status == "In-Progress" ? 1 : 0).Sum(),
             //another way
             InProgress = g.Where(x => x.Status == "In-Progress").Count(),
             //another way, not sure if it works in linq to entities
             InProgress = g.Count(x => x.Status == "In-Progress"),

             //same 3 ways possible for completed, the first here
             Completed = g.Select(x => x.status == "Success" ? 1 : 0).Sum(),

             //for the count
             //not sure if this one works in linq to entities
             Total = g.Count(x => statuses.Contains(x.Status))
             //alternative
             Total = g.Where(x => statuses.Contains(x.Status)).Count()
          });

【讨论】：

【解决方案3】：

 string statuses = new {"In-Progress","Success"};

var results = from item in TableName
              group item by p.ProcessName into g
              select new yourReturnType 
              { 
                 ProcessName = g.Key,
                 In-Progress = g.Where(l=>l.Status == "In-Progress").Count(),
                 Completed = g.Where(l=>l.Status == "Success").Count(),                     
                 Total = g.Where(x => statuses.Contains(x.Status)).Count()

              };

【讨论】：