LINQ do SQL - Grupowanie według godzin


11

Jak mogę grupować wynik zapytania LINQ do zapytania SQL przez godziny, biorąc pod uwagę, że typem kolumny jest DateTime?

23

Oto rozwiązanie dla godzin technicznych (bezkontekstowych).

var query = myDC.Orders 
    .GroupBy(x => x.OrderDate.Hour) 
    .Select(g => new { 
     Hour = g.Key, 
     Amount = g.Sum(x => x.OrderAmount) 
    }); 

który generuje to:

SELECT SUM([t1].[OrderAmount]) AS [Amount], [t1].[value] AS [Hour] 
FROM (
    SELECT DATEPART(Hour, [t0].[OrderDate]) AS [value], [t0].[OrderAmount] 
    FROM [dbo].[Orders] AS [t0] 
    ) AS [t1] 
GROUP BY [t1].[value] 

Oto rozwiązanie dla godzinach pracy (context-ful).

DateTime zeroDate = new DateTime(2008, 1, 1); 

var query = myDC.Orders 
    .GroupBy(x => System.Data.Linq.SqlClient.SqlMethods.DateDiffHour(zeroDate, x.OrderDate) 
    ) 
    .Select(g => new { Hours = g.Key, Amount = g.Sum(x => x.OrderAmount) }) 
    .Select(x => new { Hour = zeroDate.AddHours(x.Hours), Amount = x.Amount}); 

który generuje to:

SELECT DATEADD(ms, (CONVERT(BigInt,(CONVERT(Float,[t2].[value2])) * 3600000)) % 86400000, DATEADD(day, (CONVERT(BigInt,(CONVERT(Float,[t2].[value2])) * 3600000))/86400000, @p1)) AS [Hour], [t2].[value] AS [Amount] 
FROM (
    SELECT SUM([t1].[OrderAmount]) AS [value], [t1].[value] AS [value2] 
    FROM (
     SELECT DATEDIFF(Hour, @p0, [t0].[OrderDate]) AS [value], [t0].[OrderAmount] 
     FROM [dbo].[Orders] AS [t0] 
     ) AS [t1] 
    GROUP BY [t1].[value] 
    ) AS [t2] 

Fuj - że bigint/pływak/milisekundy rzeczy jest brzydkie i trudne do zweryfikowania. Wolę robić dodawanie z powrotem po stronie klienta:

var results = myDC.Orders 
    .GroupBy(x => System.Data.Linq.SqlClient.SqlMethods.DateDiffHour(zeroDate, x.OrderDate) 
    ) 
    .Select(g => new { Hours = g.Key, Amount = g.Sum(x => x.OrderAmount) }) 
    .ToList() 
    .Select(x => new { Hour = zeroDate.AddHours(x.Hours), Amount = x.Amount}); 

który generuje to:

SELECT SUM([t1].[OrderAmount]) AS [Amount], [t1].[value] AS [Hours] 
FROM (
    SELECT DATEDIFF(Hour, @p0, [t0].[OrderDate]) AS [value], [t0].[OrderAmount] 
    FROM [dbo].[Orders] AS [t0] 
    ) AS [t1] 
GROUP BY [t1].[value] 

A oto trzeci sposób robi contextful godzin. Ten jest bardzo przyjazny dla C#, ale w bazie danych jest logika łańcuchowa (fuj).

var query = myDC.Orders 
    .GroupBy(x => new DateTime(x.OrderDate.Year, x.OrderDate.Month, x.OrderDate.Day, x.OrderDate.Hour, 0, 0)) 
    .Select(g => new { Hour = g.Key, Amount = g.Sum(x => x.OrderAmount) }); 

który generuje

SELECT SUM([t1].[OrderAmount]) AS [Amount], [t1].[value] AS [Hour] 
FROM (
    SELECT CONVERT(DATETIME, (CONVERT(NCHAR(4), DATEPART(Year, [t0].[OrderDate])) + ('-' + (CONVERT(NCHAR(2), DATEPART(Month, [t0].[OrderDate])) + ('-' + CONVERT(NCHAR(2), DATEPART(Day, [t0].[OrderDate])))))) + (' ' + (CONVERT(NCHAR(2), DATEPART(Hour, [t0].[OrderDate])) + (':' + (CONVERT(NCHAR(2), @p0) + (':' + CONVERT(NCHAR(2), @p1)))))), 120) AS [value], [t0].[OrderAmount] 
    FROM [dbo].[Orders] AS [t0] 
    ) AS [t1] 
GROUP BY [t1].[value]