EntityFramework GroupBy与DBFunctions或替代

Question

I have 100 rows of data with different dates. I would like to group by the result by same dates each 30 minutes apart.

Instead:
Result 1, 2016-02-02 13:00:24
Result 1, 2016-02-02 13:01:24
Result 1, 2016-02-02 13:02:24
Result 1, 2016-02-02 13:33:24

Needed:

Result 1, 2016-02-02 13:00:24
Result 1, 2016-02-02 13:33:24

Original query:

    return await loc.Where(p => p.ReadTime >= df && p.ReadTime <= dt)
        .OrderBy(p => p.ReadTime)
        .ProjectTo<LocationModel>().ToListAsync();

What I tried:

    return await loc.Where(p => p.ReadTime >= df && p.ReadTime <= dt)
        .GroupBy(p => DbFunctions.TruncateTime(p.ReadTime))
        .Select(p => new LocationModel
        {
            Lng = p.FirstOrDefault().Lng,
            Lat = p.FirstOrDefault().Lat
        })
        .ToListAsync();

Stack trace error:

.ThrowForNonSuccess(Task task)\\r\\n at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification (Task task)\\r\\n at System.Runtime.CompilerServices.TaskAwaiter 1.GetResult()\\r\\n at System.Web.Http .Dispatcher.HttpControllerDispatcher.<SendAsync>d__1.MoveNext()","innerException":{"message":"An error has occurred.","exceptionMessage":"Timeout expired. The timeout period elapsed prior to completion of the operation or the server is not responding.","exceptionType":"System.Data.SqlClient.SqlException" ,"stackTrace":" at System.Data.SqlClient.SqlCommand.<>c.<ExecuteDbDataReaderAsync>b__167_0(Task 1.GetResult()\\r\\n at System.Web.Http .Dispatcher.HttpControllerDispatcher.<SendAsync>d__1.MoveNext()","innerException":{"message":"An error has occurred.","exceptionMessage":"Timeout expired. The timeout period elapsed prior to completion of the operation or the server is not responding.","exceptionType":"System.Data.SqlClient.SqlException" ,"stackTrace":" at System.Data.SqlClient.SqlCommand.<>c.<ExecuteDbDataReaderAsync>b__167_0(Task 1 result )\\r\\n at System.Threading.Tasks.ContinuationResultTaskFromResultTask`2.InnerInvoke()\\r\\n at System .Threading.Tasks.Task.Execute()\\r\\n--- End of stack trace from previous location where exception was thrown ---\\r\\n at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)\\r\\n at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)\\r \\n at System.Data.Entity.Core.EntityClient.Internal.EntityCommandDefinition.

d__c.MoveNext()","innerException":{"message":"An error has occurred.","exceptionMessage":"The wait operation timed out","exceptionType":"System.ComponentModel.Win32Exception","stackTrace":null}}}

Answer 1

Other than loading all of the data in to memory and manually formatting it as required, the only feasible way I can think of to do this with a single query is to put the rows in to pre-defined date "buckets" and choose a row from each bucket.

The most logical choice for the buckets would be on the hour and half past the hour since you're looking for 30 minute intervals, for example:

Result 1, 2016-02-02 13:00:24, 2016-02-02 13:00:00 (bucket 1)
Result 1, 2016-02-02 13:01:24, 2016-02-02 13:00:00 (bucket 1)
Result 1, 2016-02-02 13:02:24, 2016-02-02 13:00:00 (bucket 1)
Result 1, 2016-02-02 13:33:24, 2016-02-02 13:30:00 (bucket 2)

The next problem to tackle is selecting which row to use for each of the buckets. One approach you could use is to determine the date closest to the bucket (minimum) and use it:

from l in loc
join bucket in (
    from l in loc
    group l by new DateTime(l.ReadTime.Year, l.ReadTime.Month, l.ReadTime.Day, l.ReadTime.Hour, l.ReadTime.Minute < 30 ? 0 : 30, 0) into g
    select g.Min(m => m.ReadTime)
) on l.ReadTime equals bucket
select new LocationModel
{
    Lng = l.FirstOrDefault().Lng,
    Lat = l.FirstOrDefault().Lat
}

Care needs to be taken with this approach though as you could end up with duplicates if there are rows with identical date/time values. If the entity/table uses an identity column as its primary key and rows are inserted in to the table in a date/time sequential manner, you can use the key to determine the row per bucket as the key is effectively in ascending date order:

from l in loc
join bucket in (
    from l in loc
    group l by new DateTime(l.ReadTime.Year, l.ReadTime.Month, l.ReadTime.Day, l.ReadTime.Hour, l.ReadTime.Minute < 30 ? 0 : 30, 0) into g
    select g.Min(m => m.Id)
) on l.Id equals bucket
select new LocationModel
{
    Lng = l.FirstOrDefault().Lng,
    Lat = l.FirstOrDefault().Lat
}

This will ensure there are no duplicates. Otherwise you'll either need to filter the duplicates in code or extend the query further. I don't know enough about your particular domain though so I'll leave this part for you to solve.

I should also point out the above queries are supported by Linq to SQL, however you may need to use the DbFunctions class for the DateTime bits with EF if something isn't supported (eg DbFunctions.CreateDateTime).