如何在LINQ的DataTable上使用group by？

Question

I have DataTable with columns:

Column1, Column2, Column3, Column4

I need to select:

sum Column1, Count Distinct (Column2), Column3, Column4

group by Column3, Column4

I am using this code but with wrong result in count distinct

var queryDGroup = dtsecscaleditemsCloned2.AsEnumerable()
    .GroupBy(r => new { Col1 = r["Column3"], Col2 = r["Column4"]});

foreach (var g in queryDGroup)
{
    DataRow addedRow = dtsecscaleditemsCloned.Rows.Add();
    addedRow["Column1"] = g.Distinct ().Count();
    addedRow["Column4"] = g.Key.Col2;
    addedRow["Column3"] = g.Key.Col1;
    addedRow["store_Code"] = "";
    addedRow["Column2"] = g.Sum(r => r.Field<int>("Column2"));
}

Answer 1

As far as I could understand, you're trying to count the distinct values in the Column2 column. You need to project to that column first before doing anything else. Otherwise you're just counting distinct DataRow instances (which will have no meaning here).

I'd write it like this:

var query =
    from r in dt.AsEnumerable()
    group new
    {
        Column1 = r.Field<int>("Column1"),
        Column2 = r.Field<int>("Column2"),
    }
    by new
    {
        Column3 = r.Field<string>("Column3"),
        Column4 = r.Field<string>("Column4"),
    }
    into g
    select new
    {
        Column1 = g.Select(x => x.Column1).Distinct().Count(), // project first
        Column2 = g.Sum(x => x.Column2),
        g.Key.Column3,
        g.Key.Column4,
        StoreCode = "",
    };

Answer 2

You can try this:

var queryDGroup = dataTable.AsEnumerable()
    .GroupBy(c => new { Col3 = c["Column3"], Col4 = c["Column4"] },
             x => x,
            (k, g) => new
            {
                 Col1Sum = g.Sum(t => t.Field<int>("Column1")),
                 Col2Count = g.Select(t => t.Field<int>("Column2")).Distinct().Count()
            }).FirstOrDefault();

The object queryDGroup is an anonymously typed object with two properties: Col1Sum and Col1Count (the names are self-explanatory).

You can either create strongly typed anonymous objects as you select elements while doing group by (the second parameter to the GroupBy ), or cast to individual column's datatypes in the result selector part (the last parameter to the GroupBy ), as I have shown here.