We Keep Coding

Returning List<string> from Linq query returns query syntax not values

I have the below code to return a list of strings.
public List<string> Top5CodesForToday()
{
var date = DateTime.Now;
var resultList = new List<string>();
using (var db = new PillowContext())
{
var qry = (from d in db.DownTimes
where DbFunctions.TruncateTime(d.DateTime) == DbFunctions.TruncateTime(date)
group d by new {d.Code}
into g
let total = g.Sum(x => x.Amount)
orderby total descending
let top5 = g.Take(5).ToList()
select new {g.Key.Code, Total = total});
foreach (var item in qry)
{
int x = item.Code;
var results = from r in db.DownTimeCodes
where r.Code == x
select r.Description;
resultList.Add(results.ToString());
}
}
return resultList;
}
When I look at the contents of returnList I am seeing the correct number of items however each item is made up of the actual query syntax, not the data itself. I have seen this before and usually solve it by doing .ToList() however I am unsure how I could change my code to solve this

The problem here is that when you are calling ToString the query is not executed yet, so essentially you are calling ToString on a IQueryable object, receiving the query instead of results. You need to call something to execute the query.
You can call ToList() still:
var results = (from r in db.DownTimeCodes
where r.Code == x
select r.Description).ToList();
resultList.AddRange(results);
Or, if you expect just one result, call FirstOrDefault()/SingleOrDefault():
var results = (from r in db.DownTimeCodes
where r.Code == x
select r.Description).FirstOrDefault();
resultList.Add(results);

You are calling ToString() on List<>. As default for most complex types, it just writes out type name not the data.
This line
resultList.Add(results.ToString());
should be changed to
resultList.AddRange(results);

Linq Select Clause w/ Unknown Number of Fields

I have a linq query in which I need to be able to select an variable number of fields from a datatable. I do know all of the fields that could be included, but only two will for sure be in the datatable. I also will know which fields are included in the datatable (it will just be different depending on the user's selections). Right now I set up something like this:
var query = from item in dt.AsEnumerable()
group item by item.Field<string>("ID") into g
select new
{
ID = g.Key, //required
Status = g.Min(i => dostuff(i,"Status")), //not required
Disc = g.Min(i => dostuff(i,"Disc")), //not required
Loc = String.Join<string>(",", from i in g select i.Field<string>("Loc")) //required
};
dostuff(DataRow i,string field)
{
try
{
return i.Field<string>(field);
}
catch
{
return null;
}
}
So dostuff basically is just checking whether or not that field exists in the dataset, and then I would just need to ignore the non-existant fields when working with the query results, which would not be too difficult. However, it seems like there is probably a better way to do this, but I've had a tough time finding anything via Google about using a dynamic select clause.

You could do it with dynamic type (nb, I did not test so this might have typos.):
var query =dt.AsEnumerable().GroupBy(item => item.Field<string>("ID"))
.Select(g => {
dynamic t = new System.Dynamic.ExpandoObject();
if (g.Table.Columns.Any(c => c.ColumnName == "Status"))
t.Status = g.Field<string>("Status");
if (g.Table.Columns.Any(c => c.ColumnName == "Disc"))
t.Disc = g.Field<string>("Disc");
t.ID = g.Key;
t.Loc = String.Join<string>(",",g.Select(i => i.Field<string>("Loc")));
return t;
}

LINQ: Add RowNumber Column

How can the query below be modified to include a column for row number (ie: one-based index of results)?
var myResult = from currRow in someTable
where currRow.someCategory == someCategoryValue
orderby currRow.createdDate descending
select currRow;
EDIT1: I'm looking for the results to be {idx, col1, col2...col-n} not {idx, row}.
EDIT2: The row number should correspond to result rows not the table rows.
EDIT3: I DataBind these results to a GridView. My goal was to add a row number column to the GridView. Perhaps a different approach would be better.

Use the method-syntax where Enumerable.Select has an overload with the index:
var myResult = someTable.Select((r, i) => new { Row = r, Index = i })
.Where(x => x.Row.someCategory == someCategoryValue)
.OrderByDescending(x => x.Row.createdDate);
Note that this approach presumes that you want the original index of the row in the table and not in the filtered result since i select the index before i filter with Where.
EDIT: I'm looking for the results to be {idx, col1, col2...col-n} not
{idx, row}. The row number should correspond to result rows not
the table rows.
Then select the anonymous type with all columns you need:
var myResult = someTable.Where(r => r.someCategory == someCategoryValue)
.OrderByDescending(r => r.createdDate)
.Select((r, i) => new { idx = i, col1 = r.col1, col2 = r.col2, ...col-n = r.ColN });

Use this Select method:
Projects each element of a sequence into a new form by incorporating the element's index.
Example:
var myResult = someTable.Where(currRow => currRow.someCategory == someCategoryValue)
.OrderByDescending(currRow => currRow.createdDate)
.Select((currRow, index) => new {Row = currRow, Index = index + 1});
In response to your edit:
If you want a DataTable as result, you can go the non-Linq way by simply using a DataView and add a additional column afterwards.
someTable.DefaultView.RowFilter = String.Format("someCategory = '{0}'", someCategoryValue);
someTable.DefaultView.Sort = "createdDate";
var resultTable = someTable.DefaultView.ToTable();
resultTable.Columns.Add("Number", typeof(int));
int i = 0;
foreach (DataRow row in resultTable.Rows)
row["Number"] = ++i;

what about?
int i;
var myResult = from currRow in someTable
where currRow.someCategory == someCategoryValue
orderby currRow.createdDate descending
select new {Record = i++, currRow};

Just for fun, here's an alternative to Select with two arguments:
var resultsWithIndexes = myResult.Zip(Enumerable.Range(1, int.MaxValue - 1),
(o, i) => new { Index = i, Result = o });

According to you edit 1. NO, YOU CAN'T Linq returns the table as it is. You can build each column, but you lose the power of mapped entities.
This has been asked multiple times before: How do you add an index field to Linq results

There is no straightforward way if want to keep a flat list of columns (i.e. OP's Edit2) and also want a generic solution that works with any IEnumerable without requiring you to list out the set of expected columns.
However, there is a roundabout way to kinda go about it which is to dump the query results into a DataTable using the ToDataTable() method from here and then add a RowNumber column to that table.
var table = query.ToList().ToDataTable();
table.Columns.Add("RowNum", typeof(int));
int i = 0;
foreach (DataRow row in table.Rows)
row["RowNum"] = ++i;
This would likely cause performance issues with large datasets but it's not insanely slow either. On my machine a dataset with ~6500 rows took 33ms to process.
If your original query returned an anonymous type, then that type definition will get lost in the conversion so you'll lose the static typing on the column names of the resulting IEnumerable when you call table.AsEnumerable(). In other words, instead of being able to write something like table.AsEnumerable().First().RowNum you instead have to write table.AsEnumerable().First()["RowNum"]
However, if you don't care about performance and really want your static typing back, then you can use JSON.NET to convert the DataTable to a json string and then back to a list based on the anonymous type from the original query result. This method requires a placeholder RowNum field to be present in the original query results.
var query = (from currRow in someTable
where currRow.someCategory == someCategoryValue
orderby currRow.createdDate descending
select new { currRow.someCategory, currRow.createdDate, RowNum = -1 }).ToList();
var table = query.ToDataTable();
//Placeholder RowNum column has to already exist in query results
//So not adding a new column, but merely populating it
int i = 0;
foreach (DataRow row in table.Rows)
row["RowNum"] = ++i;
string json = JsonConvert.SerializeObject(table);
var staticallyTypedList = JsonConvert.DeserializeAnonymousType(json, query);
Console.WriteLine(staticallyTypedList.First().RowNum);
This added about 120ms to the processing time for my 6500 item dataset.
It's crazy, but it works.

I know I'm late to the party, but I wanted to show what worked for me.
I have a list of objects, and the object has an integer property on it for "row number"... or in this case, "Sequence Number". This is what I did to populate that field:
myListOfObjects = myListOfObjects.Select((o, i) => { o.SequenceNumber = i; return o; }).ToList();
I was surprised to see that this worked.

This one helped me in my case - Excel sheet extraction. anonymous type
var UploadItemList = ItemMaster.Worksheet().AsEnumerable().Select((x, index) => new
{
Code = x["Code"].Value == null ? "" : x["Code"].Value.ToString().Trim(),
Description = x["Description"].Value == null ? "" : x["Description"].Value.ToString().Trim(),
Unit = x["Unit"].Value == null ? "" : x["Unit"].Value.ToString().Trim(),
Quantity = x["Quantity"].Value == null ? "" : x["Quantity"].Value.ToString().Trim(),
Rate = x["Rate"].Value == null ? "" : x["Rate"].Value.ToString().Trim(),
Amount = x["Amount"].Value == null ? "" : x["Amount"].Value.ToString().Trim(),
RowNumber = index+1
}).ToList();

int Lc = 1;
var Lst = LstItemGrid.GroupBy(item => item.CategoryName)
.Select(group => new { CategoryName = group.Key, Items = group.ToList() ,RowIndex= Lc++ })
.ToList();

Retrieving values from a Linq GroupBy

I have a datatable which contains a load of dates. I wanted to group these by date and give each row a count.
I have managed to do this by dong the following:
IEnumerable query = from row in stats.AsEnumerable()
group row by row.Field<string>("date") into grp
select new { Date = grp.Key, Count = grp.Count(t => t["date"] != null) };
(where "stats" is the datatable)
I can see from debugging that this brings back the values all grouped as I need, but now I need to loop them and get each date and count.
My problem is I don't know how to retrieve the values!
I have a foreach loop
foreach (var rw in query)
{
string date = rw.Date; // <---- this is my problem?
}
I don't know what type my Ienumerable is to be able to reference the values in it!
So my question is how can I retrieve each date and count for each row by doing similar to the above?
I hope this makes sense!

This link on my blog should help you
http://www.matlus.com/linq-group-by-finding-duplicates/
Essentially your type is an anonymous type so you can't reference it as a type but you can access the properties like you're trying to do.
I think I see your issue. If you're trying to return it from a method, you should define a type and reuturn it like shown below:
public IEnumerable<MyType> GetQuery()
{
var query = from row in stats.AsEnumerable()
group row by row.Field<string>("date") into grp
select new { Date = grp.Key, Count = grp.Count(t => t["date"] != null) };
foreach (var rw in query)
{
yield return new MyType(rw.Date, rw.Count);
}
}
declare your "query" variable using "var" as shown above.

I guess you don't have access to the properties of the anonymous class because you're using IEnumerable query = .... Try var query = ... instead.

Going by your comment "I am returning the query from a function", which I take to mean that you want to do the query in a method, return the data to the caller, and then iterate the data in the caller, I suggest you return a Dictionary<DateTime, int>, like this:
static Dictionary<DateTime, int> GetSummarisedData()
{
var results = (
from row in stats.AsEnumerable()
group row by row.Field<string>("date") into grp
select new { Date = grp.Key, Count = grp.Count(t => t["date"] != null) })
.ToDictionary(val => val.Date, val => val.Count);
return results;
}
then in the caller you can just
foreach (var kvp in GetSummarisedData())
{
// Now kvp.Key is the date
// and kvp.Value is the count
}

String.Split in a Linq-To-SQL Query?

I have a database table that contains an nvarchar column like this:
1|12.6|18|19
I have a Business Object that has a Decimal[] property.
My LINQ Query looks like this:
var temp = from r in db.SomeTable select new BusinessObject {
// Other BusinessObject Properties snipped as they are straight 1:1
MeterValues = r.MeterValues.Split('|').Select(Decimal.Parse).ToArray()
};
var result = temp.ToArray();
This throws an NotSupportedException: Method 'System.String[] Split(Char[])' has no supported translation to SQL.
That kinda sucks :) Is there any way I can do this without having to add a string property to the business object or selecting an anonymous type and then iterating through it?
My current "solution" is:
var temp = from r in db.SomeTable select new {
mv = r.MeterValues,
bo = new BusinessObject { // all the other fields }
};
var result = new List<BusinessObject>();
foreach(var t in temp) {
var bo = t.bo;
bo.MeterValues = t.mv.Split('|').Select(Decimal.Parse).ToArray();
result.Add(bo);
}
return result.ToArray(); // The Method returns BusinessObject[]
That's kinda ugly though, with that temporary list.
I've tried adding a let mv = r.MeterValues.Split('|').Select(Decimal.Parse).ToArray() but that essentially leads to the same NotSupportedException.
This is .net 3.5SP1 if that matters.

You need to force the select clause to run on the client by calling .AsEnumerable() first:
var result = db.SomeTable.AsEnumerable().Select(r => new BusinessObject {
...
MeterValues = r.MeterValues.Split('|').Select(Decimal.Parse).ToArray()
}).ToList();

You can't use split, but in this scenario you can do the following:
// Database value is 1|12.6|18|19
string valueToFind = "19";
var temp = from r in db.SomeTable.Where(r => ("|" + r.MeterValues + "|").Contains("|" + valueToFind + "|"));
This code adds outer pipes (|) to the database value on the fly inside the query so you can do start, middle, and end value matches on the string.
For example, the above code looks for "|19|" inside "|1|12.6|18|19|", which is found and valid. This will work for any other valueToFind.

You don't need to use a temporary list:
var query = from r in db.SomeTable
select new
{
r.Id,
r.Name,
r.MeterValues,
...
};
var temp = from x in query.AsEnumerable()
select new BusinessObject
{
Id = x.Id,
Name = x.Name,
MeterValues = x.mv.Split('|').Select(Decimal.Parse).ToArray(),
...
};
return temp.ToArray();

Unfortunately its the IQueryable you are using (Linq to SQL) that is not supporting the Split function.
You are really only left with the IEnumerable (Linq to Objects) support for it in this case. You second code snippet is what you need to do, or something like...
var temp = (from r in db.SomeTable select new {
mv = r.MeterValues,
bo = new BusinessObject { // all the other fields }
}).AsEnumerable().Select(blah, blah) ;