Question

I have an XML file with the following structure:

<Entities>
    <Request> 
       <ID> A1 </ID> <!-- Can be multiple records with same values -->
       <Finance>
          <StartDate> Some Date </StartDate> <!-- Unique per ID -->
       </Finance>
    <Request>
    <Request> ... </Request>
</Entities>

There can be multiple requests with same ID but in such cases, the StartDate needs to be different per Request.

I need to take out the latest two dates per ID.

If this were an SQL table with ID and StartDate columns, I’d use the following query which is working fine:

SELECT ID, StartDate 
FROM ( SELECT ID, StartDate, RANK() 
          OVER (PARTITION BY ID ORDER BY StartDate DESC) rank
       FROM Request ) 
WHERE rank IN ('1','2')

But I have the data in XML format and the best I could come up with was ordering the data according to ID, StartDate. I still need to dug out latest two dates for every ID there is.

var cafrequests =

from request in xdoc.Descendants("Request")
orderby (int)request.Element("ID"), 
(DateTime)request.Element("Finance").Element("StartDate") ascending
select new
{
    ID = request.Element("ID"),
    StartDate = request.Element("Finance").Element("StartDate"),
};

Using Take(2) would only fetch me the top 2 lines of data, not top 2 per ID.

So can anybody tell me what is the equivalent of the above SQL statement in LINQ? I don’t want to parse and manage XML using loops and conditionals in C# and I’m quite new to LINQ (read about it yesterday and started using it) and I’m still going through the documentation.

Answer 1

This works, I tested it:

        XDocument doc = XDocument.Load(@"Data.xml");

        var result2 = doc.Element("Entities")
            .Elements("Request")
            .GroupBy(key => key.Element("ID").Value, el => DateTime.Parse(el.Element("Finance").Element("StartDate").Value))
            .Select(el =>
                new
                {
                    id = el.Key,
                    max2 = el.OrderByDescending(date => date).Take(2)
                });

• doc.Element("Entities") – gets element named Entity which is a root element of the document, another way of retrieving this element would be doc.Root
• Elements("Request") – gets elements named Request that are children of Entity
• GroupBy – is a method somewhat similar to GROUP BY in SQL, first parameter is element that is going to be used for grouping, it is a child of a Request element, second parameter is an element selector function that parses DateTime from StartDate element (which is a child of Finance element which is a child of Request element)
• .Select(el => new...) – method that creates anonymous type that holds ID and 2 records that have highest dates for ID

Here is a query that gets the same result, but written in an query syntax (example above is method syntax):

        var result = from el in doc.Root.Elements("Request")
                     group DateTime.Parse(el.Element("Finance").Element("StartDate").Value) by el.Element("ID").Value into grouped
                     select new { 
                         id = grouped.Key,
                         max2 = (from el in grouped
                                 orderby el descending
                                 select el)
                                 .Take(2)
                     };