Question

I have a query that returns data like the following:

Event | Sum | Date 
sent  | 400 | 1/1/12 
open  | 200 | 1/1/12
click | 50  | 1/1/12 
sent  | 300 | 1/8/12
open  | 150 | 1/8/12
click | 30  | 1/8/12

The SQL query:

select event, sum(thecount), rundate
from send_open_click_counts
group by event, rundate
order by rundate, event desc 

To send these values to google visualization api to render a graph I need to arrange the values like this:

['date' , 'sent' , 'opened' , 'clicked']
['1/1/12', '400' , '200' , '50']
['1/8/12', '300', '150' , '30']

I’m not really familiar with this stuff, but to me that is pivoting by date.

Here is the linq query I used to simulate my original SQL

    var dbLinqObj= from r in db.SEND_OPEN_CLICK_COUNTS
                group r by new { r.EVENT, r.RUNDATE } into g
                select new {
                    EVENT = g.Key.EVENT,
                    THECOUNT = g.Sum(r => r.THECOUNT),
                    RUNDATE = g.Key.RUNDATE

                } ;
    dbLinqObj= dbLinqObj.OrderBy(r => r.RUNDATE);

==== EDIT ==== SOLUTION FOUND =====

After digging into this more and getting a few suggestions from the comments I found an elegant solution using LINQ.

Here is the LINQ that will pivot the result properly, so that it can easily be transformed into a JSON string.

        var query = from q in db.SEND_OPEN_CLICK_COUNTS
                    group q by q.RUNDATE  into g
                    select new
      {
          Date = g.Key,
          Send = g.Where(x => x.EVENT == "sent").Sum(x => x.THECOUNT),
          Open = g.Where(x => x.EVENT == "opened").Sum(x => x.THECOUNT),
          Click = g.Where(x => x.EVENT == "clicked").Sum(x => x.THECOUNT)
      };

Answer 1

The linq pivot you’ve used is generally how it’s done, instead of converting to an array you could try using the Aggregate linq method to build the JSON.

For example:

var dbLinqObj= from r in db.SEND_OPEN_CLICK_COUNTS
            group r by new { r.EVENT, r.RUNDATE } into g
            select new {
                EVENT = g.Key.EVENT,
                THECOUNT = g.Sum(r => r.THECOUNT),
                RUNDATE = g.Key.RUNDATE

            } ;
var dbLinqObjJSON = dbLinqObj.OrderBy(r => r.RUNDATE).Aggregate((json, x) => json + "['" + x.RUNDATE + ", "'" + x.THECOUNT + "', '" + ...  + "]";

Note that Aggregate does not use a StringBuilder, so there will be a performance hit for large sequences. This is currently a problem in your posted code, since you are using String instead of StringBuilder to create your JSON. Each time you use the + operator on a String the .NET runtime needs to create a new String.

If you want to make it as fast as possible, there is an override of Aggregate that will take a StringBuilder:

var dbLinqObjJSON = dbLinqObj.OrderBy(r => r.RUNDATE).Aggregate(new StringBuilder("['date' , 'sent' , 'opened' , 'clicked'] , "),
                         (json, x) => json.append("['" + x.RUNDATE + "', '" + x.THECOUNT + "', '" + ...  + "], "),
                         (json) => json.ToString());