charting success

In my continued exploration of C# I recently had the opportunity to investigate the Google Chart API using Google Chart Sharp. Both the C# API and Google Chart itself are great. For those unfamiliar with Google Chart here is a quick summary.

Google Chart allows you to dynamically generate many of the popular types of chart – currently the API includes support for Line Charts, Bar Charts, Pie Charts, Venn Diagrams, plus many more. Check out the full list of chart types here.

Charts are generated via a GET request using a simple URL, where various GET parameters can be supplied to customize the chart. For example, to construct a pie chart that describes web browser usage for a particular page in which FireFox has 60%, and IE 40% of the total views, we use the following URL:

http://chart.apis.google.com/chart?cht=p3&chd=t:60,40&chs=250x100&chl=FireFox|IE

Typing this URL into your browser will show the chart below:

Pie Chart with two data points

Pie Chart with two data points

I’m sure many of you will recognise the chart design from Google Analytics.

Not only is the Chart API free to use (but not to abuse!), it also generates pretty nifty looking charts.

The charts can also be customised in many different ways including size, colour, axis, etc. Check out the documentation for this on the Google Chart API page linked to above. To show how simple it is to change the colours, take a look at the chart below:

Pie Chart with custom colours

Pie Chart with custom colours

Here the only change required to add custom colours is an additional HTML GET parameter – I simply added chco=9913CE,D3A4E5 to the end of the URL given above.

Now on to the C# side of things.

As you can probably guess, the C# wrapper for Google Chart simply generates a URL. For example, to create the chart above using our new colour scheme we do the following:

1
2
3
4
5
PieChart chart = new PieChart(250, 100, PieChartType.ThreeD);
chart.SetData(new int [] { 60, 40 });
chart.SetPieChartLabels(new string [] { "FireFox", "IE" });
chart.SetDatasetColors(new string [] { "9913CE", "D3A4E5" });
string url= chart.GetUrl();

You can then use the URL directly in an HTML document or download the image locally using the following code:

1
2
3
4
WebRequest request = WebRequest.Create(url);
WebResponse response = request.GetResponse();
Image image = Image.FromStream(response.GetResponseStream());
image.Save("mychart.png");

Yep it’s that easy. The other charts can be created just as quickly in both the Google Chart API and also with the C# wrapper. So if you are looking to create good looking charts and don’t want to pay much (any) money I suggest that you check this out.

dynamic queries with LINQ

So, I’m a master C# programmer having been using it for a grand total of 4 weeks! My learning process began by ignoring the provincial and proverbial “Hello World” program, and instead I decided to learn the language features that are particular to C#.

As it happens, C# has quite a few modern language features that Java is sadly lacking (I won’t go into this here, but a comparison between the two is something I plan to blog about in the future). Along with many other new features, C# 3.0 gave us LINQ, which will be the topic of this post.

LINQ, I think, is reasonably straightforward to both understand and comprehend. Basic queries are extremely simple to create, and LINQ almost gives C# a functional programming feel. An example LINQ query is shown below:

from s in students where s.classyear == "2008" && s.courseyear == "2" select s;

That said, not everything is plain sailing with LINQ. The application I was writing required the where condition in my LINQ statement be dynamic. That is, a column specified in the where condition is only known at runtime. Achieving this functionality using the syntax shown above is not possible (which if you try it out I’m sure it will be clear why). To achieve this functionality the learning curve soon ramped up.

If you have searched on Google for dynamic queries in LINQ, you probably found these two articles, ScottGu’s Dynamic Linq and tomasp.net’s Building LINQ Queries at Runtime in C# – there were others, but few proved to be much help. ScottGu’s article was very good, and in essence solved my problem, but didn’t really help me understand the problem. The second article just went way over my head. Why? Well it came down to another new feature of C# that I didn’t know about, which was Expressions and Expression Trees, which, as it turns out, proved to be the key to solving this problem.

In addition to using the query syntax with LINQ (as shown above), you can also use query expressions. Query expressions are simply static methods that allow you to express a LINQ query. The example given above can be rewritten as follows using a query expression:

student.Where(s => s.courseyear == "2008" && s.classyear == "2");

On first looking at this example, it is still not obvious how runtime queries can be generated using query expressions. However, by taking a look at the type of the argument passed to Where helps us out:

Expression<Func<TSource, bool>> predicate

After some thought, this begins to make sense. The Where method takes a lambda expression whose parameter is of type TSource and returns a boolean. Therefore, surely we can create a lambda expression dynamically by building an expression tree for it? And yes, this is exactly what you are supposed to do.

C# experts are probably howling at me now and can’t understand why it took me so long to get there. However, this article is not aimed at you; it’s aimed at a newbie like me, and is simply trying to detail my thought process.

So now it was time to learn about Expression Trees.

Although I understood the concept of Expression Trees, they took a bit of getting used to – just don’t give up if you are struggling with it. In order to show how we can use Expression Trees to create dynamic queries, I will jump straight in with the code:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
public Expression<Func<student, bool>> GetWhereLambda(string courseyear,
                                                      string classyear,
                                                      string myDynaColumn)
{
    ParameterExpression param = Expression.Parameter(typeof(student), "s");
 
    Expression courseExpr = GetEqualsExpr(param, "courseyear", courseyear);
    Expression classExpr = GetEqualsExpr(param, "classyear", classyear);
    Expression cond = Expression.And(courseExpr, classExpr);
 
    // This is where we create the expression for the dynamic column.
    // Obviously the value could have been dynamic as well but it 
    // saves a little visual complexity this way.
    cond = Expression.And(cond, GetEqualsExpr(param, myDynaColumn, "YES"));
 
    return Expression.Lambda<Func<student, bool>>(cond, param);
}
 
private Expression GetEqualsExpr(ParameterExpression param,
                                 string property,
                                 string value)
{
     Expression prop = Expression.Property(param, property);
     Expression val = Expression.Constant(value);
     return Expression.Equal(prop, val);
}

Seeing the code makes this look pretty simple, but there are quite a few concepts that a programmer new to C# has to get the hang of. However, let’s just step through the code – well one clause anyway, the rest follows from that.

Let’s look at how the clause (s.courseyear == "2008"), from out original where condition, converts to an Expression. We first create a parameter expression (Line 5), that is, the argument (parameter) supplied to the lambda expression (this parameter must be used for all sub-expressions that make up the lambda expression). Next, we need to create an expression to represent the right hand-side (s.courseyear) and left hand-side ("2008") of our clause. Since courseyear is a property of student (property as in C# property – think neat getters and setters) we create a property expression, i.e. prop = Expression.Property(param, "courseyear"), and as "2008" is really a constant, we create constant expression using value = Expression.Constant("2008"). Finally, we want to express the fact that the property should be equal to the constant in our query, we use the Expression.Equal(prop,value). That’s it. Essentially what expressions do is model our code as data.

Now it becomes obvious how we can represent a column in the where clause that is supplied at runtime – we simply create a property expression, as shown at Line 20, where property is the dynamic column name specified at runtime.

To combine the clauses in the where condition we simply use Expression.And, as shown in Lines 9 and 11. The resulting expression and the parameter is then used to create a lambda expression, which is in turn passed as an argument to Where, e.g.

Expression<Func<student, bool>> myLambda = GetWhereLambda("2008", "2", "active");
IEnumerable<student> filtered = students.Where(myLambda);

There you have it. It’s not that straightforward for those new to C# like me, but hopefully this article at least gives some code that can easily be used by people in the same position as I was.

generating a unique range of numbers

Often at work I find the need to generate a set of n unique integers in a specified range. In order to do this as efficiently and as easily as possible, I created a small C# class, which a colleague thought may be of general interest. Hence, I’m posting it here. I’m sure someone else has come up with a similar (or better) way to do this in the past, but I’m sharing my way regardless 😀 .

The code is shown below (and you download the C# class here):

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
 class UniqueSetGenerator
 {
      private int[] store_;
      private int size_;
      private Random random_;
 
      public UniqueSetGenerator(int size, int start)
      {
          size_ = size;
          store_ = new int[size];
          random_ = new Random();
          PopulateArray(start);
      }
 
      private void PopulateArray(int start)
      {
          for (int i = 0; i < size_; i++)
              store_[i] = start++;
      }
 
      private int Delete(int pos)
      {
          int val = store_[pos];
          store_[pos] = store_[--size_];
          return val;
      }
 
      public int GetRandomNumber()
      {
          if (size_ <= 0)
             return -1;
 
          return Delete(random_.Next(size_));
      }
 }

I think it’s pretty easy to see how this works, so I will not go into it in much detail discussing it, but the sequence of figures below shows the basic operations.

First populate the array with the values.

First populate the array with the values.


Use the Random class to obtain our first random number - in this case 6. Now copy the value at position size_ in the array to position 6, and decrement size_. Note not we don't actually delete anything from the array.

Use the Random class to obtain our first random number - in this case 6. Now copy the value at position size_ in the array to position 6, and decrement size_. Note: not we don't actually delete anything from the array.


We then use the Random class to generate another random number in our reduced range - in this case 3. We then copy the value at position size_ (i.e. 9) to position 3, and then decrement the size_ count as before. This process continues until size_ is reduced to 0.

We then use the Random class to generate another random number in our reduced range - in this case 3. We then copy the value at position size_ (i.e. 9) to position 3, and then decrement the size_ count as before. This process continues until size_ is reduced to 0.

So when would you require something like this? Well, say you need to generate unique random IDs with values 1 to 10, then you can use the class as follows:

1
2
3
4
5
6
7
8
    UniqueSetGenerator uniqueSet = new UniqueSetGenerator(10, 1);
 
    for(int i = 0; i < 9; i++)
    {
        int id = uniqueSet.GetRandomNumber();
 
        // Now do something with the id....
    }

For the moment I have only included a C# version but I will update this post with a Java version soon. Hope some of you find this useful.