How to write a standard calculator? - calculator

Question

I am preparing a task for a C# course I teach where the students have to write a calculator in order to learn how to work with basic GUI. However in the process I found that I cannot reproduce the behavior of a standard calculator in full. My implementation always misses some corner cases and the output differs from what the Windows Calculator does in its Standard mode.
The Wikipedia article describes the working of the calculator but it does not list how it handles corner cases. It says that a calculator has two registers X and Y and X is always displayed on the screen but it does not describe how values are pushed from one register to the other in corner cases.
Examples of corner cases include the input sequences:
"3 * 2 = =" - should display 6 and then 12
"3 - =" which results in 0 on the Windows Calculator but in -3 on my implementation
"3 * 2 = + 3"
"3 * 2 = 3 =" which strangely results in 6 on the Windows Calculator
Of course I can handle the corner cases using enough flags but I am sure standard calculators were implemented the way they were implemented because this way was the simplest. I cannot think of a simple way to implement all the cases so I am probably doing it wrong. I have a single variable called previousValue I use in place of the Y register and I parse the textbox value when doing calculation to simulate the X register. I also have a variable which indicates if the next digit input should override the textbox (X register) value. There is also a variable for the current operation which is initialized to None. Basically I force calculation every time an operator button or = is pressed and I set the input flag to true.
Here is the code for digit input (AddInput), operation input (InputOperation), = button click (btnCalculate_Click) and the methods they call into.
private void AddInput(string input)
{
if (shouldOverrideDigits)
{
previousNumber = Decimal.Parse(txtDigits.Text);
txtDigits.Text = input;
shouldOverrideDigits = false;
}
else
{
txtDigits.Text += input;
}
}
private void InputOperation(CalculatorOperations operation)
{
if (!shouldOverrideDigits)
{
Calculate();
}
currentOperation = operation;
}
private void btnCalculate_Click(object sender, EventArgs e)
{
if (shouldOverrideDigits)
{
Calculate();
}
else
{
decimal currentNumber = Decimal.Parse(txtDigits.Text);
Calculate();
previousNumber = currentNumber;
}
}
private decimal Calculate()
{
decimal currentNumber = Decimal.Parse(txtDigits.Text);
decimal result;
switch (currentOperation)
{
case CalculatorOperations.None:
{
result = currentNumber;
} break;
case CalculatorOperations.Addition:
{
result = previousNumber + currentNumber;
} break;
case CalculatorOperations.Subtraction:
{
result = previousNumber - currentNumber;
} break;
case CalculatorOperations.Multiplication:
{
result = previousNumber * currentNumber;
} break;
case CalculatorOperations.Division:
{
result = previousNumber / currentNumber;
} break;
default:
{
throw new Exception("Unexpected operation type");
}
}
shouldOverrideDigits = true;
txtDigits.Text = result.ToString();
return result;
}
Can someone help with a simple solution? If the original calculators were able to handle everything with what is effectively three variables it is unlikely that the solution in a modern language needs much more.
Interestingly every article on implementing a calculator on the Internet I could find does it wrong.