Searching for Exceptions in .NET

I recently came across a rather interesting question on StackOverflow that posed the problem of discovering all the exceptions that a given method might throw under every circumstance.

Of course, in the great majority of situations, XML  documentation for the BCL (and ideally any third-party libraries too) should provide information about any exception that might be thrown and any potential reason for it. Indeed, thisthis is generally all one needs to write largely error-safe code. However, not every exception is documented in any case, and for production-quality applications, it is often desirable to insure that there is no realistic chance of an unhandled exception ocurring. For this reason, it is sometimes desirable to do a rigorous check for all exceptions. Clearly, an application-level unhandled (fatal) exception handler would do the job to some extent, and although this is always a good fallback feature to have, it is the least elegant solution to coping with exceptions.

After some consideration, it became quite apparent that the task reduces to the halting problem. However, with a few simplifications, the problem does become relatively solvable. Most importantly, complex logic that determines whether an exception will be thrown must be ignored, and one must simply assume that any throw statement within a given method could possibly cause an exception under certain conditions.

Here is the complete code for the algorithm I wrote. The GetAllExceptions method is an extension method that returns a read-only collection of exceptions, which makes it very straightforward and efficient to use.

Notably, the code detects all of

• instantiated exceptions,
• exceptions return from method/property calls,
• exceptions stored in fields (though if the method return type or field type is non-specific, i.e. a parent class of the actual exception type thrown), this is used instead.

Exceptions are only counted when the appropiate throw instruction is encountered at some level. Also, the stack and local variables are handled correctly, as far as I can tell, so this method should work soundly in pretty much all cases. (It has been tested with some a few quite complex methods within the BCL, as well as simpler user-defined ones.)

using System;
using System.Collections.Generic;
using System.Collections.ObjectModel;
using System.Linq;
using System.Reflection;
using System.Reflection.Emit;
using System.Text;
using ClrTest.Reflection;

public static class ExceptionAnalyser
{
    public static ReadOnlyCollection<Type> GetAllExceptions(this MethodBase method)
    {
        var exceptionTypes = new HashSet<Type>();
        var visitedMethods = new HashSet<MethodBase>();
        var localVars = new Type[ushort.MaxValue];
        var stack = new Stack<Type>();
        GetAllExceptions(method, exceptionTypes, visitedMethods, localVars, stack, 0);

        return exceptionTypes.ToList().AsReadOnly();
    }

    public static void GetAllExceptions(MethodBase method, HashSet<Type> exceptionTypes,
        HashSet<MethodBase> visitedMethods, Type[] localVars, Stack<Type> stack, int depth)
    {
        var ilReader = new ILReader(method);
        var allInstructions = ilReader.ToArray();

        ILInstruction instruction;
        for (int i = 0; i < allInstructions.Length; i++)
        {
            instruction = allInstructions[i];

            if (instruction is InlineMethodInstruction)
            {
                var methodInstruction = (InlineMethodInstruction)instruction;

                if (!visitedMethods.Contains(methodInstruction.Method))
                {
                    visitedMethods.Add(methodInstruction.Method);
                    GetAllExceptions(methodInstruction.Method, exceptionTypes, visitedMethods,
                        localVars, stack, depth + 1);
                }

                var curMethod = methodInstruction.Method;
                if (curMethod is ConstructorInfo)
                    stack.Push(((ConstructorInfo)curMethod).DeclaringType);
                else if (method is MethodInfo)
                    stack.Push(((MethodInfo)curMethod).ReturnParameter.ParameterType);
            }
            else if (instruction is InlineFieldInstruction)
            {
                var fieldInstruction = (InlineFieldInstruction)instruction;
                stack.Push(fieldInstruction.Field.FieldType);
            }
            else if (instruction is ShortInlineBrTargetInstruction)
            {
            }
            else if (instruction is InlineBrTargetInstruction)
            {
            }
            else
            {
                switch (instruction.OpCode.Value)
                {
                    // ld*
                    case 0x06:
                        stack.Push(localVars[0]);
                        break;
                    case 0x07:
                        stack.Push(localVars[1]);
                        break;
                    case 0x08:
                        stack.Push(localVars[2]);
                        break;
                    case 0x09:
                        stack.Push(localVars[3]);
                        break;
                    case 0x11:
                        {
                            var index = (ushort)allInstructions[i + 1].OpCode.Value;
                            stack.Push(localVars[index]);
                            break;
                        }
                    // st*
                    case 0x0A:
                        localVars[0] = stack.Pop();
                        break;
                    case 0x0B:
                        localVars[1] = stack.Pop();
                        break;
                    case 0x0C:
                        localVars[2] = stack.Pop();
                        break;
                    case 0x0D:
                        localVars[3] = stack.Pop();
                        break;
                    case 0x13:
                        {
                            var index = (ushort)allInstructions[i + 1].OpCode.Value;
                            localVars[index] = stack.Pop();
                            break;
                        }
                    // throw
                    case 0x7A:
                        if (stack.Peek() == null)
                            break;

                        exceptionTypes.Add(stack.Pop());
                        break;
                    default:
                        switch (instruction.OpCode.StackBehaviourPop)
                        {
                            case StackBehaviour.Pop0:
                                break;
                            case StackBehaviour.Pop1:
                            case StackBehaviour.Popi:
                            case StackBehaviour.Popref:
                            case StackBehaviour.Varpop:
                                stack.Pop();
                                break;
                            case StackBehaviour.Pop1_pop1:
                            case StackBehaviour.Popi_pop1:
                            case StackBehaviour.Popi_popi:
                            case StackBehaviour.Popi_popi8:
                            case StackBehaviour.Popi_popr4:
                            case StackBehaviour.Popi_popr8:
                            case StackBehaviour.Popref_pop1:
                            case StackBehaviour.Popref_popi:
                                stack.Pop();
                                stack.Pop();
                                break;
                            case StackBehaviour.Popref_popi_pop1:
                            case StackBehaviour.Popref_popi_popi:
                            case StackBehaviour.Popref_popi_popi8:
                            case StackBehaviour.Popref_popi_popr4:
                            case StackBehaviour.Popref_popi_popr8:
                            case StackBehaviour.Popref_popi_popref:
                                stack.Pop();
                                stack.Pop();
                                stack.Pop();
                                break;
                        }

                        switch (instruction.OpCode.StackBehaviourPush)
                        {
                            case StackBehaviour.Push0:
                                break;
                            case StackBehaviour.Push1:
                            case StackBehaviour.Pushi:
                            case StackBehaviour.Pushi8:
                            case StackBehaviour.Pushr4:
                            case StackBehaviour.Pushr8:
                            case StackBehaviour.Pushref:
                            case StackBehaviour.Varpush:
                                stack.Push(null);
                                break;
                            case StackBehaviour.Push1_push1:
                                stack.Push(null);
                                stack.Push(null);
                                break;
                        }

                        break;
                }
            }
        }
    }
}

To be quite honest, I’m not sure whether I’ll need to use this code myself at any point, but I’ve posted it regardless for the benefit of anyone who might require such rigorous exception checking. It was definitely an interesting challenge, at the least.

Any further comments or suggestions would be welcome, as always.

Numerical Analysis for .NET

During my ongoing work on a computational project for university, I recently discovered the need to perform some serious numerical analysis from my C# code.  Unfortunately, I must admit that the .NET world only now seems to be catching up in terms of the free and open source libraries it offers for various tasks, and initially I was disheartened to find that there seemed to be nothing available for doing calculations on large (sparse) matrices. After a fair deal of searching, only a couple of somewhat incomplete and no longer maintained matrix libraries turned up. Being an avid user of StackOverflow, however, I decided that if anyone was aware of some library that could do what I needed, I would most likely find them there.

The result was much better than for what I was even hoping. dnAnalytics is a general-purpose package for numerical analysis in .NET that does almost everything for which I might possibly ask – and from my first impressions, does it very well indeed. This wonderful find is a well-maintained, fully open-source, library with great API documentation (not a wholly unexpected thing, but surprisingly uncommon among so many open source projects). There are several features that stand out as particularly impressive. One undoubtedly is I/O classes for Matlab and delimited files (among other formats). What is more, the library seems to offer both a fully managed version and one that wraps the Intel® Math Kernel Library. I’m not sure how the performance compares between the two (I haven’t yet tried the latter), but it is surely nice to have the pair of options available, quite similarly to how you have alternatives of cryptographic algorithms in the .NET BCL, that is to say, a) a fully managed version, v) a version based on top of the Windows Crypto API, c) a version that uses the CNG (Next Generation) API introduced with Vista. Perhaps what appeals to me the greatest about this library is that the developers have clearly gone to an effort to make it user-friendly, not only with regards to the documentation, but also by adding an interface friendly to F# coders (likely to be a language of choice for future mathematical/scientific programming), and even visual debuggers for Visual Studio (possibly the only library to date I’ve seen include them).

My particular usage of the library requires me to use the linear algebra (specifically, sparse matrix) classes. Although I must point out that the specific algorithm that I was intending to employ for the project was not available (see my later discussion), it did include a host of other ones, primarily focusing on direct and iterative matrix decomposition, which would appear to be quite handy in many circumstances. I haven’t yet had a chance to play with the other areas of the library, but I have noticed that it offers some statistical functions and methods as well as a number of modern pseudo-RNG algorithms such as the Mersenne Twister.

To conclude, I should come back to the point that the most important part of the analysis I require was not (at least direclty) contained by the library – finding the eigenvalues or eigendecomposition of large (1000s of rows/columns) matrices, which happens to be in relation to spectral theory, in case you’re curious. Even so, being such a complex field and one fraught with difficulties when it comes to implementation (numerical instability is a huge problem), I was not surprised to find that an implementation of the Arnoldi or Lanczos algorithm was not present. Fortunately, after a bit more searching around (by this point I knew specifically what I was looking for), I came across the ARPACK library, written in the archaic Fortran77 language. It did however seem to be exactly what I was looking for: a set of fast routines to find the eigenvalues of large (either dense or sparse) matrices of various types. After only a small amount of pain messing about with MinGW, I managed to get the code compiled nicely into a DLL. At this point, I am of course perfectly able just to use the P/Invoke capabilities of .NET and do some hackery to integrate the ARPACK stuff with my existing code and dnAnalytics. Yet, I am also inclined to do this whole task properly and basically write a managed wrapper for ARPACK that is tightly conforms with dnAnalytics. I could then perhaps submit these wrapper types (along with a few unit tests?) as a repository patch to the dnAnalytics team in the hope that they’ll take it and add it to the next release. As with most other projects at this time, I will have to see what time permits me, though I would certainly hope to contribute something substantial to what truly is a terrific project that I would love to see expand further.