static void TestCppCodeGen(Regex[] regexes)
{
Automaton<BDD>[] automata = new Automaton<BDD>[regexes.Length];
Automaton<BDD>[] Cautomata = new Automaton<BDD>[regexes.Length];
var solver = new CharSetSolver();
#region convert the regexes to automata
Console.Write("Converting {0} regexes to automata and minimizing the automata ...", regexes.Length);
int t = System.Environment.TickCount;
Func<Automaton<BDD>, bool> IsFull = (a => a.StateCount == 1 && a.IsFinalState(a.InitialState) && a.IsLoopState(a.InitialState) && a.GetMovesCountFrom(a.InitialState) == 1 && a.GetMoveFrom(a.InitialState).Label.Equals(solver.True));
for (int i = 0; i < regexes.Length; i++)
{
try
{
var aut = CppCodeGenerator.Regex2Automaton(solver, regexes[i]);
automata[i] = aut;
if (IsFull(automata[i]) || automata[i].IsEmpty)
{
Console.WriteLine("\nReplacing trivial regex \"{0}\" with \"^dummy$\"", i, regexes[i]);
regexes[i] = new Regex("^dummy$");
automata[i] = CppCodeGenerator.Regex2Automaton(solver, regexes[i]);
}
}
catch (Exception e)
{
Console.WriteLine("\nCoverting regex {0}: '{1}' failed, reason: {2}, replacing with \"^dummy$\"", i, regexes[i], e.Message);
regexes[i] = new Regex("^dummy$");
automata[i] = CppCodeGenerator.Regex2Automaton(solver, regexes[i]);
}
}
t = System.Environment.TickCount - t;
Console.WriteLine(string.Format(" done ({0}ms)", t));
#endregion
#region complement the automata
t = System.Environment.TickCount;
Console.Write("Creating complements of autmata ...");
for (int i = 0; i < regexes.Length; i++)
{
Cautomata[i] = automata[i].Complement().Minimize();
}
t = System.Environment.TickCount - t;
Console.WriteLine(string.Format(" done ({0}ms)", t));
#endregion
#region generate positive test strings
Console.Write(string.Format("Generating a positive test set for all automata ", NrOfStrings));
t = System.Environment.TickCount;
List<string[]> members = new List<string[]>();
List<string[]> Cmembers = new List<string[]>();
for (int id = 0; id < automata.Length; id++)
{
Console.Write(".");
var M = automata[id].Intersect(solver.Convert("^[\0-\x7F]{0," + CodeGenTests.MaxStringLength + "}$", RegexOptions.Singleline)).Determinize();
var tmp = new string[NrOfStrings];
int time = System.Environment.TickCount;
for (int i = 0; i < NrOfStrings; i++)
{
tmp[i] = solver.GenerateMemberUniformly(M);
//if (i % 10 == 0)
// Console.Write(".");
}
time = System.Environment.TickCount - time;
members.Add(tmp);
}
t = System.Environment.TickCount - t;
Console.WriteLine(string.Format(" done ({0}ms)", t));
#endregion
#region generate negative test strings
t = System.Environment.TickCount;
Console.Write(string.Format("Generating a negative test set for all automata ", NrOfStrings));
for (int id = 0; id < Cautomata.Length; id++)
{
Console.Write(".");
//var M = Cautomata[id].Intersect(solver.Convert("^[^\uD800-\uDFFF]{0,100}$", RegexOptions.Singleline), solver).Determinize(solver);
var M = Cautomata[id].Intersect(solver.Convert("^[\0-\uFFFF]{0,100}$", RegexOptions.Singleline)).Determinize();
var tmp = new string[NrOfStrings];
for (int i = 0; i < NrOfStrings; i++)
{
tmp[i] = solver.GenerateMemberUniformly(M);
//if (i % 10 == 0)
// Console.Write(".");
}
Cmembers.Add(tmp);
}
t = System.Environment.TickCount - t;
Console.WriteLine(string.Format(" done ({0}ms)", t));
#endregion
#region generate c++
int t2 = System.Environment.TickCount;
CppTest.Compile(automata, solver, true);
t2 = System.Environment.TickCount - t2;
Console.WriteLine(string.Format(" done ({0}ms)", t2));
#endregion
#region convert the test strings to UTF8
List<byte[][]> membersUTF8 = new List<byte[][]>();
List<byte[][]> CmembersUTF8 = new List<byte[][]>();
for (int id = 0; id < automata.Length; id++)
{
var tmp = new byte[NrOfStrings][];
for (int i = 0; i < NrOfStrings; i++)
tmp[i] = Encoding.UTF8.GetBytes(members[id][i]);
membersUTF8.Add(tmp);
}
for (int id = 0; id < Cautomata.Length; id++)
{
var tmp = new byte[NrOfStrings][];
for (int i = 0; i < NrOfStrings; i++)
tmp[i] = Encoding.UTF8.GetBytes(Cmembers[id][i]);
CmembersUTF8.Add(tmp);
}
#endregion
#region compute tot nr of bits
double bits = 0;
for (int id = 0; id < automata.Length; id++)
{
int nrBytes = 0;
for (int i = 0; i < NrOfStrings; i++)
nrBytes += membersUTF8[id][i].Length + CmembersUTF8[id][i].Length;
bits += (nrBytes * 8.0);
}
bits = bits * CodeGenTests.Repetitions; //repeated Reps times
#endregion
#region run c++ tests
Console.Write("Running c++ tests ... ");
double totsec_cpp = 0;
for (int id = 0; id < automata.Length; id++)
{
double sec_cpp = 0;
int accepted = CppTest.Test(true, id, membersUTF8[id], members[id], out sec_cpp);
totsec_cpp += sec_cpp;
int Caccepted = CppTest.Test(false, id, CmembersUTF8[id], Cmembers[id], out sec_cpp);
totsec_cpp += sec_cpp;
}
double bps_cpp = bits / totsec_cpp;
double mbps_cpp = (bps_cpp / 1000000.0);
int Mbps_cpp = (int)Math.Round(mbps_cpp);
Console.WriteLine(string.Format("{0}sec, throughput = {1}Mbps", totsec_cpp, Mbps_cpp));
#endregion
#region run .NET tests
Console.Write("Running .NET tests ... ");
double totsec_net = 0;
for (int id = 0; id < automata.Length; id++)
{
DotNetTest.Compile(regexes[id]); //make sure each regex is precompiled
double sec_net;
int accepted2 = DotNetTest.Test(true, members[id], out sec_net);
totsec_net += sec_net;
int Caccepted2 = DotNetTest.Test(false, Cmembers[id], out sec_net);
totsec_net += sec_net;
}
double bps_net = bits / totsec_net;
double mbps_net = (bps_net / 1000000.0);
int Mbps_net = (int)Math.Round(mbps_net);
Console.WriteLine(string.Format("{0}sec, throughput = {1}Mbps", totsec_net, Mbps_net));
#endregion
Console.WriteLine(string.Format("speedup (.NET-time/c++-time) = {0}X", ((int)Math.Round(totsec_net / totsec_cpp))));
}
