public override Regexp Normalize()
{
var disj = GetDisjuncts();
disj.Sort();
Regexp re = null;
foreach (var d in disj)
{
if (re == null)
{
re = d.Normalize();
}
else
{
re = new REUnion(d.Normalize(), re);
}
}
if (re.GetDisjuncts().Count == disj.Count)
{
return(re);
}
else
{
return(re.Normalize());
}
}
private static bool CorrectOnPosSet(Regexp regexp, IEnumerable <string> testSet)
{
foreach (var test in testSet)
{
if (!regexp.HasModel(test))
{
return(false);
}
}
return(true);
}
private static Automaton <BDD> getDfa(Regexp regexp)
{
var re = regexp.Normalize();
if (!memoDfa.Keys.Contains(re.ToString()))
{
memoDfa[re.ToString()] = re.getDFA(alph, solver);
}
return(memoDfa[re.ToString()]);
}
public override Regexp Normalize()
{
var r1 = r.Normalize();
Regexp dj = null;
foreach (var d in r1.GetDisjuncts())
{
Regexp sol = null;
if (!(d is REStar) && !(d is REPlus) && !(d is REQMark))
{
sol = d;
}
else
{
if (d is REStar)
{
var v = d as REStar;
sol = v.r;
}
else
{
if (d is REPlus)
{
var v = d as REPlus;
sol = v.r;
}
else
{
var v = d as REQMark;
sol = v.r;
}
}
}
if (dj == null)
{
dj = sol;
}
else
{
dj = new REUnion(sol, dj);
}
}
if (dj.GetDisjuncts().Count != r.GetDisjuncts().Count)
{
return(new REStar(dj).Normalize());
}
return(new REStar(dj));
}
public static IEnumerable <Regexp> SynthesizeRegexp(HashSet <char> alphabet, Automaton <BDD> dfa, CharSetSolver s, StringBuilder sb, long timeout)
{
using (System.IO.StreamWriter file = new System.IO.StreamWriter(@"..\..\..\regexpenum.txt"))
{
solver = s;
numStates = dfa.StateCount;
alph = alphabet;
#region test variables
StringBuilder sb1 = new StringBuilder();
int lim = 0;
Stopwatch membershipTimer = new Stopwatch();
Stopwatch equivTimer = new Stopwatch();
timer = new Stopwatch();
timer.Start();
#endregion
#region TestSets for equiv
var mytests = DFAUtilities.MyHillTestGeneration(alphabet, dfa, solver);
var posMN = mytests.First;
var negMN = mytests.Second;
var tests = DFAUtilities.GetTestSets(dfa, alphabet, solver);
var positive = tests.First;
var negative = tests.Second;
foreach (var t in posMN)
{
positive.Remove(t);
}
foreach (var t in negMN)
{
negative.Remove(t);
}
#endregion
#region Sigma Star
bool fst = true;
foreach (var c in alph)
{
if (fst)
{
fst = false;
sigmaStar = new RELabel(c);
}
else
{
sigmaStar = new REUnion(sigmaStar, new RELabel(c));
}
}
sigmaPlus = new REPlus(sigmaStar);
sigmaStar = new REStar(sigmaStar);
#endregion
#region Accessories vars
maxWidthC = 0;
maxSigmaStarC = 0;
var isSubset = true;
HashSet <string> visited = new HashSet <string>();
HashSet <string> newReg = new HashSet <string>();
currUnionEls = new Dictionary <string, Automaton <BDD> >();
memoDfa = new Dictionary <string, Automaton <BDD> >();
List <Regexp> subsetReg = new List <Regexp>();
#endregion
for (maxWidth = 1; true; maxWidth++)
{
newReg = new HashSet <string>();
maxSigmaStar = 2;
foreach (var regexp in EnumerateRegexp())
{
#region run for at most timeout
if (timer.ElapsedMilliseconds > timeout)
{
sb.AppendLine("| Timeout");
timer.Stop();
yield break;
}
#endregion
var re = regexp.Normalize();
if (!(visited.Contains(re.ToString())))
{
visited.Add(re.ToString());
sb1 = new StringBuilder();
sb1.Append(re.ToString());
file.WriteLine(sb1);
lim++;
#region Membership test
membershipTimer.Start();
isSubset = CorrectOnNegSet(regexp, negMN);
membershipTimer.Stop();
#endregion
#region equivalence check
if (isSubset)
{
membershipTimer.Start();
if (CorrectOnNegSet(regexp, negative))
{
if (CorrectOnPosSet(regexp, posMN) && CorrectOnPosSet(regexp, positive))
{
membershipTimer.Stop();
equivTimer.Start();
var rDfa = getDfa(regexp);
memoDfa[regexp.ToString()] = rDfa;
if (rDfa.IsEquivalentWith(dfa, solver))
{
isSubset = false;
equivTimer.Stop();
timer.Stop();
sb.Append("| ");
regexp.ToString(sb);
sb.AppendLine("|");
sb.AppendLine(string.Format("| elapsed time: \t {0} ms", timer.ElapsedMilliseconds));
sb.AppendLine(string.Format("| equivalence cost:\t {0} ms", equivTimer.ElapsedMilliseconds));
sb.AppendLine(string.Format("| membership cost: \t {0} ms", membershipTimer.ElapsedMilliseconds));
sb.AppendLine(string.Format("| attempts: \t {0}", lim));
yield return(regexp);
}
else
{
Console.WriteLine("used dfa");
equivTimer.Stop();
}
}
else
{
membershipTimer.Stop();
}
}
else
{
membershipTimer.Stop();
isSubset = false;
}
}
#endregion
//#region Subsets
//if (isSubset)
//{
// foreach (var reg1 in subsetReg)
// {
// var union = (reg1.CompareTo(regexp) > 0) ? (new REUnion(reg1, regexp)) : (new REUnion(regexp, reg1));
// visited.Add(union.ToString());
// sb1 = new StringBuilder();
// sb1.Append(union + " From union");
// file.WriteLine(sb1);
// lim++;
// membershipTimer.Start();
// if (CorrectOnPosSet(union, posMN) && CorrectOnPosSet(union, positive))
// {
// membershipTimer.Stop();
// equivTimer.Start();
// var rDfa = getDfa(union);
// memoDfa[union.ToString()] = rDfa;
// if (rDfa.IsEquivalentWith(dfa, solver))
// {
// equivTimer.Stop();
// timer.Stop();
// sb.Append("| ");
// union.ToString(sb);
// sb.AppendLine("|");
// sb.AppendLine(string.Format("| elapsed time: \t {0} ms", timer.ElapsedMilliseconds));
// sb.AppendLine(string.Format("| equivalence cost:\t {0} ms", equivTimer.ElapsedMilliseconds));
// sb.AppendLine(string.Format("| membership cost: \t {0} ms", membershipTimer.ElapsedMilliseconds));
// sb.AppendLine(string.Format("| attempts: \t {0}", lim));
// yield return union;
// }
// else
// {
// Console.WriteLine("used dfa");
// equivTimer.Stop();
// }
// }
// else
// {
// membershipTimer.Stop();
// }
// }
// subsetReg.Add(regexp);
//}
//#endregion
}
}
visited = new HashSet <string>(visited.Union(newReg));
}
}
}
public REUnion(Regexp left, Regexp right)
{
this.left = left;
this.right = right;
this.str = string.Format(@"(({0})|({1}))", left.str, right.str);
}
public REQMark(Regexp rexp)
{
this.r = rexp;
this.str = string.Format(@"({0})?", rexp.str);
}
public REPlus(Regexp rexp)
{
this.r = rexp;
this.str = string.Format(@"({0})+", rexp.str);
}
public REStar(Regexp rexp)
{
this.r = rexp;
this.str = string.Format(@"({0})*", rexp.str);
}
public REConcatenation(Regexp left, Regexp right)
{
this.left = left;
this.right = right;
this.str = string.Format(@"{0}{1}", left.str, right.str);
}
