private IList<string> DoSubmatch(string re, string input)
{
var tokenStream = new TokenStream(new StringReader(re));
var parser = new RDParser(tokenStream);
var ast = parser.Parse();
var nfa = new NFAGraph(ast);
var dfa = new DFAGraph(nfa);
return dfa.SimulateSubmatch(input);
}
public DFAGraph(NFAGraph nfa)
{
Alphabet = nfa.Alphabet;
FinalStates = new HashSet<DFAState>();
AdjList = new Dictionary<DFAState, IDictionary<CharacterClassElement, DFAState>>();
var stateVisitQueue = new Queue<DFAState>();
var marked = new HashSet<DFAState>();
//kgo subset construction
var startEClosure = nfa.EClosure(nfa.StartState);
StartState = new DFAState(startEClosure);
stateVisitQueue.Enqueue(StartState);
if (StartState.Tags.Contains(FAStateTags.Final))
FinalStates.Add(StartState);
AdjList[StartState] = new Dictionary<CharacterClassElement, DFAState>();
while (stateVisitQueue.Count > 0)
{
var thisState = stateVisitQueue.Dequeue();
marked.Add(thisState);
foreach (var symbol in Alphabet)
{
var u = nfa.EClosure(nfa.Move(thisState.Constituents.ToList(), symbol));
if (u.Count > 0)
{
var dfaU = new DFAState(u);
if (!AdjList.Keys.Contains(dfaU))
{
if (dfaU.Tags.Contains(FAStateTags.Final))
FinalStates.Add(dfaU);
stateVisitQueue.Enqueue(dfaU);
AdjList[dfaU] = new Dictionary<CharacterClassElement, DFAState>();
}
LinkStates(thisState, dfaU, symbol);
}
}
}
}
private bool DoMatch(string re, string input)
{
var tokenStream = new TokenStream(new StringReader(re));
var parser = new RDParser(tokenStream);
var ast = parser.Parse();
var nfa = new NFAGraph(ast);
return nfa.RecursiveMatch(input);
}
public NFAGraph(RegexNode rxNode, StateFactory fac = null)
{
Alphabet = CharacterClassMapper.NormaliseAST(rxNode);
StateFac = fac ?? new StateFactory();
AdjList = new Dictionary<NFAState, IDictionary<CharacterClassElement, ISet<NFAState>>>();
//begin from the start
StartState = MakeState();
StartState.Tags.Add(FAStateTags.Start);
FinalState = MakeState();
FinalState.Tags.Add(FAStateTags.Final);
//every alternative is connected to this
for (var altNode = rxNode.FirstAlternative; altNode != null; altNode = altNode.Next)
{
//node for this alt
var altPosNode = MakeState();
LinkStates(StartState, altPosNode, Empty);
NFAState oldPosNode = null;
//Iter the and matches
for (var matchNode = altNode.FirstFactor; matchNode != null; matchNode = matchNode.Next)
{
oldPosNode = altPosNode;
if (matchNode is CharacterClassMatchNode)
{
var ccn = (CharacterClassMatchNode) matchNode;
var thisState = MakeState();
LinkStates(altPosNode, thisState, ccn.MatchingCharacterClass);
altPosNode = thisState;
}
else if (matchNode is GroupMatchNode)
{
var gn = (GroupMatchNode) matchNode;
//Create a NFA for the group and join it in
var groupGraph = new NFAGraph(gn.Body, StateFac);
//Merge all their states into ours
foreach (var oState in groupGraph.AdjList.Keys)
AdjList[oState] = groupGraph.AdjList[oState];
LinkStates(altPosNode, groupGraph.StartState, Empty);
//tag it
groupGraph.StartState.Tags.Remove(FAStateTags.Start);
groupGraph.StartState.Tags.Add(FAStateTags.PushSubmatch);
groupGraph.FinalState.Tags.Remove(FAStateTags.Final); //destroys final tag as intended
groupGraph.FinalState.Tags.Add(FAStateTags.PopSubmatch);
altPosNode = groupGraph.FinalState;
}
//operators?
switch (matchNode.OpType)
{
case UnaryOperatorType.NoneMany: //kleene star
//link the state we just made back via e
LinkStates(altPosNode, oldPosNode, Empty);
var newTermination = MakeState();
LinkStates(oldPosNode, newTermination, Empty);
oldPosNode = altPosNode;
altPosNode = newTermination;
break;
case UnaryOperatorType.Optional:
LinkStates(oldPosNode, altPosNode, Empty); //skip path we just made
break;
case UnaryOperatorType.OneMany:
LinkStates(altPosNode, oldPosNode, Empty); //make a cycle
break;
}
}
//join it into the end
LinkStates(altPosNode, FinalState, Empty);
}
}