private NfaState NewNfaState(ushort?acceptState = null)
{
var res = new NfaState(_nextNfaStateNumber, acceptState);
_nfaStates.Add(res);
++_nextNfaStateNumber;
return(res);
}
private void GetEpsilonMoves(SortedSet <NfaState> stateCollection, NfaState currentState)
{
foreach (var pair in currentState.Map)
{
if (pair.Key != RegExInputRange.EPS)
{
continue;
}
var state = pair.Value;
if (stateCollection.Contains(state))
{
continue;
}
stateCollection.Add(state);
GetEpsilonMoves(stateCollection, state);
}
}
private void Apply(RegExNode regExNode, NfaState begin, NfaState end)
{
switch (regExNode.RegExType)
{
case RegExType.Blank:
Transition(begin, end, RegExInputRange.EPS);
break;
case RegExType.Union:
var union = (RegExNodeUnion)regExNode;
foreach (var exp in union.Expressions)
{
Apply(exp, begin, end);
}
break;
case RegExType.Concat:
var concat = (RegExNodeConcat)regExNode;
var lastConcat = begin;
for (var i = 0; i < concat.Expressions.Count; i++)
{
var tmp = (i + 1) < concat.Expressions.Count ? NewNfaState() : end;
Apply(concat.Expressions[i], lastConcat, tmp);
lastConcat = tmp;
}
break;
case RegExType.Ranges:
var prim = (RegExNodeRanges)regExNode;
foreach (var range in prim.Ranges)
{
Transition(begin, end, range);
}
break;
case RegExType.Sequence:
var seq = (RegExNodeSequence)regExNode;
var lastSeq = begin;
switch (seq.Casing)
{
case RegExCasing.Insensitive:
for (var i = 0; i < seq.Sequence.Length; i++)
{
var tmp = (i + 1 < seq.Sequence.Length) ? NewNfaState() : end;
var c1 = char.ToLowerInvariant(seq.Sequence[i]);
var c2 = char.ToUpperInvariant(c1);
Transition(lastSeq, tmp, new RegExInputRange(c1));
if (c2 != c1)
{
Transition(lastSeq, tmp, new RegExInputRange(c2));
}
lastSeq = tmp;
}
break;
case RegExCasing.Sensitive:
for (var i = 0; i < seq.Sequence.Length; i++)
{
var tmp = (i + 1 < seq.Sequence.Length) ? NewNfaState() : end;
Transition(lastSeq, tmp, new RegExInputRange(seq.Sequence[i]));
lastSeq = tmp;
}
break;
}
break;
case RegExType.Repeat:
var rep = (RegExNodeRepeat)regExNode;
switch (rep.RegExRepeatType)
{
case RegExRepeatType.ZeroOrOne:
Apply(rep.Expression, begin, end);
Transition(begin, end, RegExInputRange.EPS);
break;
case RegExRepeatType.ZeroOrMore:
{
var s1 = NewNfaState();
var s2 = NewNfaState();
Transition(begin, end, RegExInputRange.EPS);
Transition(begin, s1, RegExInputRange.EPS);
Apply(rep.Expression, s1, s2);
Transition(s2, s1, RegExInputRange.EPS);
Transition(s2, end, RegExInputRange.EPS);
}
break;
case RegExRepeatType.OneOrMore:
{
var s1 = NewNfaState();
var s2 = NewNfaState();
Transition(begin, s1, RegExInputRange.EPS);
Apply(rep.Expression, s1, s2);
Transition(s2, s1, RegExInputRange.EPS);
Transition(s2, end, RegExInputRange.EPS);
}
break;
}
break;
case RegExType.Except:
var except = (RegExNodeExcept)regExNode;
break;
case RegExType.Accept:
// Add an accept state.
var accept = (RegExNodeAccept)regExNode;
var term = NewNfaState(accept.AcceptState);
Apply(accept.Expression, begin, term);
Transition(term, end, RegExInputRange.EPS);
break;
}
}
private static void Transition(NfaState from, NfaState to, RegExInputRange via)
{
from.Map.Add(new KeyValuePair <RegExInputRange, NfaState>(via, to));
}
