public static NFA Create(ShuntingYard yard)
{
var stack = new Stack <NFA>();
foreach (var token in yard.ShuntedTokens())
{
switch (token.Type)
{
case RegExToken.TokenType.OperatorMul:
stack.Push(RepeatZeroOrMore(stack.Pop()));
break;
case RegExToken.TokenType.OperatorQuestion:
stack.Push(RepeatZeroOrOnce(stack.Pop()));
break;
case RegExToken.TokenType.OperatorOr:
stack.Push(Or(stack.Pop(), stack.Pop()));
break;
case RegExToken.TokenType.OperatorPlus:
stack.Push(RepeatOnceOrMore(stack.Pop()));
break;
case RegExToken.TokenType.Accept:
stack.Push(Accept(token.Characters));
break;
case RegExToken.TokenType.OperatorConcat:
// & is not commutative, and the stack is reversed.
var second = stack.Pop();
var first = stack.Pop();
stack.Push(And(first, second));
break;
case RegExToken.TokenType.NumberedRepeat:
stack.Push(NumberedRepeat(stack.Pop(), token.MinRepetitions, token.MaxRepetitions));
break;
default:
throw new LexerConstructionException("Unknown operator!");
}
}
// We should end up with only ONE NFA on the stack or the expression
// is malformed.
if (stack.Count() != 1)
{
throw new LexerConstructionException("Malformed regexp expression");
}
// Pop it from the stack, and assign each state a number, primarily for debugging purposes,
// they dont _really_ need it. The state numbers actually used are the one used in the DFA.
var nfa = stack.Pop();
nfa.AssignStateNumbers();
return(nfa);
}
public static NFA Create(ShuntingYard yard)
{
var stack = new Stack<NFA>();
foreach (var token in yard.ShuntedTokens())
{
switch (token.Type)
{
case RegExToken.TokenType.OperatorMul:
stack.Push(RepeatZeroOrMore(stack.Pop()));
break;
case RegExToken.TokenType.OperatorQuestion:
stack.Push(RepeatZeroOrOnce(stack.Pop()));
break;
case RegExToken.TokenType.OperatorOr:
stack.Push(Or(stack.Pop(), stack.Pop()));
break;
case RegExToken.TokenType.OperatorPlus:
stack.Push(RepeatOnceOrMore(stack.Pop()));
break;
case RegExToken.TokenType.Accept:
stack.Push(Accept(token.Characters));
break;
case RegExToken.TokenType.OperatorConcat:
// & is not commutative, and the stack is reversed.
var second = stack.Pop();
var first = stack.Pop();
stack.Push(And(first, second));
break;
case RegExToken.TokenType.NumberedRepeat:
stack.Push(NumberedRepeat(stack.Pop(), token.MinRepetitions, token.MaxRepetitions));
break;
default:
throw new LexerConstructionException("Unknown operator!");
}
}
// We should end up with only ONE NFA on the stack or the expression
// is malformed.
if (stack.Count() != 1)
{
throw new LexerConstructionException("Malformed regexp expression");
}
// Pop it from the stack, and assign each state a number, primarily for debugging purposes,
// they dont _really_ need it. The state numbers actually used are the one used in the DFA.
var nfa = stack.Pop();
nfa.AssignStateNumbers();
return nfa;
}
public static NFA Create(ShuntingYard yard, bool ignorecase)
{
Stack <NFA> stack = new Stack <NFA>();
foreach (RegExToken token in yard.ShuntedTokens(ignorecase))
{
try
{
switch (token.Type)
{
case RegExToken.TokenType.OperatorMul:
stack.Push(RepeatZeroOrMore(stack.Pop()));
break;
case RegExToken.TokenType.OperatorQuestion:
stack.Push(RepeatZeroOrOnce(stack.Pop()));
break;
case RegExToken.TokenType.OperatorOr:
stack.Push(Or(stack.Pop(), stack.Pop()));
break;
case RegExToken.TokenType.OperatorPlus:
stack.Push(RepeatOnceOrMore(stack.Pop()));
break;
case RegExToken.TokenType.Accept:
stack.Push(Accept(token.Characters));
break;
case RegExToken.TokenType.OperatorConcat:
// & is not commutative, and the stack is reversed.
NFA second = stack.Pop();
NFA first = stack.Pop();
stack.Push(And(first, second));
break;
case RegExToken.TokenType.NumberedRepeat:
stack.Push(NumberedRepeat(stack.Pop(), token.MinRepetitions, token.MaxRepetitions));
break;
default:
throw new LexerConstructionException("Unknown operator!");
}
}
catch (InvalidOperationException) when(stack.Count == 0) // stack popping failed
{
FieldInfo field = typeof(StringReader).GetField("_s", BindingFlags.Instance | BindingFlags.NonPublic);
TextReader reader = yard.lexer.input;
throw new LexerConstructionException($"Malformed regexp expression: '{field.GetValue(reader)}'");
}
}
// We should end up with only ONE NFA on the stack or the expression is malformed.
if (stack.Count != 1)
{
throw new LexerConstructionException("Malformed regexp expression!");
}
// Pop it from the stack, and assign each state a number, primarily for debugging purposes,
// they dont _really_ need it. The state numbers actually used are the one used in the DFA.
NFA nfa = stack.Pop();
nfa.AssignStateNumbers();
return(nfa);
}
