static NumberLexerRule()
{
var states = new DfaState[5];
for (int i = 0; i < states.Length; i++)
states[i] = new DfaState(i==4 || i== 2);
var zeroThroughNine = new RangeTerminal('0', '9');
var plusOrMinusTo1 = new DfaTransition(new SetTerminal('+', '-'), states[1]);
var dotTo3 = new DfaTransition(new CharacterTerminal('.'), states[3]);
var zeroThroughNineTo2 = new DfaTransition(zeroThroughNine, states[2]);
var zeroThroughNineTo4 = new DfaTransition(zeroThroughNine, states[4]);
states[0].AddTransition(dotTo3);
states[0].AddTransition(plusOrMinusTo1);
states[0].AddTransition(zeroThroughNineTo2);
states[1].AddTransition(dotTo3);
states[1].AddTransition(zeroThroughNineTo2);
states[2].AddTransition(zeroThroughNineTo2);
states[2].AddTransition(dotTo3);
states[3].AddTransition(zeroThroughNineTo4);
states[4].AddTransition(zeroThroughNineTo4);
_start = states[0];
}
static WhitespaceLexerRule()
{
_start = new DfaState();
var end = new DfaState(isFinal: true);
var transition = new DfaTransition(
new WhitespaceTerminal(),
end);
_start.AddTransition(transition);
end.AddTransition(transition);
}
private static ILexerRule Whitespace()
{
var start = new DfaState();
var end = new DfaState(isFinal: true);
var transition = new DfaTransition(
new WhitespaceTerminal(),
end);
start.AddTransition(transition);
end.AddTransition(transition);
return new DfaLexerRule(start, "\\w+");
}
public void DfaLexemeShouldMatchOneOrMoreRandomWhitespaceCharacters()
{
var randomWhitespace = "\t\f\v \r\n";
var dfa = new DfaState();
var final = new DfaState(true);
dfa.AddTransition(new DfaTransition(new WhitespaceTerminal(), final));
final.AddTransition(new DfaTransition(new WhitespaceTerminal(), final));
var dfaLexerRule = new DfaLexerRule(dfa, new TokenType("whitespace"));
var whitespaceLexeme = new DfaLexeme(dfaLexerRule);
for (int i = 0; i < randomWhitespace.Length; i++)
Assert.IsTrue(whitespaceLexeme.Scan(randomWhitespace[i]));
}
public void DfaLexemeGivenCharacerLexemeNumberShouldFail()
{
var numberInput = "0";
var dfa = new DfaState();
var final = new DfaState(true);
dfa.AddTransition(new DfaTransition(new RangeTerminal('a', 'z'), final));
final.AddTransition(new DfaTransition(new RangeTerminal('a', 'z'), final));
var dfaLexerRule = new DfaLexerRule(dfa, new TokenType("lowerCase"));
var letterLexeme = new DfaLexeme(dfaLexerRule);
Assert.IsFalse(letterLexeme.Scan(numberInput[0]));
Assert.AreEqual(string.Empty, letterLexeme.Capture);
}
public void DfaLexemeShouldMatchMixedCaseWord()
{
var wordInput = "t90vAriabl3";
var dfa = new DfaState();
var final = new DfaState(true);
dfa.AddTransition(new DfaTransition(new RangeTerminal('a', 'z'), final));
dfa.AddTransition(new DfaTransition(new RangeTerminal('A', 'Z'), final));
final.AddTransition(new DfaTransition(new RangeTerminal('a', 'z'), final));
final.AddTransition(new DfaTransition(new RangeTerminal('A', 'Z'), final));
final.AddTransition(new DfaTransition(new DigitTerminal(), final));
var dfaLexerRule = new DfaLexerRule(dfa, new TokenType("Identifier"));
var indentifierLexeme = new DfaLexeme(dfaLexerRule);
for (int i = 0; i < wordInput.Length; i++)
Assert.IsTrue(indentifierLexeme.Scan(wordInput[i]));
}
private static BaseLexerRule CreateWhitespaceLexerRule()
{
var whitespaceTerminal = new WhitespaceTerminal();
var startWhitespace = new DfaState();
var finalWhitespace = new DfaState(true);
startWhitespace.AddTransition(new DfaTransition(whitespaceTerminal, finalWhitespace));
finalWhitespace.AddTransition(new DfaTransition(whitespaceTerminal, finalWhitespace));
var whitespace = new DfaLexerRule(startWhitespace, "whitespace");
return whitespace;
}
private static BaseLexerRule CreateSettingIdentifierLexerRule()
{
// /:[a-zA-Z][a-zA-Z0-9]*/
var start = new DfaState();
var oneLetter = new DfaState();
var zeroOrMoreLetterOrDigit = new DfaState(true);
start.AddTransition(
new DfaTransition(
new CharacterTerminal(':'),
oneLetter));
oneLetter.AddTransition(
new DfaTransition(
new CharacterClassTerminal(
new RangeTerminal('a', 'z'),
new RangeTerminal('A', 'Z')),
zeroOrMoreLetterOrDigit));
zeroOrMoreLetterOrDigit.AddTransition(
new DfaTransition(
new CharacterClassTerminal(
new RangeTerminal('a', 'z'),
new RangeTerminal('A', 'Z'),
new DigitTerminal()),
zeroOrMoreLetterOrDigit));
return new DfaLexerRule(start, "settingIdentifier");
}
private static BaseLexerRule CreateNotSingleQuoteLexerRule()
{
var start = new DfaState();
var final = new DfaState(true);
var terminal = new NegationTerminal(new CharacterTerminal('\''));
var edge = new DfaTransition(terminal, final);
start.AddTransition(edge);
final.AddTransition(edge);
return new DfaLexerRule(start, new TokenType(@"([^']|(\\.))+"));
}
private static BaseLexerRule CreateNotDoubleQuoteLexerRule()
{
// ([^"]|(\\.))*
var start = new DfaState();
var escape = new DfaState();
var final = new DfaState(true);
var notDoubleQuoteTerminal = new NegationTerminal(
new CharacterTerminal('"'));
var escapeTerminal = new CharacterTerminal('\\');
var anyTerminal = new AnyTerminal();
var notDoubleQuoteEdge = new DfaTransition(notDoubleQuoteTerminal, final);
start.AddTransition(notDoubleQuoteEdge);
final.AddTransition(notDoubleQuoteEdge);
var escapeEdge = new DfaTransition(escapeTerminal, escape);
start.AddTransition(escapeEdge);
final.AddTransition(escapeEdge);
var anyEdge = new DfaTransition(anyTerminal, final);
escape.AddTransition(anyEdge);
return new DfaLexerRule(start, new TokenType(@"([^""]|(\\.))+"));
}
private static BaseLexerRule CreateMultiLineCommentLexerRule()
{
var pattern = @"\/[*]([*][^\/]|[^*])*[*][\/]";
var states = new DfaState[5];
for (int i = 0; i < states.Length; i++)
states[i] = new DfaState();
var slash = new CharacterTerminal('/');
var star = new CharacterTerminal('*');
var notStar = new NegationTerminal(star);
var notSlash = new NegationTerminal(slash);
var firstSlash = new DfaTransition(slash, states[1]);
var firstStar = new DfaTransition(star, states[2]);
var repeatNotStar = new DfaTransition(notStar, states[2]);
var lastStar = new DfaTransition(star, states[3]);
var goBackNotSlash = new DfaTransition(notSlash, states[2]);
var lastSlash = new DfaTransition(slash, states[4]);
states[0].AddTransition(firstSlash);
states[1].AddTransition(firstStar);
states[2].AddTransition(repeatNotStar);
states[2].AddTransition(lastStar);
states[3].AddTransition(goBackNotSlash);
states[3].AddTransition(lastSlash);
return new DfaLexerRule(states[0], pattern);
}
private static BaseLexerRule CreateIdentifierLexerRule()
{
// /[a-zA-Z][a-zA-Z0-9-_]*/
var identifierState = new DfaState();
var zeroOrMoreLetterOrDigit = new DfaState(true);
identifierState.AddTransition(
new DfaTransition(
new CharacterClassTerminal(
new RangeTerminal('a', 'z'),
new RangeTerminal('A', 'Z')),
zeroOrMoreLetterOrDigit));
zeroOrMoreLetterOrDigit.AddTransition(
new DfaTransition(
new CharacterClassTerminal(
new RangeTerminal('a', 'z'),
new RangeTerminal('A', 'Z'),
new DigitTerminal(),
new SetTerminal('-', '_')),
zeroOrMoreLetterOrDigit));
var identifier = new DfaLexerRule(identifierState, "identifier");
return identifier;
}
private static BaseLexerRule CreateEscapeCharacterLexerRule()
{
var start = new DfaState();
var escape = new DfaState();
var final = new DfaState(true);
start.AddTransition(new DfaTransition(new CharacterTerminal('\\'), escape));
escape.AddTransition(new DfaTransition(new AnyTerminal(), final));
return new DfaLexerRule(start, "escape");
}
private static BaseLexerRule CreateWhitespaceLexerRule()
{
var whitespaceTerminal = new WhitespaceTerminal();
var startState = new DfaState();
var finalState = new DfaState(true);
var whitespaceTransition = new DfaTransition(whitespaceTerminal, finalState);
startState.AddTransition(whitespaceTransition);
finalState.AddTransition(whitespaceTransition);
return new DfaLexerRule(startState, new TokenType("[\\s]+"));
}
private static ILexerRule CreateRuleNameLexerRule()
{
var ruleNameState = new DfaState();
var zeroOrMoreLetterOrDigit = new DfaState(true);
ruleNameState.AddTransition(
new DfaTransition(
new CharacterClassTerminal(
new RangeTerminal('a', 'z'),
new RangeTerminal('A', 'Z')),
zeroOrMoreLetterOrDigit));
zeroOrMoreLetterOrDigit.AddTransition(
new DfaTransition(
new CharacterClassTerminal(
new RangeTerminal('a', 'z'),
new RangeTerminal('A', 'Z'),
new DigitTerminal(),
new SetTerminal('-', '_')),
zeroOrMoreLetterOrDigit));
var ruleName = new DfaLexerRule(ruleNameState, new TokenType("rule-name"));
return ruleName;
}
public NfaClosure(HashSet<INfaState> closure, bool isFinal)
{
Closure = closure;
_hashCode = HashCode.ComputeHash(closure);
State = new DfaState(isFinal);
}
