private GrammarParser(Lexer lex, Func <char, T> mkExprinal)
{
lexer = lex;
this.mkExprinal = mkExprinal;
startvar = null;
productions = new List <Grammars.Production>();
}
private GrammarParser(Lexer lex, Func <string, Automaton <T> > parseRegex)
{
lexer = lex;
this.parseRegex = parseRegex;
startvar = null;
productions = new List <Grammars.Production>();
terminalMap = new Dictionary <string, GrammarSymbol[]>();
this.parsedRegexes = new Dictionary <Nonterminal, Automaton <T> >();
}
private void Parse()
{
bool done = false;
Token cur = null;
Token last = null;
Grammars.Nonterminal curlhs = ExpectNT();
startvar = curlhs;
ExpectArrow();
List <Grammars.GrammarSymbol> currhs = new List <Grammars.GrammarSymbol>();
while (!done)
{
last = cur;
cur = lexer.Next();
switch (cur.t)
{
case TokenType.NT:
currhs.Add(new Grammars.Nonterminal(cur.content));
break;
case TokenType.T:
currhs.Add(new Exprinal <T>(mkExprinal(cur.content[0]), cur.content));
break;
case TokenType.OR:
productions.Add(new Grammars.Production(curlhs, currhs.ToArray()));
currhs.Clear();
break;
case TokenType.ARR:
if (currhs.Count < 1)
{
throw new ParseException();
}
if (last.t != TokenType.NT)
{
throw new ParseException();
}
// downcast :(
Grammars.Nonterminal newlhs = (Grammars.Nonterminal)currhs[currhs.Count - 1];
currhs.RemoveAt(currhs.Count - 1);
productions.Add(new Grammars.Production(curlhs, currhs.ToArray()));
currhs.Clear();
curlhs = newlhs;
break;
case TokenType.EOS:
productions.Add(new Grammars.Production(curlhs, currhs.ToArray()));
currhs.Clear();
done = true;
break;
default:
throw new ParseException();
}
}
}
private void Parse()
{
bool done = false;
Token cur = null;
Token last = null;
Grammars.Nonterminal curlhs = ExpectNT();
startvar = curlhs;
ExpectArrow();
List <Grammars.GrammarSymbol> currhs = new List <Grammars.GrammarSymbol>();
while (!done)
{
last = cur;
cur = lexer.Next();
switch (cur.t)
{
case TokenType.NT:
currhs.Add(Grammars.Nonterminal.CreateByParser(cur.content));
break;
case TokenType.T:
{
GrammarSymbol[] symbs;
if (!terminalMap.TryGetValue(cur.content, out symbs))
{
var aut = parseRegex(cur.content);
#region parse this terminal-regex as an automaton and compute symbs or set symbs to top nonterminal
int seq_length = -1;
if (aut.IsEpsilon)
{
symbs = new GrammarSymbol[] { };
}
else if (aut.InitialStateIsSource && aut.HasSingleFinalSink && aut.MoveCount == 1)
{
//just a single terminal
var move = aut.GetMoveFrom(aut.InitialState);
symbs = new GrammarSymbol[] { new Terminal <T>(move.Label) };
}
else if (aut.CheckIfSequence(out seq_length) && aut.HasSingleFinalSink && aut.IsEpsilonFree)
{
//collect all the elements and map them to individual terminals
//inline the automaton as sequence of terminals
symbs = new GrammarSymbol[seq_length];
int q = aut.InitialState;
int i = 0;
while (!aut.IsFinalState(q))
{
var move = aut.GetMoveFrom(q);
q = move.TargetState;
symbs[i] = new Terminal <T>(move.Label);
i += 1;
}
}
else
{
//introduce new nonterminal for the automaton
int id = __regexId++;
var nt = Nonterminal.MkNonterminalForRegex(id);
parsedRegexes[nt] = aut;
symbs = new GrammarSymbol[] { nt };
}
terminalMap[cur.content] = symbs;
#endregion
}
currhs.AddRange(symbs);
//---
break;
}
case TokenType.OR:
productions.Add(new Grammars.Production(curlhs, currhs.ToArray()));
currhs.Clear();
break;
case TokenType.ARR:
if (currhs.Count < 1)
{
throw new ParseException();
}
if (last.t != TokenType.NT)
{
throw new ParseException();
}
// downcast :(
Grammars.Nonterminal newlhs = (Grammars.Nonterminal)currhs[currhs.Count - 1];
currhs.RemoveAt(currhs.Count - 1);
productions.Add(new Grammars.Production(curlhs, currhs.ToArray()));
currhs.Clear();
curlhs = newlhs;
break;
case TokenType.EOS:
productions.Add(new Grammars.Production(curlhs, currhs.ToArray()));
currhs.Clear();
done = true;
break;
default:
throw new ParseException();
}
}
}