// can get production from nonterminal and terminal as usual
public ISymbol[] Get(Nonterminal var, Terminal term)
{
if (!table.ContainsKey(var))
return null;
if (!table[var].ContainsKey(term))
return null;
return table[var][term];
}
public Parser(CFG grammar, Terminal syncTerm)
{
this.grammar = grammar;
this.table = grammar.CreateLL1ParseTable();
this.start = grammar.StartSymbol;
this.syncTerm = syncTerm;
if (table == null)
throw new Exception("GRAMMAR NOT LL(1)");
}
// Add a production for some (nonterminal, terminal) pair to the table
public void Add(Nonterminal var, Terminal term, ISymbol[] production)
{
if (!table.ContainsKey(var))
table[var] = new Dictionary<Terminal, ISymbol[]>();
table[var][term] = production;
}
private MiniPL()
{
// Create NFA-type things for tokens using regular operations
Regex reBlockCommentStart = Regex.Concat("/*");
Regex reBlockCommentEnd = Regex.Concat("*/");
Regex reLineComment = Regex.Concat("//").Concat(Regex.Not('\n').Star());
Regex reWhitespace = Regex.Union(" \t\r\n").Star().Union(reLineComment);
Regex reString = Regex.Char('"').Concat(Regex.Char('\\').Concat(Regex.Any()).Union(Regex.Not('"', '\\')).Star()).Concat(Regex.Char('"'));
Regex reBinaryOperator = Regex.Union("+-*/<=&");
Regex reUnaryOperator = Regex.Char('!');
Regex reKeyword = Regex.Union(Regex.Concat("var"), Regex.Concat("for"), Regex.Concat("end"), Regex.Concat("in"),
Regex.Concat("do"), Regex.Concat("read"), Regex.Concat("print"), Regex.Concat("assert"));
Regex reType = Regex.Union(Regex.Concat("bool"), Regex.Concat("int"), Regex.Concat("string"));
Regex reParenRight = Regex.Char(')'),
reParenLeft = Regex.Char('('),
reColon = Regex.Char(':'),
reSemicolon = Regex.Char(';'),
reAssignment = Regex.Concat(":="),
reDots = Regex.Concat("..");
Regex reIdentifier = Regex.Union(Regex.Range('A', 'Z'), Regex.Range('a', 'z'))
.Concat(Regex.Union(Regex.Range('A', 'Z'), Regex.Range('a', 'z'), Regex.Range('0', '9'), Regex.Char('_')).Star());
Regex reInteger = Regex.Range('0', '9').Plus();
// Define token types
tokenTypes["block_comment_start"] = new TokenType("block_comment_start", reBlockCommentStart);
tokenTypes["block_comment_end"] = new TokenType("block_comment_end", reBlockCommentEnd);
tokenTypes["int"] = new TokenType("int", reInteger);
tokenTypes["whitespace"] = new TokenType("whitespace", reWhitespace, priority: TokenType.Priority.Whitespace);
tokenTypes["string"] = new TokenType("string", reString);
tokenTypes["binary_op"] = new TokenType("binary op", reBinaryOperator);
tokenTypes["unary_op"] = new TokenType("unary op", reUnaryOperator);
tokenTypes["keyword"] = new TokenType("keyword", reKeyword, priority: TokenType.Priority.Keyword);
tokenTypes["type"] = new TokenType("type", reType, priority: TokenType.Priority.Keyword);
tokenTypes["left_paren"] = new TokenType("left paren", reParenLeft);
tokenTypes["right_paren"] = new TokenType("right paren", reParenRight);
tokenTypes["colon"] = new TokenType("colon", reColon);
tokenTypes["semicolon"] = new TokenType("semicolon", reSemicolon);
tokenTypes["assignment"] = new TokenType("assignment", reAssignment);
tokenTypes["dots"] = new TokenType("dots", reDots);
tokenTypes["identifier"] = new TokenType("identifier", reIdentifier);
// create combined automaton and scanner object
TokenAutomaton automaton = TokenType.CombinedAutomaton(tokenTypes.Values.ToArray());
scanner = new Scanner(automaton, tokenTypes["block_comment_start"], tokenTypes["block_comment_end"]);
// Define nonterminal variables of CFG
nonterminals["program"] = new Nonterminal("PROG");
nonterminals["statements"] = new Nonterminal("STMTS");
nonterminals["statements_head"] = new Nonterminal("STMTS_HEAD");
nonterminals["statements_tail"] = new Nonterminal("STMTS_TAIL");
nonterminals["statement"] = new Nonterminal("STMT");
nonterminals["declaration"] = new Nonterminal("DECL");
nonterminals["declaration_assignment"] = new Nonterminal("DECL_ASSIGN");
nonterminals["expression"] = new Nonterminal("EXPR");
nonterminals["unary_operation"] = new Nonterminal("UNARY_OP");
nonterminals["binary_operation"] = new Nonterminal("BINARY_OP");
nonterminals["operand"] = new Nonterminal("OPND");
// Define terminal variables of CFG
terminals["identifier"] = new Terminal(tokenTypes["identifier"]);
terminals["assert"] = new Terminal("assert");
terminals["print"] = new Terminal("print");
terminals["read"] = new Terminal("read");
terminals["for"] = new Terminal("for");
terminals["in"] = new Terminal("in");
terminals["end"] = new Terminal("end");
terminals["do"] = new Terminal("do");
terminals["var"] = new Terminal("var");
terminals["type"] = new Terminal(tokenTypes["type"]);
terminals["string"] = new Terminal(tokenTypes["string"]);
terminals["int"] = new Terminal(tokenTypes["int"]);
terminals[")"] = new Terminal(")");
terminals["("] = new Terminal("(");
terminals[".."] = new Terminal("..");
terminals[":="] = new Terminal(":=");
terminals[":"] = new Terminal(":");
terminals[";"] = new Terminal(";");
terminals["binary_operator"] = new Terminal(tokenTypes["binary_op"]);
terminals["unary_operator"] = new Terminal(tokenTypes["unary_op"]);
// Create the Mini-PL grammar
grammar = new CFG(nonterminals["program"], terminals.Values, nonterminals.Values);
// define production rules for the grammar
grammar.AddProductionRule(nonterminals["program"], new ISymbol[] { nonterminals["statements"] });
grammar.AddProductionRule(nonterminals["statements"], new ISymbol[] { nonterminals["statements_head"], nonterminals["statements_tail"] });
grammar.AddProductionRule(nonterminals["statements_head"], new ISymbol[] { nonterminals["statement"], terminals[";"] });
grammar.AddProductionRule(nonterminals["statements_tail"], new ISymbol[] { nonterminals["statements_head"], nonterminals["statements_tail"] });
grammar.AddProductionRule(nonterminals["statements_tail"], new ISymbol[] { Terminal.EPSILON });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { nonterminals["declaration"] });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { terminals["identifier"], terminals[":="], nonterminals["expression"] });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { terminals["for"], terminals["identifier"], terminals["in"], nonterminals["expression"], terminals[".."], nonterminals["expression"], terminals["do"],
nonterminals["statements"], terminals["end"], terminals["for"] });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { terminals["read"], terminals["identifier"] });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { terminals["print"], nonterminals["expression"] });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { terminals["assert"], terminals["("], nonterminals["expression"], terminals[")"] });
grammar.AddProductionRule(nonterminals["declaration"], new ISymbol[] { terminals["var"], terminals["identifier"], terminals[":"], terminals["type"], nonterminals["declaration_assignment"] });
grammar.AddProductionRule(nonterminals["declaration_assignment"], new ISymbol[] { terminals[":="], nonterminals["expression"] });
grammar.AddProductionRule(nonterminals["declaration_assignment"], new ISymbol[] { Terminal.EPSILON });
grammar.AddProductionRule(nonterminals["expression"], new ISymbol[] { nonterminals["unary_operation"] });
grammar.AddProductionRule(nonterminals["expression"], new ISymbol[] { nonterminals["operand"], nonterminals["binary_operation"] });
grammar.AddProductionRule(nonterminals["unary_operation"], new ISymbol[] { terminals["unary_operator"], nonterminals["operand"] });
grammar.AddProductionRule(nonterminals["binary_operation"], new ISymbol[] { terminals["binary_operator"], nonterminals["operand"] });
grammar.AddProductionRule(nonterminals["binary_operation"], new ISymbol[] { Terminal.EPSILON });
grammar.AddProductionRule(nonterminals["operand"], new ISymbol[] { terminals["int"] });
grammar.AddProductionRule(nonterminals["operand"], new ISymbol[] { terminals["string"] });
grammar.AddProductionRule(nonterminals["operand"], new ISymbol[] { terminals["identifier"] });
grammar.AddProductionRule(nonterminals["operand"], new ISymbol[] { terminals["("], nonterminals["expression"], terminals[")"] });
// use ; as synchronizing token for Mini-PL
parser = new Parser(grammar, terminals[";"]);
}
public ParseLeaf(Terminal terminal)
{
this.terminal = terminal;
}
// Parse the given stream of tokens
public ParseTree Parse(IEnumerable <Token> tokenSource)
{
isValidParseTree = true;
errors = new List <Error>();
Stack <ISymbol> symbolStack = new Stack <ISymbol>();
symbolStack.Push(Terminal.EOF);
symbolStack.Push(start);
ParseTree parseTree = new ParseTree(start);
Stack <IParseNode> treeStack = new Stack <IParseNode>();
treeStack.Push(new ParseLeaf(Terminal.EOF));
treeStack.Push(parseTree);
IEnumerator <Token> tokenStream = tokenSource.GetEnumerator();
tokenStream.MoveNext();
while (symbolStack.Count > 0)
{
if (Program.debug)
{
Console.WriteLine("=========================================================");
Console.WriteLine(" PARSE: Stack " + SymbolsToString(symbolStack));
Console.WriteLine(" PARSE: expecting " + symbolStack.Peek());
Console.WriteLine(" PARSE: token " + tokenStream.Current);
}
// ignore error tokens
if (tokenStream.Current.Type == TokenType.ERROR)
{
if (Program.debug)
{
Console.WriteLine(" PARSE: skipping error token");
}
errors.Add(new LexicalError(tokenStream.Current));
tokenStream.MoveNext();
continue;
}
if (symbolStack.Peek() is Terminal)
{
Terminal term = symbolStack.Peek() as Terminal;
ParseLeaf leaf = treeStack.Peek() as ParseLeaf;
if (term == Terminal.EPSILON)
{
// epsilon production was used, exclude from parse tree
if (Program.debug)
{
Console.WriteLine(" PARSE: ignore epsilon");
}
symbolStack.Pop();
treeStack.Pop();
}
else if (term.Matches(tokenStream.Current))
{
// current token matches the top of the parse stack, add it to parse tree
if (Program.debug)
{
Console.WriteLine(" PARSE: Terminal match");
}
leaf.Token = tokenStream.Current;
tokenStream.MoveNext();
symbolStack.Pop();
treeStack.Pop();
}
else
{
// current token does no match, recover from error
if (Program.debug)
{
Console.WriteLine(" PARSE: Error, Terminal mismatch");
}
errors.Add(new SyntaxError(tokenStream.Current));
Synchronize(symbolStack, treeStack, tokenStream);
}
}
else // top of stack is a nonterminal
{
Nonterminal var = symbolStack.Pop() as Nonterminal;
IParseNode popped = treeStack.Pop();
ParseTree subtree = popped as ParseTree;
ISymbol[] production = table.Get(var, tokenStream.Current);
if (production == null)
{
// cannot derive the current token from the nonterminal at the top of the stack
if (Program.debug)
{
Console.WriteLine(" PARSE: Error, No such production");
}
symbolStack.Push(var);
treeStack.Push(popped);
errors.Add(new SyntaxError(tokenStream.Current));
Synchronize(symbolStack, treeStack, tokenStream);
}
else
{
// use the production specified by the parse table, add node to parse tree
if (Program.debug)
{
Console.WriteLine(" PARSE: Using production " + SymbolsToString(production));
}
for (int i = production.Length - 1; i >= 0; i--)
{
IParseNode treeChild;
if (production[i] is Terminal)
{
treeChild = new ParseLeaf(production[i] as Terminal);
}
else
{
treeChild = new ParseTree(production[i] as Nonterminal);
}
subtree.Children.Insert(0, treeChild);
treeStack.Push(treeChild);
symbolStack.Push(production[i]);
}
}
}
}
if (Program.debug)
{
Console.WriteLine(parseTree);
}
return(parseTree);
}
// Uses phrase-level error recovery with First and Follow sets to recover from bad state
private void Synchronize(Stack <ISymbol> symbolStack, Stack <IParseNode> treeStack, IEnumerator <Token> tokenStream)
{
// Loop until good state found (return statements) or no more symbols on stack.
sync : while (symbolStack.Count > 0)
{
if (Program.debug)
{
Console.WriteLine(" PARSE: Synchronize token " + tokenStream.Current + " symbol " + symbolStack.Peek());
}
// no recovery from end of file, empty the parse stack
if (tokenStream.Current.Type == TokenType.EOF)
{
while (symbolStack.Count > 0)
{
isValidParseTree = false;
symbolStack.Pop();
treeStack.Pop();
}
if (Program.debug)
{
Console.WriteLine("PARSE: Unexpected EOF");
}
return;
}
if (symbolStack.Peek() is Terminal)
{
Terminal t = symbolStack.Peek() as Terminal;
if (t.Matches(tokenStream.Current))
{
// good state reached
if (Program.debug)
{
Console.WriteLine(" PARSE: Token matches");
}
return;
}
if (syncTerm.Matches(tokenStream.Current))
{
// special case: we have a synchronising token like ';' that we do not want to skip over
// so remove symbol from parse stack
if (Program.debug)
{
Console.WriteLine(" PARSE: Discarding symbol " + symbolStack.Peek());
}
isValidParseTree = false;
symbolStack.Pop();
treeStack.Pop();
continue;
}
// normal case: discard token and continue with recovery
if (Program.debug)
{
Console.WriteLine(" PARSE: Discard token");
}
if (!tokenStream.MoveNext())
{
throw new Exception("OUT OF TOKENS");
}
continue;
}
else
{
Nonterminal v = symbolStack.Peek() as Nonterminal;
if (table.Get(v, tokenStream.Current) != null)
{
// good state reached (current token in First set of symbol at top of stack)
if (Program.debug)
{
Console.WriteLine(" PARSE: Valid production exists");
}
return;
}
// if current terminal could be matched by something in Follow set of
// symbol at top of stack, then skip the current symbol
foreach (ISymbol sym in grammar.Follow(v))
{
if (sym is Terminal)
{
Terminal followTerm = sym as Terminal;
if (followTerm.Matches(tokenStream.Current))
{
if (Program.debug)
{
Console.WriteLine(" PARSE: Discarding symbol " + symbolStack.Peek());
}
isValidParseTree = false;
symbolStack.Pop();
treeStack.Pop();
goto sync;
}
}
if (sym is Nonterminal)
{
Nonterminal followVar = sym as Nonterminal;
if (table.Get(followVar, tokenStream.Current) != null)
{
if (Program.debug)
{
Console.WriteLine(" PARSE: Discarding symbol " + symbolStack.Peek());
}
isValidParseTree = false;
symbolStack.Pop();
treeStack.Pop();
goto sync;
}
}
}
// default action: skip current terminal
if (Program.debug)
{
Console.WriteLine(" PARSE: Discard token");
}
if (!tokenStream.MoveNext())
{
throw new Exception("OUT OF TOKENS");
}
}
}
}
