public List <ISymbol[]> GetProductionRules(Nonterminal v)
{
List <ISymbol[]> productionRules = null;
productions.TryGetValue(v, out productionRules);
return(productionRules);
}
// create a context free grammar object with some starting symbol and sets of terminals and nonterminals
public CFG(Nonterminal start, ICollection <Terminal> terminals, ICollection <Nonterminal> nonterminals)
{
this.startSymbol = start;
this.terminals = terminals.ToList();
this.terminals.Add(Terminal.EOF);
this.nonterminals = nonterminals;
}
// create a context free grammar object with some starting symbol and sets of terminals and nonterminals
public CFG(Nonterminal start, ICollection<Terminal> terminals, ICollection<Nonterminal> nonterminals)
{
this.startSymbol = start;
this.terminals = terminals.ToList();
this.terminals.Add(Terminal.EOF);
this.nonterminals = nonterminals;
}
// 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];
}
// 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;
}
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)");
}
// or from a nonterminal and a token, in which case we need to check
// all the terminals for the given nonterminal and see if one matches the token
public ISymbol[] Get(Nonterminal var, Token token)
{
Dictionary<Terminal, ISymbol[]> tableRow = table[var];
foreach (Terminal term in tableRow.Keys)
if (term.Matches(token))
return tableRow[term];
return null;
}
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)");
}
}
// 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 void AddProductionRule(Nonterminal v, ISymbol[] result)
{
if (productions.ContainsKey(v))
{
List <ISymbol[]> vProductions = productions[v];
vProductions.Add(result);
}
else
{
List <ISymbol[]> vProductions = new List <ISymbol[]>();
vProductions.Add(result);
productions.Add(v, vProductions);
}
}
// or from a nonterminal and a token, in which case we need to check
// all the terminals for the given nonterminal and see if one matches the token
public ISymbol[] Get(Nonterminal var, Token token)
{
Dictionary <Terminal, ISymbol[]> tableRow = table[var];
foreach (Terminal term in tableRow.Keys)
{
if (term.Matches(token))
{
return(tableRow[term]);
}
}
return(null);
}
public void AddProductionRule(Nonterminal v, ISymbol[] result)
{
if (productions.ContainsKey(v))
{
List<ISymbol[]> vProductions = productions[v];
vProductions.Add(result);
}
else
{
List<ISymbol[]> vProductions = new List<ISymbol[]>();
vProductions.Add(result);
productions.Add(v, vProductions);
}
}
// iteratively computer the follow set for each grammar nonterminal
private void ComputeFollowSets()
{
if (Program.debug)
{
Console.WriteLine("=========================================================");
Console.WriteLine("CFG: Computing follow sets");
}
// initialize follow sets for each nonterminal by going through every production
// and finding terminals that come immediately after nonterminals in the productions
foreach (Nonterminal A in nonterminals)
{
follow[A] = new HashSet <Terminal>();
if (A == startSymbol)
{
follow[A].Add(Terminal.EOF);
}
foreach (Nonterminal B in nonterminals)
{
foreach (ISymbol[] prod in productions[B])
{
for (int i = 0; i < prod.Length - 1; i++)
{
if (prod[i] == A && prod[i + 1] is Terminal)
{
follow[A].Add(prod[i + 1] as Terminal);
}
}
}
}
if (Program.debug)
{
Console.WriteLine("CFG: Follow " + A + " = " + SymbolsToString(follow[A]));
}
}
if (Program.debug)
{
Console.WriteLine("CFG: Initial step done");
}
// loop until follow sets unchanged
bool converged = false;
while (!converged)
{
converged = true;
if (Program.debug)
{
Console.WriteLine("--iteration--");
}
// foreach nonterminal and for each of its productions
foreach (Nonterminal B in nonterminals)
{
foreach (ISymbol[] prod in productions[B])
{
// for every nonterminal in the production
for (int i = 0; i < prod.Length; i++)
{
if (prod[i] is Nonterminal)
{
// A is a nonterminal symbol in a production rule for B
Nonterminal A = prod[i] as Nonterminal;
int tmp = follow[A].Count;
// A is last symbol in a production of B, add Follow(B) to Follow(A)
if (i == prod.Length - 1)
{
follow[A].UnionWith(follow[B]);
}
else
{
ISymbol[] w = prod.Skip(i + 1).ToArray(); // rest of the production after A
ISet <Terminal> fw = First(w);
// if epsilon can be derived from symbols following A in the production
// add Follow(B) to Follow(A)
if (fw.Contains(Terminal.EPSILON))
{
follow[A].UnionWith(follow[B]);
fw.Remove(Terminal.EPSILON);
}
// add first set of the symbols following A in the production to Follow(A) in any case
follow[A].UnionWith(fw);
}
if (follow[A].Count > tmp)
{
// a follow set was changed
if (Program.debug)
{
Console.WriteLine("CFG: used rule " + B + " -> " + SymbolsToString(prod) + "for " + A);
Console.WriteLine("CFG: Follow " + A + " = " + SymbolsToString(follow[A]));
Console.WriteLine();
}
converged = false;
}
}
}
}
}
if (Program.debug)
{
foreach (Nonterminal A in nonterminals)
{
Console.WriteLine("CFG: Follow " + A + " = " + SymbolsToString(follow[A]));
}
}
}
if (Program.debug)
{
Console.WriteLine("CFG: follow sets converged");
}
}
// 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);
}
// 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;
}
// get the follow set for a grammar nonterminal
public ISet <Terminal> Follow(Nonterminal startVar)
{
return(follow[startVar]);
}
// get the follow set for a grammar nonterminal
public ISet<Terminal> Follow(Nonterminal startVar)
{
return follow[startVar];
}
public ParseTree(Nonterminal var)
{
this.nonterminal = var;
}
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 Runnable ProcessParseTree(ParseTree parseTree, IEnumerable<Error> parseErrors, bool isValidParseTree)
{
Runnable prog = new Runnable();
foreach (Error err in parseErrors)
{
prog.errors.Add(err);
}
// can't construct AST if parse tree is bad
if (!isValidParseTree)
return prog;
// first remove unnecessary symbols ; : .. ( ) := and epsilons
String[] pruneTokens = { "(", ")", ";", ":", "..", ":=", "var", "in", "for", "end", "do" };
Predicate<IParseNode> isUnnecessaryTerminal =
n => (n is ParseLeaf) ? (n as ParseLeaf).Token == null || pruneTokens.Contains((n as ParseLeaf).Token.Lexeme) : false;
parseTree.RemoveNodes(isUnnecessaryTerminal);
// remove any tree nodes with no children
Predicate<IParseNode> isEmptyNonterminal =
v => (v is ParseTree) ? (v as ParseTree).Children.Count == 0 : false;
parseTree.RemoveNodes(isEmptyNonterminal);
// refactor
// STMTS->STMTS_HEAD STMTS_TAIL to STMTS->(STMT)+
// DECL->"var" <IDENT> ":" <TYPE> ASSIGN to DECL->"var" <IDENT> ":" <TYPE> [":=" <EXPR>]
// EXPR->UNARY|OPND BINARY to EXPR-> unary_op OPND | OPND | OPND binary_op OPND
// OPND-><INT>|<STRING>|<IDENT>|<EXPR> to just <INT>|<STRING>|<IDENT>|<EXPR>
Nonterminal[] pruneVariables = new Nonterminal[] {
nonterminals["statements_head"], nonterminals["statements_tail"], nonterminals["unary_operation"],
nonterminals["binary_operation"], nonterminals["declaration_assignment"], nonterminals["operand"] };
Predicate<IParseNode> isUnnecessaryNonterminal =
n => (n is ParseTree) ? pruneVariables.Contains((n as ParseTree).Nonterminal) : false;
parseTree.RemoveNodes(isUnnecessaryNonterminal);
if (Program.debug) Console.WriteLine(parseTree);
// AST is formed at this point, so do semantic checks
// find declarations, produce errors if identifier declared multiple times
foreach (IParseNode node in parseTree.Nodes())
{
if (node is ParseTree)
{
ParseTree subtree = node as ParseTree;
if (subtree.Nonterminal == nonterminals["declaration"])
{
ParseLeaf idLeaf = (subtree.Children[0] as ParseLeaf);
ParseLeaf typeLeaf = (subtree.Children[1] as ParseLeaf);
Token idToken = idLeaf.Token;
Token typeToken = typeLeaf.Token;
string identifier = idToken.Lexeme;
ValueType type = Value.TypeFromString(typeToken.Lexeme);
Statement.DeclarationStmt declaration;
switch (subtree.Children.Count)
{
case 2: // simple declaration
declaration = new Statement.DeclarationStmt(identifier, type, idToken);
break;
case 3: // declaration with assignment
ParseLeaf valueLeaf = (subtree.Children[2] as ParseLeaf);
Expression expr = Expression.FromTreeNode(subtree.Children[2], terminals, nonterminals);
declaration = new Statement.DeclarationStmt(identifier, type, idToken, expr);
break;
default:
throw new Exception("BAD AST STRUCTURE");
}
if (prog.declarations.ContainsKey(identifier))
prog.errors.Add(new SemanticError(idToken, identifier + " multiply defined"));
else
prog.declarations[identifier] = declaration;
}
}
}
// check that variables are defined before use
foreach (IParseNode node in parseTree.Nodes())
{
if (node is ParseLeaf)
{
ParseLeaf leaf = node as ParseLeaf;
Token leafToken = leaf.Token;
if (leafToken.Type == tokenTypes["identifier"])
{
string identifier = leafToken.Lexeme;
Position idPosition = leafToken.TextPosition;
if (!prog.declarations.ContainsKey(identifier))
prog.errors.Add(new SemanticError(leafToken, identifier + " never defined"));
else if (idPosition.CompareTo(prog.declarations[identifier].Token.TextPosition) < 0)
prog.errors.Add(new SemanticError(leafToken, identifier + " not defined before use"));
}
}
}
// add statements to runnable
ParseTree statementListNode = parseTree.Children[0] as ParseTree;
foreach (IParseNode statementNode in statementListNode.Children)
prog.statements.Add(Statement.FromTreeNode(statementNode, terminals, nonterminals));
// check that for-loop control variables are not modified inside the for-loop
foreach (Statement stmt in prog.statements)
{
if (stmt is Statement.ForStmt)
{
Statement.ForStmt forStmt = stmt as Statement.ForStmt;
Stack<Statement> stmtStack = new Stack<Statement>();
foreach (Statement substmt in forStmt.Block)
stmtStack.Push(substmt);
while (stmtStack.Count != 0)
{
Statement s = stmtStack.Pop();
if (s is Statement.AssignStmt)
{
Statement.AssignStmt assignment = s as Statement.AssignStmt;
if (assignment.Identifier == forStmt.Identifier)
prog.errors.Add(new SemanticError(assignment.Token, forStmt.Identifier + " cannot be modified inside for-loop"));
}
else if (s is Statement.DeclarationStmt)
{
Statement.DeclarationStmt declaration = s as Statement.DeclarationStmt;
if (declaration.Identifier == forStmt.Identifier)
prog.errors.Add(new SemanticError(declaration.Token, forStmt.Identifier + " cannot be modified inside for-loop"));
}
else if (s is Statement.ForStmt)
{
Statement.ForStmt nestedFor = s as Statement.ForStmt;
if (nestedFor.Identifier == forStmt.Identifier)
prog.errors.Add(new SemanticError(nestedFor.Token, forStmt.Identifier + " cannot be modified inside for-loop"));
foreach (Statement substmt in nestedFor.Block)
stmtStack.Push(substmt);
}
}
}
}
// typecheck each statement
foreach (Statement stmt in prog.statements)
stmt.TypeCheck(prog);
return prog;
}
public List<ISymbol[]> GetProductionRules(Nonterminal v)
{
List<ISymbol[]> productionRules = null;
productions.TryGetValue(v, out productionRules);
return productionRules;
}
public ParseTree(Nonterminal var)
{
this.nonterminal = var;
}
// 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");
}
}
}
}
