/// <name>Parser</name>
/// <type>Constructor</type>
/// <summary>
/// Creates a parser object.
/// </summary>
public Parser()
: base()
{
// Initalize class variables
currentOperator = new Token(TokenType.NO_TOKEN, String.Empty);
nextOperator = new Token(TokenType.NO_TOKEN, String.Empty);
operand = null;
literal = null;
symbol = null;
inExpression = false;
nextLocation = 1;
symbolTable = new Dictionary<string, Symbol>();
literalTable = new Dictionary<string, Literal>();
operatorStack = new Stack<Token>();
structureStack = new Stack<Structure>();
forwardRefStack = new Stack<ForwardReference>();
}
/// <name>Scanner</name>
/// <type>Constructor</type>
/// <summary>
/// Creates a scanner object and initializes the operator lookup table and
/// global token object for use during execution of the class's public methods.
/// </summary>
public Scanner()
: base()
{
operatorTable = new Dictionary<string, TokenType>();
operatorTable["+"] = TokenType.ADD_OP;
operatorTable["-"] = TokenType.ADD_OP;
operatorTable["*"] = TokenType.MULT_OP;
operatorTable["/"] = TokenType.MULT_OP;
operatorTable["("] = TokenType.L_PAREN;
operatorTable[")"] = TokenType.R_PAREN;
operatorTable["["] = TokenType.L_BRACKET;
operatorTable["]"] = TokenType.R_BRACKET;
operatorTable[";"] = TokenType.SEMICOLON;
operatorTable["<"] = TokenType.REL_OP;
operatorTable["<="] = TokenType.REL_OP;
operatorTable["="] = TokenType.REL_OP;
operatorTable[">="] = TokenType.REL_OP;
operatorTable[">"] = TokenType.REL_OP;
operatorTable["!="] = TokenType.REL_OP;
operatorTable[","] = TokenType.COMMA;
operatorTable[":="] = TokenType.ASSIGN;
operatorTable["call"] = TokenType.CALL;
operatorTable["else"] = TokenType.ELSE;
operatorTable["endif"] = TokenType.END;
operatorTable["endproc"] = TokenType.END;
operatorTable["endprogram"] = TokenType.END;
operatorTable["endwhile"] = TokenType.END;
operatorTable["if"] = TokenType.IF_WHILE;
operatorTable["while"] = TokenType.IF_WHILE;
operatorTable["proc"] = TokenType.PROC;
operatorTable["then"] = TokenType.THEN_DO;
operatorTable["do"] = TokenType.THEN_DO;
operatorTable["write"] = TokenType.WRITE;
operatorTable["program"] = TokenType.PROGRAM;
operatorTable["const"] = TokenType.CONST;
operatorTable["var"] = TokenType.VAR;
operatorTable["{"] = TokenType.L_BRACE;
operatorTable["}"] = TokenType.R_BRACE;
currentToken = new Token(TokenType.NO_TOKEN, String.Empty);
}
public object Clone()
{
Token copyToken = new Token(this.Type, this.Lexeme);
return copyToken;
}
/// <name>GetNextToken</name>
/// <type>Method</type>
/// <summary>
/// Returns the next token in the file and consumes the token.
/// </summary>
/// <returns>Next valid token from the input file</returns>
public Token GetNextToken()
{
if (sourceFile.EndOfStream)
{
// If we're at the end of the file stream then return end file token
currentToken.Type = TokenType.END_FILE;
currentToken.Lexeme = String.Empty;
return currentToken;
}
else
{
int nextChar = sourceFile.Read();
// Skip over white space but process newline characters by updating
// the line count (static variable to the compiler class)
while (Char.IsWhiteSpace((char)nextChar))
{
if (sourceFile.EndOfStream)
{
currentToken.Type = TokenType.END_FILE;
currentToken.Lexeme = String.Empty;
return currentToken;
}
else if (nextChar == '\n')
{
Compiler.LineNumber++;
}
nextChar = sourceFile.Read();
}
if (Char.IsLetter((char)nextChar))
{
// Symbol token - call ReadSymbol to handle the rest
currentToken = ReadSymbol(nextChar);
}
else if (Char.IsDigit((char)nextChar))
{
// Literal token - call ReadLiteral to handle the rest
currentToken = ReadLiteral(nextChar);
}
else if (nextChar == '\'')
{
// Literal token in apostrophes - get the integer representation
// of the token and skip over the second apostrophe
nextChar = sourceFile.Read();
currentToken.Lexeme = nextChar.ToString();
currentToken.Type = TokenType.LITERAL;
sourceFile.Read();
}
else if (nextChar == '/')
{
// We either have a comment or an operator - run a series of checks
// to determine which token type we've encountered
if (sourceFile.Peek() == '/')
{
// Comment - skip over the rest of the line then get the next token
while (nextChar != '\n')
{
nextChar = sourceFile.Read();
}
Compiler.LineNumber++;
currentToken = GetNextToken();
}
else if (sourceFile.Peek() == '*')
{
// Multi-line comment - keep reading until we encounter the end
// of the comment (and update line number on newline characters)
// then get the next token
while (nextChar != '*' || sourceFile.Peek() != '/')
{
nextChar = sourceFile.Read();
if (nextChar == '\n')
{
Compiler.LineNumber++;
}
}
sourceFile.Read();
currentToken = GetNextToken();
}
else
{
// Operator token - check the lookup table to get the type
string op = ((char)nextChar).ToString();
if (operatorTable.ContainsKey(op))
{
currentToken.Lexeme = op;
currentToken.Type = operatorTable[op];
}
else
{
Compiler.ThrowCompilerException("Illegal operator");
}
}
}
else if (sourceFile.Peek() == '=')
{
// Two-character operator token - check the lookup table to get the type
string op = ((char)nextChar).ToString() + ((char)sourceFile.Read()).ToString();
if (operatorTable.ContainsKey(op))
{
currentToken.Lexeme = op;
currentToken.Type = operatorTable[op];
}
else
{
Compiler.ThrowCompilerException("Illegal operator");
}
}
else
{
// Remaining operator token - check the lookup table to get the type
string op = ((char)nextChar).ToString();
if (operatorTable.ContainsKey(op))
{
currentToken.Lexeme = op;
currentToken.Type = operatorTable[op];
}
else
{
Compiler.ThrowCompilerException("Illegal operator");
}
}
return currentToken;
}
}
/// <name>ReadSymbol</name>
/// <type>Method</type>
/// <summary>
/// Advances through the input file and tokenizes character sets classified
/// as symbols. The input file is read until a non-letter character is
/// encountered and the entire token is returned.
/// </summary>
/// <param name="firstChar">First character of the token</param>
/// <returns>Token object with the type and lexeme of the symbol</returns>
private Token ReadSymbol(int firstChar)
{
Token newToken = new Token();
string symbol = Char.ToLower(((char)firstChar)).ToString();
// Keep reading letters until we have the entire literal (converting each
// letter to lowercase along the way)
while (Char.IsLetter((char)sourceFile.Peek()))
{
symbol += Char.ToLower(((char)sourceFile.Read())).ToString();
}
// Check to see if we have a reserved word
if (operatorTable.ContainsKey(symbol))
{
newToken.Type = operatorTable[symbol];
}
else
{
newToken.Type = TokenType.SYMBOL;
}
// Return the token
newToken.Lexeme = symbol;
return newToken;
}
/// <name>ReadLiteral</name>
/// <type>Method</type>
/// <summary>
/// Advances through the input file and tokenizes literal values. Literals are
/// numbers that can be in base 10, base 8 (octal) or base 16 (hex). Return value
/// is always in base 10. If the literal is already in base 10, a string is returned
/// representing that number. If the literal is in base 8 or base 16, it is converted
/// to base 10 and then returned.
/// </summary>
/// <param name="firstChar">First character of the token</param>
/// <returns>Token object with the type and lexeme of the literal</returns>
private Token ReadLiteral(int firstChar)
{
Token newToken = new Token();
string literal = String.Empty, tempString = String.Empty;
int numberBase = 0, power = 0, result = 0;
// If the token starts with a 0 then it is octal, 0x and it is hex
if (firstChar == '0' && (sourceFile.Peek() == 'x' || sourceFile.Peek() == 'X'))
{
sourceFile.Read();
numberBase = 16;
}
else if (firstChar == '0')
{
numberBase = 8;
}
else
{
// First char is part of the literal in base 10 so append it
tempString += ((char)firstChar).ToString();
numberBase = 10;
}
// Keep reading digits (or hex characters) until we have the entire literal
while (Char.IsDigit((char)sourceFile.Peek()) || IsHexDigit((char)sourceFile.Peek()))
{
tempString += ((char)sourceFile.Read()).ToString();
}
if (numberBase == 8)
{
// If the number is octal then convert it to base 10 and set the literal
// to the string value of the base 10 number
for (int i = tempString.Length - 1; i >= 0; i--)
{
int digit = 0;
try
{
digit = Convert.ToInt32(tempString[i].ToString());
}
catch
{
Compiler.ThrowCompilerException("Invalid literal");
}
if (digit < 8)
{
result += digit * (int)Math.Pow(8, power);
power++;
}
else
{
Compiler.ThrowCompilerException("Invalid literal");
}
}
literal = result.ToString();
}
else if (numberBase == 16)
{
// If the number is hex then convert it to base 10 and set the literal
// to the string value of the base 10 number
for (int i = tempString.Length - 1; i >= 0; i--)
{
char tempChar = Char.ToLower(tempString[i]);
int digit = 0;
switch (tempChar)
{
case 'a':
digit = 10;
break;
case 'b':
digit = 11;
break;
case 'c':
digit = 12;
break;
case 'd':
digit = 13;
break;
case 'e':
digit = 14;
break;
case 'f':
digit = 15;
break;
default:
digit = Convert.ToInt32(tempChar.ToString());
break;
}
result += digit * (int)Math.Pow(16, power);
power++;
}
literal = result.ToString();
}
else
{
// Already in base 10 so no conversion is needed
for (int i = tempString.Length - 1; i >= 0; i--)
{
if (Char.IsDigit((char)tempString[i]) == false)
{
Compiler.ThrowCompilerException("Invalid literal");
}
}
literal = tempString;
}
// Return the token
newToken.Lexeme = literal;
newToken.Type = TokenType.LITERAL;
return newToken;
}
/// <name>Parentheses</name>
/// <type>Method</type>
/// <summary>
/// Parentheses code generator.
/// </summary>
private void Parentheses()
{
Advance();
currentOperator = operatorStack.Pop();
status = Status.FREEZE;
}
/// <name>EqualityForOperators</name>
/// <type>Method</type>
/// <summary>
/// Equality for operators code generator.
/// </summary>
private void EqualityForOperators()
{
switch (currentOperator.Lexeme)
{
case "+":
jvm.Emit(Opcode.IADD);
break;
case "-":
jvm.Emit(Opcode.ISUB);
break;
case "*":
jvm.Emit(Opcode.IMUL);
break;
case "/":
jvm.Emit(Opcode.IDIV);
break;
default:
Compiler.ThrowCompilerException("Invalid expression operator");
break;
}
currentOperator = operatorStack.Pop();
}
/// <name>CompileVariables</name>
/// <type>Method</type>
/// <summary>
/// Generates code for all variable declarations in the program.
/// </summary>
private void CompileVariables()
{
// Keep processing until we reach the end of the declarations
while (nextOperator.Type != TokenType.SEMICOLON)
{
// Advance to the next variable declaration and record the address
Advance();
if (symbol.Address == -1)
{
symbolTable[operand.Lexeme].Address = nextLocation;
}
else
{
Compiler.ThrowCompilerException("Symbol has already been defined");
}
// Now generate bytecode for the declaration
if (nextOperator.Type == TokenType.L_BRACKET)
{
// Array declaration
Token array = new Token(TokenType.SYMBOL, operand.Lexeme);
symbolTable[operand.Lexeme].SymbolType = SymbolType.ARRAY_TYPE;
Advance();
// Push the number representing the size of the array
if (literal.NumberValue > 0 && literal.NumberValue < 101)
{
PushConstant(literal.NumberValue);
}
else
{
Compiler.ThrowCompilerException("Illegal array declaration");
}
jvm.Emit(Opcode.NEWARRAY, (byte)10);
jvm.ChooseOp(Opcode.ASTORE, Opcode.ASTORE_0, nextLocation);
Advance();
if (nextOperator.Type == TokenType.REL_OP)
{
// If the next operator is an equal sign then we have an initalization
Advance();
int arrayIndex = 0;
while (nextOperator.Type != TokenType.R_BRACE)
{
// Keep going until we come to the closing bracket
Advance();
PushOperand(array);
PushConstant(arrayIndex);
PushOperand(operand);
jvm.Emit(Opcode.IASTORE);
arrayIndex++;
}
// Advance past the right brace and comma
Advance();
}
}
else
{
// Variable declaration
symbolTable[operand.Lexeme].SymbolType = SymbolType.VAR_TYPE;
if (nextOperator.Lexeme == "=")
{
// If the next operator is an equal sign then we have an initalization
Advance();
if (operand.Type == TokenType.LITERAL)
{
// Assignment is a literal
PushConstant(literalTable[operand.Lexeme].NumberValue);
}
else
{
// Assignment is a previously-defined symbol
PushOperand(operand);
}
}
else if (nextOperator.Type == TokenType.COMMA || nextOperator.Type == TokenType.SEMICOLON)
{
// Variable declaration with no initalization so initalize to zero
PushConstant(0);
}
else
{
Compiler.ThrowCompilerException("Invalid operator found in variable declarations");
}
// Generate code to store the inital value at the variable's address
jvm.ChooseOp(Opcode.ISTORE, Opcode.ISTORE_0, nextLocation);
}
// Finally increase the location index since we assigned the current address
nextLocation++;
}
Advance();
}
/// <name>CompileExpression</name>
/// <type>Method</type>
/// <summary>
/// Compile arithmetic expression.
/// </summary>
private void CompileExpression()
{
// Perform initalization and generate code for the first operand
inExpression = true;
status = Status.FREEZE;
if (operand != null)
{
if (operand.Lexeme == "length")
{
// First we need to save the new current operator
operatorStack.Push(currentOperator);
// Push array length onto the stack
Advance();
jvm.ChooseOp(Opcode.ALOAD, Opcode.ALOAD_0, symbol.Address);
jvm.Emit(Opcode.ARRAYLENGTH);
Advance();
// Now retrieve the current operator that we started with
currentOperator = operatorStack.Pop();
}
else
{
PushOperand(operand);
}
}
// Loop continues until we reach the end of the expression
while (inExpression == true)
{
if (status == Status.EXIT)
{
inExpression = false;
}
else if (status == Status.FREEZE)
{
status = Status.CONTINUE;
}
else
{
// Advance but save the current operator on the stack
operatorStack.Push(currentOperator);
Advance();
// Generate code to push operand onto the stack
if (operand != null)
{
if (operand.Lexeme == "length")
{
// First we need to save the new current operator
operatorStack.Push(currentOperator);
// Push array length onto the stack
Advance();
jvm.ChooseOp(Opcode.ALOAD, Opcode.ALOAD_0, symbol.Address);
jvm.Emit(Opcode.ARRAYLENGTH);
Advance();
// Now retrieve the current operator that we started with
currentOperator = operatorStack.Pop();
}
else
{
PushOperand(operand);
}
}
}
// Call the appropriate code generator for the current operators
CodeGenerator(currentOperator, nextOperator);
}
}
/// <name>CodeGenerator</name>
/// <type>Method</type>
/// <summary>
/// Calls the appropriate code generator function based on the set of
/// two-character codes used in the CONO table. Each valid operator
/// type is converted to an integer representing a position in the table,
/// which returns the string that identifies which code generator to call.
/// </summary>
/// <param name="currentOp">Current operator token</param>
/// <param name="nexttOp">Next operator token</param>
private void CodeGenerator(Token currentOp, Token nextOp)
{
if (currentOp.Type == TokenType.CALL && nextOp.Type != TokenType.SEMICOLON)
{
Compiler.ThrowCompilerException("No semicolon found following procedure call");
}
else if (currentOp.Type == TokenType.PROC && nextOp.Type != TokenType.SEMICOLON)
{
Compiler.ThrowCompilerException("No semicolon found following procedure definition");
}
else if (currentOp.Type == TokenType.END && nextOp.Type != TokenType.SEMICOLON)
{
Compiler.ThrowCompilerException("No semicolon found following end of control block");
}
int first = GetTokenTypeIndex(currentOp.Type);
int second = GetTokenTypeIndex(nextOp.Type);
string code = CONO[first,second];
switch (code)
{
case "AS":
Assignment();
break;
case "BC":
BeginCondition();
break;
case "CA":
CallProcedure();
break;
case "EB":
EndBlock();
break;
case "EE":
EndExpression();
break;
case "EQ":
EqualityForOperators();
break;
case "GT":
GreaterThanForOperators();
break;
case "LT":
LessThanForOperators();
break;
case "NO":
NoOperation();
break;
case "PA":
Parentheses();
break;
case "PR":
ProcedureDefinition();
break;
case "SU":
Subscript();
break;
case "WR":
Write();
break;
default:
Error();
break;
}
}
/// <name>Advance</name>
/// <type>Method</type>
/// <summary>
/// Advances the current and next operators from the input file and
/// populates the operand token between the operators if one exists.
/// </summary>
private void Advance()
{
// Advance one token
currentOperator = (Token)nextOperator.Clone();
Token peekToken = source.GetNextToken();
// Make sure operand, symbol and literal are null
operand = null;
symbol = null;
literal = null;
if (peekToken.Type == TokenType.SYMBOL)
{
// If the new token is a symbol then set operand to that token
operand = (Token)peekToken.Clone();
// Now check to see if the symbol is already defined - if
// it is then assign to symbol, otherwise add it to the table
if (symbolTable.ContainsKey(peekToken.Lexeme))
{
symbol = symbolTable[peekToken.Lexeme];
}
else
{
symbolTable.Add(peekToken.Lexeme, new Symbol(peekToken.Lexeme));
symbol = symbolTable[peekToken.Lexeme];
}
// Finally get the next token
nextOperator = source.GetNextToken();
}
else if (peekToken.Type == TokenType.LITERAL)
{
// If the new token is a literal then set operand to that token
operand = (Token)peekToken.Clone();
// Now check to see if the literal is already defined - if
// it is then assign to literal, otherwise add it to the table
if (literalTable.ContainsKey(peekToken.Lexeme))
{
literal = literalTable[peekToken.Lexeme];
}
else
{
literalTable.Add(peekToken.Lexeme, new Literal(peekToken.Lexeme));
literal = literalTable[peekToken.Lexeme];
}
// Finally get the next token
nextOperator = source.GetNextToken();
}
else
{
// Set the next operator to the token we just retrieved
nextOperator = (Token)peekToken.Clone();
}
// Output debug information
Compiler.WriteToDebug("Parser - Advance called (line " + Compiler.LineNumber + ")");
Compiler.WriteToDebug(" current operator:\t" + currentOperator.Lexeme);
if (operand != null)
{
Compiler.WriteToDebug(" operand:\t\t" + operand.Lexeme);
}
Compiler.WriteToDebug(" next operator:\t" + nextOperator.Lexeme);
Compiler.WriteToDebug(String.Empty);
}
/// <name>PushOperand</name>
/// <type>Method</type>
/// <summary>
/// Push an operand onto the runtime stack.
/// </summary>
private void PushOperand(Token op)
{
if (op.Type == TokenType.LITERAL)
{
// Literal value
PushConstant(literalTable[op.Lexeme].NumberValue);
}
else
{
// Previously-declared symbol
if (symbolTable[op.Lexeme].Address != -1)
{
Symbol s = symbolTable[op.Lexeme];
if (s.SymbolType == SymbolType.ARRAY_TYPE)
{
// Emit code to load the array
jvm.ChooseOp(Opcode.ALOAD, Opcode.ALOAD_0, s.Address);
}
else
{
if (nextOperator.Type != TokenType.L_BRACKET)
{
// Emit code to load the constant or variable
jvm.ChooseOp(Opcode.ILOAD, Opcode.ILOAD_0, s.Address);
}
else
{
Compiler.ThrowCompilerException("Subscript given for non-array symbol");
}
}
}
else
{
Compiler.ThrowCompilerException("Reference to undefined symbol");
}
}
}
public object Clone()
{
Token copyToken = new Token(this.Type, this.Lexeme);
return(copyToken);
}
