private SymTabEntry parseRoutineName(Token token, String dummyName) throws Exception {
SymTabEntry routineId = null;
// Parse the routine name identifier.
if (token.getType() == IDENTIFIER) {
String routineName = token.getText().toLowerCase();
routineId = symTabStack.lookupLocal(routineName);
// Not already defined locally: Enter into the local symbol table.
if (routineId == null) {
routineId = symTabStack.enterLocal(routineName);
routineId.appendLineNumber(token.getLineNumber());
}
// If already defined, it should be a forward definition.
else {
routineId = null;
errorHandler.flag(token, IDENTIFIER_REDEFINED, this);
}
token = nextToken(); // consume routine name identifier
} else {
errorHandler.flag(token, MISSING_IDENTIFIER, this);
}
// If necessary, create a dummy routine name symbol table entry.
if (routineId == null) {
routineId = symTabStack.enterLocal(dummyName);
}
return routineId;
}
protected void parseFormalParameters(Token token, SymTabEntry routineId) throws Exception {
// Parse the formal parameters if there is an opening left parenthesis.
token = synchronize(LEFT_PAREN_SET);
if (token.getType() == LEFT_PAREN) {
token = nextToken(); // consume (
TokenType tokenType1 = token.getType();
ArrayList<SymTabEntry> parms = new ArrayList<SymTabEntry>();
while (tokenType1 != RIGHT_PAREN) {
parms.addAll(parseParmSublist(token, routineId));
token = currentToken();
tokenType1 = token.getType();
}
// This sync needs to check for Int or Char
token = synchronize(PARAMETER_SET);
// Closing right parenthesis.
if (token.getType() == RIGHT_PAREN) {
token = nextToken(); // consume )
} else {
errorHandler.flag(token, MISSING_RIGHT_PAREN, this);
}
routineId.setAttribute(ROUTINE_PARMS, parms);
}
}
/**
* Compile or interpret a Pascal source program.
*
* @param operation either "compile" or "execute".
* @param filePath the source file path.
* @param flags the command line flags.
*/
public Pascal(String operation, String sourcePath, String inputPath, String flags) {
try {
intermediate = flags.indexOf('i') > -1;
xref = flags.indexOf('x') > -1;
source = new Source(new BufferedReader(new FileReader(sourcePath)));
source.addMessageListener(new SourceMessageListener());
parser = FrontendFactory.createParser("Pascal", "top-down", source);
parser.addMessageListener(new ParserMessageListener());
backend = BackendFactory.createBackend(operation, inputPath);
backend.addMessageListener(new BackendMessageListener());
parser.parse();
source.close();
if (parser.getErrorCount() == 0) {
symTabStack = parser.getSymTabStack();
SymTabEntry programId = symTabStack.getProgramId();
iCode = (ICode) programId.getAttribute(ROUTINE_ICODE);
if (xref) {
CrossReferencer crossReferencer = new CrossReferencer();
crossReferencer.print(symTabStack);
}
if (intermediate) {
ParseTreePrinter treePrinter = new ParseTreePrinter(System.out);
treePrinter.print(symTabStack);
}
backend.process(iCode, symTabStack);
}
} catch (Exception ex) {
System.out.println("***** Internal translator error. *****");
ex.printStackTrace();
}
}
/**
* 解析标识符
*
* @param token
* @return 对应符号表项
* @throws Exception
*/
private SymTabEntry parseIdentifier(Token token) throws Exception {
SymTabEntry id = null;
if (token.getType() == IDENTIFIER) {
String name = token.getText().toLowerCase();
id = symTabStack.lookupLocal(name);
// 申明必须得没有，否则重定义了
if (id == null) {
id = symTabStack.enterLocal(name);
id.setDefinition(definition);
id.appendLineNumber(token.getLineNumber());
} else {
errorHandler.flag(token, IDENTIFIER_REDEFINED, this);
}
token = nextToken();
} else {
errorHandler.flag(token, MISSING_IDENTIFIER, this);
}
return id;
}
/**
* 解析标识符子表及类型说明
*
* @param token the current token.
* @return 一个申明中的子表标识符列表
* @throws Exception
*/
protected ArrayList<SymTabEntry> parseIdentifierSublist(Token token) throws Exception {
ArrayList<SymTabEntry> sublist = new ArrayList<SymTabEntry>();
do {
token = synchronize(IDENTIFIER_START_SET);
SymTabEntry id = parseIdentifier(token);
if (id != null) {
sublist.add(id);
}
token = synchronize(COMMA_SET);
TokenType tokenType = token.getType();
// 找逗号
if (tokenType == COMMA) {
token = nextToken(); // 有逗号吞掉以便下一个
if (IDENTIFIER_FOLLOW_SET.contains(token.getType())) {
errorHandler.flag(token, MISSING_IDENTIFIER, this);
}
} else if (IDENTIFIER_START_SET.contains(tokenType)) {
errorHandler.flag(token, MISSING_COMMA, this);
}
} while (!IDENTIFIER_FOLLOW_SET.contains(token.getType()));
// 冒号后面的类型
TypeSpec type = parseTypeSpec(token);
// 将子表中的每个变量的类型设置上，比如a,b,c:int，先解析了a,b,c，然后解析了int类型
// 那么这儿就会给a,b,c设置上int类型
for (SymTabEntry variableId : sublist) {
variableId.setTypeSpec(type);
}
return sublist;
}
private ArrayList<SymTabEntry> parseParmSublist(Token token, SymTabEntry routineId)
throws Exception {
// name = main, definition == function
Definition parmDefn = null;
TokenType tokenType = token.getType();
parmDefn = VALUE_PARM;
SymTabEntry sublist = null;
if (tokenType == INT) {
IntDeclarationsParser intDeclarationsParser = new IntDeclarationsParser(this);
intDeclarationsParser.setDefinition(parmDefn);
token = nextToken();
sublist = intDeclarationsParser.parseIdentifier(token);
sublist.setTypeSpec(integerType);
} else if (tokenType == CHAR) {
IntDeclarationsParser intDeclarationsParser = new IntDeclarationsParser(this);
intDeclarationsParser.setDefinition(parmDefn);
token = nextToken();
sublist = intDeclarationsParser.parseIdentifier(token);
sublist.setTypeSpec(charType);
}
token = nextToken();
token = currentToken();
if (token.getType() == COMMA) {
token = nextToken();
}
tokenType = token.getType();
token = synchronize(PARAMETER_SET);
ArrayList<SymTabEntry> SC = new ArrayList<SymTabEntry>();
SC.add(sublist);
return SC;
}
/**
* Parse a Triangle SingleDeclarationParser.
*
* <p>Single-Declaration ::= const Identifier ~ Expression | var Identifier : Type-Denoter | proc
* Identifier (Formal-Parameter-Sequence) ~ Single-Command | func Identifier
* (Formal-Parameter-Sequence) : Type-Denoter ~ Expression | Type Identifier ~ Type-Denoter
*
* <p>To be overridden by the specialized command parse subclasses.
*
* @param token the initial token.
* @return the root node of the generated parse tree.
* @throws Exception if an error occurred.
*/
public void parse(Token token) throws Exception {
SingleCommandParser singleCommand = null;
ExpressionParser expression = null;
FormalParameterSequenceParser formalParameterSequence = null;
TypeDenoterParser typeDenoter = null;
EnumSet<TriangleTokenType> syncSet = null;
TypeSpec identifierType = TrianglePredefined.undefinedType;
Token identifierToken = null;
ICode routineICode = null;
token = currentToken();
switch ((TriangleTokenType) token.getType()) {
case CONST:
identifierToken = nextToken();
syncSet = EnumSet.of(TILDE);
syncSet.addAll(ExpressionParser.FIRST_FOLLOW_SET);
synchronize(IDENTIFIER, MISSING_IDENTIFIER, syncSet);
syncSet.remove(TILDE);
token = synchronize(TILDE, MISSING_TILDE, syncSet);
expression = new ExpressionParser(this);
ICodeNode expressionNode = expression.parse(token);
SymTabEntry constantId = symTabStack.lookupLocal(identifierToken.getText().toLowerCase());
// Enter the new identifier into the symbol table
// but don't set how it's defined yet.
if (constantId == null) {
constantId = symTabStack.enterLocal(identifierToken.getText().toLowerCase());
constantId.setDefinition(DefinitionImpl.CONSTANT);
constantId.setAttribute(CONSTANT_VALUE, expressionNode);
constantId.appendLineNumber(identifierToken.getLineNumber());
constantId.setTypeSpec(expressionNode.getTypeSpec());
} else {
errorHandler.flag(identifierToken, IDENTIFIER_REDEFINED, this);
}
break;
case VAR:
identifierToken = nextToken();
syncSet = EnumSet.of(COLON);
syncSet.addAll(TypeDenoterParser.FIRST_FOLLOW_SET);
synchronize(IDENTIFIER, MISSING_IDENTIFIER, syncSet);
syncSet.remove(COLON);
token = synchronize(COLON, MISSING_COLON, syncSet);
typeDenoter = new TypeDenoterParser(this);
identifierType = typeDenoter.parse(token);
SymTabEntry variableId = symTabStack.lookupLocal(identifierToken.getText().toLowerCase());
// Enter the new identifier into the symbol table
// but don't set how it's defined yet.
if (variableId == null) {
variableId = symTabStack.enterLocal(identifierToken.getText().toLowerCase());
variableId.setDefinition(DefinitionImpl.VARIABLE);
variableId.appendLineNumber(identifierToken.getLineNumber());
variableId.setTypeSpec(identifierType);
} else {
errorHandler.flag(identifierToken, IDENTIFIER_REDEFINED, this);
}
break;
case PROC:
identifierToken = nextToken();
routineICode = ICodeFactory.createICode();
syncSet = EnumSet.of(LEFT_PAREN);
syncSet.addAll(FormalParameterSequenceParser.FIRST_FOLLOW_SET);
token = synchronize(IDENTIFIER, MISSING_IDENTIFIER, syncSet);
syncSet.remove(LEFT_PAREN);
token = synchronize(LEFT_PAREN, MISSING_LEFT_PAREN, syncSet);
SymTabEntry procId = symTabStack.lookupLocal(identifierToken.getText().toLowerCase());
// Enter the new identifier into the symbol table
// but don't set how it's defined yet.
if (procId == null) {
procId = symTabStack.enterLocal(identifierToken.getText().toLowerCase());
procId.setTypeSpec(TrianglePredefined.undefinedType);
procId.setDefinition(DefinitionImpl.PROCEDURE);
procId.appendLineNumber(identifierToken.getLineNumber());
procId.setAttribute(ROUTINE_SYMTAB, symTabStack.push());
procId.setAttribute(ROUTINE_ICODE, routineICode);
} else {
errorHandler.flag(identifierToken, IDENTIFIER_REDEFINED, this);
procId = null;
}
formalParameterSequence = new FormalParameterSequenceParser(this);
ArrayList<SymTabEntry> procParamList = formalParameterSequence.parse(token);
syncSet = EnumSet.of(TILDE);
syncSet.addAll(SingleCommandParser.FIRST_FOLLOW_SET);
token = synchronize(RIGHT_PAREN, MISSING_RIGHT_PAREN, syncSet);
syncSet.remove(TILDE);
token = synchronize(TILDE, MISSING_TILDE, syncSet);
singleCommand = new SingleCommandParser(this);
routineICode.setRoot(singleCommand.parse(token));
if (procId != null) {
procId.setAttribute(ROUTINE_PARMS, procParamList);
}
symTabStack.pop();
break;
case FUNC:
identifierToken = nextToken();
routineICode = ICodeFactory.createICode();
SymTabEntry funcId = symTabStack.lookupLocal(identifierToken.getText().toLowerCase());
// Enter the new identifier into the symbol table
// but don't set how it's defined yet.
if (funcId == null) {
funcId = symTabStack.enterLocal(identifierToken.getText().toLowerCase());
funcId.setDefinition(DefinitionImpl.FUNCTION);
funcId.appendLineNumber(identifierToken.getLineNumber());
funcId.setTypeSpec(TrianglePredefined.undefinedType);
funcId.setAttribute(ROUTINE_SYMTAB, symTabStack.push());
funcId.setAttribute(ROUTINE_ICODE, routineICode);
} else {
errorHandler.flag(identifierToken, IDENTIFIER_REDEFINED, this);
funcId = null;
}
syncSet = EnumSet.of(LEFT_PAREN);
syncSet.addAll(FormalParameterSequenceParser.FIRST_FOLLOW_SET);
token = synchronize(IDENTIFIER, MISSING_IDENTIFIER, syncSet);
syncSet.remove(LEFT_PAREN);
token = synchronize(LEFT_PAREN, MISSING_LEFT_PAREN, syncSet);
formalParameterSequence = new FormalParameterSequenceParser(this);
ArrayList<SymTabEntry> funcParamList = formalParameterSequence.parse(token);
syncSet = EnumSet.of(COLON);
syncSet.addAll(TypeDenoterParser.FIRST_FOLLOW_SET);
synchronize(RIGHT_PAREN, MISSING_RIGHT_PAREN, syncSet);
syncSet.remove(COLON);
token = synchronize(COLON, MISSING_COLON, syncSet);
Token typeToken = currentToken();
typeDenoter = new TypeDenoterParser(this);
TypeSpec funcType = typeDenoter.parse(token);
syncSet = ExpressionParser.FIRST_FOLLOW_SET.clone();
token = synchronize(TILDE, MISSING_TILDE, syncSet);
expression = new ExpressionParser(this);
routineICode.setRoot(expression.parse(token));
if (funcId != null) {
funcId.setTypeSpec(funcType);
funcId.setAttribute(ROUTINE_PARMS, funcParamList);
if (!routineICode.getRoot().getTypeSpec().equals(funcType)) {
errorHandler.flag(typeToken, RETURN_TYPE_MISMATCH, this);
}
}
symTabStack.pop();
break;
case TYPE:
identifierToken = nextToken();
syncSet = EnumSet.of(TILDE);
syncSet.addAll(TypeDenoterParser.FIRST_FOLLOW_SET);
synchronize(IDENTIFIER, MISSING_IDENTIFIER, syncSet);
syncSet.remove(TILDE);
token = synchronize(TILDE, MISSING_TILDE, syncSet);
typeDenoter = new TypeDenoterParser(this);
TypeSpec typeType = typeDenoter.parse(token);
SymTabEntry typeId = symTabStack.lookupLocal(identifierToken.getText().toLowerCase());
// Enter the new identifier into the symbol table
// but don't set how it's defined yet.
if (typeId == null) {
typeId = symTabStack.enterLocal(identifierToken.getText().toLowerCase());
typeId.setDefinition(DefinitionImpl.TYPE);
typeId.appendLineNumber(identifierToken.getLineNumber());
typeId.setTypeSpec(typeType);
} else {
errorHandler.flag(token, IDENTIFIER_REDEFINED, this);
}
break;
default:
errorHandler.flag(token, MISSING_DECLARATION, this);
break;
}
}
/**
* Parse a factor.
*
* @param token the initial token.
* @return the root of the generated parse subtree.
* @throws Exception if an error occurred.
*/
private ICodeNode parseFactor(Token token) throws Exception {
TokenType tokenType = token.getType();
ICodeNode rootNode = null;
switch ((PascalTokenType) tokenType) {
case IDENTIFIER:
{
// Look up the identifier in the symbol table stack.
// Flag the identifier as undefined if it's not found.
String name = token.getText().toLowerCase();
SymTabEntry id = symTabStack.lookup(name);
if (id == null) {
errorHandler.flag(token, IDENTIFIER_UNDEFINED, this);
id = symTabStack.enterLocal(name);
}
rootNode = ICodeFactory.createICodeNode(VARIABLE);
rootNode.setAttribute(ID, id);
id.appendLineNumber(token.getLineNumber());
token = nextToken(); // consume the identifier
break;
}
case INTEGER:
{
// Create an INTEGER_CONSTANT node as the root node.
rootNode = ICodeFactory.createICodeNode(INTEGER_CONSTANT);
rootNode.setAttribute(VALUE, token.getValue());
token = nextToken(); // consume the number
break;
}
case REAL:
{
// Create an REAL_CONSTANT node as the root node.
rootNode = ICodeFactory.createICodeNode(REAL_CONSTANT);
rootNode.setAttribute(VALUE, token.getValue());
token = nextToken(); // consume the number
break;
}
case STRING:
{
String value = (String) token.getValue();
// Create a STRING_CONSTANT node as the root node.
rootNode = ICodeFactory.createICodeNode(STRING_CONSTANT);
rootNode.setAttribute(VALUE, value);
token = nextToken(); // consume the string
break;
}
case NOT:
{
token = nextToken(); // consume the NOT
// Create a NOT node as the root node.
rootNode = ICodeFactory.createICodeNode(ICodeNodeTypeImpl.NOT);
// Parse the factor. The NOT node adopts the
// factor node as its child.
rootNode.addChild(parseFactor(token));
break;
}
case LEFT_PAREN:
{
token = nextToken(); // consume the (
// Parse an expression and make its node the root node.
rootNode = parseExpression(token);
// Look for the matching ) token.
token = currentToken();
if (token.getType() == RIGHT_PAREN) {
token = nextToken(); // consume the )
} else {
errorHandler.flag(token, MISSING_RIGHT_PAREN, this);
}
break;
}
case LEFT_BRACKET:
{
rootNode = parseSet(token);
break;
}
default:
{
errorHandler.flag(token, UNEXPECTED_TOKEN, this);
break;
}
}
return rootNode;
}
/**
* Parse type definitions.
*
* @param token the initial token.
* @throws Exception if an error occurred.
*/
public void parse(Token token) throws Exception {
token = synchronize(IDENTIFIER_SET);
// Loop to parse a sequence of type definitions
// separated by semicolons.
while (token.getType() == IDENTIFIER) {
String name = token.getText().toLowerCase();
SymTabEntry typeId = symTabStack.lookupLocal(name);
// Enter the new identifier into the symbol table
// but don't set how it's defined yet.
if (typeId == null) {
typeId = symTabStack.enterLocal(name);
typeId.appendLineNumber(token.getLineNumber());
} else {
errorHandler.flag(token, IDENTIFIER_REDEFINED, this);
typeId = null;
}
token = nextToken(); // consume the identifier token
// Synchronize on the = token.
token = synchronize(EQUALS_SET);
if (token.getType() == EQUALS) {
token = nextToken(); // consume the =
} else {
errorHandler.flag(token, MISSING_EQUALS, this);
}
// Parse the type specification.
TypeSpecificationParser typeSpecificationParser = new TypeSpecificationParser(this);
TypeSpec type = typeSpecificationParser.parse(token);
// Set identifier to be a type and set its type specificationt.
if (typeId != null) {
typeId.setDefinition(TYPE);
}
// Cross-link the type identifier and the type specification.
if (type != null && typeId != null) {
if (type.getIdentifier() == null) {
type.setIdentifier(typeId);
}
typeId.setTypeSpec(type);
} else {
token = synchronize(FOLLOW_SET);
}
token = currentToken();
TokenType tokenType = token.getType();
// Look for one or more semicolons after a definition.
if (tokenType == SEMICOLON) {
while (token.getType() == SEMICOLON) {
token = nextToken(); // consume the ;
}
}
// If at the start of the next definition or declaration,
// then missing a semicolon.
else if (NEXT_START_SET.contains(tokenType)) {
errorHandler.flag(token, MISSING_SEMICOLON, this);
}
token = synchronize(IDENTIFIER_SET);
}
}
public SymTabEntry parse(Token token, SymTabEntry parentId) throws Exception {
Token FunctionReturnToken = token;
Definition routineDefn = null;
String dummyName = null;
SymTabEntry routineId = null;
TokenType routineType = token.getType();
// Initialize.
switch ((WookieTokenType) routineType) {
case INT:
{
// save the int data type for use later
FunctionReturnToken = token;
token = nextToken(); // consume Int
routineDefn = DefinitionImpl.FUNCTION;
dummyName = "DummyProgramName".toLowerCase() + String.format("%03d", ++dummyCounter);
break;
}
case CHAR:
{
// save the char data type for use later
FunctionReturnToken = token;
token = nextToken(); // consume Char
routineDefn = DefinitionImpl.FUNCTION;
dummyName = "DummyProgramName".toLowerCase() + String.format("%03d", ++dummyCounter);
break;
}
case VOID:
{
// save the void data type for use later
FunctionReturnToken = token;
token = nextToken(); // consume PROGRAM
routineDefn = DefinitionImpl.PROCEDURE;
dummyName = "DummyProgramName".toLowerCase() + String.format("%03d", ++dummyCounter);
break;
}
default:
{
routineDefn = DefinitionImpl.PROGRAM;
dummyName = "DummyProgramName".toLowerCase();
break;
}
}
// Parse the routine name.
routineId = parseRoutineName(token, dummyName);
routineId.setDefinition(routineDefn);
token = currentToken();
// Create new intermediate code for the routine.
ICode iCode = ICodeFactory.createICode();
routineId.setAttribute(ROUTINE_ICODE, iCode);
routineId.setAttribute(ROUTINE_ROUTINES, new ArrayList<SymTabEntry>());
SymTab symTab = symTabStack.push();
routineId.setAttribute(ROUTINE_SYMTAB, symTab);
((SymTabImpl) symTab).setfuncName(routineId.getName());
parseFormalParameters(token, routineId);
token = currentToken();
if (FunctionReturnToken.getType() == INT) {
routineId.setTypeSpec(integerType);
} else if (FunctionReturnToken.getType() == CHAR) {
routineId.setTypeSpec(charType);
}
// Parse the routine's block declaration.
if (token.getType() == LEFT_BRACE) {
routineId.setAttribute(ROUTINE_CODE, DECLARED);
BlockParser blockParser = new BlockParser(this);
ICodeNode rootNode = blockParser.parse(token, routineId);
iCode.setRoot(rootNode);
}
ArrayList<SymTabEntry> subroutines =
(ArrayList<SymTabEntry>) parentId.getAttribute(ROUTINE_ROUTINES);
subroutines.add(routineId);
// Pop the routine's symbol table off the stack.
symTabStack.pop();
return routineId;
}
private ICodeNode parseIdentifier(Token token) throws Exception {
ICodeNode rootNode = null;
// Look up the identifier in the symbol table stack.
// Flag the identifier as undefined if it's not found.
String name = token.getText().toLowerCase();
SymTabEntry id = symTabStack.lookup(name);
// Undefined.
if (id == null) {
errorHandler.flag(token, IDENTIFIER_UNDEFINED, this);
id = symTabStack.enterLocal(name);
id.setDefinition(UNDEFINED);
id.setTypeSpec(Predefined.undefinedType);
}
Definition defnCode = id.getDefinition();
switch ((DefinitionImpl) defnCode) {
case CONSTANT:
{
Object value = id.getAttribute(CONSTANT_VALUE);
TypeSpec type = id.getTypeSpec();
if (value instanceof Integer) {
rootNode = ICodeFactory.createICodeNode(INTEGER_CONSTANT);
rootNode.setAttribute(VALUE, value);
} else if (value instanceof Float) {
rootNode = ICodeFactory.createICodeNode(REAL_CONSTANT);
rootNode.setAttribute(VALUE, value);
} else if (value instanceof String) {
rootNode = ICodeFactory.createICodeNode(STRING_CONSTANT);
rootNode.setAttribute(VALUE, value);
}
id.appendLineNumber(token.getLineNumber());
token = nextToken(); // consume the constant identifier
if (rootNode != null) {
rootNode.setTypeSpec(type);
}
break;
}
case ENUMERATION_CONSTANT:
{
Object value = id.getAttribute(CONSTANT_VALUE);
TypeSpec type = id.getTypeSpec();
rootNode = ICodeFactory.createICodeNode(INTEGER_CONSTANT);
rootNode.setAttribute(VALUE, value);
id.appendLineNumber(token.getLineNumber());
token = nextToken(); // consume the enum constant identifier
rootNode.setTypeSpec(type);
break;
}
case FUNCTION:
{
CallParser callParser = new CallParser(this);
rootNode = callParser.parse(token);
break;
}
default:
{
VariableParser variableParser = new VariableParser(this);
rootNode = variableParser.parse(token, id);
break;
}
}
return rootNode;
}
