const CType* CAstStringConstant::GetType(void) const
{
CTypeManager *tm = CTypeManager::Get();
// string is array of character.
return tm->GetArray(CToken::unescape(GetValueStr()).length()+1, tm->GetChar());
}
CAstStringConstant::CAstStringConstant(CToken t, const string value,
CAstScope *s)
: CAstOperand(t)
{
CTypeManager *tm = CTypeManager::Get();
_type = tm->GetArray(strlen(CScanner::unescape(value).c_str())+1,
tm->GetChar());
_value = new CDataInitString(value);
ostringstream o;
o << "_str_" << ++_idx;
_sym = new CSymGlobal(o.str(), _type);
_sym->SetData(_value);
s->GetSymbolTable()->AddSymbol(_sym);
}
const CType* CAstSpecialOp::GetType(void) const
{
CTypeManager *tm = CTypeManager::Get();
EOperation oper = GetOperation();
if(GetOperand()->GetType() == NULL)
return NULL;
// case '&'
if(oper == opAddress) {
return tm->GetPointer(GetOperand()->GetType());
}
// we don't implement '*' , '(cast)'.
else if(oper == opDeref) {
return NULL;
}
else {
return NULL;
}
}
const CType* CAstBinaryOp::GetType(void) const
{
CTypeManager *tm = CTypeManager::Get();
EOperation oper = GetOperation();
// If lhs or rhs is null, then null.
if(_left->GetType() == NULL || _right->GetType() == NULL)
return NULL;
// case '+', '-', '*', '/'
if(oper == opAdd || oper == opSub || oper == opMul || oper == opDiv) {
if(_left->GetType()->IsInt() && _right->GetType()->IsInt())
return tm->GetInt();
else return NULL;
}
// case '&&' , '||'
else if(oper == opAnd || oper == opOr) {
if(_left->GetType()->IsBoolean() && _right->GetType()->IsBoolean())
return tm->GetBool();
else return NULL;
}
// case '=', '#'
else if(oper == opEqual || oper == opNotEqual) {
if(_left->GetType()->IsBoolean() && _right->GetType()->IsBoolean() ||
_left->GetType()->IsChar() && _right->GetType()->IsChar() ||
_left->GetType()->IsInt() && _right->GetType()->IsInt())
return tm->GetBool();
else return NULL;
}
// case '>', '>=', '<', '<='
else {
if(_left->GetType()->IsInt() && _right->GetType()->IsInt() ||
_left->GetType()->IsChar() && _right->GetType()->IsChar())
return tm->GetBool();
else return NULL;
}
}
const CType* CAstUnaryOp::GetType(void) const
{
CTypeManager *tm = CTypeManager::Get();
EOperation oper = GetOperation();
CAstExpression *e = GetOperand();
if(e->GetType() == NULL)
return NULL;
if(oper == opNeg || oper == opPos) { //case '+' || '-'
if(e->GetType()->IsInt())
return tm->GetInt();
else return NULL;
}
else { // case '!'
if(e->GetType()->IsBoolean())
return tm->GetBool();
else return NULL;
}
}
void CParser::InitSymbolTable(CSymtab *s)
{
CTypeManager *tm = CTypeManager::Get();
CSymProc *fun;
// function DIM(array: pointer to array; dim: integer): integer
fun = new CSymProc("DIM", tm->GetInt());
fun->AddParam(new CSymParam(0, "arr", tm->GetPointer(tm->GetNull())));
fun->AddParam(new CSymParam(1, "dim", tm->GetInt()));
s->AddSymbol(fun);
// function DOFS(array: pointer to array): integer;
fun = new CSymProc("DOFS", tm->GetInt());
fun->AddParam(new CSymParam(0, "arr", tm->GetPointer(tm->GetNull())));
s->AddSymbol(fun);
// function ReadInt() : integer;
fun = new CSymProc("ReadInt", tm->GetInt());
s->AddSymbol(fun);
// procedure WriteInt(i: integer);
fun = new CSymProc("WriteInt", tm->GetNull());
fun->AddParam(new CSymParam(0, "i", tm->GetInt()));
s->AddSymbol(fun);
// procedure WriteChar(c: char);
fun = new CSymProc("WriteChar", tm->GetNull());
fun->AddParam(new CSymParam(0, "c", tm->GetChar()));
s->AddSymbol(fun);
// procedure WriteStr(string: char[]);
fun = new CSymProc("WriteStr", tm->GetNull());
fun->AddParam(new CSymParam(0, "str", tm->GetPointer(tm->GetArray(CArrayType::OPEN, tm->GetChar()))));
s->AddSymbol(fun);
// procedure WriteLn();
fun = new CSymProc("WriteLn", tm->GetNull());
s->AddSymbol(fun);
}
const CType* CParser::type(bool isParam, bool isFunction) {
//
// type ::= basetype | type "[" [ number ] "]".
// basetype ::= "boolean" | "char" | "integer".
//
CToken t;
CToken tk;
EToken tt = _scanner->Peek().GetType();
CTypeManager *tm = CTypeManager::Get();
const CType *type;
vector<int> nelems;
// basetype cases.
switch(tt) {
case tBoolean:
Consume(tBoolean, &t);
tk = t;
type = dynamic_cast<const CType *>(tm->GetBool());
break;
case tChar:
Consume(tChar, &t);
tk = t;
type = dynamic_cast<const CType *>(tm->GetChar());
break;
case tInteger:
Consume(tInteger, &t);
tk = t;
type = dynamic_cast<const CType *>(tm->GetInt());
break;
default:
SetError(_scanner->Peek(), "basetype expected.");
break;
}
assert(type != NULL);
// array case.
while(_scanner->Peek().GetType() == tLSBrak) {
Consume(tLSBrak);
CToken nelem;
if(_scanner->Peek().GetType() == tNumber) {
Consume(tNumber, &nelem);
nelems.push_back(atoi(nelem.GetValue().c_str()));
}
else {
// open arrays are allowed only in parameter.
if(isParam)
nelems.push_back(-1);
else
SetError(_scanner->Peek(), "expected 'tNumber', got '" + _scanner->Peek().GetName() + "'");
}
Consume(tRSBrak);
}
for(int i = nelems.size()-1; i >= 0; i--) {
type = dynamic_cast<const CType *>(tm->GetArray(nelems[i], type));
assert(type != NULL);
nelems.pop_back();
// Arrays are addressed in parameter.
if(i == 0 && isParam)
type = dynamic_cast<const CType *>(tm->GetPointer(type));
}
if(isFunction && !type->IsScalar()) {
SetError(tk, "invalid composite type for function.");
}
return type;
}
void CParser::subroutineDecl(CAstScope *s) {
//
// subroutineDecl ::= (procedureDecl | functionDecl) subroutineBody ident ";".
// procedureDecl ::= "procedure" ident [ formalParam ] ";".
// functionDecl ::= "function" ident [ formalParam ] ":" type ";".
// formalParam ::= "(" [ varDeclSequence ] ")".
// subroutineBody ::= varDeclaration "begin" statSequence "end".
// varDeclSequence ::= varDecl { ";" varDecl }.
// varDecl ::= ident { "," ident } ":" type.
//
CToken subroutine_name;
CToken subroutine_name_check;
CTypeManager *tm = CTypeManager::Get();
// procedureDecl
if(_scanner->Peek().GetType() == tProcedure) {
CToken t;
CAstStatement *statseq = NULL;
Consume(tProcedure);
Consume(tIdent, &subroutine_name);
// Check duplicate procedure declaration.
if(dynamic_cast<CAstModule *>(s)) {
if(s->GetSymbolTable()->FindSymbol(subroutine_name.GetValue(), sGlobal) != NULL)
SetError(subroutine_name, "duplicate procedure/function declaration '" + subroutine_name.GetValue() + "'.");
}
else {
if(s->GetSymbolTable()->FindSymbol(subroutine_name.GetValue(), sLocal) != NULL)
SetError(subroutine_name, "duplicate procedure/function declaration '" + subroutine_name.GetValue() + "'.");
}
CSymProc *symproc = new CSymProc(subroutine_name.GetValue(), tm->GetNull());
CAstProcedure *procedure = new CAstProcedure(subroutine_name, subroutine_name.GetValue(), s, symproc);
// formalParam
if(_scanner->Peek().GetType() == tLBrak) {
Consume(tLBrak);
if(_scanner->Peek().GetType() == tIdent) {
varDeclParam(procedure, symproc, 0);
while(1) {
if(_scanner->Peek().GetType() == tSemicolon) {
Consume(tSemicolon);
varDeclParam(procedure, symproc, symproc->GetNParams());
}
else break;
}
}
Consume(tRBrak);
}
s->GetSymbolTable()->AddSymbol(symproc);
Consume(tSemicolon);
// subroutineBody from here.
// subroutineBody ::= varDeclaration "begin" statSequence "end".
// varDeclaration
if(_scanner->Peek().GetType() == tVar) {
Consume(tVar);
varDecl(procedure);
Consume(tSemicolon);
while(1) {
if(_scanner->Peek().GetType() == tIdent) {
varDecl(procedure);
Consume(tSemicolon);
}
else break;
}
}
Consume(tBegin);
// statSequence
statseq = statSequence(procedure);
procedure->SetStatementSequence(statseq);
Consume(tEnd);
Consume(tIdent, &subroutine_name_check);
// Check procedure names are equal.
if(subroutine_name.GetValue() != subroutine_name_check.GetValue()) {
SetError(subroutine_name_check, "procedure/function identifier mismatch ('" + subroutine_name.GetValue() + "' != '" + subroutine_name_check.GetValue() + "').");
}
Consume(tSemicolon);
}
// functionDecl
else {
CToken tmp;
// Parameters are pushed into params. Then they will be added into symproc later.
vector<CSymParam *> params;
const CType *func_type;
CAstStatement *statseq = NULL;
Consume(tFunction);
Consume(tIdent, &subroutine_name);
// Check duplicate function declaration.
if(dynamic_cast<CAstModule *>(s)) {
if(s->GetSymbolTable()->FindSymbol(subroutine_name.GetValue(), sGlobal) != NULL)
SetError(subroutine_name, "duplicate procedure/function declaration '" + subroutine_name.GetValue() + "'.");
}
else {
if(s->GetSymbolTable()->FindSymbol(subroutine_name.GetValue(), sLocal) != NULL)
SetError(subroutine_name, "duplicate procedure/function declaration '" + subroutine_name.GetValue() + "'.");
}
// formalParam
if(_scanner->Peek().GetType() == tLBrak) {
int idx = 0;
CToken t;
vector<string> vars;
const CType *var_type;
Consume(tLBrak);
// Get parameters.
if(_scanner->Peek().GetType() == tIdent) {
Consume(tIdent, &t);
vars.push_back(t.GetValue());
while(_scanner->Peek().GetType() == tComma) {
Consume(tComma);
Consume(tIdent, &t);
// Check duplicate parameter declaration.
for(int i = 0; i < vars.size(); i++) {
if(vars[i] == t.GetValue()) {
SetError(t, "duplicate variable declaration '" + t.GetValue() + "'.");
break;
}
}
vars.push_back(t.GetValue());
}
Consume(tColon);
var_type = type(true);
for(int i = vars.size() - 1; i >= 0; i--) {
CSymParam *param = new CSymParam(i, vars[i], var_type);
params.push_back(param);
idx++;
}
// for multiple parameter declaration.
while(1) {
if(_scanner->Peek().GetType() == tSemicolon) {
Consume(tSemicolon);
CToken t_loop;
vector<string> vars_loop;
const CType *var_type_loop;
Consume(tIdent, &t_loop);
// Check duplicate parameter declaration.
for(int i = 0; i < params.size(); i++) {
if(params[i]->GetName() == t_loop.GetValue()) {
SetError(t_loop, "duplicate variable declaration '" + t_loop.GetValue() + "'.");
break;
}
}
vars_loop.push_back(t_loop.GetValue());
while(_scanner->Peek().GetType() == tComma) {
Consume(tComma);
Consume(tIdent, &t_loop);
for(int i = 0; i < params.size(); i++) {
if(params[i]->GetName() == t_loop.GetValue()) {
SetError(t_loop, "duplicate variable declaration '" + t_loop.GetValue() + "'.");
break;
}
}
// Check duplicate parameter declaration.
for(int i = 0; i < vars_loop.size(); i++) {
if(vars_loop[i] == t_loop.GetValue()) {
SetError(t_loop, "duplicate variable declaration '" + t_loop.GetValue() + "'.");
break;
}
}
vars_loop.push_back(t_loop.GetValue());
}
Consume(tColon);
var_type_loop = type(true);
int tmp_idx = idx;
// Add parameters into vector 'params'.
for(int i = vars_loop.size() - 1; i >= 0; i--) {
CSymParam *param_loop = new CSymParam(i+tmp_idx, vars_loop[i], var_type_loop);
params.push_back(param_loop);
vars_loop.pop_back();
idx++;
}
}
else break;
}
}
Consume(tRBrak);
}
Consume(tColon);
func_type = type(false, true);
Consume(tSemicolon);
CSymProc *symproc = new CSymProc(subroutine_name.GetValue(), func_type);
CAstProcedure *function = new CAstProcedure(subroutine_name, subroutine_name.GetValue(), s, symproc);
assert(symproc != NULL);
// Add parameters into symproc.
for(int i = 0; i < params.size(); i++) {
CSymParam *param = params[i];
symproc->AddParam(param);
function->GetSymbolTable()->AddSymbol(param);
}
while(!params.empty())
params.pop_back();
s->GetSymbolTable()->AddSymbol(symproc);
// subroutineBody from here.
// subroutineBody ::= varDeclaration "begin" statSequence "end".
// varDeclaration
if(_scanner->Peek().GetType() == tVar) {
Consume(tVar);
varDecl(function);
Consume(tSemicolon);
while(1) {
if(_scanner->Peek().GetType() == tIdent) {
varDecl(function);
Consume(tSemicolon);
}
else break;
}
}
Consume(tBegin);
// statSequence
statseq = statSequence(function);
function->SetStatementSequence(statseq);
Consume(tEnd);
Consume(tIdent, &subroutine_name_check);
// Check function names are equal.
if(subroutine_name.GetValue() != subroutine_name_check.GetValue()) {
SetError(subroutine_name_check, "procedure/function identifier mismatch ('" + subroutine_name.GetValue() + "' != '" + subroutine_name_check.GetValue() + "').");
}
Consume(tSemicolon);
}
}
CAstExpression* CParser::factor(CAstScope *s) {
//
// factor ::= qualident | number | boolean | char | string | "(" expression ")" | subroutineCall | "!" factor.
// FIRST(factor) = { tIdent, tNumber, tTrue, tFalse, tCharacter, tString, tLBrak, tEMark }.
//
CToken t;
EToken tt = _scanner->Peek().GetType();
CTypeManager *tm = CTypeManager::Get();
CAstExpression *n = NULL;
CSymtab *symtab = s->GetSymbolTable();
switch(tt) {
case tTrue:
Consume(tTrue, &t);
n = new CAstConstant(t, tm->GetBool(), 1);
break;
case tFalse:
Consume(tFalse, &t);
n = new CAstConstant(t, tm->GetBool(), 0);
break;
case tNumber:
{
Consume(tNumber, &t);
errno = 0;
long long v = strtoll(t.GetValue().c_str(), NULL, 10);
if (errno != 0) SetError(t, "invalid number.");
// integer range validation check.
if(v > 2147483648)
SetError(t, "integer constant outside valid range.");
n = new CAstConstant(t, tm->GetInt(), v);
break;
}
case tIdent:
{
Consume(tIdent, &t);
if(_scanner->Peek().GetType() == tLBrak) { // subroutineCall
Consume(tLBrak);
CAstFunctionCall *f;
// If subroutineCall calls undefined procedure/function, then set error.
if(symtab->FindSymbol(t.GetValue(),sLocal) == NULL && symtab->FindSymbol(t.GetValue(), sGlobal) != NULL) {
// Check if symbol is procedure.
if(dynamic_cast<const CSymProc *>(symtab->FindSymbol(t.GetValue(), sGlobal)) == NULL)
SetError(t, "invalid procedure/function identifier.");
f = new CAstFunctionCall(t, dynamic_cast<const CSymProc *>(symtab->FindSymbol(t.GetValue(), sGlobal)));
}
else if(symtab->FindSymbol(t.GetValue(), sLocal) != NULL) {
// Check if symbol is procedure.
if(dynamic_cast<const CSymProc *>(symtab->FindSymbol(t.GetValue(), sLocal)) == NULL)
SetError(t, "invalid procedure/function identifier.");
f = new CAstFunctionCall(t, dynamic_cast<const CSymProc *>(symtab->FindSymbol(t.GetValue(), sLocal)));
}
else
SetError(t, "undefined identifier.");
assert(f != NULL);
if(isExpr(_scanner->Peek())) {
CAstExpression *expr = expression(s);
assert(expr != NULL);
// If array, then addressed.
if(expr->GetType()->IsArray())
expr = new CAstSpecialOp(expr->GetToken(), opAddress, expr, NULL);
f->AddArg(expr);
while(_scanner->Peek().GetType() == tComma) {
Consume(tComma);
expr = expression(s);
assert(expr != NULL);
// If array, then addressed.
if(expr->GetType()->IsArray())
expr = new CAstSpecialOp(expr->GetToken(), opAddress, expr, NULL);
f->AddArg(expr);
}
}
Consume(tRBrak);
n = f;
}
else { // qualident
if(_scanner->Peek().GetType() == tLSBrak) { // It means array
CAstArrayDesignator *f;
if(symtab->FindSymbol(t.GetValue(), sLocal) == NULL && symtab->FindSymbol(t.GetValue(), sGlobal) != NULL) {
// Check if symbol is procedure.
if(dynamic_cast<const CSymProc *>(symtab->FindSymbol(t.GetValue(), sGlobal)) != NULL)
SetError(t, "designator expected.");
f = new CAstArrayDesignator(t, symtab->FindSymbol(t.GetValue(), sGlobal));
}
else if(symtab->FindSymbol(t.GetValue(), sLocal) != NULL) {
// Check if symbol is procedure.
if(dynamic_cast<const CSymProc *>(symtab->FindSymbol(t.GetValue(), sLocal)) != NULL)
SetError(t, "designator expected");
f = new CAstArrayDesignator(t, symtab->FindSymbol(t.GetValue(), sLocal));
}
else {
SetError(t, "undefined identifier.");
}
while(_scanner->Peek().GetType() == tLSBrak) {
Consume(tLSBrak);
CAstExpression *expr = expression(s);
assert(expr != NULL);
f->AddIndex(expr);
Consume(tRSBrak);
}
f->IndicesComplete();
n = f;
} // non-array case.
else {
if(symtab->FindSymbol(t.GetValue(), sLocal) == NULL && symtab->FindSymbol(t.GetValue(), sGlobal) != NULL) {
// Check if symbol is procedure.
if(dynamic_cast<const CSymProc *>(symtab->FindSymbol(t.GetValue(), sGlobal)) != NULL) {
SetError(t, "designator expected.");
}
n = new CAstDesignator(t, symtab->FindSymbol(t.GetValue(), sGlobal));
}
else if(symtab->FindSymbol(t.GetValue(), sLocal) != NULL) {
// Check if symbol is procedure.
if(dynamic_cast<const CSymProc *>(symtab->FindSymbol(t.GetValue(), sLocal)) != NULL) {
SetError(t, "designator expected.");
}
n = new CAstDesignator(t, symtab->FindSymbol(t.GetValue(), sLocal));
}
else
SetError(t, "undefined identifier.");
}
}
break;
}
case tString:
Consume(tString, &t);
n = new CAstStringConstant(t, t.GetValue(), s);
break;
case tCharacter:
{
Consume(tCharacter, &t);
char res;
// If character length is 0, it means character is '\0' because invalid character cases are handled in scanner.
if(t.GetValue().length() == 0)
res = '\0';
else if(t.GetValue().length() == 1) {
if(t.GetValue().at(0) != '\\')
res = t.GetValue().at(0);
else SetError(t, "wrong character");
}
// unescape.
else if(t.GetValue().length() == 2) {
if(t.GetValue().at(0) == '\\') {
if(t.GetValue().at(1) == 'n') {
res = '\n';
}
else if(t.GetValue().at(1) == 't') {
res = '\t';
}
else if(t.GetValue().at(1) == '"') {
res = '"';
}
else if(t.GetValue().at(1) == '\'') {
res = '\'';
}
else if(t.GetValue().at(1) == '\\') {
res = '\\';
}
else if(t.GetValue().at(1) == '0') {
res = '\0';
}
else
SetError(t, "wrong character");
}
}
else SetError(t, "wrong character");
n = new CAstConstant(t, tm->GetChar(), (long long)res);
break;
}
case tLBrak:
Consume(tLBrak);
n = expression(s);
Consume(tRBrak);
break;
case tEMark:
Consume(tEMark, &t);
n = new CAstUnaryOp(t, opNot, factor(s));
break;
default:
// cout << "got " << _scanner->Peek() << endl;
SetError(_scanner->Peek(), "factor expected.");
break;
}
return n;
}
void CParser::InitSymbolTable(CSymtab *s)
{
CTypeManager *tm = CTypeManager::Get();
// predefined functions:
// - DIM(a: ptr to array; dim: integer): integer
// - DOFS(a: ptr to array): integer
// - ReadInt(): integer
// - WriteInt(i:integer): void
// - WriteChar(c:char): void
// - WriteStr(str: char[]): void
// - WriteLn():void
// - DIM(a: ptr to array; dim: integer): integer
CSymProc *DIM = new CSymProc("DIM", tm->GetInt());
CSymParam *DIM_a = new CSymParam(0, "a", tm->GetPointer(tm->GetNull()));
CSymParam *DIM_dim = new CSymParam(1, "dim", tm->GetInt());
DIM->AddParam(DIM_a);
DIM->AddParam(DIM_dim);
s->AddSymbol(DIM);
// - DOFS(a: ptr to array): integer
CSymProc *DOFS = new CSymProc("DOFS", tm->GetInt());
CSymParam *DOFS_a = new CSymParam(0, "a", tm->GetPointer(tm->GetNull()));
DOFS->AddParam(DOFS_a);
s->AddSymbol(DOFS);
// - ReadInt(): integer
CSymProc *ReadInt = new CSymProc("ReadInt", tm->GetInt());
s->AddSymbol(ReadInt);
// - WriteInt(i:integer): void
CSymProc *WriteInt = new CSymProc("WriteInt", tm->GetNull());
CSymParam *WriteInt_i = new CSymParam(0, "i", tm->GetInt());
WriteInt->AddParam(WriteInt_i);
s->AddSymbol(WriteInt);
// - WriteChar(c:char): void
CSymProc *WriteChar = new CSymProc("WriteChar", tm->GetNull());
CSymParam *WriteChar_c = new CSymParam(0, "c", tm->GetChar());
WriteChar->AddParam(WriteChar_c);
s->AddSymbol(WriteChar);
// - WriteStr(str: char[]): void
CSymProc *WriteStr = new CSymProc("WriteStr", tm->GetNull());
CSymParam *WriteStr_str = new CSymParam(0, "str", tm->GetPointer(tm->GetArray(CArrayType::OPEN, tm->GetChar())));
WriteStr->AddParam(WriteStr_str);
s->AddSymbol(WriteStr);
// - WriteLn():void
CSymProc *WriteLn = new CSymProc("WriteLn", tm->GetNull());
s->AddSymbol(WriteLn);
}