def p_postfix_expression_4(p):
# struct ref
'''
postfix_expression : postfix_expression ARROW IDENTIFIER
'''
if p[1].type == "error":
p[0] = Node(type="error")
return
if p[1].type.class_type != "PointerType" or p[
1].type.type.class_type != "StructType":
p[0] = Node(type="error")
Errors(errorType='TypeError',
errorText='not pointer of struct type',
token_object=p.slice[-1])
return
# arg_dict = p[1].type.type.arg_dict
# success = arg_dict.get(p[3])
struct_sym = p[1].type.type.symbol_table
success = struct_sym._look_up(p[3],
token_object=p.slice[3],
in_struct=True,
no_error=True)
if success == None:
Errors(errorType='DeclarationError',
errorText='variable ' + p[3] + ' not declared in struct',
token_object=p.slice[-1])
p[0] = Node(type="error")
return
p[3] = Node(name="id", value=p[3])
p[0] = Node(name="struct ref",
value=p[2],
type=success.type,
children=[p[1], p[3]])
def type_check_relational(node1, node2, op, token, is_bool=False):
allowed_class = {'BasicType'}
allowed_base = {'int', 'long', 'char', 'double', 'float'}
if node1.type == "error" or node2.type == "error":
return Node(type="error")
if node1.type.class_type not in allowed_class or node2.type.class_type not in allowed_class:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
if is_bool:
if node1.type.type == "bool" and node2.type.type == "bool":
return Node(name="binary_op",
value="bool" + op,
children=[node1, node2],
type=BasicType('bool'))
if node1.type.type not in allowed_base or node2.type.type not in allowed_base:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
node1, node2, typ = implicit_casting(node1, node2)
return Node(name="binary_op",
value=typ.stype + op,
children=[node1, node2],
type=BasicType('bool'))
def p_conditional_expression(p):
'''
conditional_expression : logical_or_expression
| logical_or_expression QUES_MARK expression COLON conditional_expression
'''
if len(p) == 2:
p[0] = p[1]
else:
if p[1].type == "error" or p[3].type == "error" or p[5].type == "error":
p[0] = Node(type="error")
return
allowed_class = {'BasicType'}
allowed_base = {'int', 'long', 'char', 'bool', 'double', 'float'}
if p[1].type.class_type not in allowed_class or p[
1].type.type not in allowed_base:
Errors(errorType='TypeError',
errorText=p[2] + ' not support type ' + p[1].type.stype,
token_object=p.slice[2])
p[0] = Node(type="error")
return
if p[3].type != p[5].type:
Errors(errorType='TypeError',
errorText='type should be same',
token_object=p.slice[2])
p[0] = Node(type="error")
p[1] = Node(name="type_cast",
value='bool',
children=[p[1]],
type=BasicType('bool'))
p[0] = Node("ternary_op", children=[p[1], p[3], p[5]], type=p[3].type)
def type_check_multi(node1, node2, op, token, decimal=True):
allowed_class = {'BasicType'}
allowed_base = {'int', 'long', 'char'}
if decimal: allowed_base = {'int', 'long', 'char', 'float', 'double'}
if node1.type == "error" or node2.type == "error":
return Node(type="error")
if node1.type.class_type not in allowed_class or node2.type.class_type not in allowed_class:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
if node1.type.type not in allowed_base or node2.type.type not in allowed_base:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
node1, node2, typ = implicit_casting(node1, node2)
return Node(name="binary_op",
value=typ.stype + op,
children=[node1, node2],
type=typ)
def p_func_rparen_2(p):
'''
func_rparen_2 : R_PAREN
'''
if p.stack[-2].value == "error":
return
func_name = p.stack[-1].value[0].value
token = p.stack[-1].value[0].data['token']
func_type = p.stack[-1].value[1]
func_type.param_list = []
success = sym_table.curr_symbol_table.parent._look_up(name=func_name,
token_object=token,
no_error=True)
if success:
if success.type.defined == True:
Errors(errorType='DeclarationError',
errorText='Function is already defined',
token_object=p.slice[-1])
#checking if same parameters
elif func_type.is_same(success.type) == False:
Errors(
errorType='DeclarationError',
errorText=
'Function is declared with different parameters/return type',
token_object=p.slice[-1])
else:
success.type.symbol_table.unused = True
success.type.symbol_table = sym_table.curr_symbol_table
else:
sym_table.curr_symbol_table.parent._add_entry(name=func_name,
type=func_type,
token_object=token)
def type_check_bit(node1, node2, op, token):
allowed_class = {'BasicType'}
allowed_base = {'int', 'long', 'char', 'bool'}
if node1.type == "error" or node2.type == "error":
return Node(type="error")
if node1.type.class_type not in allowed_class or node2.type.class_type not in allowed_class:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
if node1.type.type not in allowed_base or node2.type.type not in allowed_base:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
node1, node2, typ = implicit_casting(node1, node2)
node = Node(name="binary_op",
value=typ.stype + op,
children=[node1, node2],
type=typ)
node.code = node1.code + node2.code
tmp, code = get_opcode(op=typ.stype + op,
place1=node1.place,
place2=node2.place,
type=typ)
node.code += [code]
node.place = tmp
return node
def p_func_rparen_1(p):
'''
func_rparen_1 : R_PAREN
'''
if p.stack[-2].value == "error":
return
func_name = p.stack[-2].value[0].value
token = p.stack[-2].value[0].data['token']
func_type = p.stack[-2].value[1]
param_list = p.stack[-1].value
func_type.add_param_list(param_list)
#adding function to global table after creating paramlist
success = sym_table.curr_symbol_table.parent._look_up(
name=func_name, token_object=p.slice[-1], no_error=True)
if success and success.type.class_type == "FunctionType":
if success.type.defined == True:
Errors(errorType='DeclarationError',
errorText='Function is already defined',
token_object=p.slice[-1])
#checking if same parameters
elif func_type.is_same(success.type) == False:
Errors(
errorType='DeclarationError',
errorText=
'Function is declared with different parameters/return type',
token_object=p.slice[-1])
else:
#removing sym table of decl
success.type.symbol_table.unused = True
success.type.symbol_table = sym_table.curr_symbol_table
else:
sym_table.curr_symbol_table.parent._add_entry(name=func_name,
type=func_type,
token_object=token)
def type_check_logical(node1, node2, op, token):
allowed_class = {'BasicType'}
allowed_base = {'int', 'long', 'char', 'double', 'float', 'bool'}
if node1.type == "error" or node2.type == "error":
return Node(type="error")
if node1.type.class_type not in allowed_class or node2.type.class_type not in allowed_class:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
if node1.type.type not in allowed_base or node2.type.type not in allowed_base:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
if node1.type.type != "bool":
node1 = Node(name="type_cast",
value='bool',
children=[node1],
type=BasicType('bool'))
if node2.type.type != "bool":
node2 = Node(name="type_cast",
value='bool',
children=[node2],
type=BasicType('bool'))
return Node(name="binary_op",
value=op,
children=[node1, node2],
type=BasicType('bool'))
def type_check_add(node1, node2, op, token):
allowed_class = [('PointerType', 'BasicType'),
('BasicType', 'PointerType'), ('BasicType', 'BasicType')]
allowed_base = [{'int', 'long', 'char'}, {'int', 'long', 'char'},
{'int', 'long', 'char', 'double', 'float'}]
if node1.type == "error" or node2.type == "error":
return Node(type="error")
class1 = node1.type.class_type
class2 = node2.type.class_type
if (class1, class2) not in allowed_class:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
i = allowed_class.index((class1, class2))
if i == 0:
if node2.type.type not in allowed_base[i]:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype +
',' + node2.type.stype,
token_object=token)
return Node(type="error")
return Node(name="binary_op",
value=node1.type.stype + op,
type=node1.type,
children=[node1, node2])
if i == 1:
if node1.type.type not in allowed_base[i]:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype +
',' + node2.type.stype,
token_object=token)
return Node(type="error")
return Node(name="binary_op",
value=node2.type.stype + op,
type=node2.type,
children=[node2, node1])
if i == 2:
if node1.type.type not in allowed_base[
i] or node2.type.type not in allowed_base[i]:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype +
',' + node2.type.stype,
token_object=token)
return Node(type="error")
node1, node2, typ = implicit_casting(node1, node2)
return Node(name="binary_op",
value=typ.stype + op,
children=[node1, node2],
type=typ)
def type_check_assign(node1, node2, token):
if node1.type == "error" or node2.type == "error":
return Node(type="error")
if "const@" in node1.place:
Errors(errorType='TypeError',
errorText="cannot asign to constant",
token_object=token)
return Node(type="error")
if "sconst@" in node2.place:
Errors(errorType='TypeError',
errorText="can asign string constant only in declaration",
token_object=token)
return Node(type="error")
if node1.type.class_type == "PointerType" and node1.type.is_array == True:
Errors(errorType='TypeError',
errorText="cannot asign to array",
token_object=token)
return Node(type="error")
if node2.type.is_convertible_to(node1.type):
node2 = typecast(node2, type=node1.type)
node = Node("binary_op",
node1.type.stype + "=",
children=[node1, node2],
type=node1.type)
node.code = node1.code + node2.code
if "load$" in node1.place:
place = node1.place.split("load$")[-1]
node.code += [
gen(op=get_type(node2) + "_eq",
place1=node2.place,
place3=place)
]
else:
node.code += [
gen(op=get_type(node2) + "=",
place1=node2.place,
place3=node1.place)
]
if "const@" in node2.place:
const_use(node2.place)
node.place = node1.place
node = load_place(node)
return node
Errors(errorType='TypeError',
errorText="cannot assign " + node2.type.stype + " to " +
node1.type.stype,
token_object=token)
return Node(type="error")
def p_postfix_expression(p):
'''
postfix_expression : primary_expression
| postfix_expression INCREMENT
| postfix_expression DECREMENT
'''
if len(p) == 2:
p[0] = p[1]
else:
allowed_base = {'int', 'float', 'double', 'char', 'long'}
allowed_class = {'PointerType'}
if p[1].type == 'error':
p[0] = p[1]
elif p[1].type.class_type == 'BasicType' and p[
1].type.type in allowed_base:
p[0] = Node(name="unary_op",
value=str(p[1].type) + ': p' + p[2],
children=[p[1]],
type=p[1].type)
elif p[1].type.class_type in allowed_class:
p[0] = Node(name="unary_op",
value=str(p[1].type) + ': p' + p[2],
children=[p[1]],
type=p[1].type)
else:
p[0] = p[1]
p[0].type = 'error'
Errors(errorType='TypeError',
errorText='increment/decrement not possible',
token_object=p.slice[-1])
def p_function_definition_1(p):
'''
function_definition : type_specifier declarator func_scope parameter_type_list R_PAREN SEMI_COLON pop_sym
| type_specifier declarator func_scope R_PAREN SEMI_COLON pop_sym
'''
p[0] = []
#make unused sym = True
# if len(p) == 8: p[7].unused = True
# else: p[6].unused = True
success = sym_table.look_up(name=p[2].value,
token_object=p[2].data['token'],
no_error=True)
if success != None:
Errors(errorType='DeclarationError',
errorText='Function is already declared/defined',
token_object=p[2].data['token'])
else:
if len(p) == 8:
sym_table.add_entry(name=p[2].value,
type=FunctionType(return_type=p[2].type,
param_list=p[4],
defined=False,
symbol_table=p[7]))
else:
sym_table.add_entry(name=p[2].value,
type=FunctionType(return_type=p[2].type,
param_list=[],
defined=False,
symbol_table=p[6]))
def op_on_const(op, place1, place2):
value1 = get_const_value(place1)
value2 = get_const_value(place2)
#length 2 op <= , => , == , >> , << , !=
if op[-2:] == "<=": return int(value1 <= value2)
if op[-2:] == ">=": return int(value1 >= value2)
if op[-2:] == "==": return int(value1 == value2)
if op[-2:] == "!=": return int(value1 != value2)
if op[-2:] == ">>": return value1 >> value2
if op[-2:] == "<<": return value1 << value2
#length 1 +,-,/,*,%,^,&,|
if op[-1] == "+": return value1 + value2
if op[-1] == "-": return value1 - value2
if op[-1] == "/":
if value2 == 0:
Errors(
errorType='RuntimeError',
errorText='division by zero',
)
return 0
return value1 / value2
if op[-1] == "*": return value1 * value2
if op[-1] == "%": return value1 % value2
if op[-1] == "^": return value1 ^ value2
if op[-1] == "&": return value1 & value2
if op[-1] == "|": return value1 | value2
if op[-1] == ">": return int(value1 > value2)
if op[-1] == "<": return int(value1 < value2)
assert False, op + "not in list"
def p_postfix_expression_2(p):
# function ref
'''
postfix_expression : postfix_expression L_PAREN R_PAREN
| postfix_expression L_PAREN argument_expression_list R_PAREN
'''
if p[1].type == "error":
p[0] = Node(type="error")
return
if p[1].type.class_type != "FunctionType":
p[0] = Node(type="error")
Errors(errorType='TypeError',
errorText='not function type',
token_object=p.slice[-1])
return
param_list = p[1].type.param_list
return_type = p[1].type.return_type
if len(p) == 4:
if len(param_list) != 0:
p[0] = Node(type="error")
Errors(errorType='TypeError',
errorText='Function require more arguments',
token_object=p.slice[-1])
else:
p[0] = Node(name="func_call", type=return_type, children=[p[1]])
else:
arg_list = p[3].data['args_type']
if len(arg_list) != len(param_list):
p[0] = Node(type="error")
Errors(errorType='TypeError',
errorText='No of arguments is not matching',
token_object=p.slice[-1])
# check for type cast if possible
else:
p[0] = Node(name="func_call",
type=return_type,
children=[p[1], p[3]])
def typecast(node1, type, token=None, hard=False):
node1 = load_place(node1)
assert isinstance(type, Type), "not of class Type"
assert type.class_type in {"BasicType", "PointerType",
"StructType"}, "not valid type"
# assert node1.type.class_type in {"BasicType","PointerType"}, "not valid type"
# assert "sconst@" not in node1.place, "string in typecast"
if node1.type.stype == type.stype:
if hard: node1.type = type
return node1
elif "sconst@" in node1.place:
# print(str(node1.type.stype),str(type.stype))
if token:
Errors(errorType='TypeError',
errorText="cannot assign string constant to type " +
type.stype,
token_object=token)
return Node(type="error")
else:
assert "string in typecast"
else:
node = Node(name="type_cast",
value=type.stype,
children=[node1],
type=type)
node.code = node1.code
node.place = node1.place
if type.class_type == 'PointerType':
type1 = 'long'
else:
type1 = type.type
if node1.type.class_type == 'PointerType':
type2 = 'long'
else:
type2 = node1.type.type
if type1 == type2:
return node
if type1 in {"long", "int"} and type2 in {"long", "int"}:
return node
if "const@" in node1.place:
node.place = get_const(const=get_const_value(node1.place),
type=BasicType(type1))
return node
node.place = get_newtmp(type=BasicType(type1))
node.code += [
gen(op=type2 + "_to_" + type1,
place1=node1.place,
place3=node.place,
code=node.place + " = " + type2 + "_to_" + type1 + " " +
node1.place)
]
return node
def p_jump_statement_1(p):
'''
jump_statement : RETURN SEMI_COLON
| RETURN expression SEMI_COLON
'''
success = sym_table.look_up(name='return',
token_object=p.slice[1],
no_error=True)
if success:
if len(p) == 3:
if success.type != Type():
p[0] = Node(type="error")
Errors(errorType='TypeError',
errorText='return type not matching',
token_object=p.slice[2])
return
p[0] = Node(name="return", type="ok")
else:
if p[2].type == "error":
p[0] = Node(type="error")
return
if p[2].type.is_convertible_to(success.type):
if str(p[2].type) != str(success.type):
p[2] = Node(name="type_cast",
value=success.type.stype,
children=[p[2]],
type=success.type)
p[0] = Node(name="return", children=[p[2]], type="ok")
return
p[0] = Node(type="error")
Errors(errorType='TypeError',
errorText='return type not matching',
token_object=p.slice[2])
else:
p[0] = Node(type="error")
Errors(errorType='TypeError',
errorText='Not Function',
token_object=p.slice[2])
def type_check_init(init, type, token):
if isinstance(init, Node):
if init.type == "error":
return Node(type="error")
if init.type.is_convertible_to(type):
return init
else:
Errors(errorType='TypeError',
errorText="cannot assign " + init.type.stype + " to " +
type.stype,
token_object=token)
return Node(type="error")
if type.class_type != "PointerType":
Errors(errorType='TypeError',
errorText="cannot assign array to " + type.stype,
token_object=token)
return Node(type="error")
if isinstance(init[0], Node): leng = 0
else: leng = len(init[0])
size_error = False
for i in init[1:]:
if (isinstance(i, Node) and leng != 0) or (isinstance(i, list)
and leng != len(i)):
size_error = True
break
if size_error:
Errors(errorType='TypeError',
errorText="size of arrays not matching",
token_object=token)
return Node(type="error")
children = []
for i in init:
node = type_check_init(i, type.type, token)
if node.type == "error":
return Node(type="error")
children.append(node)
return Node(name="init_array", value="{}", children=children, type=type)
def p_postfix_expression_1(p):
# Array ref
'''
postfix_expression : postfix_expression L_SQBR expression R_SQBR
'''
allowed_class = {'BasicType'}
allowd_base = {'int', 'long'}
check1 = True
check2 = True
if p[3].type == "error":
check1 = False
p[0] = Node(type="error")
elif p[3].type.class_type not in allowed_class or p[
3].type.type not in allowd_base:
Errors(errorType='TypeError',
errorText='wrong index type ' + p[3].type.stype,
token_object=p.slice[-1])
check1 = False
p[0] = Node(type="error")
allowed_class = {'PointerType'}
if p[1].type == "error":
check2 = False
p[0] = Node(type="error")
elif p[1].type.class_type not in allowed_class:
Errors(errorType='TypeError',
errorText='cannot reference type ' + p[1].type.stype,
token_object=p.slice[-1])
check2 = False
p[0] = Node(type="error")
if check1 and check2:
p[0] = Node(name="array_ref",
value="[]",
type=p[1].type.type,
children=[p[1], p[3]])
def p_func_scope(p):
'''
func_scope : L_PAREN
'''
decl = p.stack[-1].value #getting function name and return type
if decl.type.class_type == "PointerType" and len(
decl.type.array_size) != 0:
Errors(errorType='DeclarationError',
errorText='Function cannot have array return type',
token_object=p.slice[-1])
sym_table.start_scope(name=decl.value) #starting scope of function
sym_table.add_entry(name='return',
type=decl.type) #creating entry to check return type
p[0] = (decl,
FunctionType(return_type=decl.type,
symbol_table=sym_table.curr_symbol_table)
) #type of function
def type_check_assign(node1, node2, token):
if node1.type == "error" or node2.type == "error":
return Node(type="error")
if node2.type.is_convertible_to(node1.type):
if str(node2.type) != str(node1.type):
node2 = Node(name="type_cast",
value=node1.type.stype,
children=[node2],
type=node1.type)
return Node("binary_op",
node1.type.stype + "=",
children=[node1, node2],
type=node1.type)
Errors(errorType='TypeError',
errorText="cannot assign " + node2.type.stype + " to " +
node1.type.stype,
token_object=token)
return Node(type="error")
def type_check_assign_op(node1, node2, op, token):
if node1.type == "error" or node2.type == "error":
return Node(type="error")
if op == "=":
return type_check_assign(node1, node2, token)
op = op[:-1]
if op == "+" or op == "-":
node = type_check_add(node1, node2, op, token)
elif op == "/" or op == "*":
node = type_check_multi(node1, node2, op, token)
elif op == "%":
node = type_check_multi(node1, node2, op, token, decimal=False)
elif op == "^" or op == "&" or op == "|":
node = type_check_bit(node1, node2, op, token)
else:
Errors(errorType='SyntaxError',
errorText="invalid " + op + "=",
token_object=token)
return Node(type="error")
return type_check_assign(node1, node, token)
def p_cast_expression(p):
'''
cast_expression : unary_expression
| L_PAREN type_name R_PAREN cast_expression
'''
if len(p) == 2:
p[0] = p[1]
else:
if p[2].type == 'error' or p[4].type == 'error':
p[0] = Node(type='error')
else:
if p[4].type.is_convertible_to(p[2].type):
p[0] = Node(name="type_cast",
value=p[2].type.stype,
type=p[2].type,
children=[p[4]])
else:
Errors(errorType='TypeError',
errorText="cannot typecast " + p[4].types.stype +
" to " + p[2].type.stype,
token_object=p.slice[1])
p[0] = Node(type='error')
def type_check_logical(node1, node2, op, token):
allowed_class = {'BasicType', 'PointerType'}
allowed_base = {'int', 'long', 'char', 'float', 'bool'}
if node1.type == "error" or node2.type == "error":
return Node(type="error")
if "sconst@" in node1.place or "sconst@" in node2.place:
Errors(errorType='TypeError',
errorText=op + ' not support string constant',
token_object=token)
return Node(type="error")
if node1.type.class_type not in allowed_class or node2.type.class_type not in allowed_class:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
if node1.type.class_type == "BasicType" and node1.type.type not in allowed_base:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
if node2.type.class_type == "BasicType" and node2.type.type not in allowed_base:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
# if node1.type.type != "bool":
# node1 = typecast(node1,BasicType("bool"))
# if node2.type.type != "bool":
# node2 = typecast(node2,BasicType("bool"))
node = Node(name="binary_op",
value=op,
children=[node1, node2],
type=BasicType('bool'))
# node.code = node1.code + node2.code
# tmp,code = get_opcode(op=op,place1=node1.place,place2=node2.place,type=BasicType('bool'))
# node.code += [code]
# node.place = tmp
if op == "&&":
if "const@" in node1.place and "const@" in node2.place:
value = 1 if get_const_value(node1.place) and get_const_value(
node2.place) else 0
node.place = get_const(value, "bool")
return node
if "const@" in node1.place:
const_use(node1.place)
if "const@" in node2.place:
const_use(node2.place)
node.place = get_newtmp(BasicType('bool'))
label1 = get_newlabel()
label2 = get_newlabel()
node.code = node1.code
node.code += [gen(op="ifz", place1=node1.place, place2=label1)]
node.code += node2.code
node.code += [gen(op="ifz", place1=node2.place, place2=label1)]
node.code += [
gen(op="bool=",
place1=get_const(1, type='bool', use=True),
place3=node.place)
]
node.code += [gen(op="goto", place1=label2)]
node.code += [gen(op="label", place1=label1)]
node.code += [
gen(op="bool=",
place1=get_const(0, type='bool', use=True),
place3=node.place)
]
node.code += [gen(op="label", place1=label2)]
else:
if "const@" in node1.place and "const@" in node2.place:
value = 1 if get_const_value(node1.place) or get_const_value(
node2.place) else 0
node.place = get_const(value, "bool")
return node
if "const@" in node1.place:
const_use(node1.place)
if "const@" in node2.place:
const_use(node2.place)
node.place = get_newtmp(BasicType('bool'))
label1 = get_newlabel()
label2 = get_newlabel()
node.code = node1.code
node.code += [gen(op="ifnz", place1=node1.place, place2=label1)]
node.code += node2.code
node.code += [gen(op="ifnz", place1=node2.place, place2=label1)]
node.code += [
gen(op="bool=",
place1=get_const(0, type='bool', use=True),
place3=node.place)
]
node.code += [gen(op="goto", place1=label2)]
node.code += [gen(op="label", place1=label1)]
node.code += [
gen(op="bool=",
place1=get_const(1, type='bool', use=True),
place3=node.place)
]
node.code += [gen(op="label", place1=label2)]
return node
def main():
"""Driver code for Abstract syntax tree
Generation"""
#read source code provided by user
arg_parser = argparse.ArgumentParser(description="C compiler for x86_64")
arg_parser.add_argument('source_code', help="location of source code file")
arg_parser.add_argument('-o',
help="output file name, \{default a.out\}",
default="a.out")
arg_parser.add_argument(
'-f',
help="name of file for additional files, \{default a\}",
default="a")
arg_parser.add_argument('-c',
action='store_true',
help="output object file")
arg_parser.add_argument('-d', action='store_true', help="output assembly")
arg_parser.add_argument('-a', action='store_true', help="output ast")
arg_parser.add_argument('-s',
action='store_true',
help="output symbol table")
arg_parser.add_argument('-t',
action='store_true',
help="output 3 address code")
arg_parser.add_argument('-l',
action='store_true',
help="output lexeme table")
arg_parser.add_argument(
'-stdc',
action='store_true',
help=
"linker method, if specified it'll uses custom elf entry else from standard X86-64-linux.so"
)
arg_parser.add_argument(
'-n',
action="store_false",
help="only create till asm, , do not create executable")
args = arg_parser.parse_args()
try:
source_code = open(args.source_code, "r").read()
except FileNotFoundError:
print(
"source file cannot be open/read.\nCheck the file name or numbers of arguments!!"
)
sys.exit(-1)
source_file = args.source_code
file_name = args.f
grammar = get_grammar(source_file, source_code, debug=1)
if len(Errors.get_all_error()):
for error in Errors.get_all_error():
print(error)
return
tac_code = grammar.code
tac_code = remove_none(tac_code)
#to remove redundant labels
tac_code = remove_label(tac_code)
print_asm(tac_code, stdc=args.stdc)
#assembly
if args.d:
os.system("cp temp.asm {}.asm".format(file_name))
if args.n:
asm_file = "temp.asm"
os.system('yasm -g dwarf2 -f elf64 temp.asm 2> temp')
os.system('touch temp')
if args.c:
os.system("cp temp.o " + file_name + ".o")
if args.stdc:
os.system(
"ld -dynamic-linker /lib64/ld-linux-x86-64.so.2 -o {} temp.o -lc -lm 2> temp"
.format(args.o))
else:
os.system(
"ld -o {} -dynamic-linker /lib64/ld-linux-x86-64.so.2 /usr/lib/x86_64-linux-gnu/crt1.o /usr/lib/x86_64-linux-gnu/crti.o -lc temp.o /usr/lib/x86_64-linux-gnu/crtn.o -lm 2> temp"
.format(args.o))
os.system("rm -rf temp.asm temp.o temp")
else:
os.system("rm -rf temp.asm")
# ast
if args.a:
Graph = draw_ast(grammar)
Graph.draw(file_name + '.png', format='png')
# symbol table
if args.s:
print_csv(sym_table=sym_table, filename=file_name + ".csv")
# 3AC
if args.t:
print_code(tac_code, filename=file_name + ".3ac")
# lexeme
if args.l:
print_lexeme(source_code, file_name + ".lex")
def type_check_unary(node1, op, token, is_typename=False):
allowed_base = {'int', 'float', 'char', 'long'}
allowed_base_1 = {'int', 'char', 'long'}
error = False
error_const = False
if node1.type == "error":
return Node(type="error")
const = False
if 'const@' in node1.place:
const = True
if (op == "++" or op == "--"):
if const:
error_const = True
elif (node1.type.class_type == "BasicType" and node1.type.type
in allowed_base) or node1.type.class_type == "PointerType":
node = Node(name="unary_op",
value=node1.type.stype + op,
children=[node1],
type=node1.type)
node.place = node1.place
const_place = get_const(const=1, type=BasicType("long"))
node_1 = Node(type=BasicType("long"))
node_1.place = const_place
node_assign = type_check_assign_op(node1,
node_1,
op=op[0] + "=",
token=token)
if node_assign.type == "error":
return Node(type="error")
node.code = node_assign.code
return node
# elif node1.type.class_type == "BasicType" and node1.type.type in allowed_base:
# node = Node(name="unary_op",value=node1.type.stype+op,children=[node1],type=node1.type)
# node.code = node1.code
# const_place = get_const(const=1,type=node1.type,use=True)
# node.code += [gen(op=node1.type.stype+op[0],place1=node1.place,place2=const_place,place3=node1.place)]
# node.place = node1.place
# return node
# elif node1.type.class_type == "PointerType":
# node = Node(name="unary_op",value=node1.type.stype+op,children=[node1],type=node1.type)
# node.code = node1.code
# width = node1.type.type_size
# const_place = get_const(const=width,type="long",use=True)
# node.code += [gen(op="long"+op[0],place1=node1.place,place2=const_place,place3=node1.place)]
# node.place = node1.place
# return node
else:
error = True
elif op == "+" or op == "-":
node1 = load_place(node1)
if node1.type.class_type == "BasicType" and node1.type.type in allowed_base:
node = Node(name="unary_op",
value=node1.type.stype + op,
children=[node1],
type=node1.type)
if op == "+":
node.place = node1.place
return node
if "const@" in node1.place:
neg = -1 if op == "-" else 1
node.place = get_const(neg * get_const_value(node1.place),
type=node1.type)
return node
node.code = node1.code
node.place = get_newtmp(type=node1.type)
node.code += [
gen(op="u" + node1.type.stype + op,
place1=node1.place,
place3=node.place)
]
return node
error = True
elif op == "&":
if const:
error_const = True
else:
node = Node(name="unary_op",
value=op,
children=[node1],
type=PointerType(node1.type))
node.code = node1.code
if "load$" in node1.place:
node.place = node1.place.split("load$")[-1]
return node
node.place = get_newtmp()
node.code += [
gen(op="addr",
place1=node1.place,
place3=node.place,
code=node.place + " = " + "addr(" + node1.place + ")")
]
return node
elif op == "*":
node1 = load_place(node1)
if const:
error_const = True
elif node1.type.class_type == "PointerType":
node = Node(name="unary_op",
value=op,
children=[node1],
type=node1.type.type)
if node.type.class_type == "PointerType":
node.type.array_size = node1.type.array_size[1:]
node.code = node1.code
# node.place = get_newtmp(node1.type.type)
# node.code += [gen(op="load",place1=node1.place,place3=node.place,code=node.place+" = "+"load("+node1.place+")")]
node.place = "load$" + node1.place
return node
else:
Errors(errorType='TypeError',
errorText="cannot dereference non-pointer type " +
node1.type.stype,
token_object=token)
return Node(type="error")
elif op == "~":
node1 = load_place(node1)
if node1.type.class_type == "BasicType" and node1.type.type in allowed_base_1:
node = Node(name="unary_op",
value=node1.type.stype + op,
children=[node1],
type=node1.type)
if "const@" in node1.place:
node.place = get_const(~get_const_value(node1.place),
type=node1.type)
return node
node.code = node1.code
node.place = get_newtmp(type=node1.type)
node.code += [
gen(op=node1.type.stype + op,
place1=node1.place,
place3=node.place)
]
return node
error = True
elif op == "!":
node1 = load_place(node1)
if node1.type.is_convertible_to(BasicType('bool')):
node = Node(name="unary_op",
value=node1.type.stype + op,
children=[node1],
type=BasicType('bool'))
if "const@" in node1.place:
value = 0 if get_const_value(node1.place) == 0 else 1
node.place = get_const(value, type='bool')
return node
node.code = node1.code
node.place = get_newtmp(type=BasicType('bool'))
node.code += [
gen(op="not_bool", place1=node1.place, place3=node.place)
]
return node
error = True
elif op == "sizeof":
if is_typename:
node = Node(name="unary_op",
value=op + ':' + node1.type.stype,
type=BasicType(type='long'))
node.place = get_const(const=node1.type.width, type="long")
# node.code += [gen(op="=",place1=str(node1.type.width),place3=node.place)]
return node
if isinstance(node1.type, Type) == False:
Errors(errorType='TypeError',
errorText="cannot do sizeof on non type",
token_object=token)
return Node(type="error")
if node1.type.class_type in {'BasicType', 'PointerType', 'StructType'}:
node = Node(name="unary_op",
value=op,
type=BasicType(type='long'),
children=[node1])
node.place = get_const(const=node1.type.width, type="long")
# node.code += [gen(op="=",place1=str(node1.type.width),place3=node.place)]
return node
error = True
if error_const:
Errors(errorType='TypeError',
errorText=op + " not valid on constant",
token_object=token)
return Node(type="error")
if error:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype,
token_object=token)
return Node(type="error")
def main():
"""Driver code for Abstract syntax tree
Generation"""
#read source code provided by user
arg_parser = argparse.ArgumentParser(
description="Lexer for Source Language C")
arg_parser.add_argument('source_code', help="source code file location")
arg_parser.add_argument('-o',
help="take the name of dot script",
default="ast.dot")
arg_parser.add_argument('-f',
help="take the name of png file",
default="ast.png")
arg_parser.add_argument('-d',
help="take the name of csv file",
default="dump.csv")
arg_parser.add_argument('-p',
action='store_true',
help="output dot script to console")
arg_parser.add_argument('-l',
action='store_true',
help="output lexeme table")
args = arg_parser.parse_args()
try:
# source_code = open(sys.argv[1],"r").read()
source_code = open(args.source_code, "r").read()
except FileNotFoundError:
print(
"source file cannot be open/read.\nCheck the file name or numbers of arguments!!"
)
sys.exit(-1)
if args.l:
print_lexeme(source_code)
parser = yacc.yacc(debug=1)
lexer.lexer.filename = args.source_code
result = parser.parse(source_code, lexer=lexer.lexer)
#print(sym_table)
if len(Errors.get_all_error()):
for error in Errors.get_all_error():
print(error)
return
Graph = draw_ast(result)
# print(args)
if args.p:
Graph.draw(args.f, format='png')
print(Graph.string())
return
Graph.draw(args.f, format='png')
file = open(args.o, 'w')
file.write(Graph.string())
file.close()
# print(sym_table)
# print(args.d)
print_csv(sym_table=sym_table, filename=args.d)
def type_check_unary(node1, op, token, is_typename=False):
allowed_base = {'int', 'float', 'double', 'char', 'long'}
allowed_base_1 = {'int', 'char', 'long'}
error = False
if node1.type == "error":
return Node(type="error")
if op == "++" or op == "--":
if node1.type.class_type == "BasicType" and node1.type.type in allowed_base:
return Node(name="unary_op",
value=node1.type.stype + op,
children=[node1],
type=node1.type)
if node1.type.class_type == "PointerType":
return Node(name="unary_op",
value=node1.type.stype + op,
children=[node1],
type=node1.type)
error = True
elif op == "+" or op == "-":
if node1.type.class_type == "BasicType" and node1.type.type in allowed_base:
return Node(name="unary_op",
value=node1.type.stype + op,
children=[node1],
type=node1.type)
error = True
elif op == "&":
return Node(name="unary_op",
value=op,
children=[node1],
type=PointerType(node1.type))
elif op == "*":
if node1.type.class_type == "PointerType":
return Node(name="unary_op",
value=op,
children=[node1],
type=node1.type.type)
Errors(errorType='TypeError',
errorText="cannot dereference non-pointer type " +
node1.type.stype,
token_object=token)
return Node(type="error")
elif op == "~":
if node1.type.class_type == "BasicType" and node1.type.type in allowed_base_1:
return Node(name="unary_op",
value=node1.type.stype + op,
children=[node1],
type=node1.type)
error = True
elif op == "!":
if node1.type.is_convertible_to(BasicType('bool')):
return Node(name="unary_op",
value=node1.type.stype + op,
children=[node1],
type=BasicType('bool'))
error = True
elif op == "sizeof":
if is_typename:
return Node(name="unary_op",
value=op + ':' + node1.type.stype,
type=BasicType(type='long'))
if isinstance(node1.type, Type) == False:
Errors(errorType='TypeError',
errorText="cannot do sizeof on non type",
token_object=token)
return Node(type="error")
if node1.type.class_type in {'BasicType', 'PointerType', 'StructType'}:
return Node(name="unary_op",
value=op,
type=BasicType(type='long'),
children=[node1])
error = True
if error:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype,
token_object=token)
return Node(type="error")
def type_check_add(node1, node2, op, token):
allowed_class = [('PointerType', 'BasicType'),
('BasicType', 'PointerType'), ('BasicType', 'BasicType')]
allowed_base = [{'int', 'long', 'char'}, {'int', 'long', 'char'},
{'int', 'long', 'char', 'float', 'bool'}]
if node1.type == "error" or node2.type == "error":
return Node(type="error")
if "sconst@" in node1.place or "sconst@" in node2.place:
Errors(errorType='TypeError',
errorText=op + ' not support string constant',
token_object=token)
return Node(type="error")
class1 = node1.type.class_type
class2 = node2.type.class_type
if (class1, class2) not in allowed_class:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype + ',' +
node2.type.stype,
token_object=token)
return Node(type="error")
i = allowed_class.index((class1, class2))
if i == 0:
if node2.type.type not in allowed_base[i]:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype +
',' + node2.type.stype,
token_object=token)
return Node(type="error")
node = Node(name="binary_op",
value=node1.type.stype + op,
type=node1.type,
children=[node1, node2])
node.code = node1.code + node2.code
width = node1.type.type_size
const_place = get_const(width, type="long")
tmp, code = get_opcode(op="long*",
place1=node2.place,
place2=const_place,
type="long")
node.code += [code]
tmp, code = get_opcode(op="long" + op,
place1=node1.place,
place2=tmp,
type="long")
node.code += [code]
node.place = tmp
return node
if i == 1:
if node1.type.type not in allowed_base[i]:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype +
',' + node2.type.stype,
token_object=token)
return Node(type="error")
node = Node(name="binary_op",
value=node2.type.stype + op,
type=node2.type,
children=[node2, node1])
node.code = node1.code + node2.code
width = node2.type.type_size
const_place = get_const(width, type="long")
tmp, code = get_opcode(op="long*",
place1=node1.place,
place2=const_place,
type="long")
node.code += [code]
tmp, code = get_opcode(op="long" + op,
place1=node2.place,
place2=tmp,
type="long")
node.code += [code]
node.place = tmp
return node
if i == 2:
if node1.type.type not in allowed_base[
i] or node2.type.type not in allowed_base[i]:
Errors(errorType='TypeError',
errorText=op + ' not support type ' + node1.type.stype +
',' + node2.type.stype,
token_object=token)
return Node(type="error")
node1, node2, typ = implicit_casting(node1, node2)
node = Node(name="binary_op",
value=typ.stype + op,
children=[node1, node2],
type=typ)
node.code = node1.code + node2.code
tmp, code = get_opcode(op=typ.stype + op,
place1=node1.place,
place2=node2.place,
type=typ)
node.code += [code]
node.place = tmp
return node