something's screwed up with im im handling string variables :/

This commit is contained in:
SpookyDervish
2025-09-06 20:54:17 +10:00
parent 459f53a4e1
commit 85331b4067
7 changed files with 199 additions and 18 deletions

BIN
a.out Normal file

Binary file not shown.

View File

@@ -13,10 +13,10 @@ class Generator:
self.constants_reverse = {} self.constants_reverse = {}
self.constant_counter = 0 self.constant_counter = 0
def add_constant(self, value: Any): def add_constant(self, value: Any, no_string: bool = False):
existing_constant_name = self.constants_reverse.get(value, None) existing_constant_name = self.constants_reverse.get(value, None)
if existing_constant_name != None: return f"[.{existing_constant_name}]" if existing_constant_name != None: return f"[.{existing_constant_name}]"
self.constants["LC" + str(self.constant_counter)] = value self.constants["LC" + str(self.constant_counter)] = {"value": value, "no_string": no_string}
self.constants_reverse[value] = "LC" + str(self.constant_counter) self.constants_reverse[value] = "LC" + str(self.constant_counter)
self.constant_counter += 1 self.constant_counter += 1
return "[.LC" + str(self.constant_counter-1) + "]" return "[.LC" + str(self.constant_counter-1) + "]"

View File

@@ -23,12 +23,12 @@ class X86_64Generator(Generator):
self.lines.append("pop " + reg + "\n\t") self.lines.append("pop " + reg + "\n\t")
self.stack_size -= 1 self.stack_size -= 1
def get_variable(self, var_name: str, reg: str, float: bool = False): def get_variable(self, var_name: str, reg: str, float: bool = False, offset: int = 0, no_stack_pop: bool = False):
var = self.variables.get(var_name, None) var = self.variables.get(var_name, None)
var_pos = self.get_var_pos(var_name) var_pos = self.get_var_pos(var_name)
try: try:
#print(var["type"]) #print(var["type"])
if var["type"] == FloatNode or float: if var["type"] == FloatNode:
conversion = { conversion = {
"rax": "xmm0", "rax": "xmm0",
"rbx": "xmm1", "rbx": "xmm1",
@@ -36,12 +36,15 @@ class X86_64Generator(Generator):
# ... # ...
} }
self.lines.append(f"movsd {conversion[reg]}, [rsp + {var_pos}]\n\t") self.lines.append(f"movsd {conversion[reg]}, [rsp + {var_pos + offset}]\n\t")
self.lines.append("add rsp, 8\n\t") self.lines.append("add rsp, 8\n\t")
#self.stack_size -= 1 #self.stack_size -= 1
elif var["type"] == IntNode: elif var["type"] in [IntNode,StringNode]:
if no_stack_pop:
self.lines.append(f"mov {reg}, [rsp + {var_pos + offset}]\n\t")
else:
self.push( self.push(
f"QWORD [rsp + {var_pos}]" f"QWORD [rsp + {var_pos + offset}]"
) )
self.pop(reg) self.pop(reg)
@@ -76,6 +79,14 @@ class X86_64Generator(Generator):
self.lines.append("movsd [rsp], xmm0\n\t") self.lines.append("movsd [rsp], xmm0\n\t")
self.stack_size += 1 self.stack_size += 1
elif type(starting_value) == StringNode:
string_pointer = self.add_constant(starting_value.value)
string_len = self.add_constant(f"equ $ - {string_pointer[1:-1]}", no_string=True)
self.lines.append(f"lea rax, {string_pointer}\n\t")
self.push("rax")
self.lines.append(f"lea rax, {string_len[1:-1]}\n\t")
self.push("rax")
elif type(starting_value) == str: elif type(starting_value) == str:
if starting_value.startswith("xmm"): # floating point stuff if starting_value.startswith("xmm"): # floating point stuff
self.lines.append("sub rsp, 8\n\t") # make space self.lines.append("sub rsp, 8\n\t") # make space
@@ -99,6 +110,7 @@ class X86_64Generator(Generator):
self.lines.append(f"mov QWORD [rsp + {var_pos}], {new_value}\n\t") self.lines.append(f"mov QWORD [rsp + {var_pos}], {new_value}\n\t")
def generate_InstructionNode(self, node: InstructionNode): def generate_InstructionNode(self, node: InstructionNode):
### MISC ###
if node.instruction == "end": if node.instruction == "end":
if len(node.arguments) == 0: # example: "end" if len(node.arguments) == 0: # example: "end"
traceback(self.code, "TypeError", "end expects atleast 1 argument.") traceback(self.code, "TypeError", "end expects atleast 1 argument.")
@@ -111,7 +123,7 @@ class X86_64Generator(Generator):
if isinstance(node.arguments[0], IntNode): if isinstance(node.arguments[0], IntNode):
self.lines.append("mov rdi, " + str(node.arguments[0].value) + "\n\t") self.lines.append("mov rdi, " + str(node.arguments[0].value) + "\n\t")
elif isinstance(node.arguments[0], VarRefNode): elif isinstance(node.arguments[0], VarRefNode):
var_type = self.get_variable(node.arguments[0].var_name, "rdi") var_type = self.get_variable(node.arguments[0].var_name, "rdi", no_stack_pop=True)
if var_type == FloatNode: if var_type == FloatNode:
self.lines.append("cvttsd2si rdi, xmm0\n\t") self.lines.append("cvttsd2si rdi, xmm0\n\t")
else: else:
@@ -120,6 +132,7 @@ class X86_64Generator(Generator):
#self.lines.append("mov rdi, " + str(self.get_variable(node.arguments[0].var_name)) + "\n\t") #self.lines.append("mov rdi, " + str(self.get_variable(node.arguments[0].var_name)) + "\n\t")
self.lines.append("syscall\n\t") self.lines.append("syscall\n\t")
### VARIABLE INSTRUCTIONS ###
elif node.instruction == "set": elif node.instruction == "set":
if len(node.arguments) < 2: # example: "set" or "set &hi" if len(node.arguments) < 2: # example: "set" or "set &hi"
traceback(self.code, "TypeError", "set expects atleast 2 arguments.") traceback(self.code, "TypeError", "set expects atleast 2 arguments.")
@@ -127,7 +140,7 @@ class X86_64Generator(Generator):
traceback(self.code, "TypeError", "set expects only 2 arguments.") traceback(self.code, "TypeError", "set expects only 2 arguments.")
if not isinstance(node.arguments[0], VarPointerNode): if not isinstance(node.arguments[0], VarPointerNode):
traceback(self.code, "TypeError", f"the first argument of set should be a variable pointer, not \"{node.arguments[0]}\"") traceback(self.code, "TypeError", f"the first argument of set should be a variable pointer, not \"{node.arguments[0]}\"")
if type(node.arguments[1]) not in [IntNode, VarRefNode, FloatNode]: if type(node.arguments[1]) not in [IntNode, VarRefNode, FloatNode, StringNode]:
traceback(self.code, "TypeError", f"variables can't be of type \"{type(node.arguments[1])}\"") traceback(self.code, "TypeError", f"variables can't be of type \"{type(node.arguments[1])}\"")
variable_exists = self.variables.get(node.arguments[0].var_name, None) != None variable_exists = self.variables.get(node.arguments[0].var_name, None) != None
@@ -137,6 +150,7 @@ class X86_64Generator(Generator):
else: # modify the existing one else: # modify the existing one
self.change_variable(node.arguments[0].var_name, node.arguments[1]) self.change_variable(node.arguments[0].var_name, node.arguments[1])
### MATH INSTRUCTIONS ###
elif node.instruction == "add": elif node.instruction == "add":
if len(node.arguments) < 3: # example: "add" or "add 1" or "add 1 2" if len(node.arguments) < 3: # example: "add" or "add 1" or "add 1 2"
traceback(self.code, "TypeError", "add expects atleast 3 arguments.") traceback(self.code, "TypeError", "add expects atleast 3 arguments.")
@@ -191,7 +205,6 @@ class X86_64Generator(Generator):
self.create_variable(node.arguments[2].var_name, starting_reg, IntNode if is_integer else FloatNode) self.create_variable(node.arguments[2].var_name, starting_reg, IntNode if is_integer else FloatNode)
else: else:
self.change_variable(node.arguments[2].var_name, starting_reg) self.change_variable(node.arguments[2].var_name, starting_reg)
elif node.instruction == "subtract": elif node.instruction == "subtract":
if len(node.arguments) < 3: # example: "subtract" or "subtract 1" or "subtract 1 2" if len(node.arguments) < 3: # example: "subtract" or "subtract 1" or "subtract 1 2"
traceback(self.code, "TypeError", "subtract expects atleast 3 arguments.") traceback(self.code, "TypeError", "subtract expects atleast 3 arguments.")
@@ -246,6 +259,147 @@ class X86_64Generator(Generator):
self.create_variable(node.arguments[2].var_name, starting_reg, IntNode if is_integer else FloatNode) self.create_variable(node.arguments[2].var_name, starting_reg, IntNode if is_integer else FloatNode)
else: else:
self.change_variable(node.arguments[2].var_name, starting_reg) self.change_variable(node.arguments[2].var_name, starting_reg)
elif node.instruction == "multiply":
if len(node.arguments) < 3: # example: "multiply" or "multiply 1" or "multiply 1 2"
traceback(self.code, "TypeError", "multiply expects atleast 3 arguments.")
elif len(node.arguments) > 3: # example: "multiply 1 2 3 4"
traceback(self.code, "TypeError", "multiply expects only 3 arguments.")
elif type(node.arguments[2]) != VarPointerNode:
traceback(self.code, "TypeError", f"the destination of the multiply command must be a variable pointer, not \"{node.arguments[2]}\"")
# bro this entire god damn instruction is just error handling 😔
number1_type = None
number2_type = None
arg2 = "rbx"
if isinstance(node.arguments[0], VarRefNode):
number1_type = self.get_variable(node.arguments[0].var_name, "rax", isinstance(node.arguments[1], FloatNode))
elif isinstance(node.arguments[0], FloatNode) or isinstance(node.arguments[1], FloatNode):
number1_type = FloatNode
constant_name = self.add_constant(node.arguments[0].value)
self.lines.append(f"movsd xmm0, {constant_name}\n\t")
elif isinstance(node.arguments[0], IntNode):
number1_type = IntNode
#arg1 = node.arguments[0].value
self.lines.append(f"mov rax, {node.arguments[0].value}\n\t")
else:
traceback(self.code, "TypeError", f"expected a variable reference or number for argument 1 of multiply, got {node.arguments[0]}")
if isinstance(node.arguments[1], VarRefNode):
number2_type = self.get_variable(node.arguments[1].var_name, "rbx", number1_type == FloatNode)
elif number1_type == FloatNode or isinstance(node.arguments[1], FloatNode):
number2_type = FloatNode
constant_name = self.add_constant(node.arguments[1].value)
self.lines.append(f"movsd xmm1, {constant_name}\n\t")
elif isinstance(node.arguments[1], IntNode):
number2_type = IntNode
arg2 = node.arguments[1].value
#self.lines.append(f"mov rbx, {node.arguments[1].value}\n\t")
else:
traceback(self.code, "TypeError", f"expected a variable reference or number for argument 2 of multiply, got {node.arguments[1]}")
# TODO: numbers can be added to numbers, but numbers cant be added to strings. but strings can be added to strings, etc...
if number1_type not in [IntNode, FloatNode] or number2_type not in [IntNode, FloatNode]:
traceback(self.code, "TypeError", f"Unsupported operation \"multiply\" for \"{node.arguments[0]}\" and \"{node.arguments[1]}\".")
if number1_type == IntNode and number2_type == IntNode:
self.lines.append(f"imul rax, {arg2}\n\t")
else:
self.lines.append(f"mulsd xmm0, xmm1\n\t")
is_integer = number1_type == IntNode and number2_type == IntNode
starting_reg = "rax" if is_integer else "xmm0"
if self.variables.get(node.arguments[2].var_name, None) == None: # we need to create a variable for the destination
self.create_variable(node.arguments[2].var_name, starting_reg, IntNode if is_integer else FloatNode)
else:
self.change_variable(node.arguments[2].var_name, starting_reg)
elif node.instruction == "divide":
if len(node.arguments) < 3: # example: "divide" or "divide 1" or "divide 1 2"
traceback(self.code, "TypeError", "divide expects atleast 3 arguments.")
elif len(node.arguments) > 3: # example: "divide 1 2 3 4"
traceback(self.code, "TypeError", "divide expects only 3 arguments.")
elif type(node.arguments[2]) != VarPointerNode:
traceback(self.code, "TypeError", f"the destination of the divide command must be a variable pointer, not \"{node.arguments[2]}\"")
# bro this entire god damn instruction is just error handling 😔
arg2 = "xmm1"
number1_type = None
number2_type = None
if isinstance(node.arguments[0], VarRefNode):
number1_type = self.get_variable(node.arguments[0].var_name, "rax", True)
elif type(node.arguments[0]) in [IntNode, FloatNode]:
number1_type = FloatNode
constant_name = self.add_constant(node.arguments[0].value)
self.lines.append(f"movsd xmm0, {constant_name}\n\t")
else:
traceback(self.code, "TypeError", f"expected a variable reference or number for argument 1 of divide, got {node.arguments[0]}")
if isinstance(node.arguments[1], VarRefNode):
number2_type = self.get_variable(node.arguments[1].var_name, "rbx", True)
elif type(node.arguments[1]) in [IntNode, FloatNode]:
number2_type = FloatNode
constant_name = self.add_constant(node.arguments[1].value)
self.lines.append(f"movsd xmm1, {constant_name}\n\t")
else:
traceback(self.code, "TypeError", f"expected a variable reference or number for argument 2 of divide, got {node.arguments[1]}")
# TODO: numbers can be added to numbers, but numbers cant be added to strings. but strings can be added to strings, etc...
if number1_type not in [IntNode, FloatNode] or number2_type not in [IntNode, FloatNode]:
traceback(self.code, "TypeError", f"Unsupported operation \"divide\" for \"{node.arguments[0]}\" and \"{node.arguments[1]}\".")
self.lines.append(f"divsd xmm0, xmm1\n\t")
if self.variables.get(node.arguments[2].var_name, None) == None: # we need to create a variable for the destination
self.create_variable(node.arguments[2].var_name, "xmm0", FloatNode)
else:
self.change_variable(node.arguments[2].var_name, "xmm0")
elif node.instruction == "stdout":
if len(node.arguments) < 1: # example: "stdout"
traceback(self.code, "TypeError", "stdout expects atleast 1 argument.")
elif len(node.arguments) > 1: # example: "stdout "hi" 123"
traceback(self.code, "TypeError", "stdout expects at most 1 argument.")
arg = node.arguments[0]
printed_value = arg.__str__()
if isinstance(arg, VarRefNode):
self.get_variable(arg.var_name, "rax")
string_pointer = "rax"
else:
string_pointer = self.add_constant(printed_value)[1:-1]
string_len = self.add_constant(f"equ $ - {string_pointer}", True)[1:-1]
self.lines.append("mov rax, 1\n\t") # sys_write syscall
self.lines.append("mov rdi, 1\n\t") # a file descriptor of 1 is stdout
self.lines.append(f"mov rsi, {string_pointer}\n\t") # pointer
self.lines.append(f"mov rdx, {string_len}\n\t") # length
self.lines.append("syscall\n\t")
elif node.instruction == "stdlnout":
if len(node.arguments) < 1: # example: "stdlnout"
traceback(self.code, "TypeError", "stdlnout expects atleast 1 argument.")
elif len(node.arguments) > 1: # example: "stdlnout "hi" 123"
traceback(self.code, "TypeError", "stdlnout expects at most 1 argument.")
arg = node.arguments[0]
printed_value = arg.__str__() + "\n"
if isinstance(arg, VarRefNode):
self.get_variable(arg.var_name, "rsi", False, 0, True)
self.get_variable(arg.var_name, "rdx", False, 8, True)
else:
string_pointer = self.add_constant(printed_value)[1:-1]
string_len = self.add_constant(f"equ $ - {string_pointer}", True)[1:-1]
self.lines.append(f"mov rsi, {string_pointer}\n\t")
self.lines.append(f"mov rdx, {string_len}\n\t") # length
print(string_pointer)
self.lines.append("mov rax, 1\n\t") # sys_write syscall
self.lines.append("mov rdi, 1\n\t") # a file descriptor of 1 is stdout
self.lines.append("syscall\n\t")
else: else:
self.lines.append("; FUCK\n\t") self.lines.append("; FUCK\n\t")
@@ -255,11 +409,15 @@ class X86_64Generator(Generator):
with open(self.output_path + ".asm", "w") as f: with open(self.output_path + ".asm", "w") as f:
f.write("section .data\n") f.write("section .data\n")
for name, value in self.constants.items(): for name, const in self.constants.items():
value = const["value"]
f.write("." + name + ": ") f.write("." + name + ": ")
value_type = type(value) value_type = type(value)
if value_type == str: if value_type == str:
f.write(f"db \"{value}\", 0") if not const["no_string"]:
f.write(f"db \"{value.replace("\n","\", 10, \"")}\", 0".replace(", \"\", ", ", "))
else:
f.write(value)
elif value_type == float or value_type == int: elif value_type == float or value_type == int:
f.write(f"dq {float(value)}") f.write(f"dq {float(value)}")
f.write("\n") f.write("\n")

View File

@@ -18,6 +18,8 @@ class StringNode:
value: str value: str
def __repr__(self): def __repr__(self):
return "String" return "String"
def __str__(self):
return self.value
@dataclass @dataclass
class IntNode: class IntNode:
value: float value: float

BIN
out

Binary file not shown.

22
out.asm
View File

@@ -1,9 +1,27 @@
section .data section .data
.LC0: db "Hello, World!", 0
.LC1: equ $ - .LC0
section .text section .text
global _start global _start
_start: _start:
; InstructionNode(instruction='divide', parent=RootNode(statements=[...]), arguments=[]) ; InstructionNode(instruction='set', parent=RootNode(statements=[..., InstructionNode(instruction='set', parent=..., arguments=[VariablePointer, Int]), InstructionNode(instruction='stdlnout', parent=..., arguments=[VariableReference]), InstructionNode(instruction='end', parent=..., arguments=[VariableReference])]), arguments=[VariablePointer, String])
; FUCK lea rax, [.LC0]
push rax
lea rax, .LC1
push rax
; InstructionNode(instruction='set', parent=RootNode(statements=[InstructionNode(instruction='set', parent=..., arguments=[VariablePointer, String]), ..., InstructionNode(instruction='stdlnout', parent=..., arguments=[VariableReference]), InstructionNode(instruction='end', parent=..., arguments=[VariableReference])]), arguments=[VariablePointer, Int])
mov rax, 123
push rax
; InstructionNode(instruction='stdlnout', parent=RootNode(statements=[InstructionNode(instruction='set', parent=..., arguments=[VariablePointer, String]), InstructionNode(instruction='set', parent=..., arguments=[VariablePointer, Int]), ..., InstructionNode(instruction='end', parent=..., arguments=[VariableReference])]), arguments=[VariableReference])
mov rsi, [rsp + 16]
mov rdx, [rsp + 24]
mov rax, 1
mov rdi, 1
syscall
; InstructionNode(instruction='end', parent=RootNode(statements=[InstructionNode(instruction='set', parent=..., arguments=[VariablePointer, String]), InstructionNode(instruction='set', parent=..., arguments=[VariablePointer, Int]), InstructionNode(instruction='stdlnout', parent=..., arguments=[VariableReference]), ...]), arguments=[VariableReference])
mov rax, 60
mov rdi, [rsp + 0]
syscall

View File

@@ -1 +1,4 @@
divide set &x "Hello, World!"
set &y 123
stdlnout $x
end $y