Files
GroundPY/generators/x86_64.py

723 lines
29 KiB
Python
Raw Normal View History

2025-09-10 18:56:45 +10:00
from generators.generator import Generator, SymbolTable
from ground_ast import *
from error import traceback, warning
from optimizers.x86_64 import X86_64Optimizer
class X86_64Generator(Generator):
2025-09-02 06:42:58 +10:00
def __init__(self, ast, code, output_path):
super().__init__(ast, code, output_path)
self.stack_size = 0
2025-09-10 07:30:58 +10:00
self.function_lines = []
2025-09-02 06:42:58 +10:00
def init(self):
2025-09-07 11:58:01 +10:00
self.lines.append("global _start\n")
self.lines.append("_start:\n\t")
2025-09-02 06:42:58 +10:00
# generate code
self.generate()
2025-09-02 06:42:58 +10:00
self.write()
2025-09-10 07:30:58 +10:00
def push(self, reg: str, lines):
lines.append("push " + reg + "\n\t")
2025-09-02 06:42:58 +10:00
self.stack_size += 1
2025-09-10 07:30:58 +10:00
def pop(self, reg: str, lines):
lines.append("pop " + reg + "\n\t")
2025-09-02 06:42:58 +10:00
self.stack_size -= 1
def get_variable(self, lines, var_name: str, reg: str, float: bool = False, offset: int = 0, no_stack_pop: bool = True, scope: SymbolTable = None):
scope = scope or self.current_var_scope
var = scope.lookup(var_name)
var_pos = self.get_var_pos(var_name, scope)
if isinstance(var_pos, str): # in a reg, not on the stack
2025-09-13 07:08:11 +10:00
waste = {
"rax": "eax",
"rdi": "edi",
"rsi": "esi"
}
if waste[var_pos] == reg: # literally useless
return
2025-09-12 06:28:45 +10:00
if offset == 0:
lines.append(f"mov {reg}, {var_pos}\n\t")
elif offset == -8:
if var_pos == "rax": # according to the SysV ABI, the second return value of a function is stored in RDX
var_pos = "rdx"
lines.append(f"mov {reg}, {var_pos}\n\t")
else:
try:
if var["type"] == FloatNode:
conversion = {
"rax": "xmm0",
"rbx": "xmm1",
"rdi": "xmm0"
# ...
}
lines.append(f"movsd {conversion[reg]}, [rsp + {var_pos + offset}]\n\t")
lines.append("add rsp, 8\n\t")
self.stack_size += 1
elif var["type"] in [IntNode,StringNode]:
if no_stack_pop:
lines.append(f"mov {reg}, [rsp + {var_pos + offset}]\n\t")
else:
self.push(
f"QWORD [rsp + {var_pos + offset}]",
lines
)
self.pop(reg, lines)
elif var["type"] == BoolNode:
if no_stack_pop:
lines.append(f"mov {reg}, [rsp + {var_pos + offset}]\n\t")
else:
self.push(
f"QWORD [rsp + {var_pos + offset}]",
lines
)
self.pop(reg, lines)
except TypeError as e: # variable doesnt exist
traceback(self.code, "NameError", f"\"{var_name}\" is not defined.")
2025-09-06 21:18:22 +10:00
scope.table[var_name]["used"] = True
return var["type"]
def get_var_pos(self, var_name: str, scope: SymbolTable = None):
scope = scope or self.current_var_scope
2025-09-07 07:25:11 +10:00
try:
stack_loc = scope.lookup(var_name)['stack_loc']
if isinstance(stack_loc, str):
return stack_loc # its in a reg not on the stack
return (self.stack_size - stack_loc - 1) * 8
except TypeError as e: # not defined
2025-09-07 07:25:11 +10:00
traceback(self.code, "TypeError", f"\"{var_name}\" is not defined.")
2025-09-02 16:57:56 +10:00
def create_variable(self, lines, var_name: str, starting_value, var_type: Any = None, scope = None):
var_type = var_type or type(starting_value)
scope = scope or self.current_var_scope
2025-09-04 07:45:20 +10:00
stack_location = self.stack_size
if type(starting_value) == IntNode:
2025-09-10 07:30:58 +10:00
lines.append(f"mov rax, {starting_value.value}\n\t")
self.push("rax", lines)
2025-09-02 16:57:56 +10:00
elif type(starting_value) == VarRefNode:
var_type = self.get_variable(lines, starting_value.var_name, "rax", scope=scope)
2025-09-04 07:45:20 +10:00
if var_type == FloatNode:
2025-09-10 07:30:58 +10:00
lines.append("sub rsp, 8\n\t")
lines.append("movsd [rsp], xmm0\n\t")
2025-09-04 07:45:20 +10:00
else:
2025-09-10 07:30:58 +10:00
self.push("rax", lines)
2025-09-04 07:45:20 +10:00
elif type(starting_value) == FloatNode:
name = self.add_constant(starting_value.value)
2025-09-10 07:30:58 +10:00
lines.append("sub rsp, 8\n\t") # make space on the stack
lines.append(f"movsd xmm0, {name}\n\t")
lines.append("movsd [rsp], xmm0\n\t")
2025-09-04 07:45:20 +10:00
self.stack_size += 1
elif type(starting_value) == StringNode:
2025-09-06 21:18:22 +10:00
string_pointer = self.add_constant(
starting_value.value
)
string_len = self.add_constant(f"equ $ - {string_pointer[1:-1]}", no_string=True)
2025-09-10 07:30:58 +10:00
lines.append(f"lea rax, {string_pointer}\n\t")
self.push("rax", lines)
lines.append(f"mov rax, {string_len[1:-1]}\n\t")
self.push("rax", lines)
2025-09-07 13:38:16 +10:00
elif type(starting_value) == BoolNode:
2025-09-10 07:30:58 +10:00
self.push("1" if starting_value.value else "0", lines)
2025-09-07 13:38:16 +10:00
elif type(starting_value) == str:
2025-09-04 07:45:20 +10:00
if starting_value.startswith("xmm"): # floating point stuff
2025-09-10 07:30:58 +10:00
lines.append("sub rsp, 8\n\t") # make space
lines.append(f"movsd [rsp], {starting_value}\n\t")
2025-09-04 07:45:20 +10:00
self.stack_size += 1
else:
2025-09-13 15:48:41 +10:00
if starting_value.startswith("BUF"):
self.push(starting_value, lines) # push buffer
self.push("rax", lines) # push length
else:
self.push(starting_value, lines)
scope.define(var_name, {"stack_loc": stack_location, "type": var_type, "used": False})
2025-09-02 16:57:56 +10:00
def change_variable(self, lines, var_name: str, new_value, scope: SymbolTable = None):
scope = scope or self.current_var_scope
var_pos = self.get_var_pos(var_name, scope)
2025-09-02 16:26:19 +10:00
old_var_type = scope.table[var_name]["type"]
2025-09-10 20:45:22 +10:00
if isinstance(var_pos, str):
if type(new_value) == IntNode: # we're changing a variable to a number
#lines.append(f"mov QWORD [rsp + {var_pos}], {new_value.value}\n\t")
lines.append(f"mov {var_pos}, {new_value.value}\n\t")
scope.table[var_name]["type"] = IntNode
elif type(new_value) == VarRefNode: # we're changing a variable to the value of another variable
var_type = self.get_variable(lines, new_value.var_name, "rax")
#lines.append(f"mov QWORD [rsp + {var_pos}], rax\n\t")
lines.append(f"mov {var_pos}, rax\n\t")
scope.table[var_name]["type"] = var_type
elif type(new_value) == StringNode: # we're changing a variable to a string
lines.append(f"mov QWORD [rsp + {var_pos}], 0\n\t")
2025-09-02 16:26:19 +10:00
2025-09-10 20:45:22 +10:00
string_pointer = self.add_constant(new_value.value)
string_len = self.add_constant(f"equ $ - {string_pointer[1:-1]}", no_string=True)
#lines.append(f"lea QWORD [rsp + {var_pos}], {string_pointer}\n\t")
##lines.append(f"mov QWORD [rsp + {var_pos + 8}], {string_len[1:-1]}\n\t")
2025-09-07 07:25:11 +10:00
2025-09-10 20:45:22 +10:00
scope.table[var_name]["stack_loc"] = self.stack_size
lines.append(f"lea rax, {string_pointer}\n\t")
self.push("rax", lines)
lines.append(f"mov rax, {string_len[1:-1]}\n\t")
self.push("rax", lines)
2025-09-07 07:25:11 +10:00
2025-09-10 20:45:22 +10:00
scope.table[var_name]["type"] = StringNode
2025-09-07 07:25:11 +10:00
2025-09-10 20:45:22 +10:00
elif type(new_value) == BoolNode:
#lines.append(f"mov QWORD [rsp + {var_pos}], {'1' if new_value.value else '0'}\n\t")
lines.append(f"mov {var_pos}, {int(new_value.value)}\n\t")
scope.table[var_name]["type"] = BoolNode
elif type(new_value) == str: # we're changing a variable to the value of a register
#lines.append(f"mov QWORD [rsp + {var_pos}], {new_value}\n\t")
lines.append(f"mov {var_pos}, {new_value}\n\t")
scope.table[var_name]["type"] = IntNode
else:
if type(new_value) == IntNode: # we're changing a variable to a number
lines.append(f"mov QWORD [rsp + {var_pos}], {new_value.value}\n\t")
scope.table[var_name]["type"] = IntNode
elif type(new_value) == VarRefNode: # we're changing a variable to the value of another variable
var_type = self.get_variable(lines, new_value.var_name, "rax")
lines.append(f"mov QWORD [rsp + {var_pos}], rax\n\t")
scope.table[var_name]["type"] = var_type
elif type(new_value) == StringNode: # we're changing a variable to a string
lines.append(f"mov QWORD [rsp + {var_pos}], 0\n\t")
string_pointer = self.add_constant(new_value.value)
string_len = self.add_constant(f"equ $ - {string_pointer[1:-1]}", no_string=True)
#lines.append(f"lea QWORD [rsp + {var_pos}], {string_pointer}\n\t")
##lines.append(f"mov QWORD [rsp + {var_pos + 8}], {string_len[1:-1]}\n\t")
scope.table[var_name]["stack_loc"] = self.stack_size
lines.append(f"lea rax, {string_pointer}\n\t")
self.push("rax", lines)
lines.append(f"mov rax, {string_len[1:-1]}\n\t")
self.push("rax", lines)
2025-09-07 07:25:11 +10:00
2025-09-10 20:45:22 +10:00
scope.table[var_name]["type"] = StringNode
2025-09-02 16:26:19 +10:00
2025-09-10 20:45:22 +10:00
elif type(new_value) == BoolNode:
lines.append(f"mov QWORD [rsp + {var_pos}], {'1' if new_value.value else '0'}\n\t")
scope.table[var_name]["type"] = BoolNode
2025-09-07 13:38:16 +10:00
2025-09-10 20:45:22 +10:00
elif type(new_value) == str: # we're changing a variable to the value of a register
lines.append(f"mov QWORD [rsp + {var_pos}], {new_value}\n\t")
scope.table[var_name]["type"] = IntNode
if scope.table[var_name]["type"] != old_var_type:
warning(self.code, f"Changing the type of \"{var_name}\" at runtime is considered bad practice.")
2025-09-07 11:58:01 +10:00
2025-09-10 07:30:58 +10:00
def generate_LabelDecNode(self, node: LabelDecNode, lines):
2025-09-07 11:58:01 +10:00
self.labels.append(node.name)
2025-09-10 07:30:58 +10:00
lines.append("." + node.name + ":\n\t")
def generate_FunctionNode(self, node: FunctionNode, lines):
if node.return_type == None:
traceback(self.code, "SyntaxError", "Functions require a return type.")
if node.name == None:
traceback(self.code, "SyntaxError", "Functions require a name.")
func = self.add_function(node)
self.current_var_scope = func["scope"]
2025-09-10 18:56:45 +10:00
# function boilerplate
2025-09-12 06:28:45 +10:00
self.function_lines.append(node.name + ":\n\t")
2025-09-10 18:56:45 +10:00
self.push("rbp", self.function_lines)
self.function_lines.append("mov rbp, rsp\n\t")
for i, arg in enumerate(node.args):
try:
stack_loc = ["rdi","rsi","rdx","rcx","r8","r9"][i] # ""
except IndexError:
traceback(self.code, "CallError", "Functions with more than 6 args aren't supported yet, sorry...")
self.current_var_scope.define(arg.name, {"stack_loc": stack_loc, "type": self.ground_type_to_node(arg.arg_type), "used": False})
2025-09-10 07:30:58 +10:00
for inst in node.statements:
#self.function_lines.append(f"; {inst}\n\t")
2025-09-13 07:08:11 +10:00
node_type = str(type(inst))[19:-2]
getattr(self, f"generate_{node_type}")(inst, self.function_lines)
#self.generate_InstructionNode(inst, self.function_lines)
2025-09-10 18:56:45 +10:00
2025-09-10 07:30:58 +10:00
def generate_InstructionNode(self, node: InstructionNode, lines = None):
lines = lines or self.lines
### MISC ###
if node.instruction == "end":
2025-09-08 05:54:13 +10:00
self.clamp_instruction_args(node, 1, 1)
2025-09-04 07:45:20 +10:00
if not type(node.arguments[0]) in [IntNode, VarRefNode]: # example: "end true"
traceback(self.code, "TypeError", f"end expects an integer, not {node.arguments[0]}")
2025-09-07 13:38:16 +10:00
if type(node.arguments[0]) in [IntNode,BoolNode]:
2025-09-10 07:30:58 +10:00
lines.append("mov rdi, " + str(node.arguments[0].value) + "\n\t")
2025-09-04 07:45:20 +10:00
elif isinstance(node.arguments[0], VarRefNode):
2025-09-10 07:30:58 +10:00
var_type = self.get_variable(lines, node.arguments[0].var_name, "rdi", no_stack_pop=True)
2025-09-04 07:45:20 +10:00
if var_type == FloatNode:
2025-09-10 07:30:58 +10:00
lines.append("cvttsd2si rdi, xmm0\n\t")
2025-09-04 07:45:20 +10:00
else:
2025-09-07 13:38:16 +10:00
if var_type not in [IntNode,BoolNode]:
2025-09-04 07:45:20 +10:00
traceback(self.code, "TypeError", f"end expects an integer, not \"{var_type}\"")
lines.append("mov rax, 60\n\t")
2025-09-10 07:30:58 +10:00
lines.append("syscall\n\t")
2025-09-02 06:42:58 +10:00
### VARIABLE INSTRUCTIONS ###
2025-09-02 06:42:58 +10:00
elif node.instruction == "set":
2025-09-07 20:00:35 +10:00
self.clamp_instruction_args(node, 2, 2)
2025-09-02 06:42:58 +10:00
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]}\"")
2025-09-07 13:38:16 +10:00
if type(node.arguments[1]) not in [IntNode, VarRefNode, FloatNode, StringNode, BoolNode]:
2025-09-02 06:42:58 +10:00
traceback(self.code, "TypeError", f"variables can't be of type \"{type(node.arguments[1])}\"")
2025-09-10 18:56:45 +10:00
variable_exists = self.current_var_scope.lookup(node.arguments[0].var_name) != None
2025-09-02 16:26:19 +10:00
if not variable_exists: # create a new variable
2025-09-10 07:30:58 +10:00
self.create_variable(lines, node.arguments[0].var_name, node.arguments[1])
2025-09-02 16:26:19 +10:00
else: # modify the existing one
2025-09-10 07:30:58 +10:00
self.change_variable(lines, node.arguments[0].var_name, node.arguments[1])
2025-09-02 16:26:19 +10:00
### MATH INSTRUCTIONS ###
2025-09-07 20:00:35 +10:00
elif node.instruction in ["add", "subtract", "multiply"]:
self.clamp_instruction_args(node, 3, 3)
if type(node.arguments[2]) != VarPointerNode:
traceback(self.code, "TypeError", f"the destination of the {node.instruction} 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"
2025-09-04 07:45:20 +10:00
if isinstance(node.arguments[0], VarRefNode):
2025-09-10 07:30:58 +10:00
number1_type = self.get_variable(lines, node.arguments[0].var_name, "rax")
2025-09-04 07:45:20 +10:00
elif isinstance(node.arguments[0], FloatNode) or isinstance(node.arguments[1], FloatNode):
number1_type = FloatNode
constant_name = self.add_constant(node.arguments[0].value)
2025-09-10 07:30:58 +10:00
lines.append(f"movsd xmm0, {constant_name}\n\t")
2025-09-04 07:45:20 +10:00
elif isinstance(node.arguments[0], IntNode):
number1_type = IntNode
#arg1 = node.arguments[0].value
2025-09-10 07:30:58 +10:00
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 add, got {node.arguments[0]}")
2025-09-04 07:45:20 +10:00
if isinstance(node.arguments[1], VarRefNode):
2025-09-10 07:30:58 +10:00
number2_type = self.get_variable(lines, node.arguments[1].var_name, "rbx")
2025-09-04 07:45:20 +10:00
elif number1_type == FloatNode or isinstance(node.arguments[1], FloatNode):
number2_type = FloatNode
constant_name = self.add_constant(node.arguments[1].value)
2025-09-10 07:30:58 +10:00
lines.append(f"movsd xmm1, {constant_name}\n\t")
2025-09-04 07:45:20 +10:00
elif isinstance(node.arguments[1], IntNode):
number2_type = IntNode
arg2 = node.arguments[1].value
2025-09-10 07:30:58 +10:00
#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 add, 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...
2025-09-04 07:45:20 +10:00
if number1_type not in [IntNode, FloatNode] or number2_type not in [IntNode, FloatNode]:
2025-09-07 20:00:35 +10:00
traceback(self.code, "TypeError", f"Unsupported operation \"f{node.instruction}\" for \"{node.arguments[0]}\" and \"{node.arguments[1]}\".")
2025-09-04 07:45:20 +10:00
if number1_type == IntNode and number2_type == IntNode:
2025-09-07 20:00:35 +10:00
if node.instruction == "add":
2025-09-10 07:30:58 +10:00
lines.append(f"add rax, {arg2}\n\t")
2025-09-07 20:00:35 +10:00
elif node.instruction == "subtract":
2025-09-10 07:30:58 +10:00
lines.append(f"sub rax, {arg2}\n\t")
2025-09-07 20:00:35 +10:00
elif node.instruction == "multiply":
2025-09-10 07:30:58 +10:00
lines.append(f"imul rax, {arg2}\n\t")
2025-09-04 07:45:20 +10:00
else:
2025-09-07 20:00:35 +10:00
if node.instruction == "add":
2025-09-10 07:30:58 +10:00
lines.append(f"addsd xmm0, xmm1\n\t")
2025-09-07 20:00:35 +10:00
elif node.instruction == "subtract":
2025-09-10 07:30:58 +10:00
lines.append(f"subsd xmm0, xmm1\n\t")
2025-09-07 20:00:35 +10:00
elif node.instruction == "multiply":
2025-09-10 07:30:58 +10:00
lines.append(f"mulsd xmm0, xmm1\n\t")
2025-09-04 07:45:20 +10:00
is_integer = number1_type == IntNode and number2_type == IntNode
starting_reg = "rax" if is_integer else "xmm0"
2025-09-10 18:56:45 +10:00
if self.current_var_scope.lookup(node.arguments[2].var_name) == None: # we need to create a variable for the destination
2025-09-10 07:30:58 +10:00
self.create_variable(lines, node.arguments[2].var_name, starting_reg, IntNode if is_integer else FloatNode)
else:
2025-09-10 07:30:58 +10:00
self.change_variable(lines, node.arguments[2].var_name, starting_reg)
elif node.instruction == "divide":
2025-09-08 05:54:13 +10:00
self.clamp_instruction_args(node, 3, 3)
if 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):
2025-09-10 07:30:58 +10:00
number1_type = self.get_variable(lines, 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)
2025-09-10 07:30:58 +10:00
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):
2025-09-10 07:30:58 +10:00
number2_type = self.get_variable(lines, 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)
2025-09-10 07:30:58 +10:00
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]}\".")
2025-09-10 07:30:58 +10:00
lines.append(f"divsd xmm0, xmm1\n\t")
2025-09-10 18:56:45 +10:00
if self.current_var_scope.lookup(node.arguments[2].var_name) == None: # we need to create a variable for the destination
2025-09-10 07:30:58 +10:00
self.create_variable(lines, node.arguments[2].var_name, "xmm0", FloatNode)
else:
2025-09-10 07:30:58 +10:00
self.change_variable(lines, node.arguments[2].var_name, "xmm0")
elif node.instruction == "stdout":
2025-09-08 05:54:13 +10:00
self.clamp_instruction_args(node, 1, 1)
arg = node.arguments[0]
printed_value = arg.__str__()
if isinstance(arg, VarRefNode):
2025-09-10 07:30:58 +10:00
var_type = self.get_variable(lines, arg.var_name, "rsi", offset=0, no_stack_pop=True)
2025-09-07 07:25:11 +10:00
if var_type == StringNode:
2025-09-10 07:30:58 +10:00
self.get_variable(lines, arg.var_name, "rdx", offset=-8, no_stack_pop=True)
2025-09-07 07:25:11 +10:00
else:
traceback(self.code, "TypeError", f"You can't print \"{var_type(None).__repr__()}\", try converting it to a string first.")
else:
string_pointer = self.add_constant(printed_value)[1:-1]
string_len = self.add_constant(f"equ $ - {string_pointer}", True)[1:-1]
2025-09-10 07:30:58 +10:00
lines.append(f"mov rsi, {string_pointer}\n\t")
lines.append(f"mov rdx, {string_len}\n\t") # length
self.push("rdi", lines) # save rdi
2025-09-10 07:30:58 +10:00
lines.append("mov rax, 1\n\t") # sys_write syscall
lines.append("mov rdi, 1\n\t") # a file descriptor of 1 is stdout
lines.append("syscall\n\t")
self.pop("rdi", lines) # restore rdi
2025-09-13 15:48:41 +10:00
elif node.instruction == "stdin":
self.clamp_instruction_args(node, 1, 1)
arg = node.arguments[0]
if not isinstance(arg, VarPointerNode):
traceback(self.code, "TypeError", f"Argument 1 of stdin has to be a variable pointer, not \"{arg}\"")
buf_size = 255
buffer = self.add_buffer(buf_size)
lines.append("mov rax, 0\n\t") # sys_read syscall
lines.append("mov rdi, 0\n\t") # a file descriptor of 0 is stdin
lines.append(f"mov rsi, {buffer}\n\t")
lines.append(f"mov rdx, {buf_size}")
lines.append("syscall")
if not self.current_var_scope.table.get(arg.var_name):
self.create_variable(lines, arg.var_name, buffer, StringNode)
else:
self.change_variable(lines, arg.var_name, buffer)
2025-09-07 11:58:01 +10:00
elif node.instruction == "jump":
2025-09-08 05:54:13 +10:00
self.clamp_instruction_args(node, 1, 1)
if not isinstance(node.arguments[0], LabelRefNode):
2025-09-07 11:58:01 +10:00
traceback(self.code, "TypeError", f"jump expects a label reference as the first argument, not \"{node.arguments[0]}\"")
2025-09-10 07:30:58 +10:00
lines.append(f"jmp .{node.arguments[0].name}\n\t")
2025-09-07 13:38:16 +10:00
elif node.instruction == "if":
2025-09-08 05:54:13 +10:00
self.clamp_instruction_args(node, 2, 2)
if not type(node.arguments[0]) in [VarRefNode,BoolNode,StringNode,FloatNode,IntNode]:
2025-09-07 13:38:16 +10:00
traceback(self.code, "TypeError", f"if expects a value or variable refernce as the first argument, not \"{node.arguments[0]}\"")
elif not isinstance(node.arguments[1], LabelRefNode):
traceback(self.code, "TypeError", f"if expects a label reference as the second argument, not \"{node.arguments[1]}\"")
if isinstance(node.arguments[0], BoolNode):
if node.arguments[0].value:
2025-09-10 07:30:58 +10:00
lines.append(f"jmp .{node.arguments[1].name}\n\t")
#lines.append("mov eax, 1")
#lines.append(f"cmp eax, {1 if node.arguments[0].value else 0}")
2025-09-07 13:38:16 +10:00
elif type(node.arguments[0]) in [IntNode,FloatNode]:
if node.arguments[0].value != 0:
2025-09-10 07:30:58 +10:00
lines.append(f"jmp .{node.arguments[1].name}\n\t")
2025-09-07 13:38:16 +10:00
elif isinstance(node.arguments[0], VarRefNode):
2025-09-13 07:08:11 +10:00
self.current_var_scope.table[node.arguments[0].var_name]["used"] = True # putting a variable in an if statement means we're accessing it
2025-09-10 07:30:58 +10:00
self.get_variable(lines, node.arguments[0].var_name, "eax")
lines.append(f"test eax, eax\n\t")
lines.append(f"jnz .{node.arguments[1].name}\n\t")
elif node.instruction in ["equal", "inequal", "greater", "lesser"]:
2025-09-08 05:54:13 +10:00
self.clamp_instruction_args(node, 3, 3)
if not type(node.arguments[0]) in [VarRefNode,BoolNode,FloatNode,IntNode]:
traceback(self.code, "TypeError", f"{node.instruction} expects a value or variable refernce as the first argument, not \"{node.arguments[0].__repr__()}\"")
2025-09-07 13:38:16 +10:00
elif not type(node.arguments[1]) in [VarRefNode,BoolNode,FloatNode,IntNode]:
2025-09-08 05:54:13 +10:00
traceback(self.code, "TypeError", f"{node.instruction} expects a value or variable refernce as the second argument, not \"{node.arguments[1].__repr__()}\"")
2025-09-07 13:38:16 +10:00
elif not isinstance(node.arguments[2], VarPointerNode):
2025-09-08 05:54:13 +10:00
traceback(self.code, "TypeError", f"the third argument of {node.instruction} should be a variable pointer, not \"{node.arguments[2].__repr__()}\"")
2025-09-07 13:38:16 +10:00
arg1 = None
arg2 = None
if isinstance(node.arguments[0], BoolNode):
2025-09-10 07:30:58 +10:00
lines.append(f"mov rax, {int(node.arguments[0].value)}\n\t")
2025-09-07 13:38:16 +10:00
arg1 = "rax"
elif isinstance(node.arguments[0], IntNode):
2025-09-10 07:30:58 +10:00
lines.append(f"mov rax, {node.arguments[0].value}\n\t")
2025-09-07 13:38:16 +10:00
arg1 = "rax"
elif isinstance(node.arguments[0], FloatNode):
const_name = self.add_constant(node.arguments[0].value)
2025-09-10 07:30:58 +10:00
lines.append(f"movsd xmm0, {const_name}\n\t")
2025-09-07 13:38:16 +10:00
arg1 = "xmm0"
elif isinstance(node.arguments[0], VarRefNode):
2025-09-10 07:30:58 +10:00
self.get_variable(lines, node.arguments[0].var_name, "rax")
2025-09-07 13:38:16 +10:00
arg1 = "rax"
if isinstance(node.arguments[1], BoolNode):
2025-09-10 07:30:58 +10:00
lines.append(f"mov rbx, {int(node.arguments[1].value)}\n\t")
2025-09-07 13:38:16 +10:00
arg2 = "rbx"
elif isinstance(node.arguments[1], IntNode):
2025-09-10 07:30:58 +10:00
lines.append(f"mov rbx, {node.arguments[1].value}\n\t")
2025-09-07 13:38:16 +10:00
arg2 = "rbx"
elif isinstance(node.arguments[1], FloatNode):
const_name = self.add_constant(node.arguments[1].value)
2025-09-10 07:30:58 +10:00
lines.append(f"movsd xmm1, {const_name}\n\t")
2025-09-07 13:38:16 +10:00
arg2 = "xmm1"
elif isinstance(node.arguments[1], VarRefNode):
2025-09-10 07:30:58 +10:00
self.get_variable(lines, node.arguments[1].var_name, "rbx")
2025-09-07 13:38:16 +10:00
arg2 = "rbx"
2025-09-10 07:30:58 +10:00
lines.append(f"cmp {arg1}, {arg2}\n\t")
2025-09-07 13:38:16 +10:00
instructions = {
"equal": "sete",
"inequal": "setne",
"greater": "setg",
"lesser": "setl"
}
2025-09-10 07:30:58 +10:00
lines.append(f"{instructions[node.instruction]} al\n\t")
2025-09-13 07:08:11 +10:00
lines.append("movzx eax, al\n\t")
2025-09-07 13:38:16 +10:00
var_name = node.arguments[2].var_name
2025-09-10 18:56:45 +10:00
if self.current_var_scope.lookup(var_name) == None:
2025-09-10 07:30:58 +10:00
self.create_variable(lines, var_name, "rax", BoolNode)
2025-09-07 13:38:16 +10:00
else:
2025-09-10 07:30:58 +10:00
self.change_variable(lines, var_name, "rax")
elif node.instruction == "not":
self.clamp_instruction_args(node, 2, 2)
if type(node.arguments[0]) not in [VarRefNode,BoolNode]:
traceback(self.code, "TypeError", f"Argument 1 of \"not\" must be a boolean or variable reference, not \"{node.arguments[0]}\"")
elif not isinstance(node.arguments[1], VarPointerNode):
traceback(self.code, "TypeError", f"Argument 2 of \"not\" must be a variable pointer, not \"{node.arguments[1]}\"")
result = ""
if isinstance(node.arguments[0], BoolNode):
result = BoolNode(not node.arguments[0].value)
elif isinstance(node.arguments[0], VarRefNode):
var_type = self.get_variable(lines, node.arguments[0].var_name, "rax")
if var_type != BoolNode:
traceback(self.code, "TypeError", f"Argument 1 of \"not\" must be a boolean or variable reference, not \"{var_type(0).__repr__()}\"")
result = "rax"
lines.append("xor rax, 1\n\t")
if self.current_var_scope.table.get(node.arguments[1].var_name):
self.change_variable(lines, node.arguments[1].var_name, result)
else:
self.create_variable(lines, node.arguments[1].var_name, result, BoolNode)
2025-09-10 07:30:58 +10:00
elif node.instruction == "endfun":
return
elif node.instruction == "return":
self.clamp_instruction_args(node, 0, 1)
func = self.functions.get(node.parent.name) # the return statement will be a parent of a function node, so we can get the current scope using that
2025-09-12 06:28:45 +10:00
func_scope: SymbolTable = func["scope"]
if len(node.arguments) == 1:
if isinstance(node.arguments[0], IntNode):
lines.append(f"mov rax, {node.arguments[0].value}")
elif isinstance(node.arguments[0], BoolNode):
2025-09-13 07:08:11 +10:00
lines.append(f"mov eax, {int(node.arguments[0].value)}")
elif isinstance(node.arguments[0], FloatNode):
lines.append(f"mov xmm0, {node.arguments[0].value}")
2025-09-10 18:56:45 +10:00
#self.get_variable(lines, node.arguments[0].var_name, "rax")
2025-09-12 06:28:45 +10:00
elif isinstance(node.arguments[0], StringNode):
string_pointer = self.add_constant(node.arguments[0].value)
length = self.add_constant(f"equ $ - {string_pointer[1:-1]}", no_string=True)
lines.append(f"lea rax, [rel {string_pointer[1:-1]}]\n\t")
lines.append(f"mov rdx, {length[1:-1]}\n\t")
2025-09-12 06:28:45 +10:00
#self.push("rax", lines)
elif isinstance(node.arguments[0], VarRefNode):
2025-09-12 06:28:45 +10:00
var = func_scope.lookup(node.arguments[0].var_name)
if var["type"] == StringNode:
self.get_variable(self.function_lines, node.arguments[0].var_name, "rax", offset=0, scope=func["scope"])
self.get_variable(self.function_lines, node.arguments[0].var_name, "rdx", offset=-8, scope=func["scope"])
2025-09-13 07:08:11 +10:00
elif var["type"] == BoolNode:
self.get_variable(self.function_lines, node.arguments[0].var_name, "eax", scope=func["scope"])
2025-09-12 06:28:45 +10:00
else:
self.get_variable(self.function_lines, node.arguments[0].var_name, "rax", scope=func["scope"])
else:
lines.append("mov rax, 0\n\t")
size = 0
for name, var in self.current_var_scope.table.items():
if type(var["stack_loc"]) == str:
continue # its in a register, we dont need to free up the stack
if var["type"] == StringNode:
size += 2
else:
size += 1
lines.append(f"add rsp, {8 * (size)}\n\t")
self.stack_size -= size - 1
lines.append("pop rbp")
2025-09-10 07:30:58 +10:00
lines.append("ret\n\t")
# warn about unused variables
for name, var in self.current_var_scope.table.items():
if not var["used"]:
warning(self.code, f"\"{name}\" was defined but never used.")
2025-09-10 18:56:45 +10:00
old_scope = self.current_var_scope
self.current_var_scope = self.current_var_scope.parent
del old_scope
2025-09-10 07:30:58 +10:00
elif node.instruction == "call":
self.clamp_instruction_args(node, 1, 2)
2025-09-10 07:30:58 +10:00
if not isinstance(node.arguments[0], FunctionCallNode):
2025-09-12 06:28:45 +10:00
traceback(self.code, "TypeError", "Argument 2 of call needs to be a function reference.")
2025-09-10 07:30:58 +10:00
2025-09-10 18:56:45 +10:00
func = self.functions.get(node.arguments[0].func_name, None)
if not func:
traceback(self.code, "TypeError", f"Function \"{node.arguments[0].func_name}\" is not defined.")
func_scope: SymbolTable = func["scope"]
func: FunctionNode = func["func"]
2025-09-10 18:56:45 +10:00
if len(self.arg_list) != len(func.args):
traceback(self.code, "TypeError", f"Function \"{node.arguments[0].func_name}\" takes {len(func.args)} arguments, but got {len(self.arg_list)}")
# stack alignment
if self.stack_size % 2 != 0:
2025-09-10 18:56:45 +10:00
lines.append("sub rsp, 8\n\t") # align the stack to 16 bytes
2025-09-12 06:28:45 +10:00
self.stack_size -= 1
2025-09-10 18:56:45 +10:00
for i, arg in enumerate(self.arg_list):
#self.create_variable(lines, func.args[i].name, arg, func.args[i].arg_type)
value = ""
var_type = ""
2025-09-10 18:56:45 +10:00
if isinstance(arg, IntNode):
value = arg.value
var_type = IntNode
2025-09-10 18:56:45 +10:00
elif isinstance(arg, StringNode):
2025-09-10 18:56:45 +10:00
value = self.add_constant(arg.value)
else:
traceback(self.code, "CallError", f"Can't pass {arg} to function.")
arg_name = func.args[i].name
#self.change_variable(lines, arg_name, value, func_scope)
2025-09-10 18:56:45 +10:00
if i == 0:
#func_scope.define(arg_name, {"stack_loc": "rdi", "type": type(arg)})
2025-09-10 18:56:45 +10:00
lines.append(f"mov rdi, {value}")
elif i == 1:
lines.append(f"mov rsi, {value}")
elif i == 2:
lines.append(f"mov rdx, {value}")
elif i == 3:
lines.append(f"mov rcx, {value}")
elif i == 4:
lines.append(f"mov r8, {value}")
elif i == 5:
lines.append(f"mov r9, {value}")
else:
traceback(self.code, "CallError", "Functions with more than 6 args aren't supported yet, sorry...")
2025-09-10 07:30:58 +10:00
lines.append(f"call {node.arguments[0].func_name}\n\t")
2025-09-10 18:56:45 +10:00
#if len(self.arg_list) > 0:
# lines.append(f"add rsp, {len(self.arg_list) * 8}")
2025-09-12 06:28:45 +10:00
self.stack_size += len(self.arg_list)
2025-09-10 18:56:45 +10:00
self.arg_list.clear()
if len(node.arguments) == 2:
2025-09-10 18:56:45 +10:00
if not isinstance(node.arguments[1], VarPointerNode):
traceback(self.code, "TypeError", "Argument 1 of call needs to be a variable pointer.")
if self.current_var_scope.lookup(node.arguments[1].var_name):
self.change_variable(lines, node.arguments[1].var_name, "rax")
else:
self.current_var_scope.define(node.arguments[1].var_name, {"stack_loc": "rax", "type": self.ground_type_to_node(func.return_type), "used": False})
2025-09-12 06:28:45 +10:00
#self.create_variable(lines, node.arguments[1].var_name, "rax", self.ground_type_to_node(self.functions.get(node.arguments[0].func_name)["func"].return_type))
2025-09-10 18:56:45 +10:00
elif node.instruction == "pusharg":
self.clamp_instruction_args(node, 1, 1)
#if type(node.arguments[0]) not in [IntNode]:
# traceback(self.code, "TypeError", f"A {node.arguments[0]} can't be passed as an argument.")
2025-09-07 13:38:16 +10:00
2025-09-10 18:56:45 +10:00
self.arg_list.append(node.arguments[0])
2025-09-02 06:42:58 +10:00
else:
2025-09-09 17:47:43 +10:00
raise NotImplementedError(f"A generate method hasn't been made for the \"{node.instruction}\" instruction.")
2025-09-04 07:45:20 +10:00
2025-09-07 13:38:16 +10:00
2025-09-04 07:45:20 +10:00
def write(self):
with open(self.output_path + ".asm", "w") as f:
2025-09-06 21:18:22 +10:00
f.write("; ~~~ Auto generated by the GroundPY compiler for Linux x86_64 targets. ~~~\n\n")
2025-09-04 07:45:20 +10:00
f.write("section .data\n")
for name, const in self.constants.items():
value = const["value"]
2025-09-13 07:08:11 +10:00
f.write(name + ": ")
2025-09-04 07:45:20 +10:00
value_type = type(value)
if value_type == str:
if not const["no_string"]:
2025-09-10 07:30:58 +10:00
final = f'db "' + value.replace("\\n", "\", 10, \"") + "\", 0"
final = final.replace(", \"\", ", ", ")
f.write(final)
else:
f.write(value)
2025-09-04 07:45:20 +10:00
elif value_type == float or value_type == int:
f.write(f"dq {float(value)}")
f.write("\n")
2025-09-13 15:48:41 +10:00
f.write("section .bss\n")
for buf, size in self.buffers.items():
f.write(f"{buf} resb {size}\n")
2025-09-07 11:58:01 +10:00
f.write("section .text\n")
2025-09-10 07:30:58 +10:00
optimizer = X86_64Optimizer(self.lines + self.function_lines)
2025-09-09 07:20:50 +10:00
f.writelines(optimizer.optimize_until_stable())