from-scratch/lang/libcompiler/src/lib.rs
2025-11-13 14:39:45 +01:00

631 lines
21 KiB
Rust

#![feature(debug_closure_helpers, box_as_ptr, allocator_api)]
#[cfg(test)]
mod tests;
pub mod ffi {
#![allow(
non_camel_case_types,
dead_code,
non_upper_case_globals,
improper_ctypes
)]
include!(concat!(env!("OUT_DIR"), "/bindings.rs"));
}
#[repr(C)]
#[derive(Debug, PartialEq, Eq, Copy, Clone)]
pub struct FFISlice {
pub ptr: *const u8,
pub len: usize,
}
#[repr(transparent)]
#[derive(Debug, PartialEq, Eq, Copy, Clone)]
pub struct MaybeFFISlice {
inner: FFISlice,
}
impl MaybeFFISlice {
pub fn is_none(&self) -> bool {
self.inner.ptr.is_null()
}
pub fn into_option(self) -> Option<FFISlice> {
if self.is_none() {
None
} else {
Some(self.inner)
}
}
}
impl FFISlice {
/// # Safety
/// The caller must ensure that the slice is valid for type T, and lasts for 'a.
pub unsafe fn as_slice_unchecked<'a, T: Sized>(self) -> &'a [T] {
// SAFETY: The caller ensures that the FFISlice is valid for type T.
unsafe { core::slice::from_raw_parts(self.ptr.cast(), self.len) }
}
/// # Safety
/// The caller ensures that the slice is valid byte slice.
/// Namely, the pointer must be well-aligned and point to `len` bytes, and
/// must last for at least 'a.
pub fn as_u8s_unchecked<'a>(self) -> &'a [u8] {
// SAFETY: The FFISlice is guaranteed to be a valid byte slice.
unsafe { self.as_slice_unchecked() }
}
/// # Safety
/// The caller must ensure that the slice is a valid utf8 string.
/// Furthermore, the pointer must be well-aligned, point to `len` bytes, and
/// must last for at least 'a.
pub unsafe fn as_str_unchecked<'a>(self) -> &'a str {
// SAFETY: The caller ensures that the FFISlice is a valid utf8 string.
unsafe { core::str::from_utf8_unchecked(self.as_u8s_unchecked()) }
}
}
pub mod vec {
#![allow(dead_code)]
impl Default for BlobVec {
fn default() -> Self {
Self {
data: core::ptr::null_mut(),
len: 0,
cap: 0,
elem_size: 0,
drop: None,
}
}
}
unsafe impl Send for BlobVec {}
unsafe impl Sync for BlobVec {}
use super::ffi::*;
#[repr(transparent)]
#[derive(Debug)]
pub struct Vec<T> {
pub vec: BlobVec,
_marker: core::marker::PhantomData<T>,
}
impl<T> Default for Vec<T> {
fn default() -> Self {
Self::new()
}
}
impl<T> Drop for Vec<T> {
fn drop(&mut self) {
// SAFETY: The vec is valid and owned by self.
unsafe {
vec_drop(&mut self.vec);
}
}
}
impl<T> Vec<T> {
pub fn new() -> Self {
Self::new_with(32)
}
pub fn new_with(capacity: usize) -> Self {
let mut vec = BlobVec {
data: core::ptr::null_mut(),
len: 0,
cap: 0,
elem_size: 0,
drop: None,
};
unsafe extern "C" fn drop_fn<T>(ptr: *mut ()) {
unsafe {
core::ptr::drop_in_place::<T>(ptr as *mut T);
}
}
unsafe {
vec_init_with(
&mut vec,
core::mem::size_of::<T>(),
Some(drop_fn::<T>),
capacity,
);
}
Self {
vec,
_marker: core::marker::PhantomData,
}
}
pub fn as_slice(&self) -> &[T] {
assert_eq!(self.vec.elem_size, core::mem::size_of::<T>());
unsafe { core::slice::from_raw_parts(self.vec.data as *const T, self.vec.len) }
}
pub fn as_slice_mut(&mut self) -> &mut [T] {
assert_eq!(self.vec.elem_size, core::mem::size_of::<T>());
unsafe { core::slice::from_raw_parts_mut(self.vec.data as *mut T, self.vec.len) }
}
pub fn extend(&mut self, elements: Box<[T]>) {
unsafe {
let elements =
core::mem::transmute::<Box<[T]>, Box<[core::mem::ManuallyDrop<T>]>>(elements);
vec_extend(&mut self.vec, elements.as_ptr() as *const _, elements.len());
}
}
pub fn push(&mut self, value: T) {
let value = core::mem::ManuallyDrop::new(value);
unsafe {
vec_push(&mut self.vec, &raw const value as *const T as *const _);
}
}
pub fn insert(&mut self, value: T, index: usize) {
if index > self.vec.len {
return;
}
let value = core::mem::ManuallyDrop::new(value);
unsafe {
vec_insert(
&mut self.vec,
index,
&raw const value as *const T as *const _,
);
}
}
pub fn insert_many(&mut self, index: usize, elements: Box<[T]>) {
unsafe {
let elements =
core::mem::transmute::<Box<[T]>, Box<[core::mem::ManuallyDrop<T>]>>(elements);
vec_insert_many(
&mut self.vec,
index,
elements.as_ptr() as *const _,
elements.len(),
);
}
}
pub fn pop(&mut self) -> Option<T> {
if self.vec.len == 0 {
return None;
}
unsafe {
let ptr = vec_get(&mut self.vec, self.vec.len - 1) as *mut T;
let value = ptr.read();
vec_pop(&mut self.vec);
Some(value)
}
}
pub fn get(&self, index: usize) -> Option<&T> {
if index >= self.vec.len {
return None;
}
unsafe {
let ptr = vec_get(&raw const self.vec as *mut _, index) as *mut T;
Some(&*ptr)
}
}
pub fn get_mut(&mut self, index: usize) -> Option<&mut T> {
if index >= self.vec.len {
return None;
}
unsafe {
let ptr = vec_get(&raw mut self.vec, index) as *mut T;
Some(&mut *ptr)
}
}
pub fn remove(&mut self, index: usize) {
if index >= self.vec.len {
return;
}
unsafe {
vec_remove(&mut self.vec, index);
}
}
pub fn len(&self) -> usize {
self.vec.len
}
pub fn is_empty(&self) -> bool {
self.vec.len == 0
}
pub fn position<F>(&self, elem: &T, mut cmp: F) -> Option<usize>
where
F: FnMut(&T, &T) -> bool,
{
extern "C" fn cmp_trampoline<T, F: FnMut(&T, &T) -> bool>(
f: *const (),
a: *const (),
b: *const (),
) -> bool {
let f = unsafe { &mut *(f as *mut F) };
let a = unsafe { &*(a as *const T) };
let b = unsafe { &*(b as *const T) };
f(a, b)
}
unsafe {
let index = vec_find(
&raw const self.vec as *mut _,
elem as *const T as *const _,
cmp_trampoline::<T, F>,
&raw mut cmp as *mut (),
);
if index == usize::MAX {
None
} else {
Some(index)
}
}
}
pub fn binary_search_by<F>(&self, elem: &T, mut cmp: F) -> Result<usize, usize>
where
F: FnMut(&T, &T) -> i32,
{
extern "C" fn cmp_trampoline<T, F: FnMut(&T, &T) -> i32>(
f: *const (),
a: *const (),
b: *const (),
) -> i32 {
let f = unsafe { &mut *(f as *mut F) };
let a = unsafe { &*(a as *const T) };
let b = unsafe { &*(b as *const T) };
f(a, b)
}
unsafe {
let (index, vacant) = vec_binary_search_by(
&raw const self.vec as *mut _,
elem as *const T as *const _,
cmp_trampoline::<T, F>,
&raw mut cmp as *mut (),
);
if vacant { Err(index) } else { Ok(index) }
}
}
pub fn insert_sorted<F>(&mut self, elem: T, mut cmp: F) -> Result<usize, usize>
where
F: FnMut(&T, &T) -> i32,
{
extern "C" fn cmp_trampoline<T, F: FnMut(&T, &T) -> i32>(
f: *const (),
a: *const (),
b: *const (),
) -> i32 {
let f = unsafe { &mut *(f as *mut F) };
let a = unsafe { &*(a as *const T) };
let b = unsafe { &*(b as *const T) };
f(a, b)
}
let mut elem = core::mem::ManuallyDrop::new(elem);
unsafe {
let (index, _inserted) = vec_insert_sorted(
&raw const self.vec as *mut _,
&raw mut elem as *const _,
cmp_trampoline::<T, F>,
&raw mut cmp as *mut (),
);
Ok(index)
}
}
}
}
mod display {
use super::ffi::{self, Ast, AstNode};
impl core::fmt::Display for AstNode {
fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
use crate::ffi::{
AST_ADDRESS_OF, AST_ARG, AST_AS, AST_ASSIGNMENT, AST_BINARY_OP, AST_BLOCK,
AST_CALL, AST_DEREF, AST_FUNCTION, AST_IF, AST_NUMBER, AST_PLACE_TO_VALUE,
AST_RETURN_STATEMENT, AST_VALUE_TO_PLACE, AST_VAR_DECL, AST_VAR_REF, BinaryExpr,
};
match self.kind {
AST_NUMBER => {
write!(f, "Number({})", self.data as usize)
}
AST_DEREF => {
write!(f, "Deref(expr: {})", self.data as usize)
}
AST_ADDRESS_OF => {
write!(f, "AddressOf(expr: {})", self.data as usize)
}
AST_CALL => {
let call = unsafe { self.data.cast::<crate::ffi::AstCallExpr>().read() };
write!(f, "Call(callee: {}, params: {:?})", call.callee, unsafe {
std::slice::from_raw_parts(call.params.cast::<u64>(), call.params_len)
},)
}
AST_AS => {
let as_expr = unsafe { self.data.cast::<crate::ffi::AstAsExpr>().read() };
write!(f, "As(expr: {}, target_type: {})", as_expr.expr, as_expr.ty)
}
AST_ARG => {
let arg = unsafe { self.data.cast::<crate::ffi::AstArgument>().read() };
write!(
f,
"Arg(name: {:?}, arg_type: {})",
unsafe {
std::str::from_utf8(std::slice::from_raw_parts(arg.name, arg.name_len))
},
arg.arg_type,
)
}
AST_VAR_REF => {
let var_ref = unsafe { self.data.cast::<crate::ffi::AstVarRef>().read() };
if var_ref.resolved != u64::MAX {
write!(f, "VarRef({})", var_ref.resolved)
} else {
write!(f, "VarRef(name: {:?})", unsafe {
std::str::from_utf8(std::slice::from_raw_parts(
var_ref.name,
var_ref.name_len,
))
},)
}
}
AST_VAR_DECL => {
let var_decl = unsafe { self.data.cast::<crate::ffi::AstVarDecl>().read() };
write!(
f,
"VarDecl(name: {:?}, var_type: {})",
unsafe {
std::str::from_utf8(std::slice::from_raw_parts(
var_decl.name,
var_decl.name_len,
))
},
var_decl.var_type,
)
}
AST_ASSIGNMENT => {
write!(
f,
"Assignment(dest: {}, src: {})",
self.data as usize, self.extra
)
}
AST_BINARY_OP => {
let BinaryExpr {
left,
operator,
right,
} = unsafe { self.data.cast::<crate::ffi::BinaryExpr>().read() };
write!(
f,
"BinaryOp(op: {}, left: {}, right: {})",
operator, left, right
)
}
AST_RETURN_STATEMENT => {
let return_expr_id = self.data as usize;
write!(f, "ReturnStatement(expr: {})", return_expr_id)
}
AST_FUNCTION => {
let func = unsafe { self.data.cast::<crate::ffi::AstFunction>().read() };
write!(
f,
"Function(name: {:?}, args: {:?}, return_type: {}, body: {})",
unsafe {
std::str::from_utf8(std::slice::from_raw_parts(
func.name,
func.name_len,
))
},
unsafe {
std::slice::from_raw_parts(func.args.cast::<u64>(), func.args_len)
},
func.return_type,
func.body
)
}
AST_BLOCK => {
write!(f, "Block(statements: {:?})", unsafe {
std::slice::from_raw_parts(self.data.cast::<u64>(), self.extra)
})
}
AST_IF => {
let if_node = unsafe { self.data.cast::<crate::ffi::AstIfExpr>().read() };
write!(
f,
"If(cond: {}, then_branch: {}, else_branch: {:?})",
if_node.condition,
if_node.then,
match if_node.else_ {
u64::MAX => None,
v => Some(v),
}
)
}
AST_PLACE_TO_VALUE => {
write!(f, "PlaceToValue(place: {})", self.data as usize)
}
AST_VALUE_TO_PLACE => {
write!(f, "ValueToPlace(value: {})", self.data as usize)
}
kind => write!(f, "UnknownNode(kind: {kind})"),
}
}
}
impl core::fmt::Display for Ast {
fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
writeln!(f, "[")?;
for (i, item) in self.nodes.as_slice().iter().enumerate() {
if i > 0 {
writeln!(f, ", ")?;
}
write!(f, "\t{i}: {}", item)?;
}
write!(f, "\n]")
}
}
impl core::fmt::Display for ffi::SymEntry {
fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
f.debug_struct("SymEntry")
.field_with("key", |f| {
f.debug_struct("Key")
.field_with("kind", |f| {
f.write_str(match self.key.kind {
ffi::SYM_KEY_SCOPE => "Scope",
ffi::SYM_KEY_SCOPE_NAME => "ScopeName",
ffi::SYM_KEY_PARENT_SCOPE => "ParentScope",
ffi::SYM_KEY_ARG => "Argument",
ffi::SYM_KEY_VAR => "Variable",
ffi::SYM_KEY_FUNCTION => "Function",
_ => "Unknown",
})
})
.field("scope", &self.key.scope_index)
.field("span", &self.key.span)
.field_with("ident", |f| {
f.write_str(unsafe {
core::str::from_utf8_unchecked(core::slice::from_raw_parts(
self.key.ident,
self.key.ident_len,
))
})
})
.finish()
})
.field_with("value", |f| {
let stct = &mut f.debug_struct("Value");
if self.extra == 0 {
stct.field("ast_index", &self.index).finish()
} else if self.index != 0 {
stct.field_with("ident", |f| {
f.write_str(unsafe {
core::str::from_utf8_unchecked(core::slice::from_raw_parts(
self.index as *const u8,
self.extra as usize,
))
})
})
.finish()
} else {
stct.field("index", &self.index)
.field("extra", &self.extra)
.finish()
}
})
.finish()
}
}
impl core::fmt::Display for ffi::Type {
fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
use ffi::{
TYPE_BOOL, TYPE_F32, TYPE_F64, TYPE_I8, TYPE_I16, TYPE_I32, TYPE_I64, TYPE_ISIZE,
TYPE_POINTER, TYPE_STR, TYPE_U8, TYPE_U16, TYPE_U32, TYPE_U64, TYPE_USIZE,
TYPE_VOID,
};
match self.kind {
TYPE_VOID => {
write!(f, "void")
}
TYPE_BOOL => {
write!(f, "bool")
}
TYPE_I32 => {
write!(f, "i32")
}
TYPE_U32 => {
write!(f, "u32")
}
TYPE_STR => {
write!(f, "str")
}
TYPE_POINTER => {
let pointee = unsafe { (self.data as *const ffi::Type).read() };
write!(f, "*{pointee}",)
}
TYPE_I8 => {
write!(f, "i8")
}
TYPE_U8 => {
write!(f, "u8")
}
TYPE_I16 => {
write!(f, "i16")
}
TYPE_U16 => {
write!(f, "u16")
}
TYPE_I64 => {
write!(f, "i64")
}
TYPE_U64 => {
write!(f, "u64")
}
TYPE_F32 => {
write!(f, "f32")
}
TYPE_F64 => {
write!(f, "f64")
}
TYPE_USIZE => {
write!(f, "usize")
}
TYPE_ISIZE => {
write!(f, "isize")
}
_ => {
write!(f, "UnknownType")
}
}
}
}
#[repr(transparent)]
pub struct Displayed<T: core::fmt::Display>(pub T);
impl<T: core::fmt::Display> core::fmt::Debug for Displayed<T> {
fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
write!(f, "{}", self.0)
}
}
pub trait DisplayedSliceExt {
type Displayed: core::fmt::Debug;
fn displayed(self) -> Self::Displayed;
}
impl<'a, T: core::fmt::Display> DisplayedSliceExt for &'a [T] {
type Displayed = &'a [Displayed<T>];
fn displayed(self) -> Self::Displayed {
unsafe { core::mem::transmute(self) }
}
}
}
pub use display::{Displayed, DisplayedSliceExt};
impl Default for ffi::Ast {
fn default() -> Self {
ffi::Ast {
nodes: vec::Vec::default(),
}
}
}