Compare commits

...

18 Commits

Author SHA1 Message Date
0e7ea508cd add memory.fill/memory.copy intrinsics 2022-06-11 23:31:02 +02:00
0a0d90c801 add some docs, allow leading zeros in literals 2022-05-08 19:21:19 +02:00
c22297ea82 fix <| not parsing 2022-05-08 00:06:38 +02:00
c59b35f9c6 fix #<= and #>= not parsing 2022-05-07 23:59:40 +02:00
2cf47085c1 add support for +:= etc. 2022-05-07 23:51:25 +02:00
01d64baaab implement binop-assignment (+= etc.) 2022-05-07 20:30:16 +02:00
a52fe53a01 fix let lazy/inline chains re-using the same variable 2022-05-07 18:33:26 +02:00
1e746be750 add support for 'if else', change hex escapes to '\ab' 2022-05-06 09:03:13 +02:00
4cfc7ae8a8 add support for escaped strings + (multi-)char literatls 2022-05-06 01:23:31 +02:00
aac7bbd878 implement _f literal suffix as shortcut for NUM as f32 2022-04-10 23:58:18 +02:00
ebc701e2f2 fix != operator failing to parse 2022-04-09 08:51:23 +02:00
557c3a8426 always return dependencies, even when hitting an error 2022-03-05 21:02:58 +01:00
896385654a return dependencies for compiling cwa source 2022-02-27 21:26:55 +01:00
cda3eb868b fix "i?SCREEN = 0" being parsed as "i?(SCREEN = 0)" 2022-02-26 21:00:01 +01:00
5f316cf17d add new features to readme 2022-02-26 12:14:31 +01:00
e608d3bb4b implement const 2022-02-26 11:46:38 +01:00
b41b7f250c add include support 2022-02-25 23:38:50 +01:00
71de622634 add source filename to errors to prepare for includes 2022-02-25 23:13:11 +01:00
18 changed files with 1074 additions and 456 deletions

12
Cargo.lock generated
View File

@@ -13,9 +13,9 @@ dependencies = [
[[package]]
name = "anyhow"
version = "1.0.44"
version = "1.0.55"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "61604a8f862e1d5c3229fdd78f8b02c68dcf73a4c4b05fd636d12240aaa242c1"
checksum = "159bb86af3a200e19a068f4224eae4c8bb2d0fa054c7e5d1cacd5cef95e684cd"
[[package]]
name = "ariadne"
@@ -83,9 +83,9 @@ dependencies = [
[[package]]
name = "getrandom"
version = "0.2.3"
version = "0.2.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7fcd999463524c52659517fe2cea98493cfe485d10565e7b0fb07dbba7ad2753"
checksum = "d39cd93900197114fa1fcb7ae84ca742095eed9442088988ae74fa744e930e77"
dependencies = [
"cfg-if",
"libc",
@@ -106,9 +106,9 @@ checksum = "884e2677b40cc8c339eaefcb701c32ef1fd2493d71118dc0ca4b6a736c93bd67"
[[package]]
name = "libc"
version = "0.2.105"
version = "0.2.119"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "869d572136620d55835903746bcb5cdc54cb2851fd0aeec53220b4bb65ef3013"
checksum = "1bf2e165bb3457c8e098ea76f3e3bc9db55f87aa90d52d0e6be741470916aaa4"
[[package]]
name = "pico-args"

105
README.md
View File

@@ -56,6 +56,14 @@ Then run it on [MicroW8](https://exoticorn.github.io/microw8/v0.1pre2)
*/
```
### Include
Other sourcefiles can be included with the `include` top level statement:
```
include "platform_imports.cwa"
```
### Types
There are four types in WebAssembly and therefore CurlyWas:
@@ -81,16 +89,25 @@ For floating point numbers, only the most basic decimal format is currently impl
0.464, 3.141, -10.0
```
String literals exist in a very basic form. No escape character implemented yet.
String literals are used for include paths, import names and as literal strings in the data section. The following escapes are supported:
| Escape | Result | Comment |
| `\"` | `"` | |
| `\'` | `'` | |
| `\t` | 8 | |
| `\n` | 10 | |
| `\r` | 13 | |
| `\N` | 0x0N | (Can't be followed by a hex digit) |
| `\NN` | 0xNN | |
```
"env.memory", "Hello World!"
this does not work, yet:
"one line\nsecond line", "They said: \"Enough!\""
```
Character literals are enclosed in single quotes `'` and support the same escapes as strings. They can contain up to 4 characters and evaluate to the
little-endian representation of these characters. For examples: `'A'` evaluates to `0x41`, `'hi'` evaluates to 0x6968, and `'Crly'` to 0x7a6c7243.
### Imports
WebAssembly imports are specified with a module and a name. In CurlyWas you give them inside a single string literal, seperated by a dot. So a module `env` and name `printString` would be written `"env.printString"`.
@@ -138,6 +155,18 @@ can use it. However, exporting global variable is not yet supported in CurlyWas.
The type is optional, if missing it is inferred from the init value.
### Constants
Constants can be declared in the global scope:
```
const name[: type] = value;
```
`value` has to be an expression evaluating to a constant value. It may reference other constants.
The type is optional, but if given has to match the type of `value`.
### Functions
Functions look like this:
@@ -275,6 +304,9 @@ So for example this block evaluates to 12:
Blocks are used as function bodies and in flow control (`if`, `block`, `loop`), but can also used at any point inside an expression.
Variable re-assignments of the form `name = name <op> expression` can be shortened to `name <op>= expression`, for example `x += 1` to increment `x` by one. This works for all arithmetic, bit and shift operators.
The same is allowed for `name := name <op> expression`, ie. `x +:= 1` increments `x` and returns the new value.
#### Flow control
`if condition_expression { if_true_block } [else {if_false_block}]` executes the `if_true_block` if the condition evaluates to a non-zero integer and
@@ -284,6 +316,17 @@ the `if_false_block` otherwise (if it exists). It can also be used as an express
let a = if 0 { 2 } else { 3 }; // assigns 3 to a
```
If the `if_false_block` contains exactly one `if` expression or statement you may omit the curly braces, writing `else if` chains like:
```
if x == 0 {
doOneThing()
} else if x == 1 {
doThatOtherThing()
} else {
keepWaiting()
}
```
`block name { ... }` opens a named block scope. A branch statement can be used to jump to the end of the block. Currently, `block` can only be used
as a statement, returning a value from the block is not yet supported.
@@ -296,15 +339,65 @@ non-zero integer.
#### Memory load/store
To read from memory you specify a memory location as `base?offset` or `base!offset`. `?` reads a byte and `!` reads a 32bit word.
To read from memory you specify a memory location as `base?offset`, `base!offset` or `base$offset`. `?` reads a byte, `!` reads a 32bit word
and `$` reads a 32bit float.
`base` can be any expression that evaluates to an `i32` while `offset` has to be a constant `i32` value. The effective memory address is the sum of both.
Writing to memory looks just like an assignment to a memory location: `base?offset = expressoin` and `base!offset = expression`.
Writing to memory looks just like an assignment to a memory location: `base?offset = expression`, `base!offset = expression` and `base$offset = expression`.
When reading/writing 32bit words you need to make sure the address is 4-byte aligned.
These compile to `i32.load8_u`, `i32.load`, `i32.store8` and `i32.store`. Other WASM load/store instructions will be implemented as intrinsics, but aren't yet.
These compile to `i32.load8_u`, `i32.load`, `f32.load`, `i32.store8`, `i32.store` and `f32.store`.
In addition, all wasm memory instructions are available as intrinsics:
```
<load-ins>(<base-address>[, <offset>, [<align>]])
offset defaults to 0, align to the natural alignment: 0 for 8bit loads, 1 for 16bit, 2 for 32 bit and 3 for 64bit.
```
with `<load-ins>` being one of `i32.load`, `i32.load8_u`, `i32.load8_s`, `i32.load16_u`, `i32.load16_s`,
`i64.load`, `i64.load8_u`, `i64.load8_s`, `i64.load16_u`, `i64.load16_s`, `i32.load32_u`, `i32.load32_s`,
`f32.load` and `f64.load`.
```
<store-ins>(<value>, <base-address>[, <offset>, [<align>]])
offset and align defaults are the same as the load intrinsics.
```
with `<store-ins>` being one of `i32.store`, `i32.store8`, `i32.store16`, `i64.store`, `i64.store8`,
`i64.store16`, `i64.store32`, `f32.store` and `f64.store`.
#### Data
Data sections are written in `data` blocks:
```
data <address> {
...
}
```
The content of such a block is loaded at the given address at module start.
Inside the data block you can include 8, 16, 32, 64, f32 or f64 values:
```
i8(1, 255) i16(655350) i32(0x12345678) i64(0x1234567890abcdefi64) f32(1.0, 3.141) f64(0.5f64)
```
Strings:
```
"First line" i8(13, 10) "Second line"
```
And binary files:
```
file("font.bin")
```
#### Advanced sequencing

View File

@@ -1,29 +1,8 @@
import "env.memory" memory(4);
include "uw8.cwa"
import "env.pow" fn pow(f32, f32) -> f32;
import "env.sin" fn sin(f32) -> f32;
import "env.cos" fn cos(f32) -> f32;
import "env.atan2" fn atan2(f32, f32) -> f32;
import "env.tan" fn tan(f32) -> f32;
import "env.atan" fn atan(f32) -> f32;
import "env.rectangle" fn rect(f32, f32, f32, f32, i32);
const SWEETY = PALETTE + 192 * 4;
//export fn tic(time: i32) {
// let i: i32;
// loop pixels {
// let lazy x = (i % 320) as f32 - 160.5;
// let lazy y = (i / 320 - 120) as f32;
//
// let lazy dist = 4000 as f32 / sqrt(x*x + y*y + 10 as f32);
// let lazy angle = atan2(x, y) * (64.0 / 3.141);
//
// i?120 = ((((dist + time as f32 / 63 as f32) as i32 ^ angle as i32) #% 32 + 32) >> ((dist as i32 - i % 7 * 3) / 40)) + 192;
//
// branch_if (i := i + 1) < 320*240: pixels;
// }
//}
export fn tic(time: i32) {
export fn upd() {
let i: i32;
loop colors {
rect((i % 16 * 15) as f32, (i / 16 * 15) as f32, 15 as f32, 15 as f32, i);
@@ -43,7 +22,7 @@ start fn gen_palette() {
let lazy a = max(llimit, min(ulimit, c)) * (scale + 0.05);
let lazy b = scale * scale * 0.8;
let inline v = (select(i < 11*16*3, max(0 as f32, min(a + b - a * b, 1 as f32)), scale) * 255 as f32) as i32;
(i%3 + i/3*4)?(120+320*240) = v;
(i%3 + i/3*4)?PALETTE = v;
avg = (avg + c) * 0.5;
branch_if i := i - 1: gradients;
@@ -56,15 +35,15 @@ start fn gen_palette() {
let lazy first_step = index >= 32;
let inline src1 = select(first_step, index % 32 / 2, index * 2);
let inline src2 = select(first_step, (index + 1) % 32 / 2, index * 2 + 1);
let inline c1 = (src1 * 4 + channel)?(120+320*240+192*4);
let inline c2 = (src2 * 4 + channel)?(120+320*240+192*4);
i?(120+320*240+192*4) = (c1 + c2) * (3 + first_step) / 8;
let inline c1 = (src1 * 4 + channel)?SWEETY;
let inline c2 = (src2 * 4 + channel)?SWEETY;
i?SWEETY = (c1 + c2) * (3 + first_step) / 8;
branch_if (i := i - 1) >= 0: expand_sweetie;
}
}
data 120+320*240+192*4 {
data SWEETY {
i32(
0x2c1c1a,
0x5d275d,

View File

@@ -0,0 +1,24 @@
import "env.memory" memory(4);
import "env.printString" fn printString(i32);
import "env.printChar" fn printChar(i32);
export fn upd() {
printChar(12);
printChar('Test');
printChar('\1f\10\10');
printChar('abc\n');
printString(0);
let t = 32!32 / 1000 #% 3;
if t == 0 {
printChar('one');
} else if t == 1 {
printChar('two');
} else {
printChar('many');
}
}
data 0 {
"\0e\64\"Colors!!!\"\0e\1\r\n\0"
}

View File

@@ -1,24 +1,29 @@
import "env.memory" memory(4);
import "env.sin" fn sin(f32) -> f32;
import "env.time" fn time() -> f32;
import "env.setPixel" fn setPixel(i32, i32, i32);
export fn tic(time: i32) {
let i: i32;
export fn upd() {
let x: i32;
let y: i32;
loop screen {
let lazy t = time as f32 / 2000 as f32;
let lazy o = sin(t) * 0.8;
let lazy q = (i % 320) as f32 - 160.1;
let lazy w = (i / 320 - 120) as f32;
let inline t = time() / 2 as f32;
let lazy o = sin(t) * 0.75;
let inline q = x as f32 - 160.5;
let inline w = (y - 120) as f32;
let lazy r = sqrt(q*q + w*w);
let lazy z = q / r;
let lazy s = z * o + sqrt(z * z * o * o + 1 as f32 - o * o);
let lazy q2 = (z * s - o) * 10 as f32 + t;
let lazy w2 = w / r * s * 10 as f32 + t;
let lazy s2 = s * 50 as f32 / r;
i?120 = max(
let inline q2 = (z * s - o) * 10 as f32 + t;
let inline w2 = w / r * s * 10 as f32 + t;
let inline s2 = s * 100 as f32 / r;
let inline color = max(
0 as f32,
((q2 as i32 ^ w2 as i32 & ((s2 + t) * 20 as f32) as i32) & 5) as f32 *
(2 as f32 - s2) * 22 as f32
) as i32;
branch_if (i := i + 1) < 320*240: screen
((q2 as i32 ^ w2 as i32 & ((s2 + time()) * 10 as f32) as i32) & 5) as f32 *
(4 as f32 - s2) as f32
) as i32 - 32;
setPixel(x, y, color);
branch_if x := (x + 1) % 320: screen;
branch_if y := (y + 1) % 320: screen;
}
}

13
examples/microw8/uw8.cwa Normal file
View File

@@ -0,0 +1,13 @@
import "env.memory" memory(4);
import "env.pow" fn pow(f32, f32) -> f32;
import "env.sin" fn sin(f32) -> f32;
import "env.cos" fn cos(f32) -> f32;
import "env.atan2" fn atan2(f32, f32) -> f32;
import "env.tan" fn tan(f32) -> f32;
import "env.atan" fn atan(f32) -> f32;
import "env.rectangle" fn rect(f32, f32, f32, f32, i32);
const FRAMEBUFFER = 120;
const PALETTE = 0x13000;
const FONT = 0x13400;

View File

@@ -1,15 +1,16 @@
import "env.memory" memory(2);
import "env.time" fn time() -> f32;
export fn tic(time: i32) {
export fn upd() {
let i: i32;
loop pixels {
let lazy x = (i % 320) as f32 - 160.1;
let lazy y = (i / 320 - 120) as f32;
let lazy dist = 10000.0 / (x*x + y*y);
let lazy t = time as f32 / 20 as f32;
let lazy dist = 1024_f / (x*x + y*y);
let inline t = time() * 4_f;
i?120 = (x * dist + t) as i32 ^ (y * dist + t) as i32;
i?120 = (x * dist + t) as i32 ^ (y * dist + t) as i32 | -32;
branch_if (i := i + 1) < 320*240: pixels
branch_if (i +:= 1) < 320*240: pixels
}
}

View File

@@ -18,14 +18,14 @@ fn rng(state: i32) -> i32 {
}
fn set_color(color: i32) -> i32 {
?20 = color;
0?20 = color;
6
}
export fn update() {
let y: i32;
let score = pz;
let lazy pad = ?22;
let lazy pad = 0?22;
let lazy zero = 0.0;
let lazy control_speed = 0.03;
@@ -33,7 +33,7 @@ export fn update() {
f = f * 0.7;
loop lines {
?(8003-y) = (score := score / 10) % 10 + 48;
(8003-y)?0 = (score := score / 10) % 10 + 48;
let lazy z = (4000 / (y := y + 1) + pz) / 20;
let lazy x = (rng(rng(rng(rng(z)))) >> 30) as f32 - px;
let lazy w = 9 as f32 / sqrt(z as f32);

View File

@@ -1,13 +1,26 @@
use std::{fmt, path::PathBuf};
use crate::Span;
use crate::parser::Span;
#[derive(Debug)]
#[derive(Debug, Default)]
pub struct Script {
pub imports: Vec<Import>,
pub global_vars: Vec<GlobalVar>,
pub functions: Vec<Function>,
pub data: Vec<Data>,
pub includes: Vec<Include>,
pub consts: Vec<GlobalConst>,
}
impl Script {
pub fn merge(&mut self, mut other: Script) {
self.imports.append(&mut other.imports);
self.global_vars.append(&mut other.global_vars);
self.functions.append(&mut other.functions);
self.data.append(&mut other.data);
self.consts.append(&mut other.consts);
assert!(other.includes.is_empty());
}
}
#[derive(Debug)]
@@ -16,6 +29,14 @@ pub enum TopLevelItem {
GlobalVar(GlobalVar),
Function(Function),
Data(Data),
Include(Include),
Const(GlobalConst),
}
#[derive(Debug)]
pub struct Include {
pub span: Span,
pub path: String,
}
#[derive(Debug)]
@@ -49,6 +70,14 @@ pub struct GlobalVar {
pub mutable: bool,
}
#[derive(Debug)]
pub struct GlobalConst {
pub span: Span,
pub name: String,
pub value: Expression,
pub type_: Option<Type>,
}
#[derive(Debug)]
pub struct Function {
pub span: Span,
@@ -152,7 +181,7 @@ pub enum DataType {
F64,
}
#[derive(Debug)]
#[derive(Debug, Clone)]
pub struct MemoryLocation {
pub span: Span,
pub size: MemSize,
@@ -160,7 +189,7 @@ pub struct MemoryLocation {
pub right: Box<Expression>,
}
#[derive(Debug)]
#[derive(Debug, Clone)]
pub struct Expression {
pub type_: Option<Type>,
pub expr: Expr,
@@ -195,9 +224,16 @@ impl Expression {
_ => panic!("Expected F64Const"),
}
}
pub fn is_const(&self) -> bool {
match self.expr {
Expr::I32Const(_) | Expr::I64Const(_) | Expr::F32Const(_) | Expr::F64Const(_) => true,
_ => false,
}
}
}
#[derive(Debug)]
#[derive(Debug, Clone)]
pub enum Expr {
Block {
statements: Vec<Expression>,
@@ -337,7 +373,7 @@ pub enum BinOp {
pub enum MemSize {
Byte,
Word,
Float
Float,
}
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Ord, PartialOrd)]

View File

@@ -1,35 +1,103 @@
use crate::ast;
use std::collections::HashMap;
use crate::{
ast,
parser::{Sources, Span},
typecheck::{report_duplicate_definition, report_error},
};
type Result<T> = std::result::Result<T, ()>;
pub fn fold_script(script: &mut ast::Script, sources: &Sources) -> Result<()> {
let mut context = Context {
consts: HashMap::new(),
sources,
};
fold_consts(&mut context, &mut script.consts)?;
pub fn fold_script(script: &mut ast::Script) {
for var in &mut script.global_vars {
fold_expr(&mut var.value);
fold_expr(&context, &mut var.value);
}
for func in &mut script.functions {
fold_expr(&mut func.body);
fold_expr(&context, &mut func.body);
}
for data in &mut script.data {
fold_expr(&mut data.offset);
fold_expr(&context, &mut data.offset);
for values in &mut data.data {
match values {
ast::DataValues::Array { values, .. } => {
for value in values {
fold_expr(value);
fold_expr(&context, value);
}
}
ast::DataValues::String(_) | ast::DataValues::File { .. } => (),
}
}
}
Ok(())
}
fn fold_mem_location(mem_location: &mut ast::MemoryLocation) {
fold_expr(&mut mem_location.left);
fold_expr(&mut mem_location.right);
struct Context<'a> {
consts: HashMap<String, ast::Expr>,
sources: &'a Sources,
}
fn fold_expr(expr: &mut ast::Expression) {
fn fold_consts(context: &mut Context, consts: &mut [ast::GlobalConst]) -> Result<()> {
let mut spans: HashMap<&str, Span> = HashMap::new();
for cnst in consts.iter_mut() {
if let Some(prev_span) = spans.insert(&cnst.name, cnst.span.clone()) {
report_duplicate_definition(
"Const already defined",
&cnst.span,
&prev_span,
context.sources,
)?;
}
}
while context.consts.len() < consts.len() {
let mut making_progress = false;
for cnst in consts.iter_mut() {
if !context.consts.contains_key(&cnst.name) {
fold_expr(context, &mut cnst.value);
if cnst.value.is_const() {
context
.consts
.insert(cnst.name.clone(), cnst.value.expr.clone());
making_progress = true;
}
}
}
if !making_progress {
break;
}
}
let mut result = Ok(());
for cnst in consts {
if !context.consts.contains_key(&cnst.name) {
result = report_error(
&format!("Failed to fold const '{}'", cnst.name),
&cnst.span,
context.sources,
);
}
}
result
}
fn fold_mem_location(context: &Context, mem_location: &mut ast::MemoryLocation) {
fold_expr(context, &mut mem_location.left);
fold_expr(context, &mut mem_location.right);
}
fn fold_expr(context: &Context, expr: &mut ast::Expression) {
use ast::BinOp::*;
match expr.expr {
ast::Expr::Block {
@@ -37,15 +105,15 @@ fn fold_expr(expr: &mut ast::Expression) {
ref mut final_expression,
} => {
for stmt in statements {
fold_expr(stmt);
fold_expr(context, stmt);
}
if let Some(ref mut expr) = final_expression {
fold_expr(expr);
fold_expr(context, expr);
}
}
ast::Expr::Let { ref mut value, .. } => {
if let Some(ref mut expr) = value {
fold_expr(expr);
fold_expr(context, expr);
}
}
ast::Expr::Poke {
@@ -53,12 +121,12 @@ fn fold_expr(expr: &mut ast::Expression) {
ref mut value,
..
} => {
fold_mem_location(mem_location);
fold_expr(value);
fold_mem_location(context, mem_location);
fold_expr(context, value);
}
ast::Expr::Peek(ref mut mem_location) => fold_mem_location(mem_location),
ast::Expr::Peek(ref mut mem_location) => fold_mem_location(context, mem_location),
ast::Expr::UnaryOp { op, ref mut value } => {
fold_expr(value);
fold_expr(context, value);
let result = match (op, &value.expr) {
(ast::UnaryOp::Negate, ast::Expr::I32Const(value)) => {
Some(ast::Expr::I32Const(-*value))
@@ -72,6 +140,20 @@ fn fold_expr(expr: &mut ast::Expression) {
(ast::UnaryOp::Negate, ast::Expr::F64Const(value)) => {
Some(ast::Expr::F64Const(-*value))
}
(ast::UnaryOp::Negate, ast::Expr::Cast { value, type_ }) => {
if let ast::Expr::I32Const(v) = value.expr {
Some(ast::Expr::Cast {
value: Box::new(ast::Expression {
expr: ast::Expr::I32Const(-v),
span: value.span.clone(),
type_: value.type_,
}),
type_: *type_,
})
} else {
None
}
}
(ast::UnaryOp::Not, ast::Expr::I32Const(value)) => {
Some(ast::Expr::I32Const((*value == 0) as i32))
}
@@ -90,8 +172,8 @@ fn fold_expr(expr: &mut ast::Expression) {
ref mut right,
..
} => {
fold_expr(left);
fold_expr(right);
fold_expr(context, left);
fold_expr(context, right);
match (&left.expr, &right.expr) {
(&ast::Expr::I32Const(left), &ast::Expr::I32Const(right)) => {
let result = match op {
@@ -237,24 +319,28 @@ fn fold_expr(expr: &mut ast::Expression) {
ast::Expr::I32Const(_)
| ast::Expr::I64Const(_)
| ast::Expr::F32Const(_)
| ast::Expr::F64Const(_)
| ast::Expr::Variable { .. } => (),
ast::Expr::Assign { ref mut value, .. } => fold_expr(value),
ast::Expr::LocalTee { ref mut value, .. } => fold_expr(value),
ast::Expr::Loop { ref mut block, .. } => fold_expr(block),
ast::Expr::LabelBlock { ref mut block, .. } => fold_expr(block),
| ast::Expr::F64Const(_) => (),
ast::Expr::Variable { ref name, .. } => {
if let Some(value) = context.consts.get(name) {
expr.expr = value.clone();
}
}
ast::Expr::Assign { ref mut value, .. } => fold_expr(context, value),
ast::Expr::LocalTee { ref mut value, .. } => fold_expr(context, value),
ast::Expr::Loop { ref mut block, .. } => fold_expr(context, block),
ast::Expr::LabelBlock { ref mut block, .. } => fold_expr(context, block),
ast::Expr::Branch(_) => (),
ast::Expr::BranchIf {
ref mut condition, ..
} => fold_expr(condition),
ast::Expr::Cast { ref mut value, .. } => fold_expr(value),
} => fold_expr(context, condition),
ast::Expr::Cast { ref mut value, .. } => fold_expr(context, value),
ast::Expr::FuncCall {
ref name,
ref mut params,
..
} => {
for param in params.iter_mut() {
fold_expr(param);
fold_expr(context, param);
}
use ast::Expr::*;
let params: Vec<_> = params.iter().map(|e| &e.expr).collect();
@@ -269,31 +355,31 @@ fn fold_expr(expr: &mut ast::Expression) {
ref mut if_false,
..
} => {
fold_expr(condition);
fold_expr(if_true);
fold_expr(if_false);
fold_expr(context, condition);
fold_expr(context, if_true);
fold_expr(context, if_false);
}
ast::Expr::If {
ref mut condition,
ref mut if_true,
ref mut if_false,
} => {
fold_expr(condition);
fold_expr(if_true);
fold_expr(context, condition);
fold_expr(context, if_true);
if let Some(ref mut if_false) = if_false {
fold_expr(if_false);
fold_expr(context, if_false);
}
}
ast::Expr::Return {
value: Some(ref mut value),
} => fold_expr(value),
} => fold_expr(context, value),
ast::Expr::Return { value: None } => (),
ast::Expr::First {
ref mut value,
ref mut drop,
} => {
fold_expr(value);
fold_expr(drop);
fold_expr(context, value);
fold_expr(context, drop);
}
ast::Expr::Error => unreachable!(),
}

View File

@@ -289,7 +289,7 @@ struct FunctionContext<'a> {
functions: &'a HashMap<String, u32>,
locals: &'a ast::Locals,
labels: Vec<String>,
let_values: HashMap<u32, (&'a ast::Expression, ast::LetType)>,
let_values: HashMap<u32, Vec<(&'a ast::Expression, ast::LetType)>>,
intrinsics: &'a Intrinsics,
}
@@ -385,7 +385,10 @@ fn emit_expression<'a>(ctx: &mut FunctionContext<'a>, expr: &'a ast::Expression)
.instruction(&Instruction::LocalSet(local.index.unwrap()));
}
ast::LetType::Lazy | ast::LetType::Inline => {
ctx.let_values.insert(local_id.unwrap(), (value, *let_type));
ctx.let_values
.entry(local_id.unwrap())
.or_default()
.push((value, *let_type));
}
}
}
@@ -612,17 +615,17 @@ fn emit_expression<'a>(ctx: &mut FunctionContext<'a>, expr: &'a ast::Expression)
}
ast::Expr::Variable { name, local_id } => {
if let &Some(id) = local_id {
if let Some((expr, let_type)) = ctx.let_values.get(&id) {
if let Some((expr, let_type)) = ctx.let_values.get_mut(&id).and_then(|s| s.pop()) {
match let_type {
ast::LetType::Lazy => {
let expr = ctx.let_values.remove(&id).unwrap().0;
emit_expression(ctx, expr);
ctx.let_values.get_mut(&id).unwrap().clear();
ctx.function
.instruction(&Instruction::LocalTee(ctx.locals[id].index.unwrap()));
}
ast::LetType::Inline => {
let expr = *expr;
emit_expression(ctx, expr);
ctx.let_values.get_mut(&id).unwrap().push((expr, let_type));
}
_ => unreachable!(),
}

View File

@@ -1,11 +1,15 @@
use std::fs::File;
use std::io::prelude::*;
use std::path::Path;
use std::path::{Path, PathBuf};
use std::{collections::HashSet, fs::File};
use crate::ast;
use anyhow::{anyhow, Result};
pub fn resolve_includes(script: &mut ast::Script, path: &Path) -> Result<()> {
pub fn resolve_includes(
script: &mut ast::Script,
dependencies: &mut HashSet<PathBuf>,
path: &Path,
) -> Result<()> {
let script_dir = path.parent().expect("Script path has no parent");
for data in &mut script.data {
for values in &mut data.data {
@@ -21,6 +25,7 @@ pub fn resolve_includes(script: &mut ast::Script, path: &Path) -> Result<()> {
anyhow!("Failed to load data from {}: {}", full_path.display(), e)
})?
.read_to_end(data)?;
dependencies.insert(full_path);
}
}
}

View File

@@ -1,9 +1,11 @@
use crate::ast::Type;
use std::collections::HashMap;
use enc::MemArg;
use std::collections::HashMap;
use wasm_encoder as enc;
pub struct Intrinsics(HashMap<String, HashMap<Vec<Type>, (Type, enc::Instruction<'static>)>>);
pub struct Intrinsics(
HashMap<String, HashMap<Vec<Type>, (Option<Type>, enc::Instruction<'static>)>>,
);
impl Intrinsics {
pub fn new() -> Intrinsics {
@@ -12,14 +14,11 @@ impl Intrinsics {
i
}
pub fn find_types(
&self,
name: &str,
) -> Option<HashMap<Vec<Type>, Option<Type>>> {
pub fn find_types(&self, name: &str) -> Option<HashMap<Vec<Type>, Option<Type>>> {
self.0.get(name).map(|types| {
types
.iter()
.map(|(params, (ret, _))| (params.clone(), Some(*ret)))
.map(|(params, (ret, _))| (params.clone(), *ret))
.collect()
})
}
@@ -34,87 +33,121 @@ impl Intrinsics {
fn add_instructions(&mut self) {
use enc::Instruction as I;
use Type::*;
self.inst("i32.rotl", &[I32, I32], I32, I::I32Rotl);
self.inst("i32.rotr", &[I32, I32], I32, I::I32Rotr);
self.inst("i32.clz", &[I32], I32, I::I32Clz);
self.inst("i32.ctz", &[I32], I32, I::I32Ctz);
self.inst("i32.popcnt", &[I32], I32, I::I32Popcnt);
self.inst("i32.rotl", &[I32, I32], Some(I32), I::I32Rotl);
self.inst("i32.rotr", &[I32, I32], Some(I32), I::I32Rotr);
self.inst("i32.clz", &[I32], Some(I32), I::I32Clz);
self.inst("i32.ctz", &[I32], Some(I32), I::I32Ctz);
self.inst("i32.popcnt", &[I32], Some(I32), I::I32Popcnt);
self.inst("i64.rotl", &[I64, I64], I64, I::I64Rotl);
self.inst("i64.rotr", &[I64, I64], I64, I::I64Rotr);
self.inst("i64.clz", &[I64], I64, I::I64Clz);
self.inst("i64.ctz", &[I64], I64, I::I64Ctz);
self.inst("i64.popcnt", &[I64], I64, I::I64Popcnt);
self.inst("i64.rotl", &[I64, I64], Some(I64), I::I64Rotl);
self.inst("i64.rotr", &[I64, I64], Some(I64), I::I64Rotr);
self.inst("i64.clz", &[I64], Some(I64), I::I64Clz);
self.inst("i64.ctz", &[I64], Some(I64), I::I64Ctz);
self.inst("i64.popcnt", &[I64], Some(I64), I::I64Popcnt);
self.inst("f32/sqrt", &[F32], F32, I::F32Sqrt);
self.inst("f32/min", &[F32, F32], F32, I::F32Min);
self.inst("f32/max", &[F32, F32], F32, I::F32Max);
self.inst("f32/ceil", &[F32], F32, I::F32Ceil);
self.inst("f32/floor", &[F32], F32, I::F32Floor);
self.inst("f32/trunc", &[F32], F32, I::F32Trunc);
self.inst("f32/nearest", &[F32], F32, I::F32Nearest);
self.inst("f32/abs", &[F32], F32, I::F32Abs);
self.inst("f32.copysign", &[F32, F32], F32, I::F32Copysign);
self.inst("f32/sqrt", &[F32], Some(F32), I::F32Sqrt);
self.inst("f32/min", &[F32, F32], Some(F32), I::F32Min);
self.inst("f32/max", &[F32, F32], Some(F32), I::F32Max);
self.inst("f32/ceil", &[F32], Some(F32), I::F32Ceil);
self.inst("f32/floor", &[F32], Some(F32), I::F32Floor);
self.inst("f32/trunc", &[F32], Some(F32), I::F32Trunc);
self.inst("f32/nearest", &[F32], Some(F32), I::F32Nearest);
self.inst("f32/abs", &[F32], Some(F32), I::F32Abs);
self.inst("f32.copysign", &[F32, F32], Some(F32), I::F32Copysign);
self.inst("f64/sqrt", &[F64], F64, I::F64Sqrt);
self.inst("f64/min", &[F64, F64], F64, I::F64Min);
self.inst("f64/max", &[F64, F64], F64, I::F64Max);
self.inst("f64/ceil", &[F64], F64, I::F64Ceil);
self.inst("f64/floor", &[F64], F64, I::F64Floor);
self.inst("f64/trunc", &[F64], F64, I::F64Trunc);
self.inst("f64/nearest", &[F64], F64, I::F64Nearest);
self.inst("f64/abs", &[F64], F64, I::F64Abs);
self.inst("f64.copysign", &[F64, F64], F64, I::F64Copysign);
self.inst("f64/sqrt", &[F64], Some(F64), I::F64Sqrt);
self.inst("f64/min", &[F64, F64], Some(F64), I::F64Min);
self.inst("f64/max", &[F64, F64], Some(F64), I::F64Max);
self.inst("f64/ceil", &[F64], Some(F64), I::F64Ceil);
self.inst("f64/floor", &[F64], Some(F64), I::F64Floor);
self.inst("f64/trunc", &[F64], Some(F64), I::F64Trunc);
self.inst("f64/nearest", &[F64], Some(F64), I::F64Nearest);
self.inst("f64/abs", &[F64], Some(F64), I::F64Abs);
self.inst("f64.copysign", &[F64, F64], Some(F64), I::F64Copysign);
self.inst("i32.wrap_i64", &[I64], I32, I::I32WrapI64);
self.inst("i64.extend_i32_s", &[I32], I64, I::I64ExtendI32S);
self.inst("i64.extend_i32_u", &[I32], I64, I::I64ExtendI32U);
self.inst("i32.wrap_i64", &[I64], Some(I32), I::I32WrapI64);
self.inst("i64.extend_i32_s", &[I32], Some(I64), I::I64ExtendI32S);
self.inst("i64.extend_i32_u", &[I32], Some(I64), I::I64ExtendI32U);
self.inst("i32.trunc_f32_s", &[F32], I32, I::I32TruncF32S);
self.inst("i32.trunc_f64_s", &[F64], I32, I::I32TruncF64S);
self.inst("i64.trunc_f32_s", &[F32], I64, I::I64TruncF32S);
self.inst("i64.trunc_f64_s", &[F64], I64, I::I64TruncF64S);
self.inst("i32.trunc_f32_s", &[F32], Some(I32), I::I32TruncF32S);
self.inst("i32.trunc_f64_s", &[F64], Some(I32), I::I32TruncF64S);
self.inst("i64.trunc_f32_s", &[F32], Some(I64), I::I64TruncF32S);
self.inst("i64.trunc_f64_s", &[F64], Some(I64), I::I64TruncF64S);
self.inst("i32.trunc_f32_u", &[F32], I32, I::I32TruncF32U);
self.inst("i32.trunc_f64_u", &[F64], I32, I::I32TruncF64U);
self.inst("i64.trunc_f32_u", &[F32], I64, I::I64TruncF32U);
self.inst("i64.trunc_f64_u", &[F64], I64, I::I64TruncF64U);
self.inst("i32.trunc_f32_u", &[F32], Some(I32), I::I32TruncF32U);
self.inst("i32.trunc_f64_u", &[F64], Some(I32), I::I32TruncF64U);
self.inst("i64.trunc_f32_u", &[F32], Some(I64), I::I64TruncF32U);
self.inst("i64.trunc_f64_u", &[F64], Some(I64), I::I64TruncF64U);
self.inst("f32.demote_f64", &[F64], F32, I::F32DemoteF64);
self.inst("f64.promote_f32", &[F32], F64, I::F64PromoteF32);
self.inst("f32.demote_f64", &[F64], Some(F32), I::F32DemoteF64);
self.inst("f64.promote_f32", &[F32], Some(F64), I::F64PromoteF32);
self.inst("f32.convert_i32_s", &[I32], F32, I::F32ConvertI32S);
self.inst("f32.convert_i64_s", &[I64], F32, I::F32ConvertI32S);
self.inst("f64.convert_i32_s", &[I32], F64, I::F32ConvertI32S);
self.inst("f64.convert_i64_s", &[I64], F64, I::F32ConvertI32S);
self.inst("f32.convert_i32_s", &[I32], Some(F32), I::F32ConvertI32S);
self.inst("f32.convert_i64_s", &[I64], Some(F32), I::F32ConvertI32S);
self.inst("f64.convert_i32_s", &[I32], Some(F64), I::F32ConvertI32S);
self.inst("f64.convert_i64_s", &[I64], Some(F64), I::F32ConvertI32S);
self.inst("f32.convert_i32_u", &[I32], F32, I::F32ConvertI32U);
self.inst("f32.convert_i64_u", &[I64], F32, I::F32ConvertI32U);
self.inst("f64.convert_i32_u", &[I32], F64, I::F32ConvertI32U);
self.inst("f64.convert_i64_u", &[I64], F64, I::F32ConvertI32U);
self.inst("f32.convert_i32_u", &[I32], Some(F32), I::F32ConvertI32U);
self.inst("f32.convert_i64_u", &[I64], Some(F32), I::F32ConvertI32U);
self.inst("f64.convert_i32_u", &[I32], Some(F64), I::F32ConvertI32U);
self.inst("f64.convert_i64_u", &[I64], Some(F64), I::F32ConvertI32U);
self.inst("i32.reinterpret_f32", &[F32], I32, I::I32ReinterpretF32);
self.inst("i64.reinterpret_f64", &[F64], I64, I::I64ReinterpretF64);
self.inst("f32.reinterpret_i32", &[I32], F32, I::F32ReinterpretI32);
self.inst("f64.reinterpret_i64", &[I64], F64, I::F64ReinterpretI64);
self.inst(
"i32.reinterpret_f32",
&[F32],
Some(I32),
I::I32ReinterpretF32,
);
self.inst(
"i64.reinterpret_f64",
&[F64],
Some(I64),
I::I64ReinterpretF64,
);
self.inst(
"f32.reinterpret_i32",
&[I32],
Some(F32),
I::F32ReinterpretI32,
);
self.inst(
"f64.reinterpret_i64",
&[I64],
Some(F64),
I::F64ReinterpretI64,
);
self.inst("i32.extend8_s", &[I32], I32, I::I32Extend8S);
self.inst("i32.extend16_s", &[I32], I32, I::I32Extend16S);
self.inst("i64.extend8_s", &[I64], I64, I::I64Extend8S);
self.inst("i64.extend16_s", &[I64], I64, I::I64Extend16S);
self.inst("i64.extend32_s", &[I64], I64, I::I64Extend32S);
self.inst("i32.extend8_s", &[I32], Some(I32), I::I32Extend8S);
self.inst("i32.extend16_s", &[I32], Some(I32), I::I32Extend16S);
self.inst("i64.extend8_s", &[I64], Some(I64), I::I64Extend8S);
self.inst("i64.extend16_s", &[I64], Some(I64), I::I64Extend16S);
self.inst("i64.extend32_s", &[I64], Some(I64), I::I64Extend32S);
self.inst("i32.trunc_sat_f32_s", &[F32], I32, I::I32TruncSatF32S);
self.inst("i32.trunc_sat_f32_u", &[F32], I32, I::I32TruncSatF32U);
self.inst("i32.trunc_sat_f64_s", &[F64], I32, I::I32TruncSatF64S);
self.inst("i32.trunc_sat_f64_u", &[F64], I32, I::I32TruncSatF64U);
self.inst("i64.trunc_sat_f32_s", &[F32], I64, I::I64TruncSatF32S);
self.inst("i64.trunc_sat_f32_u", &[F32], I64, I::I64TruncSatF32U);
self.inst("i64.trunc_sat_f64_s", &[F64], I64, I::I64TruncSatF64S);
self.inst("i64.trunc_sat_f64_u", &[F64], I64, I::I64TruncSatF64U);
self.inst("i32.trunc_sat_f32_s", &[F32], Some(I32), I::I32TruncSatF32S);
self.inst("i32.trunc_sat_f32_u", &[F32], Some(I32), I::I32TruncSatF32U);
self.inst("i32.trunc_sat_f64_s", &[F64], Some(I32), I::I32TruncSatF64S);
self.inst("i32.trunc_sat_f64_u", &[F64], Some(I32), I::I32TruncSatF64U);
self.inst("i64.trunc_sat_f32_s", &[F32], Some(I64), I::I64TruncSatF32S);
self.inst("i64.trunc_sat_f32_u", &[F32], Some(I64), I::I64TruncSatF32U);
self.inst("i64.trunc_sat_f64_s", &[F64], Some(I64), I::I64TruncSatF64S);
self.inst("i64.trunc_sat_f64_u", &[F64], Some(I64), I::I64TruncSatF64U);
self.inst(
"memory.copy",
&[I32, I32, I32],
None,
I::MemoryCopy { src: 0, dst: 0 },
);
self.inst("memory.fill", &[I32, I32, I32], None, I::MemoryFill(0));
}
fn inst(&mut self, name: &str, params: &[Type], ret: Type, ins: enc::Instruction<'static>) {
fn inst(
&mut self,
name: &str,
params: &[Type],
ret: Option<Type>,
ins: enc::Instruction<'static>,
) {
if let Some(slash_idx) = name.find('/') {
self.insert(name[(slash_idx + 1)..].to_string(), params, ret, &ins);
let mut full_name = name[..slash_idx].to_string();
@@ -130,7 +163,7 @@ impl Intrinsics {
&mut self,
name: String,
params: &[Type],
ret: Type,
ret: Option<Type>,
ins: &enc::Instruction<'static>,
) {
self.0
@@ -157,7 +190,7 @@ impl Intrinsics {
"i64.load32_u" => MemInstruction::new(I64, I::I64Load32_U, 2),
"f32.load" => MemInstruction::new(F32, I::F32Load, 2),
"f64.load" => MemInstruction::new(F64, I::F64Load, 3),
_ => return None
_ => return None,
};
return Some(ins);
}
@@ -175,7 +208,7 @@ impl Intrinsics {
"i64.store32" => MemInstruction::new(I64, I::I64Store32, 2),
"f32.store" => MemInstruction::new(F32, I::F32Store, 2),
"f64.store" => MemInstruction::new(F64, I::F64Store, 3),
_ => return None
_ => return None,
};
return Some(ins);
}
@@ -184,13 +217,19 @@ impl Intrinsics {
pub struct MemInstruction {
pub type_: Type,
pub instruction: fn(MemArg) -> enc::Instruction<'static>,
pub natural_alignment: u32
pub natural_alignment: u32,
}
impl MemInstruction {
fn new(type_: Type, instruction: fn(MemArg) -> enc::Instruction<'static>, natural_alignment: u32) -> MemInstruction {
fn new(
type_: Type,
instruction: fn(MemArg) -> enc::Instruction<'static>,
natural_alignment: u32,
) -> MemInstruction {
MemInstruction {
type_, instruction, natural_alignment
type_,
instruction,
natural_alignment,
}
}
}

View File

@@ -1,7 +1,8 @@
use anyhow::{bail, Result};
use parser::Sources;
use std::collections::HashSet;
use std::ffi::OsStr;
use std::io::prelude::*;
use std::{fs::File, path::Path};
use std::path::{Path, PathBuf};
mod ast;
mod constfold;
@@ -11,8 +12,6 @@ mod intrinsics;
mod parser;
mod typecheck;
type Span = std::ops::Range<usize>;
#[derive(Default)]
pub struct Options {
pub(crate) debug: bool,
@@ -24,33 +23,76 @@ impl Options {
}
}
pub fn compile_file<P: AsRef<Path>>(path: P, options: Options) -> Result<Vec<u8>> {
let path = path.as_ref();
let mut input = String::new();
File::open(path)?.read_to_string(&mut input)?;
compile_str(&input, path, options)
pub struct CompiledModule {
pub wasm: Vec<u8>,
pub dependencies: Vec<PathBuf>,
}
pub fn compile_str(input: &str, path: &Path, options: Options) -> Result<Vec<u8>> {
let mut script = match parser::parse(input) {
Ok(script) => script,
Err(_) => bail!("Parse failed"),
};
pub fn compile_file<P: AsRef<Path>>(path: P, options: Options) -> (Result<Vec<u8>>, Vec<PathBuf>) {
fn compile_file_inner(
path: &Path,
options: Options,
dependencies: &mut HashSet<PathBuf>,
) -> Result<Vec<u8>> {
let mut script = ast::Script::default();
includes::resolve_includes(&mut script, path)?;
let mut sources = Sources::new();
constfold::fold_script(&mut script);
if typecheck::tc_script(&mut script, input).is_err() {
bail!("Type check failed");
let mut pending_files = vec![(path.to_path_buf(), None)];
while let Some((path, span)) = pending_files.pop() {
match sources.add(&path) {
Ok((id, true)) => {
dependencies.insert(path.clone());
let mut new_script = match parser::parse(&sources, id) {
Ok(script) => script,
Err(_) => bail!("Parse failed"),
};
includes::resolve_includes(&mut new_script, dependencies, &path)?;
for include in std::mem::take(&mut new_script.includes) {
let mut path = path
.parent()
.expect("Script path has no parent")
.to_path_buf();
path.push(include.path);
pending_files.push((path, Some(include.span)));
}
script.merge(new_script);
}
Ok((_, false)) => (), // already parsed this include
Err(err) => {
if let Some(span) = span {
let _ = typecheck::report_error(&err.to_string(), &span, &sources);
} else {
eprintln!("Failed to load script {}: {}", path.display(), err);
}
bail!("Parse failed");
}
}
}
if constfold::fold_script(&mut script, &sources).is_err() {
bail!("Constant folding failed");
}
if typecheck::tc_script(&mut script, &sources).is_err() {
bail!("Type check failed");
}
let wasm = emit::emit(
&script,
&path
.file_stem()
.unwrap_or_else(|| OsStr::new("unknown"))
.to_string_lossy(),
&options,
);
Ok(wasm)
}
let wasm = emit::emit(
&script,
&path
.file_stem()
.unwrap_or_else(|| OsStr::new("unknown"))
.to_string_lossy(),
&options,
);
Ok(wasm)
let mut dependencies = HashSet::new();
let result = compile_file_inner(path.as_ref(), options, &mut dependencies);
(result, dependencies.into_iter().collect())
}

View File

@@ -15,7 +15,7 @@ fn main() -> Result<()> {
let mut filename = args.free_from_os_str::<PathBuf, bool>(|s| Ok(s.into()))?;
let wasm = compile_file(&filename, options)?;
let wasm = compile_file(&filename, options).0?;
wasmparser::validate(&wasm)?;

View File

@@ -1,8 +1,62 @@
use crate::ast;
use crate::Span;
use ariadne::{Color, Fmt, Label, Report, ReportKind, Source};
use chumsky::{prelude::*, stream::Stream};
use std::fmt;
use anyhow::Result;
use ariadne::{Color, Fmt, Label, Report, ReportKind};
use chumsky::prelude::*;
use std::{
fmt,
fs::File,
io::Read,
ops::Range,
path::{Path, PathBuf},
};
pub type Span = (usize, Range<usize>);
pub struct SourceFile {
source: ariadne::Source,
path: PathBuf,
}
pub struct Sources(Vec<SourceFile>);
impl Sources {
pub fn new() -> Sources {
Sources(Vec::new())
}
pub fn add(&mut self, path: &Path) -> Result<(usize, bool)> {
let canonical = path.canonicalize()?;
for (index, source) in self.0.iter().enumerate() {
if source.path.canonicalize()? == canonical {
return Ok((index, false));
}
}
let mut source = String::new();
File::open(path)?.read_to_string(&mut source)?;
self.0.push(SourceFile {
source: ariadne::Source::from(source),
path: path.to_path_buf(),
});
Ok((self.0.len() - 1, true))
}
}
impl std::ops::Index<usize> for Sources {
type Output = SourceFile;
fn index(&self, idx: usize) -> &SourceFile {
&self.0[idx]
}
}
impl ariadne::Cache<usize> for &Sources {
fn fetch(&mut self, id: &usize) -> Result<&ariadne::Source, Box<dyn std::fmt::Debug + '_>> {
Ok(&self.0[*id].source)
}
fn display<'a>(&self, id: &'a usize) -> Option<Box<dyn std::fmt::Display + 'a>> {
Some(Box::new(self.0[*id].path.clone().display().to_string()))
}
}
#[derive(Clone, Debug, PartialEq, Eq, Hash)]
enum Token {
@@ -25,6 +79,7 @@ enum Token {
Str(String),
Int(i32),
Int64(i64),
IntFloat(i32),
Float(String),
Float64(String),
Op(String),
@@ -53,6 +108,7 @@ impl fmt::Display for Token {
Token::Str(s) => write!(f, "{:?}", s),
Token::Int(v) => write!(f, "{}", v),
Token::Int64(v) => write!(f, "{}", v),
Token::IntFloat(v) => write!(f, "{}_f", v),
Token::Float(v) => write!(f, "{}", v),
Token::Float64(v) => write!(f, "{}", v),
Token::Op(s) => write!(f, "{}", s),
@@ -61,8 +117,36 @@ impl fmt::Display for Token {
}
}
pub fn parse(source: &str) -> Result<ast::Script, ()> {
let tokens = match lexer().parse(source) {
type SourceStream<It> = chumsky::Stream<'static, char, Span, It>;
type TokenStream<It> = chumsky::Stream<'static, Token, Span, It>;
pub fn parse(sources: &Sources, source_id: usize) -> Result<ast::Script, ()> {
let source = &sources[source_id].source;
let source_stream = SourceStream::from_iter(
(source_id, source.len()..source.len() + 1),
Box::new(
// this is a bit of an ugly hack
// once the next version of ariadne is released
// use the source string chars as input
// and only create the ariadne::Source lazily when printing errors
source
.lines()
.flat_map(|l| {
let mut chars: Vec<char> = l.chars().collect();
while chars.len() + 1 < l.len() {
chars.push(' ');
}
if chars.len() < l.len() {
chars.push('\n');
}
chars.into_iter()
})
.enumerate()
.map(|(index, char)| (char, (source_id, index..(index + 1)))),
),
);
let tokens = match lexer().parse(source_stream) {
Ok(tokens) => tokens,
Err(errors) => {
report_errors(
@@ -70,15 +154,14 @@ pub fn parse(source: &str) -> Result<ast::Script, ()> {
.into_iter()
.map(|e| e.map(|c| c.to_string()))
.collect(),
source,
sources,
);
return Err(());
}
};
let source_len = source.chars().count();
let script = match script_parser().parse(Stream::from_iter(
source_len..source_len + 1,
let script = match script_parser().parse(TokenStream::from_iter(
(source_id, source.len()..source.len() + 1),
tokens.into_iter(),
)) {
Ok(script) => script,
@@ -88,7 +171,7 @@ pub fn parse(source: &str) -> Result<ast::Script, ()> {
.into_iter()
.map(|e| e.map(|t| t.to_string()))
.collect(),
source,
sources,
);
return Err(());
}
@@ -96,9 +179,9 @@ pub fn parse(source: &str) -> Result<ast::Script, ()> {
Ok(script)
}
fn report_errors(errors: Vec<Simple<String>>, source: &str) {
fn report_errors(errors: Vec<Simple<String, Span>>, sources: &Sources) {
for error in errors {
let report = Report::build(ReportKind::Error, (), error.span().start());
let report = Report::build(ReportKind::Error, error.span().0, error.span().1.start());
let report = match error.reason() {
chumsky::error::SimpleReason::Unclosed { span, delimiter } => report
@@ -161,30 +244,31 @@ fn report_errors(errors: Vec<Simple<String>>, source: &str) {
),
};
report.finish().eprint(Source::from(source)).unwrap();
report.finish().eprint(sources).unwrap();
}
}
fn lexer() -> impl Parser<char, Vec<(Token, Span)>, Error = Simple<char>> {
let float64 = text::int(10)
type LexerError = Simple<char, Span>;
fn lexer() -> impl Parser<char, Vec<(Token, Span)>, Error = LexerError> {
let float64 = text::digits(10)
.chain::<char, _, _>(just('.').chain(text::digits(10)))
.then_ignore(just("f64"))
.collect::<String>()
.map(Token::Float64);
let float = text::int(10)
let float = text::digits(10)
.chain::<char, _, _>(just('.').chain(text::digits(10)))
.collect::<String>()
.map(Token::Float);
let integer = just::<_, _, Simple<char>>("0x")
.ignore_then(text::int(16))
let integer = just::<_, _, LexerError>("0x")
.ignore_then(text::digits(16))
.try_map(|n, span| {
u64::from_str_radix(&n, 16).map_err(|err| Simple::custom(span, err.to_string()))
u64::from_str_radix(&n, 16).map_err(|err| LexerError::custom(span, err.to_string()))
})
.or(text::int(10).try_map(|n: String, span: Span| {
.or(text::digits(10).try_map(|n: String, span: Span| {
n.parse::<u64>()
.map_err(|err| Simple::custom(span, err.to_string()))
.map_err(|err| LexerError::custom(span, err.to_string()))
}))
.boxed();
@@ -193,28 +277,73 @@ fn lexer() -> impl Parser<char, Vec<(Token, Span)>, Error = Simple<char>> {
.then_ignore(just("i64"))
.map(|n| Token::Int64(n as i64));
let int_float = integer
.clone()
.then_ignore(just("_f"))
.map(|n| Token::IntFloat(n as i32));
let int = integer.try_map(|n, span| {
u32::try_from(n)
.map(|n| Token::Int(n as i32))
.map_err(|err| Simple::custom(span, err.to_string()))
.map_err(|err| LexerError::custom(span, err.to_string()))
});
let str_ = just('"')
.ignore_then(filter(|c| *c != '"').repeated())
.then_ignore(just('"'))
.collect::<String>()
.map(Token::Str);
let op = one_of("+-*/%&^|<=>#")
let str_ = just('\\')
.then(any())
.map(|t| vec![t.0, t.1])
.or(none_of("\"").map(|c| vec![c]))
.repeated()
.at_least(1)
.or(just(':').chain(just('=')))
.flatten()
.delimited_by(just('"'), just('"'))
.collect::<String>()
.map(Token::Op);
.map(|s| Token::Str(parse_string_escapes(s)));
let char_ = just('\\')
.then(any())
.map(|t| vec![t.0, t.1])
.or(none_of("\'").map(|c| vec![c]))
.repeated()
.flatten()
.delimited_by(just('\''), just('\''))
.collect::<String>()
.map(|s| {
let s = parse_string_escapes(s);
let mut value = 0;
for (i, c) in s.chars().enumerate() {
// TODO: generate error on overflow
if i < 4 {
value |= (c as u32) << (i * 8);
}
}
Token::Int(value as i32)
});
let op = choice((
just("#/"),
just("#%"),
just("<<"),
just(">>"),
just("#>>"),
just(">="),
just("<="),
just("=="),
just("!="),
just("#>="),
just("#<="),
just("#<"),
just("#>"),
just("->"),
just(":="),
just("<|"),
))
.map(|s| s.to_string())
.or(one_of("+-*/%&^|<=>").map(|s: char| s.to_string()))
.map(Token::Op)
.boxed();
let ctrl = one_of("(){};,:?!$").map(Token::Ctrl);
fn ident() -> impl Parser<char, String, Error = Simple<char>> + Copy + Clone {
fn ident() -> impl Parser<char, String, Error = LexerError> + Copy + Clone {
filter(|c: &char| c.is_ascii_alphabetic() || *c == '_')
.map(Some)
.chain::<char, Vec<_>, _>(
@@ -248,59 +377,117 @@ fn lexer() -> impl Parser<char, Vec<(Token, Span)>, Error = Simple<char>> {
let comment = single_line.or(multi_line);
let token = float
.or(float64)
.or(int64)
.or(int)
.or(str_)
.or(op)
.or(ctrl)
.or(ident)
.recover_with(skip_then_retry_until([]));
let token = choice((
float, float64, int64, int_float, int, str_, char_, op, ctrl, ident,
))
.recover_with(skip_then_retry_until([]));
token
.map_with_span(|tok, span| (tok, span))
.padded()
.padded_by(comment.padded().repeated())
.repeated()
.boxed()
}
fn map_token<O>(
f: impl Fn(&Token) -> Option<O> + 'static + Clone,
) -> impl Parser<Token, O, Error = Simple<Token>> + Clone {
f: impl Fn(&Token, &Span) -> Option<O> + 'static + Clone,
) -> impl Parser<Token, O, Error = ScriptError> + Clone {
filter_map(move |span, tok: Token| {
if let Some(output) = f(&tok) {
if let Some(output) = f(&tok, &span) {
Ok(output)
} else {
Err(Simple::expected_input_found(span, Vec::new(), Some(tok)))
Err(ScriptError::expected_input_found(
span,
Vec::new(),
Some(tok),
))
}
})
}
fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + Clone {
fn parse_string_escapes(s: String) -> String {
let mut result = String::new();
let mut chars = s.chars().peekable();
while let Some(c) = chars.next() {
if c != '\\' {
result.push(c);
} else if let Some(c) = chars.next() {
match c {
'0'..='9' | 'a'..='f' | 'A'..='F' => {
let mut number = c.to_string();
if let Some('0'..='9' | 'a'..='f' | 'A'..='F') = chars.peek() {
number.push(chars.next().unwrap());
}
result.push(u8::from_str_radix(&number, 16).unwrap() as char);
}
'n' => result.push('\n'),
'r' => result.push('\r'),
't' => result.push('\t'),
other => result.push(other),
}
} else {
result.push('\\');
}
}
result
}
type ScriptError = Simple<Token, Span>;
fn script_parser() -> impl Parser<Token, ast::Script, Error = ScriptError> + Clone {
let identifier = filter_map(|span, tok| match tok {
Token::Ident(id) => Ok(id),
_ => Err(Simple::expected_input_found(span, Vec::new(), Some(tok))),
_ => Err(ScriptError::expected_input_found(
span,
Vec::new(),
Some(tok),
)),
})
.labelled("identifier");
let integer = map_token(|tok| match tok {
let integer = map_token(|tok, _| match tok {
Token::Int(v) => Some(*v),
_ => None,
});
let string = map_token(|tok| match tok {
let string = map_token(|tok, _| match tok {
Token::Str(s) => Some(s.clone()),
_ => None,
});
let product_op = just(Token::Op("*".to_string()))
.to(ast::BinOp::Mul)
.or(just(Token::Op("/".to_string())).to(ast::BinOp::Div))
.or(just(Token::Op("#/".to_string())).to(ast::BinOp::DivU))
.or(just(Token::Op("%".to_string())).to(ast::BinOp::Rem))
.or(just(Token::Op("#%".to_string())).to(ast::BinOp::RemU))
.boxed();
let sum_op = just(Token::Op("+".to_string()))
.to(ast::BinOp::Add)
.or(just(Token::Op("-".to_string())).to(ast::BinOp::Sub))
.boxed();
let shift_op = just(Token::Op("<<".to_string()))
.to(ast::BinOp::Shl)
.or(just(Token::Op("#>>".to_string())).to(ast::BinOp::ShrU))
.or(just(Token::Op(">>".to_string())).to(ast::BinOp::ShrS))
.boxed();
let bit_op = just(Token::Op("&".to_string()))
.to(ast::BinOp::And)
.or(just(Token::Op("|".to_string())).to(ast::BinOp::Or))
.or(just(Token::Op("^".to_string())).to(ast::BinOp::Xor))
.boxed();
let mut expression_out = None;
let block = recursive(|block| {
let mut block_expression = None;
let expression = recursive(|expression| {
let val = map_token(|tok| match tok {
let val = map_token(|tok, span| match tok {
Token::Int(v) => Some(ast::Expr::I32Const(*v)),
Token::Int64(v) => Some(ast::Expr::I64Const(*v)),
Token::IntFloat(v) => Some(ast::Expr::Cast {
value: Box::new(ast::Expr::I32Const(*v).with_span(span.clone())),
type_: ast::Type::F32,
}),
Token::Float(v) => Some(ast::Expr::F32Const(v.parse().unwrap())),
Token::Float64(v) => Some(ast::Expr::F64Const(v.parse().unwrap())),
_ => None,
@@ -312,7 +499,11 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
name: id,
local_id: None,
}),
_ => Err(Simple::expected_input_found(span, Vec::new(), Some(tok))),
_ => Err(ScriptError::expected_input_found(
span,
Vec::new(),
Some(tok),
)),
})
.labelled("variable");
@@ -325,6 +516,36 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
})
.boxed();
let local_tee_op = identifier
.then(
product_op
.clone()
.or(sum_op.clone())
.or(shift_op.clone())
.or(bit_op.clone()),
)
.then_ignore(just(Token::Op(":=".to_string())))
.then(expression.clone())
.map_with_span(|((name, op), expr), span| ast::Expr::LocalTee {
name: name.clone(),
value: Box::new(
ast::Expr::BinOp {
left: Box::new(
ast::Expr::Variable {
name,
local_id: None,
}
.with_span(span.clone()),
),
right: Box::new(expr),
op,
}
.with_span(span),
),
local_id: None,
})
.boxed();
let loop_expr = just(Token::Loop)
.ignore_then(identifier)
.then(block.clone())
@@ -341,15 +562,25 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
block: Box::new(block),
});
let if_expr = just(Token::If)
.ignore_then(expression.clone())
.then(block.clone())
.then(just(Token::Else).ignore_then(block.clone()).or_not())
.map(|((condition, if_true), if_false)| ast::Expr::If {
condition: Box::new(condition),
if_true: Box::new(if_true),
if_false: if_false.map(Box::new),
});
let if_expr = recursive::<_, ast::Expr, _, _, _>(|if_expr| {
just(Token::If)
.ignore_then(expression.clone())
.then(block.clone())
.then(
just(Token::Else)
.ignore_then(
block
.clone()
.or(if_expr.map_with_span(|expr, span| expr.with_span(span))),
)
.or_not(),
)
.map(|((condition, if_true), if_false)| ast::Expr::If {
condition: Box::new(condition),
if_true: Box::new(if_true),
if_false: if_false.map(Box::new),
})
});
let block_expr = loop_expr.or(label_block_expr).or(if_expr).boxed();
@@ -392,16 +623,6 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
})
.boxed();
let assign = identifier
.then_ignore(just(Token::Op("=".to_string())))
.then(expression.clone())
.map(|(name, value)| ast::Expr::Assign {
name,
value: Box::new(value),
local_id: None,
})
.boxed();
let select = just(Token::Select)
.ignore_then(
expression
@@ -435,29 +656,31 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
value: value.map(Box::new),
});
let atom = val
.or(function_call)
.or(assign)
.or(local_tee)
.or(variable)
.or(block_expr)
.or(branch)
.or(branch_if)
.or(let_)
.or(select)
.or(return_)
.map_with_span(|expr, span| expr.with_span(span))
.or(expression
.clone()
.delimited_by(just(Token::Ctrl('(')), just(Token::Ctrl(')'))))
.or(block)
.recover_with(nested_delimiters(
Token::Ctrl('('),
Token::Ctrl(')'),
[(Token::Ctrl('{'), Token::Ctrl('}'))],
|span| ast::Expr::Error.with_span(span),
))
.boxed();
let atom = choice((
val,
function_call,
local_tee,
local_tee_op,
variable,
block_expr,
branch,
branch_if,
let_,
select,
return_,
))
.map_with_span(|expr, span| expr.with_span(span))
.or(expression
.clone()
.delimited_by(just(Token::Ctrl('(')), just(Token::Ctrl(')'))))
.or(block)
.recover_with(nested_delimiters(
Token::Ctrl('('),
Token::Ctrl(')'),
[(Token::Ctrl('{'), Token::Ctrl('}'))],
|span| ast::Expr::Error.with_span(span),
))
.boxed();
let unary_op = just(Token::Op("-".to_string()))
.to(ast::UnaryOp::Negate)
@@ -467,7 +690,7 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
.then(atom)
.map(|(ops, value)| {
ops.into_iter().rev().fold(value, |acc, (op, span)| {
let span = span.start..acc.span.end;
let span = (span.0, span.1.start..acc.span.1.end);
ast::Expr::UnaryOp {
op,
value: Box::new(acc),
@@ -507,7 +730,7 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
poke_op: Option<((ast::MemSize, ast::Expression), ast::Expression)>,
) -> ast::Expression {
let left = peek_ops.into_iter().fold(left, |left, (size, right)| {
let span = left.span.start..right.span.end;
let span = (left.span.0, left.span.1.start..right.span.1.end);
ast::Expr::Peek(ast::MemoryLocation {
span: span.clone(),
left: Box::new(left),
@@ -517,7 +740,7 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
.with_span(span)
});
if let Some(((size, right), value)) = poke_op {
let span = left.span.start..value.span.end;
let span = (left.span.0, left.span.1.start..value.span.1.end);
ast::Expr::Poke {
mem_location: ast::MemoryLocation {
span: span.clone(),
@@ -563,18 +786,9 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
let op_product = memory_op
.clone()
.then(
just(Token::Op("*".to_string()))
.to(ast::BinOp::Mul)
.or(just(Token::Op("/".to_string())).to(ast::BinOp::Div))
.or(just(Token::Op("#/".to_string())).to(ast::BinOp::DivU))
.or(just(Token::Op("%".to_string())).to(ast::BinOp::Rem))
.or(just(Token::Op("#%".to_string())).to(ast::BinOp::RemU))
.then(memory_op.clone())
.repeated(),
)
.then(product_op.clone().then(memory_op.clone()).repeated())
.foldl(|left, (op, right)| {
let span = left.span.start..right.span.end;
let span = (left.span.0, left.span.1.start..right.span.1.end);
ast::Expr::BinOp {
op,
left: Box::new(left),
@@ -586,15 +800,9 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
let op_sum = op_product
.clone()
.then(
just(Token::Op("+".to_string()))
.to(ast::BinOp::Add)
.or(just(Token::Op("-".to_string())).to(ast::BinOp::Sub))
.then(op_product.clone())
.repeated(),
)
.then(sum_op.clone().then(op_product.clone()).repeated())
.foldl(|left, (op, right)| {
let span = left.span.start..right.span.end;
let span = (left.span.0, left.span.1.start..right.span.1.end);
ast::Expr::BinOp {
op,
left: Box::new(left),
@@ -606,16 +814,9 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
let op_shift = op_sum
.clone()
.then(
just(Token::Op("<<".to_string()))
.to(ast::BinOp::Shl)
.or(just(Token::Op("#>>".to_string())).to(ast::BinOp::ShrU))
.or(just(Token::Op(">>".to_string())).to(ast::BinOp::ShrS))
.then(op_sum.clone())
.repeated(),
)
.then(shift_op.clone().then(op_sum.clone()).repeated())
.foldl(|left, (op, right)| {
let span = left.span.start..right.span.end;
let span = (left.span.0, left.span.1.start..right.span.1.end);
ast::Expr::BinOp {
op,
left: Box::new(left),
@@ -643,7 +844,7 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
.repeated(),
)
.foldl(|left, (op, right)| {
let span = left.span.start..right.span.end;
let span = (left.span.0, left.span.1.start..right.span.1.end);
ast::Expr::BinOp {
op,
left: Box::new(left),
@@ -655,16 +856,9 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
let op_bit = op_cmp
.clone()
.then(
just(Token::Op("&".to_string()))
.to(ast::BinOp::And)
.or(just(Token::Op("|".to_string())).to(ast::BinOp::Or))
.or(just(Token::Op("^".to_string())).to(ast::BinOp::Xor))
.then(op_cmp.clone())
.repeated(),
)
.then(bit_op.clone().then(op_cmp.clone()).repeated())
.foldl(|left, (op, right)| {
let span = left.span.start..right.span.end;
let span = (left.span.0, left.span.1.start..right.span.1.end);
ast::Expr::BinOp {
op,
left: Box::new(left),
@@ -682,7 +876,7 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
.repeated(),
)
.foldl(|left, right| {
let span = left.span.start..right.span.end;
let span = (left.span.0, left.span.1.start..right.span.1.end);
ast::Expr::First {
value: Box::new(left),
drop: Box::new(right),
@@ -696,11 +890,58 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
let block_expression = block_expression.unwrap();
let assign = identifier
.then_ignore(just(Token::Op("=".to_string())))
.then(expression.clone())
.map(|(name, value)| ast::Expr::Assign {
name,
value: Box::new(value),
local_id: None,
})
.map_with_span(|expr, span| expr.with_span(span))
.boxed();
let assign_op = identifier
.then(
product_op
.clone()
.or(sum_op.clone())
.or(shift_op.clone())
.or(bit_op.clone()),
)
.then_ignore(just(Token::Op("=".to_string())))
.then(expression.clone())
.map_with_span(|((name, op), value), span| {
ast::Expr::Assign {
name: name.clone(),
value: Box::new(
ast::Expr::BinOp {
left: Box::new(
ast::Expr::Variable {
name,
local_id: None,
}
.with_span(span.clone()),
),
right: Box::new(value),
op,
}
.with_span(span.clone()),
),
local_id: None,
}
.with_span(span)
})
.boxed();
block_expression
.clone()
.then(just(Token::Ctrl(';')).or_not())
.map_with_span(|(expr, semi), span| (expr.with_span(span), semi.is_none()))
.or(expression.clone().then(just(Token::Ctrl(';')).to(false)))
.or(assign
.or(assign_op)
.or(expression.clone())
.then(just(Token::Ctrl(';')).to(false)))
.repeated()
.then(expression.clone().or_not())
.map_with_span(|(mut statements, mut final_expression), span| {
@@ -830,6 +1071,21 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
})
.boxed();
let global_const = just(Token::Ident("const".to_string()))
.ignore_then(identifier)
.then(just(Token::Ctrl(':')).ignore_then(type_parser()).or_not())
.then(just(Token::Op("=".to_string())).ignore_then(expression.clone()))
.then_ignore(just(Token::Ctrl(';')))
.map_with_span(|((name, type_), value), span| {
ast::TopLevelItem::Const(ast::GlobalConst {
name,
type_,
value,
span,
})
})
.boxed();
let data_i8 = just(Token::Ident("i8".to_string()))
.to(ast::DataType::I8)
.or(just(Token::Ident("i16".to_string())).to(ast::DataType::I16))
@@ -875,35 +1131,43 @@ fn script_parser() -> impl Parser<Token, ast::Script, Error = Simple<Token>> + C
})
.boxed();
import.or(function).or(global).or(data).boxed()
let include =
just(Token::Ident("include".to_string())).ignore_then(string.clone().map_with_span(
|path, span| ast::TopLevelItem::Include(ast::Include { span, path }),
));
import
.or(function)
.or(global)
.or(data)
.or(include)
.or(global_const)
.boxed()
};
top_level_item.repeated().then_ignore(end()).map(|items| {
let mut script = ast::Script {
imports: Vec::new(),
global_vars: Vec::new(),
functions: Vec::new(),
data: Vec::new(),
};
let mut script = ast::Script::default();
for item in items {
match item {
ast::TopLevelItem::Import(i) => script.imports.push(i),
ast::TopLevelItem::GlobalVar(v) => script.global_vars.push(v),
ast::TopLevelItem::Function(f) => script.functions.push(f),
ast::TopLevelItem::Data(d) => script.data.push(d),
ast::TopLevelItem::Include(i) => script.includes.push(i),
ast::TopLevelItem::Const(c) => script.consts.push(c),
}
}
script
})
}
fn type_parser() -> impl Parser<Token, ast::Type, Error = Simple<Token>> + Clone {
fn type_parser() -> impl Parser<Token, ast::Type, Error = ScriptError> + Clone {
filter_map(|span, tok| match tok {
Token::Ident(id) if id == "i32" => Ok(ast::Type::I32),
Token::Ident(id) if id == "i64" => Ok(ast::Type::I64),
Token::Ident(id) if id == "f32" => Ok(ast::Type::F32),
Token::Ident(id) if id == "f64" => Ok(ast::Type::F64),
_ => Err(Simple::expected_input_found(
_ => Err(ScriptError::expected_input_found(
span,
vec![
Some(Token::Ident("i32".into())),

View File

@@ -1,9 +1,9 @@
use ariadne::{Color, Label, Report, ReportKind, Source};
use ariadne::{Color, Label, Report, ReportKind};
use std::collections::HashMap;
use crate::ast::{self, MemSize};
use crate::intrinsics::Intrinsics;
use crate::Span;
use crate::parser::{Sources, Span};
use ast::Type::*;
type Result<T> = std::result::Result<T, ()>;
@@ -15,9 +15,9 @@ struct Var {
}
type Vars = HashMap<String, Var>;
pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
pub fn tc_script(script: &mut ast::Script, sources: &Sources) -> Result<()> {
let mut context = Context {
source,
sources,
global_vars: HashMap::new(),
functions: HashMap::new(),
locals: ast::Locals::default(),
@@ -41,7 +41,7 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
"Global already defined",
&import.span,
span,
source,
sources,
);
} else {
context.global_vars.insert(
@@ -64,7 +64,7 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
"Function already defined",
&import.span,
&fnc.span,
source,
sources,
);
} else {
context.functions.insert(
@@ -83,12 +83,12 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
for v in &mut script.global_vars {
if let Some(Var { span, .. }) = context.global_vars.get(&v.name) {
result = report_duplicate_definition("Global already defined", &v.span, span, source);
result = report_duplicate_definition("Global already defined", &v.span, span, sources);
} else {
tc_const(&mut v.value, source)?;
tc_const(&mut v.value, sources)?;
if v.type_ != v.value.type_ {
if v.type_.is_some() {
result = type_mismatch(v.type_, &v.span, v.value.type_, &v.value.span, source);
result = type_mismatch(v.type_, &v.span, v.value.type_, &v.value.span, sources);
} else {
v.type_ = v.value.type_;
}
@@ -104,11 +104,26 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
}
}
for c in &mut script.consts {
tc_const(&mut c.value, sources)?;
if c.value.type_ != c.type_ {
if c.type_.is_some() {
result = type_mismatch(c.type_, &c.span, c.value.type_, &c.value.span, sources);
} else {
c.type_ = c.value.type_;
}
}
}
for f in &script.functions {
let params = f.params.iter().map(|(_, t)| *t).collect();
if let Some(fnc) = context.functions.get(&f.name) {
result =
report_duplicate_definition("Function already defined", &f.span, &fnc.span, source);
result = report_duplicate_definition(
"Function already defined",
&f.span,
&fnc.span,
sources,
);
} else {
context.functions.insert(
f.name.clone(),
@@ -132,7 +147,7 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
.or_else(|| context.global_vars.get(name).map(|v| &v.span))
{
result =
report_duplicate_definition("Variable already defined", &f.span, span, source);
report_duplicate_definition("Variable already defined", &f.span, span, sources);
} else {
context.local_vars.insert(
name.clone(),
@@ -163,7 +178,7 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
f.locals = std::mem::take(&mut context.locals);
if f.body.type_ != f.type_ {
result = type_mismatch(f.type_, &f.span, f.body.type_, &f.body.span, source);
result = type_mismatch(f.type_, &f.span, f.body.type_, &f.body.span, sources);
}
}
@@ -171,7 +186,7 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
for f in &script.functions {
if f.start {
if !f.params.is_empty() || f.type_.is_some() {
Report::build(ReportKind::Error, (), f.span.start)
Report::build(ReportKind::Error, f.span.0, f.span.1.start)
.with_message("Start function can't have params or a return value")
.with_label(
Label::new(f.span.clone())
@@ -179,7 +194,7 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
.with_color(Color::Red),
)
.finish()
.eprint(Source::from(source))
.eprint(sources)
.unwrap();
result = Err(());
@@ -189,7 +204,7 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
"Start function already defined",
&f.span,
&prev.span,
source,
sources,
);
} else {
start_function = Some(f);
@@ -198,14 +213,14 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
}
for data in &mut script.data {
tc_const(&mut data.offset, source)?;
tc_const(&mut data.offset, sources)?;
if data.offset.type_ != Some(I32) {
result = type_mismatch(
Some(I32),
&data.offset.span,
data.offset.type_,
&data.offset.span,
source,
sources,
);
}
for values in &mut data.data {
@@ -220,14 +235,14 @@ pub fn tc_script(script: &mut ast::Script, source: &str) -> Result<()> {
ast::DataType::F64 => ast::Type::F64,
};
for value in values {
tc_const(value, source)?;
tc_const(value, sources)?;
if value.type_ != Some(needed_type) {
result = type_mismatch(
Some(needed_type),
&value.span,
value.type_,
&value.span,
source,
sources,
);
}
}
@@ -247,7 +262,7 @@ struct FunctionType {
}
struct Context<'a> {
source: &'a str,
sources: &'a Sources,
global_vars: Vars,
functions: HashMap<String, FunctionType>,
locals: ast::Locals,
@@ -294,13 +309,13 @@ impl LocalVars {
}
}
fn report_duplicate_definition(
pub fn report_duplicate_definition(
msg: &str,
span: &Span,
prev_span: &Span,
source: &str,
sources: &Sources,
) -> Result<()> {
Report::build(ReportKind::Error, (), span.start)
Report::build(ReportKind::Error, span.0, span.1.start)
.with_message(msg)
.with_label(
Label::new(span.clone())
@@ -313,7 +328,7 @@ fn report_duplicate_definition(
.with_color(Color::Yellow),
)
.finish()
.eprint(Source::from(source))
.eprint(sources)
.unwrap();
Err(())
}
@@ -323,9 +338,9 @@ fn type_mismatch(
span1: &Span,
type2: Option<ast::Type>,
span2: &Span,
source: &str,
sources: &Sources,
) -> Result<()> {
Report::build(ReportKind::Error, (), span2.start)
Report::build(ReportKind::Error, span2.0, span2.1.start)
.with_message("Type mismatch")
.with_label(
Label::new(span1.clone())
@@ -348,13 +363,13 @@ fn type_mismatch(
.with_color(Color::Red),
)
.finish()
.eprint(Source::from(source))
.eprint(sources)
.unwrap();
Err(())
}
fn report_error(msg: &str, span: &Span, source: &str) -> Result<()> {
Report::build(ReportKind::Error, (), span.start)
pub fn report_error(msg: &str, span: &Span, sources: &Sources) -> Result<()> {
Report::build(ReportKind::Error, span.0, span.1.start)
.with_message(msg)
.with_label(
Label::new(span.clone())
@@ -362,29 +377,29 @@ fn report_error(msg: &str, span: &Span, source: &str) -> Result<()> {
.with_color(Color::Red),
)
.finish()
.eprint(Source::from(source))
.eprint(sources)
.unwrap();
Err(())
}
fn expected_type(span: &Span, source: &str) -> Result<()> {
fn expected_type(span: &Span, sources: &Sources) -> Result<()> {
report_error(
"Expected value but found expression of type void",
span,
source,
sources,
)
}
fn unknown_variable(span: &Span, source: &str) -> Result<()> {
report_error("Unknown variable", span, source)
fn unknown_variable(span: &Span, sources: &Sources) -> Result<()> {
report_error("Unknown variable", span, sources)
}
fn immutable_assign(span: &Span, source: &str) -> Result<()> {
report_error("Trying to assign to immutable variable", span, source)
fn immutable_assign(span: &Span, sources: &Sources) -> Result<()> {
report_error("Trying to assign to immutable variable", span, sources)
}
fn missing_label(span: &Span, source: &str) -> Result<()> {
report_error("Label not found", span, source)
fn missing_label(span: &Span, sources: &Sources) -> Result<()> {
report_error("Label not found", span, sources)
}
fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()> {
@@ -423,11 +438,11 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&expr.span,
value.type_,
&value.span,
context.source,
context.sources,
);
}
} else if value.type_.is_none() {
return expected_type(&value.span, context.source);
return expected_type(&value.span, context.sources);
} else {
*type_ = value.type_;
}
@@ -449,7 +464,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
*local_id = Some(id);
context.local_vars.insert(name.clone(), id);
} else {
return report_error("Type missing", &expr.span, context.source);
return report_error("Type missing", &expr.span, context.sources);
}
None
}
@@ -477,7 +492,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&expr.span,
value.type_,
&value.span,
context.source,
context.sources,
);
}
None
@@ -489,7 +504,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
ast::Expr::UnaryOp { op, ref mut value } => {
tc_expression(context, value)?;
if value.type_.is_none() {
return expected_type(&value.span, context.source);
return expected_type(&value.span, context.sources);
}
use ast::Type::*;
use ast::UnaryOp::*;
@@ -502,7 +517,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&expr.span,
value.type_,
&value.span,
context.source,
context.sources,
)
}
})
@@ -521,11 +536,11 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&left.span,
right.type_,
&right.span,
context.source,
context.sources,
);
}
} else {
return expected_type(&left.span, context.source);
return expected_type(&left.span, context.sources);
}
use ast::BinOp::*;
match op {
@@ -537,7 +552,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&left.span,
left.type_,
&left.span,
context.source,
context.sources,
);
} else {
left.type_
@@ -551,7 +566,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&left.span,
left.type_,
&left.span,
context.source,
context.sources,
);
} else {
Some(I32)
@@ -569,7 +584,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
} else if let Some(&Var { type_, .. }) = context.global_vars.get(name) {
Some(type_)
} else {
return unknown_variable(&expr.span, context.source);
return unknown_variable(&expr.span, context.sources);
}
}
ast::Expr::Assign {
@@ -583,7 +598,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
*local_id = Some(id);
let local = &context.locals[id];
if local.index.is_none() {
return immutable_assign(&expr.span, context.source);
return immutable_assign(&expr.span, context.sources);
}
(local.type_, &local.span)
} else if let Some(&Var {
@@ -593,15 +608,15 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
}) = context.global_vars.get(name)
{
if !mutable {
return immutable_assign(&expr.span, context.source);
return immutable_assign(&expr.span, context.sources);
}
(type_, span)
} else {
return unknown_variable(&expr.span, context.source);
return unknown_variable(&expr.span, context.sources);
};
if value.type_ != Some(type_) {
return type_mismatch(Some(type_), span, value.type_, &value.span, context.source);
return type_mismatch(Some(type_), span, value.type_, &value.span, context.sources);
}
None
}
@@ -616,7 +631,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
let local = &context.locals[id];
if local.index.is_none() {
return immutable_assign(&expr.span, context.source);
return immutable_assign(&expr.span, context.sources);
}
if value.type_ != Some(local.type_) {
@@ -625,13 +640,13 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&local.span,
value.type_,
&value.span,
context.source,
context.sources,
);
}
Some(local.type_)
} else {
return unknown_variable(&expr.span, context.source);
return unknown_variable(&expr.span, context.sources);
}
}
ast::Expr::Loop {
@@ -652,13 +667,13 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
context.block_stack.pop();
if block.type_ != None {
// TODO: implement, requires branches to optionally provide values
return type_mismatch(None, &expr.span, block.type_, &block.span, context.source);
return type_mismatch(None, &expr.span, block.type_, &block.span, context.sources);
}
None
}
ast::Expr::Branch(ref label) => {
if !context.block_stack.contains(label) {
return missing_label(&expr.span, context.source);
return missing_label(&expr.span, context.sources);
}
None
}
@@ -673,11 +688,11 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&expr.span,
condition.type_,
&condition.span,
context.source,
context.sources,
);
}
if !context.block_stack.contains(label) {
return missing_label(&expr.span, context.source);
return missing_label(&expr.span, context.sources);
}
None
}
@@ -687,7 +702,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
} => {
tc_expression(context, value)?;
if value.type_.is_none() {
return expected_type(&expr.span, context.source);
return expected_type(&expr.span, context.sources);
}
Some(type_)
}
@@ -698,7 +713,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
for param in params.iter_mut() {
tc_expression(context, param)?;
if param.type_.is_none() {
return expected_type(&param.span, context.source);
return expected_type(&param.span, context.sources);
}
}
if let Some(load) = context.intrinsics.find_load(name) {
@@ -713,11 +728,11 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&expr.span,
value.type_,
&value.span,
context.source,
context.sources,
)?;
}
} else {
return report_error("Missing parameters", &expr.span, context.source);
return report_error("Missing parameters", &expr.span, context.sources);
}
tc_memarg(context, &mut params[1..], &expr.span)?;
None
@@ -732,8 +747,9 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
{
*rtype
} else {
let mut report = Report::build(ReportKind::Error, (), expr.span.start)
.with_message("No matching function found");
let mut report =
Report::build(ReportKind::Error, expr.span.0, expr.span.1.start)
.with_message("No matching function found");
for (params, rtype) in type_map {
let param_str: Vec<_> = params.into_iter().map(|t| t.to_string()).collect();
let msg = format!(
@@ -748,17 +764,14 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
);
report = report.with_label(Label::new(expr.span.clone()).with_message(msg));
}
report
.finish()
.eprint(Source::from(context.source))
.unwrap();
report.finish().eprint(context.sources).unwrap();
return Err(());
}
} else {
return report_error(
&format!("Unknown function {}", name),
&expr.span,
context.source,
context.sources,
);
}
}
@@ -776,7 +789,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&condition.span,
condition.type_,
&condition.span,
context.source,
context.sources,
);
}
if if_true.type_.is_some() {
@@ -786,11 +799,11 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&if_true.span,
if_false.type_,
&if_false.span,
context.source,
context.sources,
);
}
} else {
return expected_type(&if_true.span, context.source);
return expected_type(&if_true.span, context.sources);
}
if_true.type_
}
@@ -809,7 +822,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&if_true.span,
if_false.type_,
&if_false.span,
context.source,
context.sources,
);
} else {
if_true.type_
@@ -827,7 +840,7 @@ fn tc_expression(context: &mut Context, expr: &mut ast::Expression) -> Result<()
&expr.span,
value.type_,
&value.span,
context.source,
context.sources,
);
}
}
@@ -851,14 +864,14 @@ fn tc_mem_location<'a>(
mem_location: &mut ast::MemoryLocation,
) -> Result<()> {
tc_expression(context, &mut mem_location.left)?;
tc_const(&mut mem_location.right, context.source)?;
tc_const(&mut mem_location.right, context.sources)?;
if mem_location.left.type_ != Some(I32) {
return type_mismatch(
Some(I32),
&mem_location.left.span,
mem_location.left.type_,
&mem_location.left.span,
context.source,
context.sources,
);
}
if mem_location.right.type_ != Some(I32) {
@@ -867,20 +880,20 @@ fn tc_mem_location<'a>(
&mem_location.right.span,
mem_location.right.type_,
&mem_location.right.span,
context.source,
context.sources,
);
}
Ok(())
}
fn tc_const(expr: &mut ast::Expression, source: &str) -> Result<()> {
fn tc_const(expr: &mut ast::Expression, sources: &Sources) -> Result<()> {
use ast::Expr::*;
expr.type_ = Some(match expr.expr {
I32Const(_) => I32,
I64Const(_) => I64,
F32Const(_) => F32,
F64Const(_) => F64,
_ => return report_error("Expected constant value", &expr.span, source),
_ => return report_error("Expected constant value", &expr.span, sources),
});
Ok(())
}
@@ -892,16 +905,16 @@ fn tc_memarg(context: &mut Context, params: &mut [ast::Expression], span: &Span)
} else {
"Too many MemArg parameters"
};
return report_error(msg, span, context.source);
return report_error(msg, span, context.sources);
}
for (index, param) in params.iter_mut().enumerate() {
tc_expression(context, param)?;
if param.type_ != Some(I32) {
return type_mismatch(Some(I32), &span, param.type_, &param.span, context.source);
return type_mismatch(Some(I32), &span, param.type_, &param.span, context.sources);
}
if index > 0 {
tc_const(param, context.source)?;
tc_const(param, context.sources)?;
}
if index == 2 {
let align = param.const_i32();
@@ -909,7 +922,7 @@ fn tc_memarg(context: &mut Context, params: &mut [ast::Expression], span: &Span)
return report_error(
&format!("Alignment {} out of range (0-4)", align),
&param.span,
context.source,
context.sources,
);
}
}

15
test/xorshift.cwa Normal file
View File

@@ -0,0 +1,15 @@
// simple test to see whether lazy/inline chains with the same variable compile correctly
fn xorshift(x: i32) -> i32 {
let lazy x = x ^ (x << 13);
let lazy x = x ^ (x #>> 17);
let inline x = x ^ (x << 5);
x
}
fn xorshift2(x: i32) -> i32 {
x ^= x << 13;
x ^= x #>> 17;
x ^= x << 5;
x
}