nicer errors
This commit is contained in:
parent
95c1ca72e6
commit
8f06f8a502
GPG key ID:
649733FCD94AFBBA
14 changed files with 672 additions and 279 deletions
164
Cargo.lock
generated
164
Cargo.lock
generated
|
|
@ -11,6 +11,12 @@ dependencies = [
|
|||
"gimli",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "adler"
|
||||
version = "1.0.2"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "f26201604c87b1e01bd3d98f8d5d9a8fcbb815e8cedb41ffccbeb4bf593a35fe"
|
||||
|
||||
[[package]]
|
||||
name = "ahash"
|
||||
version = "0.8.3"
|
||||
|
|
@ -99,6 +105,30 @@ version = "1.1.0"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "d468802bab17cbc0cc575e9b053f41e72aa36bfa6b7f55e3529ffa43161b97fa"
|
||||
|
||||
[[package]]
|
||||
name = "backtrace"
|
||||
version = "0.3.69"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "2089b7e3f35b9dd2d0ed921ead4f6d318c27680d4a5bd167b3ee120edb105837"
|
||||
dependencies = [
|
||||
"addr2line",
|
||||
"cc",
|
||||
"cfg-if",
|
||||
"libc",
|
||||
"miniz_oxide",
|
||||
"object",
|
||||
"rustc-demangle",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "backtrace-ext"
|
||||
version = "0.2.1"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "537beee3be4a18fb023b570f80e3ae28003db9167a751266b259926e25539d50"
|
||||
dependencies = [
|
||||
"backtrace",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "base64"
|
||||
version = "0.21.4"
|
||||
|
|
@ -554,6 +584,12 @@ version = "0.4.1"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "95505c38b4572b2d910cecb0281560f54b440a19336cbbcb27bf6ce6adc6f5a8"
|
||||
|
||||
[[package]]
|
||||
name = "hermit-abi"
|
||||
version = "0.3.3"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "d77f7ec81a6d05a3abb01ab6eb7590f6083d08449fe5a1c8b1e620283546ccb7"
|
||||
|
||||
[[package]]
|
||||
name = "id-arena"
|
||||
version = "2.2.1"
|
||||
|
|
@ -581,6 +617,23 @@ dependencies = [
|
|||
"serde",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "is-terminal"
|
||||
version = "0.4.9"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "cb0889898416213fab133e1d33a0e5858a48177452750691bde3666d0fdbaf8b"
|
||||
dependencies = [
|
||||
"hermit-abi",
|
||||
"rustix",
|
||||
"windows-sys",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "is_ci"
|
||||
version = "1.1.1"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "616cde7c720bb2bb5824a224687d8f77bfd38922027f01d825cd7453be5099fb"
|
||||
|
||||
[[package]]
|
||||
name = "itertools"
|
||||
version = "0.10.5"
|
||||
|
|
@ -630,6 +683,9 @@ name = "lang"
|
|||
version = "0.1.0"
|
||||
dependencies = [
|
||||
"clap",
|
||||
"leb128",
|
||||
"miette",
|
||||
"thiserror",
|
||||
"wasm-encoder 0.33.2",
|
||||
"wasmtime",
|
||||
]
|
||||
|
|
@ -691,6 +747,47 @@ dependencies = [
|
|||
"autocfg",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "miette"
|
||||
version = "5.10.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "59bb584eaeeab6bd0226ccf3509a69d7936d148cf3d036ad350abe35e8c6856e"
|
||||
dependencies = [
|
||||
"backtrace",
|
||||
"backtrace-ext",
|
||||
"is-terminal",
|
||||
"miette-derive",
|
||||
"once_cell",
|
||||
"owo-colors",
|
||||
"supports-color",
|
||||
"supports-hyperlinks",
|
||||
"supports-unicode",
|
||||
"terminal_size",
|
||||
"textwrap",
|
||||
"thiserror",
|
||||
"unicode-width",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "miette-derive"
|
||||
version = "5.10.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "49e7bc1560b95a3c4a25d03de42fe76ca718ab92d1a22a55b9b4cf67b3ae635c"
|
||||
dependencies = [
|
||||
"proc-macro2",
|
||||
"quote",
|
||||
"syn",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "miniz_oxide"
|
||||
version = "0.7.1"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "e7810e0be55b428ada41041c41f32c9f1a42817901b4ccf45fa3d4b6561e74c7"
|
||||
dependencies = [
|
||||
"adler",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "object"
|
||||
version = "0.32.1"
|
||||
|
|
@ -709,6 +806,12 @@ version = "1.18.0"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "dd8b5dd2ae5ed71462c540258bedcb51965123ad7e7ccf4b9a8cafaa4a63576d"
|
||||
|
||||
[[package]]
|
||||
name = "owo-colors"
|
||||
version = "3.5.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "c1b04fb49957986fdce4d6ee7a65027d55d4b6d2265e5848bbb507b58ccfdb6f"
|
||||
|
||||
[[package]]
|
||||
name = "paste"
|
||||
version = "1.0.14"
|
||||
|
|
@ -951,6 +1054,12 @@ version = "1.11.1"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "942b4a808e05215192e39f4ab80813e599068285906cc91aa64f923db842bd5a"
|
||||
|
||||
[[package]]
|
||||
name = "smawk"
|
||||
version = "0.3.1"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "f67ad224767faa3c7d8b6d91985b78e70a1324408abcb1cfcc2be4c06bc06043"
|
||||
|
||||
[[package]]
|
||||
name = "sptr"
|
||||
version = "0.3.2"
|
||||
|
|
@ -969,6 +1078,34 @@ version = "0.10.0"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "73473c0e59e6d5812c5dfe2a064a6444949f089e20eec9a2e5506596494e4623"
|
||||
|
||||
[[package]]
|
||||
name = "supports-color"
|
||||
version = "2.0.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "4950e7174bffabe99455511c39707310e7e9b440364a2fcb1cc21521be57b354"
|
||||
dependencies = [
|
||||
"is-terminal",
|
||||
"is_ci",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "supports-hyperlinks"
|
||||
version = "2.1.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "f84231692eb0d4d41e4cdd0cabfdd2e6cd9e255e65f80c9aa7c98dd502b4233d"
|
||||
dependencies = [
|
||||
"is-terminal",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "supports-unicode"
|
||||
version = "2.0.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "4b6c2cb240ab5dd21ed4906895ee23fe5a48acdbd15a3ce388e7b62a9b66baf7"
|
||||
dependencies = [
|
||||
"is-terminal",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "syn"
|
||||
version = "2.0.37"
|
||||
|
|
@ -986,6 +1123,27 @@ version = "0.12.11"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "9d0e916b1148c8e263850e1ebcbd046f333e0683c724876bb0da63ea4373dc8a"
|
||||
|
||||
[[package]]
|
||||
name = "terminal_size"
|
||||
version = "0.1.17"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "633c1a546cee861a1a6d0dc69ebeca693bf4296661ba7852b9d21d159e0506df"
|
||||
dependencies = [
|
||||
"libc",
|
||||
"winapi",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "textwrap"
|
||||
version = "0.15.2"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "b7b3e525a49ec206798b40326a44121291b530c963cfb01018f63e135bac543d"
|
||||
dependencies = [
|
||||
"smawk",
|
||||
"unicode-linebreak",
|
||||
"unicode-width",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "thiserror"
|
||||
version = "1.0.49"
|
||||
|
|
@ -1057,6 +1215,12 @@ version = "1.0.12"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "3354b9ac3fae1ff6755cb6db53683adb661634f67557942dea4facebec0fee4b"
|
||||
|
||||
[[package]]
|
||||
name = "unicode-linebreak"
|
||||
version = "0.1.5"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "3b09c83c3c29d37506a3e260c08c03743a6bb66a9cd432c6934ab501a190571f"
|
||||
|
||||
[[package]]
|
||||
name = "unicode-normalization"
|
||||
version = "0.1.22"
|
||||
|
|
|
|||
|
|
@ -5,5 +5,8 @@ edition = "2021"
|
|||
|
||||
[dependencies]
|
||||
clap = { version = "4.4.6", features = ["derive"] }
|
||||
leb128 = "0.2.5"
|
||||
miette = { version = "5.10.0", features = ["fancy"] }
|
||||
thiserror = "1.0.49"
|
||||
wasm-encoder = "0.33.2"
|
||||
wasmtime = "13.0.0"
|
||||
|
|
|
|||
|
|
@ -7,7 +7,10 @@ optimizations
|
|||
- [ ] don't create local variables at all if they're only used once
|
||||
*/
|
||||
|
||||
use std::io::Write;
|
||||
|
||||
use crate::data::{BinaryOp, Expr, Func, Literal, Pattern, PrefixOp, Statement, Type};
|
||||
use crate::generator::OutputFormat;
|
||||
use crate::parser::Context;
|
||||
use crate::Error;
|
||||
// use crate::generator::Generator;
|
||||
|
|
@ -15,24 +18,6 @@ use crate::Error;
|
|||
/// pipeline to convert parsed tree -> wat/wasm
|
||||
pub struct Compiler<'a> {
|
||||
output: Box<&'a mut dyn std::io::Write>,
|
||||
// generator: Generator<'a>,
|
||||
output_type: OutputType,
|
||||
}
|
||||
|
||||
#[derive(Debug, PartialEq, Eq)]
|
||||
pub enum OutputType {
|
||||
Wat,
|
||||
WatVerbose,
|
||||
Wasm,
|
||||
}
|
||||
|
||||
impl OutputType {
|
||||
pub fn is_binary(&self) -> bool {
|
||||
match self {
|
||||
Self::Wat | Self::WatVerbose => false,
|
||||
Self::Wasm => true,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
struct Allocator {
|
||||
|
|
@ -59,20 +44,14 @@ struct Allocator {
|
|||
// "#;
|
||||
|
||||
impl<'a> Compiler<'a> {
|
||||
pub fn new(output: Box<&mut dyn std::io::Write>, output_type: OutputType) -> Compiler {
|
||||
pub fn new(output: Box<&mut dyn Write>, _output_format: OutputFormat) -> Compiler {
|
||||
Compiler {
|
||||
output,
|
||||
output_type,
|
||||
}
|
||||
}
|
||||
|
||||
pub fn write_module(&mut self, stmts: &[Statement]) -> Result<(), Error> {
|
||||
if self.output_type.is_binary() {
|
||||
self.output.write_all(&[0x00, 0x61, 0x73, 0x6d])?; // magic
|
||||
self.output.write_all(&[0x01, 0x00, 0x00, 0x00])?; // version
|
||||
} else {
|
||||
writeln!(self.output, "(module")?;
|
||||
}
|
||||
writeln!(self.output, "(module")?;
|
||||
|
||||
let mut ctx = Context::new();
|
||||
// ctx.funcs.insert("print".to_string(), (vec![("message".to_string(), Type::String)], Type::empty()));
|
||||
|
|
@ -94,7 +73,7 @@ impl<'a> Compiler<'a> {
|
|||
match stmt {
|
||||
Statement::Func(func) => self.write_func(&gctx, &ctx, func)?,
|
||||
Statement::Let(..) | Statement::TailExpr(..) | Statement::Expr (..) => {
|
||||
return Err(Error::syn("incorrect top level statement"))
|
||||
return Err(Error::OtherError("incorrect top level statement".into()))
|
||||
}
|
||||
};
|
||||
}
|
||||
|
|
@ -102,15 +81,9 @@ impl<'a> Compiler<'a> {
|
|||
Ok(())
|
||||
}
|
||||
|
||||
fn write_comment(&mut self, comment: &str) -> Result<(), Error> {
|
||||
if self.output_type == OutputType::WatVerbose {
|
||||
writeln!(self.output, ";; {}", comment)?;
|
||||
}
|
||||
Ok(())
|
||||
}
|
||||
|
||||
fn write_func(&mut self, parent_gctx: &GenContext, ctx: &Context, func: &Func) -> Result<(), Error> {
|
||||
write!(self.output, "(func ${}", func.name)?;
|
||||
|
||||
if func.name == "main" {
|
||||
write!(self.output, " (export \"_start\")")?;
|
||||
} else if func.public {
|
||||
|
|
@ -134,7 +107,7 @@ impl<'a> Compiler<'a> {
|
|||
|
||||
let inferred = func.block.infer(&ctx)?;
|
||||
if func.ret != inferred {
|
||||
return Err(Error::TypeError(format!(
|
||||
return Err(Error::TypeErrorOld(format!(
|
||||
"fn should return {:?}, but instead returns {inferred:?}",
|
||||
func.ret
|
||||
)));
|
||||
|
|
|
|||
50
src/data.rs
50
src/data.rs
|
|
@ -1,6 +1,6 @@
|
|||
// TODO: TypedStatement and TypedExpression?
|
||||
|
||||
use crate::lexer::{Token, Symbol};
|
||||
use crate::lexer::{TokenContent, Symbol};
|
||||
|
||||
#[rustfmt::skip]
|
||||
#[derive(Debug, Clone)]
|
||||
|
|
@ -114,27 +114,27 @@ impl BinaryOp {
|
|||
}
|
||||
}
|
||||
|
||||
pub fn from_token(token: &Token) -> Option<Self> {
|
||||
pub fn from_token(token: &TokenContent) -> Option<Self> {
|
||||
let op = match token {
|
||||
Token::Symbol(Symbol::DoubleStar) => Self::Pow,
|
||||
Token::Symbol(Symbol::Star) => Self::Mul,
|
||||
Token::Symbol(Symbol::Slash) => Self::Div,
|
||||
Token::Symbol(Symbol::Percent) => Self::Mod,
|
||||
Token::Symbol(Symbol::Plus) => Self::Add,
|
||||
Token::Symbol(Symbol::Minus) => Self::Sub,
|
||||
Token::Symbol(Symbol::Shl) => Self::Shl,
|
||||
Token::Symbol(Symbol::Shr) => Self::Shr,
|
||||
Token::Symbol(Symbol::Less) => Self::Less,
|
||||
Token::Symbol(Symbol::LessEq) => Self::LessEq,
|
||||
Token::Symbol(Symbol::Greater) => Self::Greater,
|
||||
Token::Symbol(Symbol::GreaterEq) => Self::GreaterEq,
|
||||
Token::Symbol(Symbol::Eq) => Self::Eq,
|
||||
Token::Symbol(Symbol::Neq) => Self::Neq,
|
||||
Token::Symbol(Symbol::And) => Self::BitAnd,
|
||||
Token::Symbol(Symbol::Carat) => Self::Xor,
|
||||
Token::Symbol(Symbol::Pipe) => Self::BitOr,
|
||||
Token::Symbol(Symbol::DoubleAnd) => Self::LogicAnd,
|
||||
Token::Symbol(Symbol::DoublePipe) => Self::LogicOr,
|
||||
TokenContent::Symbol(Symbol::DoubleStar) => Self::Pow,
|
||||
TokenContent::Symbol(Symbol::Star) => Self::Mul,
|
||||
TokenContent::Symbol(Symbol::Slash) => Self::Div,
|
||||
TokenContent::Symbol(Symbol::Percent) => Self::Mod,
|
||||
TokenContent::Symbol(Symbol::Plus) => Self::Add,
|
||||
TokenContent::Symbol(Symbol::Minus) => Self::Sub,
|
||||
TokenContent::Symbol(Symbol::Shl) => Self::Shl,
|
||||
TokenContent::Symbol(Symbol::Shr) => Self::Shr,
|
||||
TokenContent::Symbol(Symbol::Less) => Self::Less,
|
||||
TokenContent::Symbol(Symbol::LessEq) => Self::LessEq,
|
||||
TokenContent::Symbol(Symbol::Greater) => Self::Greater,
|
||||
TokenContent::Symbol(Symbol::GreaterEq) => Self::GreaterEq,
|
||||
TokenContent::Symbol(Symbol::Eq) => Self::Eq,
|
||||
TokenContent::Symbol(Symbol::Neq) => Self::Neq,
|
||||
TokenContent::Symbol(Symbol::And) => Self::BitAnd,
|
||||
TokenContent::Symbol(Symbol::Carat) => Self::Xor,
|
||||
TokenContent::Symbol(Symbol::Pipe) => Self::BitOr,
|
||||
TokenContent::Symbol(Symbol::DoubleAnd) => Self::LogicAnd,
|
||||
TokenContent::Symbol(Symbol::DoublePipe) => Self::LogicOr,
|
||||
_ => return None,
|
||||
};
|
||||
Some(op)
|
||||
|
|
@ -146,11 +146,11 @@ impl PrefixOp {
|
|||
// todo!(),
|
||||
// }
|
||||
|
||||
pub fn from_token(token: &Token) -> Option<Self> {
|
||||
pub fn from_token(token: &TokenContent) -> Option<Self> {
|
||||
let op = match token {
|
||||
Token::Symbol(Symbol::Minus) => Self::Minus,
|
||||
Token::Symbol(Symbol::Not) => Self::LogicNot,
|
||||
Token::Symbol(Symbol::DoublePipe) => Self::BitNot,
|
||||
TokenContent::Symbol(Symbol::Minus) => Self::Minus,
|
||||
TokenContent::Symbol(Symbol::Not) => Self::LogicNot,
|
||||
TokenContent::Symbol(Symbol::DoublePipe) => Self::BitNot,
|
||||
_ => return None,
|
||||
};
|
||||
Some(op)
|
||||
|
|
|
|||
90
src/error.rs
90
src/error.rs
|
|
@ -1,18 +1,96 @@
|
|||
#[derive(Debug)]
|
||||
use miette::{Diagnostic, SourceSpan, NamedSource};
|
||||
use thiserror::Error;
|
||||
|
||||
#[derive(Debug, Error, Diagnostic)]
|
||||
#[error("some kind of error")]
|
||||
pub enum Error {
|
||||
SyntaxError(String),
|
||||
TypeError(String),
|
||||
SyntaxError(SyntaxError),
|
||||
TypeError(TypeError),
|
||||
TypeErrorOld(String),
|
||||
ReferenceError(String),
|
||||
IoError(std::io::Error),
|
||||
OtherError(String),
|
||||
}
|
||||
|
||||
#[derive(Debug, Error, Diagnostic)]
|
||||
#[error("")]
|
||||
#[diagnostic()]
|
||||
pub struct SyntaxWrapper {
|
||||
#[source_code]
|
||||
pub src: NamedSource,
|
||||
|
||||
#[related]
|
||||
pub syn: Vec<SyntaxError>,
|
||||
}
|
||||
|
||||
#[derive(Debug, Error, Diagnostic)]
|
||||
#[error("")]
|
||||
#[diagnostic()]
|
||||
pub struct TypeWrapper {
|
||||
#[source_code]
|
||||
pub src: NamedSource,
|
||||
|
||||
#[related]
|
||||
pub ty: Vec<TypeError>,
|
||||
}
|
||||
|
||||
#[derive(Debug, Error, Diagnostic)]
|
||||
#[error("syntax error!")]
|
||||
#[diagnostic()]
|
||||
pub struct SyntaxError {
|
||||
// #[source_code]
|
||||
// pub src: NamedSource,
|
||||
|
||||
#[help]
|
||||
pub help: String,
|
||||
|
||||
#[label("i do not know what this is")]
|
||||
pub pos: SourceSpan,
|
||||
}
|
||||
|
||||
#[derive(Debug, Error, Diagnostic)]
|
||||
#[error("type error!")]
|
||||
#[diagnostic()]
|
||||
pub enum TypeError {
|
||||
Unary {
|
||||
#[help]
|
||||
help: String,
|
||||
|
||||
#[label("i do not know what this is")]
|
||||
expr: SourceSpan,
|
||||
|
||||
#[label("i do not know what this is")]
|
||||
operator: SourceSpan,
|
||||
},
|
||||
Binary {
|
||||
#[help]
|
||||
help: String,
|
||||
|
||||
#[label("i do not know what this is")]
|
||||
expr: SourceSpan,
|
||||
|
||||
#[label("i do not know what this is")]
|
||||
operator: SourceSpan,
|
||||
},
|
||||
UnknownType {
|
||||
#[help]
|
||||
help: String,
|
||||
|
||||
#[label("i do not know what this type is")]
|
||||
unknown: SourceSpan,
|
||||
},
|
||||
}
|
||||
|
||||
impl Error {
|
||||
pub fn syn(what: &'static str) -> Error {
|
||||
Error::SyntaxError(what.to_string())
|
||||
pub fn syn(pos: (usize, usize), help: impl Into<String>) -> Error {
|
||||
Error::SyntaxError(SyntaxError {
|
||||
help: help.into(),
|
||||
pos: pos.into(),
|
||||
})
|
||||
}
|
||||
|
||||
pub fn ty(what: &'static str) -> Error {
|
||||
Error::TypeError(what.to_string())
|
||||
Error::TypeErrorOld(what.to_string())
|
||||
}
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -1,28 +1,60 @@
|
|||
// use crate::Error;
|
||||
use crate::Error;
|
||||
// use leb128::write;
|
||||
|
||||
// /// helper pipeline to convert wat/wasm -> actual text/binary
|
||||
// pub struct Generator<'a> {
|
||||
// output: Box<&'a mut dyn std::io::Write>,
|
||||
/// helper pipeline to convert wat/wasm -> actual text/binary
|
||||
pub struct Generator<'a> {
|
||||
output: Box<&'a mut dyn std::io::Write>,
|
||||
format: OutputFormat,
|
||||
}
|
||||
|
||||
#[derive(Debug, PartialEq, Eq)]
|
||||
pub enum OutputFormat {
|
||||
Wat,
|
||||
WatVerbose,
|
||||
Wasm,
|
||||
}
|
||||
|
||||
impl OutputFormat {
|
||||
pub fn is_binary(&self) -> bool {
|
||||
match self {
|
||||
Self::Wat | Self::WatVerbose => false,
|
||||
Self::Wasm => true,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
impl Generator<'_> {
|
||||
pub fn new(output: Box<&mut dyn std::io::Write>, format: OutputFormat) -> Generator {
|
||||
Generator {
|
||||
output,
|
||||
format,
|
||||
}
|
||||
}
|
||||
|
||||
pub fn write_module(&mut self) -> Result<(), Error> {
|
||||
if self.format.is_binary() {
|
||||
self.output.write_all(&[0x00, 0x61, 0x73, 0x6d])?; // magic
|
||||
self.output.write_all(&[0x01, 0x00, 0x00, 0x00])?; // version
|
||||
} else {
|
||||
writeln!(self.output, "(module")?;
|
||||
}
|
||||
|
||||
if self.format.is_binary() {
|
||||
self.output.write_all(&[0x01])?; // type section
|
||||
self.output.write_all(&[0x01])?; // 1 item
|
||||
}
|
||||
Ok(())
|
||||
}
|
||||
|
||||
pub fn write_comment(&mut self, comment: &str) -> Result<(), Error> {
|
||||
if self.format == OutputFormat::WatVerbose {
|
||||
writeln!(self.output, ";; {}", comment)?;
|
||||
}
|
||||
Ok(())
|
||||
}
|
||||
|
||||
// }
|
||||
|
||||
// #[derive(Debug)]
|
||||
// enum OutputFormat {
|
||||
// Wat, Wasm,
|
||||
// }
|
||||
|
||||
// impl Generator<'_> {
|
||||
// pub fn new(output: Box<&mut dyn std::io::Write>) -> Generator {
|
||||
// Generator { output }
|
||||
// }
|
||||
|
||||
// pub fn write_comment(&mut self, comment: &str) -> Result<(), Error> {
|
||||
// writeln!(self.output, ";; {}", comment)?;
|
||||
// Ok(())
|
||||
// }
|
||||
|
||||
// pub fn write_raw(&mut self, text: &str) -> Result<(), Error> {
|
||||
// write!(self.output, "{}", text)?;
|
||||
// Ok(())
|
||||
// }
|
||||
// }
|
||||
pub fn write_raw(&mut self, text: &str) -> Result<(), Error> {
|
||||
write!(self.output, "{}", text)?;
|
||||
Ok(())
|
||||
}
|
||||
}
|
||||
|
|
|
|||
101
src/lexer.rs
101
src/lexer.rs
|
|
@ -4,11 +4,20 @@ use crate::Error;
|
|||
pub struct Lexer {
|
||||
input: Vec<char>,
|
||||
pos: usize,
|
||||
name: String,
|
||||
src: String,
|
||||
}
|
||||
|
||||
#[rustfmt::skip]
|
||||
#[derive(Debug, Clone, PartialEq, Eq)]
|
||||
pub enum Token {
|
||||
pub struct Token {
|
||||
pub token: TokenContent,
|
||||
pub span: (usize, usize),
|
||||
}
|
||||
|
||||
#[rustfmt::skip]
|
||||
#[derive(Debug, Clone, PartialEq, Eq)]
|
||||
pub enum TokenContent {
|
||||
Number { radix: u32, text: String },
|
||||
Ident(String),
|
||||
String(String),
|
||||
|
|
@ -42,10 +51,12 @@ pub enum Symbol {
|
|||
}
|
||||
|
||||
impl Lexer {
|
||||
pub fn new(input: &str) -> Lexer {
|
||||
pub fn new(input: &str, name: &str) -> Lexer {
|
||||
Lexer {
|
||||
input: input.chars().collect(),
|
||||
pos: 0,
|
||||
src: input.to_string(),
|
||||
name: name.to_string(),
|
||||
}
|
||||
}
|
||||
|
||||
|
|
@ -53,6 +64,7 @@ impl Lexer {
|
|||
let Some(&ch) = self.input.get(self.pos) else {
|
||||
return Ok(None);
|
||||
};
|
||||
let start = self.pos;
|
||||
let tok = match ch {
|
||||
'0'..='9' => {
|
||||
let token = self.lex_number()?;
|
||||
|
|
@ -61,37 +73,37 @@ impl Lexer {
|
|||
.get(self.pos)
|
||||
.is_some_and(|c| c.is_ascii_alphanumeric())
|
||||
{
|
||||
panic!("unexpected char");
|
||||
return Err(Error::syn((self.pos, 1), "there shouldn't be extra characters after the number"));
|
||||
}
|
||||
token
|
||||
}
|
||||
'\'' => {
|
||||
self.pos += 1;
|
||||
let ch = self.lex_char()?;
|
||||
if self.input.get(self.pos).is_some_and(|c| *c != '\'') {
|
||||
panic!("expected '");
|
||||
if !self.input.get(self.pos).is_some_and(|c| *c == '\'') {
|
||||
return Err(Error::syn((start, self.pos - start), "close your character with a '"));
|
||||
}
|
||||
self.pos += 1;
|
||||
Token::Char(ch)
|
||||
TokenContent::Char(ch)
|
||||
}
|
||||
'"' => Token::String(self.lex_string()?),
|
||||
'"' => TokenContent::String(self.lex_string()?),
|
||||
ch if ch.is_alphabetic() || ch == '_' => match self.lex_ident().as_str() {
|
||||
"let" => Token::Let,
|
||||
"const" => Token::Const,
|
||||
"type" => Token::Type,
|
||||
"fn" => Token::Fn,
|
||||
"true" => Token::True,
|
||||
"false" => Token::False,
|
||||
"if" => Token::If,
|
||||
"else" => Token::Else,
|
||||
"match" => Token::Match,
|
||||
"while" => Token::While,
|
||||
"loop" => Token::Loop,
|
||||
"for" => Token::For,
|
||||
"break" => Token::Break,
|
||||
"continue" => Token::Continue,
|
||||
"pub" => Token::Pub,
|
||||
ident => Token::Ident(ident.to_string()),
|
||||
"let" => TokenContent::Let,
|
||||
"const" => TokenContent::Const,
|
||||
"type" => TokenContent::Type,
|
||||
"fn" => TokenContent::Fn,
|
||||
"true" => TokenContent::True,
|
||||
"false" => TokenContent::False,
|
||||
"if" => TokenContent::If,
|
||||
"else" => TokenContent::Else,
|
||||
"match" => TokenContent::Match,
|
||||
"while" => TokenContent::While,
|
||||
"loop" => TokenContent::Loop,
|
||||
"for" => TokenContent::For,
|
||||
"break" => TokenContent::Break,
|
||||
"continue" => TokenContent::Continue,
|
||||
"pub" => TokenContent::Pub,
|
||||
ident => TokenContent::Ident(ident.to_string()),
|
||||
},
|
||||
ch if ch.is_whitespace() => {
|
||||
self.pos += 1;
|
||||
|
|
@ -99,10 +111,10 @@ impl Lexer {
|
|||
}
|
||||
_ => self.lex_op()?,
|
||||
};
|
||||
Ok(Some(tok))
|
||||
Ok(Some(Token { token: tok, span: (start, self.pos - start) }))
|
||||
}
|
||||
|
||||
fn lex_number(&mut self) -> Result<Token, Error> {
|
||||
fn lex_number(&mut self) -> Result<TokenContent, Error> {
|
||||
let mut buffer = String::new();
|
||||
let radix = match (self.input[self.pos], self.input.get(self.pos + 1)) {
|
||||
('0', Some(ch)) if ch.is_ascii_digit() => 10,
|
||||
|
|
@ -112,11 +124,12 @@ impl Lexer {
|
|||
'x' => 16,
|
||||
'o' => 8,
|
||||
'b' => 2,
|
||||
'd' => 10,
|
||||
_ if !ch.is_ascii_alphanumeric() => {
|
||||
self.pos -= 2;
|
||||
10
|
||||
}
|
||||
_ => return Err(Error::SyntaxError(format!("unknown number radix {ch}"))),
|
||||
_ => return Err(Error::syn((self.pos - 2, 2), "use a known number radix, like 0x (hex), 0o (octal), 0b (binary), or none/0d (decimal)")),
|
||||
}
|
||||
}
|
||||
_ => 10,
|
||||
|
|
@ -130,17 +143,20 @@ impl Lexer {
|
|||
} else if ch.is_digit(radix) {
|
||||
buffer.push(ch);
|
||||
self.pos += 1;
|
||||
} else if ch.is_ascii_digit() {
|
||||
return Err(Error::syn((self.pos, 1), "you seem to have used the wrong base for this number"));
|
||||
} else {
|
||||
break;
|
||||
}
|
||||
}
|
||||
Ok(Token::Number {
|
||||
Ok(TokenContent::Number {
|
||||
radix,
|
||||
text: buffer,
|
||||
})
|
||||
}
|
||||
|
||||
fn lex_string(&mut self) -> Result<String, Error> {
|
||||
let start = self.pos;
|
||||
self.pos += 1; // take "
|
||||
let mut buffer = String::new();
|
||||
while let Some(&ch) = self.input.get(self.pos) {
|
||||
|
|
@ -150,12 +166,17 @@ impl Lexer {
|
|||
buffer.push(self.lex_char()?);
|
||||
}
|
||||
if !self.input.get(self.pos).is_some_and(|c| *c == '"') {
|
||||
panic!("expected \"");
|
||||
return Err(Error::syn((start, self.pos), "end your string with a \""));
|
||||
}
|
||||
self.pos += 1;
|
||||
Ok(buffer)
|
||||
}
|
||||
|
||||
#[inline]
|
||||
fn err_syn(&self, pos: (usize, usize), help: &str) -> Error {
|
||||
Error::syn(pos, help)
|
||||
}
|
||||
|
||||
#[inline]
|
||||
fn lex_char(&mut self) -> Result<char, Error> {
|
||||
let ch = match self.input.get(self.pos) {
|
||||
|
|
@ -164,7 +185,7 @@ impl Lexer {
|
|||
let ch = self
|
||||
.input
|
||||
.get(self.pos)
|
||||
.ok_or_else(|| Error::syn("expected escape char"))?;
|
||||
.ok_or_else(|| Error::syn((self.pos, 1), "there should be an character here (which will be escaped)"))?;
|
||||
match ch {
|
||||
'n' => '\n',
|
||||
't' => '\t',
|
||||
|
|
@ -172,11 +193,11 @@ impl Lexer {
|
|||
'"' => '\"',
|
||||
// 'x' => '\x',
|
||||
// 'u' => '\u',
|
||||
_ => return Err(Error::syn("unknown escape char")),
|
||||
_ => return Err(Error::syn((self.pos - 1, 2), &format!("i only know how to escape \\n, \\t, \\\\, or \\\", not \\{ch}"))),
|
||||
}
|
||||
}
|
||||
Some(ch) => *ch,
|
||||
None => return Err(Error::syn("expected char")),
|
||||
None => return Err(Error::syn((self.pos, 1), "there should be a character here, not an eof")),
|
||||
};
|
||||
self.pos += 1;
|
||||
Ok(ch)
|
||||
|
|
@ -192,7 +213,7 @@ impl Lexer {
|
|||
rest
|
||||
}
|
||||
|
||||
fn lex_op(&mut self) -> Result<Token, Error> {
|
||||
fn lex_op(&mut self) -> Result<TokenContent, Error> {
|
||||
let ch = self.input[self.pos];
|
||||
|
||||
macro_rules! settable {
|
||||
|
|
@ -220,12 +241,12 @@ impl Lexer {
|
|||
}
|
||||
|
||||
let token = match ch {
|
||||
'(' => Some(Token::OpenParan),
|
||||
')' => Some(Token::CloseParan),
|
||||
'[' => Some(Token::OpenBracket),
|
||||
']' => Some(Token::CloseBracket),
|
||||
'{' => Some(Token::OpenBrace),
|
||||
'}' => Some(Token::CloseBrace),
|
||||
'(' => Some(TokenContent::OpenParan),
|
||||
')' => Some(TokenContent::CloseParan),
|
||||
'[' => Some(TokenContent::OpenBracket),
|
||||
']' => Some(TokenContent::CloseBracket),
|
||||
'{' => Some(TokenContent::OpenBrace),
|
||||
'}' => Some(TokenContent::CloseBrace),
|
||||
_ => None,
|
||||
};
|
||||
|
||||
|
|
@ -339,10 +360,10 @@ impl Lexer {
|
|||
',' => Symbol::Comma,
|
||||
';' => Symbol::Semicolon,
|
||||
'?' => Symbol::Question,
|
||||
_ => return Err(Error::SyntaxError(format!("unexpected character {}", ch))),
|
||||
_ => return Err(Error::syn((self.pos, self.pos), "unexpected character here")),
|
||||
};
|
||||
self.pos += 1;
|
||||
Ok(Token::Symbol(symbol))
|
||||
Ok(TokenContent::Symbol(symbol))
|
||||
}
|
||||
}
|
||||
|
||||
|
|
|
|||
99
src/main.rs
99
src/main.rs
|
|
@ -7,8 +7,8 @@ a better way
|
|||
use std::{path::PathBuf, fs::OpenOptions};
|
||||
|
||||
use clap::Parser;
|
||||
use lang::{compiler::{self, OutputType}, lexer, parser};
|
||||
use compiler::Compiler;
|
||||
use lang::{compiler::Compiler, generator::OutputFormat, lexer, parser, error::{SyntaxWrapper, TypeWrapper}};
|
||||
use miette::NamedSource;
|
||||
|
||||
#[derive(Debug, clap::Parser)]
|
||||
enum Cli {
|
||||
|
|
@ -22,7 +22,7 @@ enum Cli {
|
|||
Compile {
|
||||
#[arg(short, long)]
|
||||
/// What format to output as
|
||||
format: Option<OutputFormat>,
|
||||
format: Option<OutputType>,
|
||||
|
||||
#[arg(short, long)]
|
||||
/// Where to output to
|
||||
|
|
@ -58,40 +58,101 @@ enum Cli {
|
|||
}
|
||||
|
||||
#[derive(Debug, Clone, clap::ValueEnum)]
|
||||
enum OutputFormat {
|
||||
enum OutputType {
|
||||
Wat, Wasm,
|
||||
}
|
||||
|
||||
fn main() {
|
||||
let args = Cli::parse();
|
||||
|
||||
let theme = miette::GraphicalTheme {
|
||||
characters: miette::ThemeCharacters {
|
||||
hbar: '─',
|
||||
vbar: '│',
|
||||
xbar: '┼',
|
||||
vbar_break: '│',
|
||||
uarrow: '┬',
|
||||
rarrow: '>',
|
||||
ltop: '┌',
|
||||
mtop: '┬',
|
||||
rtop: '┐',
|
||||
lbot: '└',
|
||||
rbot: '┘',
|
||||
mbot: '┴',
|
||||
lbox: '[',
|
||||
rbox: ']',
|
||||
lcross: '├',
|
||||
rcross: '┤',
|
||||
underbar: '┬',
|
||||
underline: '─',
|
||||
error: "error:".to_string(),
|
||||
warning: "warn:".to_string(),
|
||||
advice: "note:".to_string(),
|
||||
},
|
||||
..Default::default()
|
||||
};
|
||||
let reporter = miette::GraphicalReportHandler::new().with_theme(theme);
|
||||
|
||||
match args {
|
||||
Cli::Run { file } => {
|
||||
let source = std::fs::read_to_string(file).expect("no source!");
|
||||
let lexer = lexer::Lexer::new(&source);
|
||||
let source = std::fs::read_to_string(&file).expect("no source!");
|
||||
let lexer = lexer::Lexer::new(&source, file.to_str().unwrap());
|
||||
|
||||
let tokens = match lexer.collect::<Result<Vec<_>, _>>() {
|
||||
Ok(tokens) => tokens,
|
||||
Err(error) => {
|
||||
eprintln!("error: {:?}", error);
|
||||
match error {
|
||||
lang::Error::SyntaxError(syn) => {
|
||||
let mut s = String::new();
|
||||
let syn = SyntaxWrapper {
|
||||
src: NamedSource::new(file.to_string_lossy().to_string(), source),
|
||||
syn: vec![syn],
|
||||
};
|
||||
|
||||
reporter.render_report(&mut s, &syn).unwrap();
|
||||
eprintln!("{}", s);
|
||||
}
|
||||
_ => eprintln!("error: {:?}", error),
|
||||
}
|
||||
return;
|
||||
}
|
||||
};
|
||||
|
||||
let mut parser = parser::Parser::new(tokens);
|
||||
let mut parser = parser::Parser::new(tokens, source.clone(), file.to_string_lossy().to_string());
|
||||
let mut statements = vec![];
|
||||
loop {
|
||||
match parser.next() {
|
||||
Ok(None) => break,
|
||||
Ok(Some(tree)) => statements.push(tree),
|
||||
Err(error) => {
|
||||
eprintln!("error: {:?}", error);
|
||||
match error {
|
||||
lang::Error::SyntaxError(syn) => {
|
||||
let mut s = String::new();
|
||||
let syn = SyntaxWrapper {
|
||||
src: NamedSource::new(file.to_string_lossy().to_string(), source),
|
||||
syn: vec![syn],
|
||||
};
|
||||
reporter.render_report(&mut s, &syn).unwrap();
|
||||
eprintln!("{}", s);
|
||||
}
|
||||
lang::Error::TypeError(ty) => {
|
||||
let mut s = String::new();
|
||||
let ty = TypeWrapper {
|
||||
src: NamedSource::new(file.to_string_lossy().to_string(), source),
|
||||
ty: vec![ty],
|
||||
};
|
||||
reporter.render_report(&mut s, &ty).unwrap();
|
||||
eprintln!("{}", s);
|
||||
}
|
||||
_ => eprintln!("error: {:?}", error),
|
||||
}
|
||||
return;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
let mut wat = vec![];
|
||||
let mut gen = Compiler::new(Box::new(&mut wat), OutputType::WatVerbose);
|
||||
let mut gen = Compiler::new(Box::new(&mut wat), OutputFormat::WatVerbose);
|
||||
|
||||
if let Err(err) = gen.write_module(&statements) {
|
||||
panic!("{:?}", err);
|
||||
|
|
@ -108,17 +169,17 @@ fn main() {
|
|||
}
|
||||
Cli::Compile { file, format, output } => {
|
||||
let source = std::fs::read_to_string(&file).expect("no source!");
|
||||
let lexer = lexer::Lexer::new(&source);
|
||||
let lexer = lexer::Lexer::new(&source, file.to_str().unwrap());
|
||||
|
||||
let tokens = match lexer.collect::<Result<Vec<_>, _>>() {
|
||||
Ok(tokens) => tokens,
|
||||
Err(error) => {
|
||||
eprintln!("error: {:?}", error);
|
||||
return;
|
||||
return
|
||||
}
|
||||
};
|
||||
|
||||
let mut parser = parser::Parser::new(tokens);
|
||||
let mut parser = parser::Parser::new(tokens, source.clone(), file.to_string_lossy().to_string());
|
||||
let mut statements = vec![];
|
||||
loop {
|
||||
match parser.next() {
|
||||
|
|
@ -133,9 +194,9 @@ fn main() {
|
|||
|
||||
let format = format.unwrap_or_else(|| {
|
||||
if file.extension().is_some_and(|ext| ext == "wasm") {
|
||||
OutputFormat::Wasm
|
||||
OutputType::Wasm
|
||||
} else {
|
||||
OutputFormat::Wat
|
||||
OutputType::Wat
|
||||
}
|
||||
});
|
||||
|
||||
|
|
@ -145,15 +206,15 @@ fn main() {
|
|||
};
|
||||
|
||||
match format {
|
||||
OutputFormat::Wat => {
|
||||
let mut gen = Compiler::new(Box::new(&mut output), OutputType::WatVerbose);
|
||||
OutputType::Wat => {
|
||||
let mut gen = Compiler::new(Box::new(&mut output), OutputFormat::WatVerbose);
|
||||
|
||||
if let Err(err) = gen.write_module(&statements) {
|
||||
panic!("{:?}", err);
|
||||
}
|
||||
}
|
||||
OutputFormat::Wasm => {
|
||||
let mut gen = Compiler::new(Box::new(&mut output), OutputType::Wasm);
|
||||
OutputType::Wasm => {
|
||||
let mut gen = Compiler::new(Box::new(&mut output), OutputFormat::Wasm);
|
||||
|
||||
if let Err(err) = gen.write_module(&statements) {
|
||||
panic!("{:?}", err);
|
||||
|
|
|
|||
254
src/parser.rs
254
src/parser.rs
|
|
@ -1,12 +1,14 @@
|
|||
use std::collections::HashMap;
|
||||
|
||||
use crate::data::{BinaryOp, Block, Expr, Literal, Pattern, PrefixOp, Statement, Type, Func};
|
||||
use crate::lexer::{Token, Symbol};
|
||||
use crate::lexer::{TokenContent, Symbol, Token};
|
||||
use crate::Error;
|
||||
|
||||
pub struct Parser {
|
||||
tokens: Vec<Token>,
|
||||
pos: usize,
|
||||
src: String,
|
||||
name: String,
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone)]
|
||||
|
|
@ -16,8 +18,28 @@ pub struct Context {
|
|||
}
|
||||
|
||||
impl Parser {
|
||||
pub fn new(tokens: Vec<Token>) -> Parser {
|
||||
Parser { tokens, pos: 0 }
|
||||
pub fn new(tokens: Vec<Token>, src: String, name: String) -> Parser {
|
||||
Parser {
|
||||
src,
|
||||
name,
|
||||
tokens,
|
||||
pos: 0,
|
||||
}
|
||||
}
|
||||
|
||||
fn err_syn(&self, pos: (usize, usize), help: &str) -> Error {
|
||||
Error::syn(pos, help)
|
||||
}
|
||||
|
||||
fn err_eof(&self, wanted: &str) -> Error {
|
||||
self.err_syn(self.tokens.last().map(|tok| tok.span).unwrap_or((0, 0)), &format!("wanted {}, got eof", wanted))
|
||||
}
|
||||
|
||||
fn err_ty_unknown(&self, pos: (usize, usize), help: &str) -> Error {
|
||||
Error::TypeError(crate::error::TypeError::UnknownType {
|
||||
help: help.into(),
|
||||
unknown: pos.into(),
|
||||
})
|
||||
}
|
||||
|
||||
fn peek_tok(&self) -> Option<&Token> {
|
||||
|
|
@ -30,149 +52,143 @@ impl Parser {
|
|||
tok
|
||||
}
|
||||
|
||||
fn eat(&mut self, token: Token) -> Result<&Token, Error> {
|
||||
match self.next_tok() {
|
||||
Some(t) if t == &token => Ok(t),
|
||||
Some(t) => Err(Error::SyntaxError(format!("expected {token:?}, got {t:?}"))),
|
||||
None => Err(Error::SyntaxError(format!("expected {token:?}, got eof"))),
|
||||
}
|
||||
fn eat(&mut self, token: TokenContent) -> Result<Token, Error> {
|
||||
let result = match self.peek_tok() {
|
||||
Some(t) if t.token == token => {
|
||||
let t = t.clone();
|
||||
self.pos += 1;
|
||||
return Ok(t);
|
||||
}
|
||||
Some(t) => Err(self.err_syn(t.span, &format!("expected {token:?}, got {:?}", t.token))),
|
||||
None => Err(self.err_eof(&format!("{:?}", token))),
|
||||
};
|
||||
result
|
||||
}
|
||||
|
||||
pub fn next(&mut self) -> Result<Option<Statement>, Error> {
|
||||
match self.peek_tok() {
|
||||
Some(_) => self.parse_statement(),
|
||||
Some(_) => Some(self.parse_statement()).transpose(),
|
||||
None => Ok(None),
|
||||
}
|
||||
}
|
||||
|
||||
fn parse_statement(&mut self) -> Result<Option<Statement>, Error> {
|
||||
fn parse_statement(&mut self) -> Result<Statement, Error> {
|
||||
let Some(tok) = self.peek_tok() else {
|
||||
return Err(Error::syn("unexpected eof"));
|
||||
return Err(self.err_eof("statement"));
|
||||
};
|
||||
let stmt = match tok {
|
||||
Token::Let => {
|
||||
self.eat(Token::Let)?;
|
||||
let stmt = match tok.token {
|
||||
TokenContent::Let => {
|
||||
self.eat(TokenContent::Let)?;
|
||||
let name = self.parse_ident()?;
|
||||
self.eat(Token::Symbol(Symbol::Set))?;
|
||||
if let Ok(_) = self.eat(TokenContent::Symbol(Symbol::Colon)) {
|
||||
// TODO: types in variables
|
||||
self.parse_type()?;
|
||||
}
|
||||
self.eat(TokenContent::Symbol(Symbol::Set))?;
|
||||
let expr = self.parse_expr(0)?;
|
||||
self.eat(Token::Symbol(Symbol::Semicolon))?;
|
||||
self.eat(TokenContent::Symbol(Symbol::Semicolon))?;
|
||||
Statement::Let(name, expr)
|
||||
}
|
||||
Token::Pub | Token::Fn => {
|
||||
TokenContent::Pub | TokenContent::Fn => {
|
||||
// TODO: public things that aren't functions
|
||||
let public = if tok == &Token::Pub {
|
||||
self.eat(Token::Pub)?;
|
||||
let public = if tok.token == TokenContent::Pub {
|
||||
self.eat(TokenContent::Pub)?;
|
||||
true
|
||||
} else {
|
||||
false
|
||||
};
|
||||
self.eat(Token::Fn)?;
|
||||
self.eat(TokenContent::Fn)?;
|
||||
let name = self.parse_ident()?;
|
||||
self.eat(Token::OpenParan)?;
|
||||
self.eat(TokenContent::OpenParan)?;
|
||||
let mut params = vec![];
|
||||
while !self.peek_tok().is_some_and(|tok| tok == &Token::CloseParan) {
|
||||
while !self.peek_tok().is_some_and(|tok| tok.token == TokenContent::CloseParan) {
|
||||
let name = self.parse_ident()?;
|
||||
self.eat(Token::Symbol(Symbol::Colon))?;
|
||||
self.eat(TokenContent::Symbol(Symbol::Colon))?;
|
||||
let ty = self.parse_type()?;
|
||||
params.push((name, ty));
|
||||
|
||||
if self.peek_tok() == Some(&Token::Symbol(Symbol::Comma)) {
|
||||
self.eat(Token::Symbol(Symbol::Comma))?;
|
||||
} else {
|
||||
if self.eat(TokenContent::Symbol(Symbol::Comma)).is_err() {
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
self.eat(Token::CloseParan)?;
|
||||
let ret = if self.peek_tok().is_some_and(|tok| tok == &Token::Symbol(Symbol::ThinArrow)) {
|
||||
self.eat(Token::Symbol(Symbol::ThinArrow))?;
|
||||
self.eat(TokenContent::CloseParan)?;
|
||||
let ret = if self.eat(TokenContent::Symbol(Symbol::ThinArrow)).is_ok() {
|
||||
self.parse_type()?
|
||||
} else {
|
||||
Type::empty()
|
||||
};
|
||||
self.eat(Token::OpenBrace)?;
|
||||
self.eat(TokenContent::OpenBrace)?;
|
||||
let block = self.parse_block()?;
|
||||
self.eat(Token::CloseBrace)?;
|
||||
self.eat(TokenContent::CloseBrace)?;
|
||||
Statement::Func(Func { name, params, ret, block, public })
|
||||
}
|
||||
_ => {
|
||||
let expr = self.parse_expr(0)?;
|
||||
if self.peek_tok().is_some_and(|tk| tk == &Token::Symbol(Symbol::Semicolon)) {
|
||||
self.eat(Token::Symbol(Symbol::Semicolon))?;
|
||||
if self.peek_tok().is_some_and(|tk| tk.token == TokenContent::Symbol(Symbol::Semicolon)) {
|
||||
self.eat(TokenContent::Symbol(Symbol::Semicolon))?;
|
||||
Statement::Expr(expr)
|
||||
} else {
|
||||
Statement::TailExpr(expr)
|
||||
}
|
||||
}
|
||||
};
|
||||
Ok(Some(stmt))
|
||||
Ok(stmt)
|
||||
}
|
||||
|
||||
fn parse_type(&mut self) -> Result<Type, Error> {
|
||||
let Some(tok) = self.next_tok() else {
|
||||
return Err(Error::syn("unexpected eof"));
|
||||
let Some(tok) = self.next_tok().cloned() else {
|
||||
return Err(self.err_eof("type"));
|
||||
};
|
||||
let ty = match tok {
|
||||
Token::Ident(ident) => match ident.as_str() {
|
||||
let ty = match &tok.token {
|
||||
TokenContent::Ident(ident) => match ident.as_str() {
|
||||
"i32" => Type::Integer,
|
||||
"f64" => Type::Float,
|
||||
"bool" => Type::Boolean,
|
||||
_ => return Err(Error::TypeError(format!("unknown type {ident}"))),
|
||||
_ => return Err(self.err_ty_unknown(tok.span, "use a type that i know (currently only i32 and bool)")),
|
||||
},
|
||||
Token::OpenParan => {
|
||||
TokenContent::OpenParan => {
|
||||
let mut tys = vec![];
|
||||
while !self.peek_tok().is_some_and(|tok| tok == &Token::CloseParan) {
|
||||
while !self.peek_tok().is_some_and(|tok| tok.token == TokenContent::CloseParan) {
|
||||
tys.push(self.parse_type()?);
|
||||
|
||||
if self.peek_tok() == Some(&Token::Symbol(Symbol::Comma)) {
|
||||
self.eat(Token::Symbol(Symbol::Comma))?;
|
||||
} else {
|
||||
if self.eat(TokenContent::Symbol(Symbol::Comma)).is_err() {
|
||||
break;
|
||||
}
|
||||
}
|
||||
self.eat(Token::CloseParan)?;
|
||||
self.eat(TokenContent::CloseParan)?;
|
||||
Type::Tuple(tys)
|
||||
},
|
||||
_ => todo!(),
|
||||
_ => return Err(self.err_syn(tok.span, "this should be a pattern")),
|
||||
};
|
||||
Ok(ty)
|
||||
}
|
||||
|
||||
fn parse_ident(&mut self) -> Result<String, Error> {
|
||||
match self.next_tok() {
|
||||
Some(Token::Ident(ident)) => Ok(ident.to_string()),
|
||||
Some(tk) => {
|
||||
Err(Error::SyntaxError(format!(
|
||||
"expected identifier, got {tk:?}"
|
||||
)))
|
||||
}
|
||||
None => Err(Error::syn("expected identifier, got eof")),
|
||||
}
|
||||
let result = match self.peek_tok() {
|
||||
Some(Token { token: TokenContent::Ident(ident), .. }) => Ok(ident.to_string()),
|
||||
Some(t) => return Err(self.err_syn(t.span, &format!("expected ident, got {:?}", t.token))),
|
||||
None => return Err(self.err_eof("ident")),
|
||||
};
|
||||
self.pos += 1;
|
||||
result
|
||||
}
|
||||
|
||||
fn parse_block(&mut self) -> Result<Block, Error> {
|
||||
let mut statements = vec![];
|
||||
loop {
|
||||
match self.peek_tok() {
|
||||
Some(Token::CloseBrace) => break,
|
||||
Some(_) => (),
|
||||
None => return Err(Error::syn("missing closing brace")),
|
||||
};
|
||||
match self.parse_statement()?.unwrap() {
|
||||
if self.peek_tok().is_some_and(|tok| tok.token == TokenContent::CloseBrace) {
|
||||
break;
|
||||
}
|
||||
match self.parse_statement()? {
|
||||
stmt @ Statement::TailExpr(..) => {
|
||||
statements.push(stmt);
|
||||
match self.peek_tok() {
|
||||
Some(Token::CloseBrace) => break,
|
||||
Some(tok) => return Err(Error::SyntaxError(format!("unexpected token {tok:?}"))),
|
||||
None => return Err(Error::syn("unexpected eof")),
|
||||
};
|
||||
break;
|
||||
}
|
||||
stmt => statements.push(stmt),
|
||||
}
|
||||
match self.peek_tok() {
|
||||
Some(Token::CloseBrace) => break,
|
||||
// Some(_) => return Err(Error::SyntaxError(format!("unexpected token {tok:?}"))),
|
||||
// tok => return Err(Error::SyntaxError(format!("unexpected token {tok:?}"))),
|
||||
Some(Token { token: TokenContent::CloseBrace, .. }) => break,
|
||||
_ => (),
|
||||
};
|
||||
}
|
||||
|
|
@ -180,55 +196,55 @@ impl Parser {
|
|||
}
|
||||
|
||||
fn parse_expr(&mut self, binding: u8) -> Result<Expr, Error> {
|
||||
let tok = self.next_tok().ok_or(Error::syn("expected a token"))?;
|
||||
let mut expr = match tok {
|
||||
Token::Number { radix: _, text } => {
|
||||
let Some(tok) = self.next_tok() else {
|
||||
return Err(self.err_eof("expression"));
|
||||
};
|
||||
let tok = tok.clone();
|
||||
let mut expr = match &tok.token {
|
||||
TokenContent::Number { radix: _, text } => {
|
||||
if text.contains('.') {
|
||||
Expr::Literal(Literal::Float(text.parse().unwrap()))
|
||||
} else {
|
||||
Expr::Literal(Literal::Integer(text.parse().unwrap()))
|
||||
}
|
||||
}
|
||||
Token::Ident(ident) => {
|
||||
TokenContent::Ident(ident) => {
|
||||
let ident = ident.clone();
|
||||
if self.peek_tok().is_some_and(|t| *t == Token::OpenParan) {
|
||||
self.eat(Token::OpenParan)?;
|
||||
if self.eat(TokenContent::OpenParan).is_ok() {
|
||||
let mut params = vec![];
|
||||
while !self.peek_tok().is_some_and(|tok| tok == &Token::CloseParan) {
|
||||
while !self.peek_tok().is_some_and(|tok| tok.token == TokenContent::CloseParan) {
|
||||
params.push(self.parse_expr(0)?);
|
||||
if self.peek_tok() == Some(&Token::Symbol(Symbol::Comma)) {
|
||||
self.eat(Token::Symbol(Symbol::Comma))?;
|
||||
} else {
|
||||
if self.eat(TokenContent::Symbol(Symbol::Comma)).is_err() {
|
||||
break;
|
||||
}
|
||||
}
|
||||
self.eat(Token::CloseParan)?;
|
||||
self.eat(TokenContent::CloseParan)?;
|
||||
Expr::Call(ident, params)
|
||||
} else {
|
||||
Expr::Variable(ident)
|
||||
}
|
||||
}
|
||||
Token::False => Expr::Literal(Literal::Boolean(false)),
|
||||
Token::True => Expr::Literal(Literal::Boolean(true)),
|
||||
Token::String(s) => Expr::Literal(Literal::String(s.to_string())),
|
||||
Token::Char(ch) => Expr::Literal(Literal::Char(*ch)),
|
||||
Token::If => {
|
||||
TokenContent::False => Expr::Literal(Literal::Boolean(false)),
|
||||
TokenContent::True => Expr::Literal(Literal::Boolean(true)),
|
||||
TokenContent::String(s) => Expr::Literal(Literal::String(s.to_string())),
|
||||
TokenContent::Char(ch) => Expr::Literal(Literal::Char(*ch)),
|
||||
TokenContent::If => {
|
||||
let cond = self.parse_expr(0)?;
|
||||
self.eat(Token::OpenBrace)?;
|
||||
self.eat(TokenContent::OpenBrace)?;
|
||||
let block = self.parse_block()?;
|
||||
self.eat(Token::CloseBrace)?;
|
||||
let otherwise = if self.peek_tok().is_some_and(|t| *t == Token::Else) {
|
||||
self.eat(TokenContent::CloseBrace)?;
|
||||
let otherwise = if self.peek_tok().is_some_and(|t| t.token == TokenContent::Else) {
|
||||
self.next_tok();
|
||||
match self.peek_tok() {
|
||||
Some(Token::OpenBrace) => {
|
||||
self.eat(Token::OpenBrace)?;
|
||||
Some(Token { token: TokenContent::OpenBrace, .. }) => {
|
||||
self.eat(TokenContent::OpenBrace)?;
|
||||
let b = Some(self.parse_block()?);
|
||||
self.eat(Token::CloseBrace)?;
|
||||
self.eat(TokenContent::CloseBrace)?;
|
||||
b
|
||||
}
|
||||
Some(Token::If) => Some(Block(vec![Statement::TailExpr(self.parse_expr(0)?)])),
|
||||
Some(_) => return Err(Error::syn("unexpected token")),
|
||||
None => return Err(Error::syn("unexpected eof, wanted body for else")),
|
||||
Some(Token { token: TokenContent::If, .. }) => Some(Block(vec![Statement::TailExpr(self.parse_expr(0)?)])),
|
||||
Some(tk) => return Err(self.err_syn(tk.span, "this should be followed by an if or block")),
|
||||
None => return Err(self.err_eof("if or block")),
|
||||
}
|
||||
} else {
|
||||
None
|
||||
|
|
@ -242,50 +258,48 @@ impl Parser {
|
|||
}
|
||||
Expr::Match(Box::new(cond), map)
|
||||
}
|
||||
Token::Symbol(_) => {
|
||||
let Some(op) = PrefixOp::from_token(tok) else {
|
||||
return Err(Error::SyntaxError(format!("unexpected token {tok:?}")));
|
||||
TokenContent::Symbol(_) => {
|
||||
let Some(op) = PrefixOp::from_token(&tok.token) else {
|
||||
return Err(self.err_syn(tok.span, "this should be changed into a valid operator or removed outright"));
|
||||
};
|
||||
let expr = self.parse_expr(1)?;
|
||||
Expr::Unary(op, Box::new(expr))
|
||||
}
|
||||
Token::Match => {
|
||||
TokenContent::Match => {
|
||||
let expr = self.parse_expr(0)?;
|
||||
let mut arms = vec![];
|
||||
self.eat(Token::OpenBrace)?;
|
||||
self.eat(TokenContent::OpenBrace)?;
|
||||
loop {
|
||||
let pat = self.parse_pattern()?;
|
||||
self.eat(Token::Symbol(Symbol::FatArrow))?;
|
||||
self.eat(TokenContent::Symbol(Symbol::FatArrow))?;
|
||||
let expr = self.parse_expr(0)?;
|
||||
arms.push((pat, expr));
|
||||
|
||||
if self.peek_tok().is_some_and(|t| t == &Token::Symbol(Symbol::Comma)) {
|
||||
self.next_tok();
|
||||
} else {
|
||||
|
||||
if self.eat(TokenContent::Symbol(Symbol::Comma)).is_err() {
|
||||
break;
|
||||
}
|
||||
|
||||
if self.peek_tok().is_none() || self.peek_tok().is_some_and(|t| t == &Token::CloseBrace) {
|
||||
if self.peek_tok().is_none() || self.peek_tok().is_some_and(|t| t.token == TokenContent::CloseBrace) {
|
||||
break;
|
||||
}
|
||||
}
|
||||
self.eat(Token::CloseBrace)?;
|
||||
self.eat(TokenContent::CloseBrace)?;
|
||||
Expr::Match(Box::new(expr), arms)
|
||||
}
|
||||
Token::OpenBrace => {
|
||||
TokenContent::OpenBrace => {
|
||||
let b = Expr::Block(self.parse_block()?);
|
||||
self.eat(Token::CloseBrace)?;
|
||||
self.eat(TokenContent::CloseBrace)?;
|
||||
b
|
||||
}
|
||||
Token::OpenParan => {
|
||||
TokenContent::OpenParan => {
|
||||
let expr = self.parse_expr(0)?;
|
||||
self.eat(Token::CloseParan)?;
|
||||
self.eat(TokenContent::CloseParan)?;
|
||||
expr
|
||||
}
|
||||
_ => return Err(Error::syn("unexpected token")),
|
||||
_ => return Err(self.err_syn(tok.span, "this should be an expression")),
|
||||
};
|
||||
while let Some(next) = self.peek_tok() {
|
||||
let Some(op) = BinaryOp::from_token(next) else {
|
||||
let Some(op) = BinaryOp::from_token(&next.token) else {
|
||||
break;
|
||||
};
|
||||
let (bind_left, bind_right) = op.precedence();
|
||||
|
|
@ -300,18 +314,20 @@ impl Parser {
|
|||
}
|
||||
|
||||
fn parse_pattern(&mut self) -> Result<Pattern, Error> {
|
||||
let tok = self.next_tok().ok_or(Error::syn("expected a token"))?;
|
||||
let pat = match tok {
|
||||
Token::Number { radix: _, text } => {
|
||||
let Some(tok) = self.next_tok() else {
|
||||
return Err(self.err_eof("pattern"));
|
||||
};
|
||||
let pat = match &tok.token {
|
||||
TokenContent::Number { radix: _, text } => {
|
||||
if text.contains('.') {
|
||||
Pattern::Literal(Literal::Float(text.parse().unwrap()))
|
||||
} else {
|
||||
Pattern::Literal(Literal::Integer(text.parse().unwrap()))
|
||||
}
|
||||
}
|
||||
Token::False => Pattern::Literal(Literal::Boolean(false)),
|
||||
Token::True => Pattern::Literal(Literal::Boolean(true)),
|
||||
Token::Ident(s) if s == "_" => Pattern::Wildcard,
|
||||
TokenContent::False => Pattern::Literal(Literal::Boolean(false)),
|
||||
TokenContent::True => Pattern::Literal(Literal::Boolean(true)),
|
||||
TokenContent::Ident(s) if s == "_" => Pattern::Wildcard,
|
||||
_ => todo!("no pattern for {:?} yet", tok),
|
||||
};
|
||||
Ok(pat)
|
||||
|
|
|
|||
|
|
@ -56,7 +56,7 @@ impl Expr {
|
|||
for (arg, (name, ty)) in args.iter().zip(params) {
|
||||
let got = arg.infer(ctx)?;
|
||||
if got != *ty {
|
||||
return Err(Error::TypeError(format!("wrong type for {name}: it should've been {ty:?} but you put a {got:?}", )));
|
||||
return Err(Error::TypeErrorOld(format!("wrong type for {name}: it should've been {ty:?} but you put a {got:?}", )));
|
||||
}
|
||||
}
|
||||
|
||||
|
|
@ -92,7 +92,7 @@ impl BinaryOp {
|
|||
) => T::Boolean,
|
||||
// (B::Add | B::Sub | B::Mul | B::Div, T::Float, T::Float) => T::Float,
|
||||
(op, a, b) => {
|
||||
return Err(Error::TypeError(format!(
|
||||
return Err(Error::TypeErrorOld(format!(
|
||||
"operator {op:?} cannot be applied to {a:?} and {b:?}"
|
||||
)))
|
||||
}
|
||||
|
|
@ -112,7 +112,7 @@ impl PrefixOp {
|
|||
// (U::Minus, T::Float) => T::Float,
|
||||
(U::LogicNot, T::Boolean) => T::Boolean,
|
||||
(op, ty) => {
|
||||
return Err(Error::TypeError(format!(
|
||||
return Err(Error::TypeErrorOld(format!(
|
||||
"operator {op:?} cannot be applied to {ty:?}"
|
||||
)))
|
||||
}
|
||||
|
|
|
|||
3
test/bad
Normal file
3
test/bad
Normal file
|
|
@ -0,0 +1,3 @@
|
|||
fn main() {
|
||||
let a = 10 * true;
|
||||
}
|
||||
42
test/itoa
Normal file
42
test/itoa
Normal file
|
|
@ -0,0 +1,42 @@
|
|||
fn itoa(number: i32) -> i32 {
|
||||
let count: i32 = 0;
|
||||
let is_negative = number < 0;
|
||||
|
||||
if is_negative {
|
||||
number = -number;
|
||||
}
|
||||
|
||||
loop {
|
||||
let ch = (number % 10) + 0x30;
|
||||
store_i32_8(count + 8, ch);
|
||||
|
||||
number = number / 10;
|
||||
count = count + 1;
|
||||
|
||||
if number == 0 {
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if is_negative {
|
||||
store_i32_8(count + 8, 45);
|
||||
count = count + 1;
|
||||
}
|
||||
|
||||
let i: i32 = 0;
|
||||
loop {
|
||||
let swap_to = count - i + 7;
|
||||
let swap_from = i + 8;
|
||||
|
||||
let tmp = load_i32_8(swap_to);
|
||||
store_i32_8(swap_to, load_i32_8(swap_from));
|
||||
store_i32_8(swap_from, tmp);
|
||||
|
||||
i = i + 1;
|
||||
if i >= count / 2 {
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
count
|
||||
}
|
||||
|
|
@ -1,14 +1,14 @@
|
|||
use lang::{lexer::{Lexer, Token, Symbol}, Error};
|
||||
use lang::{lexer::{Lexer, TokenContent, Symbol}, Error};
|
||||
|
||||
#[test]
|
||||
fn test_foo() {
|
||||
let tokens: Result<Vec<_>, Error> = Lexer::new("1 * 5 / 3").collect();
|
||||
let tokens = tokens.expect("should parse");
|
||||
assert_eq!(tokens, vec![
|
||||
Token::Number { radix: 10, text: "1".into() },
|
||||
Token::Symbol(Symbol::Star),
|
||||
Token::Number { radix: 10, text: "5".into() },
|
||||
Token::Symbol(Symbol::Slash),
|
||||
Token::Number { radix: 10, text: "3".into() },
|
||||
TokenContent::Number { radix: 10, text: "1".into() },
|
||||
TokenContent::Symbol(Symbol::Star),
|
||||
TokenContent::Number { radix: 10, text: "5".into() },
|
||||
TokenContent::Symbol(Symbol::Slash),
|
||||
TokenContent::Number { radix: 10, text: "3".into() },
|
||||
])
|
||||
}
|
||||
|
|
|
|||
Loading…
Reference in a new issue