mclang/mclangc
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Cleaned up the code

Browse Source
This commit is contained in:
MCorange 2023-03-21 20:12:24 +02:00
parent 5aa2393c8f
commit 0d66ab1009
11 changed files with 279 additions and 315 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
src/bin/mcl_test_dev.rs
View File

@ -1,5 +1,5 @@
use std::path::PathBuf;
use std::path::{PathBuf, Path};
use std::process::Stdio;
use std::{process, fs};
use clap::Parser;
@ -47,10 +47,10 @@ fn run_test<P: Into<PathBuf> + std::convert::AsRef<std::ffi::OsStr>>(f_in: PathB
let mut command = process::Command::new(compiler);
command.stdout(Stdio::piped());
command.stderr(Stdio::piped());
if !compile_mode {
command.arg("-sq");
} else {
if compile_mode {
command.arg("-cqr");
} else {
command.arg("-sq");
}
command.arg("-i");
@ -63,18 +63,18 @@ fn run_test<P: Into<PathBuf> + std::convert::AsRef<std::ffi::OsStr>>(f_in: PathB
let out = child.wait_with_output()?;
let stdout = out.stdout.iter().map(|c| {
char::from_u32((*c) as u32).expect("Failed to parse stdout char").to_string()
char::from_u32(u32::from(*c)).expect("Failed to parse stdout char").to_string()
}).collect::<String>();
let stderr = out.stderr.iter().map(|c| {
char::from_u32((*c) as u32).expect("Failed to parse stderr char").to_string()
char::from_u32(u32::from(*c)).expect("Failed to parse stderr char").to_string()
}).collect::<String>();
Ok(TestOutput {
stdout: stdout,
stderr: stderr,
stdin: stdin,
stdout,
stderr,
stdin,
status: out.status.code().unwrap()
})
}
@ -91,14 +91,14 @@ fn run_tests(args: Args) -> Result<()>{
let intp = run_test(file.path(), &f_out, &args.compiler_path, false, String::new())?;
let comp = run_test(file.path(), &f_out, &args.compiler_path, true, String::new())?;
compare_results(intp, comp, file.path())?;
compare_results(&intp, &comp, &file.path())?;
}
Ok(())
}
fn compare_results(intp: TestOutput, comp: TestOutput, f_in: PathBuf) -> Result<()> {
fn compare_results(intp: &TestOutput, comp: &TestOutput, f_in: &Path) -> Result<()> {
if intp.stdout != comp.stdout {
println!("{b}[ {r}ERR{rs}{b} ]{rs} {f} compiled and interpreted stdout versions differ", r=color::FG_RED, rs=color::RESET, b=color::BRIGHT, f=f_in.display());

20
src/compile/commands.rs
View File

@ -1,9 +1,9 @@
use std::path::PathBuf;
use std::path::{PathBuf, Path};
use std::process::{Command, Stdio};
use color_eyre::Result;
use crate::info;
pub fn linux_x86_64_compile_and_link(of_a: &PathBuf, of_o: &PathBuf, of_c: &PathBuf, quiet: bool) -> Result<()> {
pub fn linux_x86_64_compile_and_link(of_a: &Path, of_o: &Path, of_c: &Path, quiet: bool) -> Result<()> {
let nasm_args = [
"-felf64",
@ -23,7 +23,7 @@ pub fn linux_x86_64_compile_and_link(of_a: &PathBuf, of_o: &PathBuf, of_c: &Path
return Ok(());
} else {
Command::new("nasm")
.args(&nasm_args)
.args(nasm_args)
.stdout(Stdio::inherit())
.stderr(Stdio::inherit())
.spawn()?
@ -42,7 +42,7 @@ pub fn linux_x86_64_compile_and_link(of_a: &PathBuf, of_o: &PathBuf, of_c: &Path
return Ok(());
} else {
Command::new("ld")
.args(&ld_args)
.args(ld_args)
.stdout(Stdio::inherit())
.stderr(Stdio::inherit())
.spawn()?
@ -60,28 +60,28 @@ pub fn linux_x86_64_compile_and_link(of_a: &PathBuf, of_o: &PathBuf, of_c: &Path
Ok(())
}
pub fn linux_x86_64_run(_bin: &PathBuf, args: Vec<String>, quiet: bool) -> Result<i32> {
pub fn linux_x86_64_run(bin: &Path, args: &[String], quiet: bool) -> Result<i32> {
let bin = PathBuf::from(
format!("./{}", _bin.to_string_lossy())
format!("./{}", bin.to_string_lossy())
);
let mut proc = if cfg!(target_os = "windows") {
return Ok(0);
} else {
Command::new(bin)
.args(&args)
Command::new(bin.clone())
.args(args)
.stdout(Stdio::inherit())
.stderr(Stdio::inherit())
.spawn()?
};
// println!("{}", quiet);
if !quiet {
info!("running {} {}", _bin.to_string_lossy(), args.join(" "));
info!("running {} {}", bin.to_string_lossy(), args.join(" "));
}
let exit = proc.wait()?;
if !quiet {
info!("{} process exited with code {}", _bin.to_string_lossy(), exit);
info!("{} process exited with code {}", bin.to_string_lossy(), exit);
}
Ok(exit.code().unwrap_or(0))

24
src/compile/linux_x86_64.rs
View File

@ -6,9 +6,9 @@ use crate::compile::commands::linux_x86_64_compile_and_link;
use super::commands::linux_x86_64_run;
pub fn compile(tokens: Vec<Operator>, args: Args) -> Result<i32>{
pub fn compile(tokens: &[Operator], args: &Args) -> Result<i32>{
let mut of_c = PathBuf::from(&args.out_file);
let (mut of_o, mut of_a) = if &args.out_file == &crate::DEFAULT_OUT_FILE.to_string() {
let (mut of_o, mut of_a) = if args.out_file == *crate::DEFAULT_OUT_FILE {
let of_o = PathBuf::from("/tmp/mclang_comp.o");
let of_a = PathBuf::from("/tmp/mclang_comp.nasm");
(of_o, of_a)
@ -22,6 +22,8 @@ pub fn compile(tokens: Vec<Operator>, args: Args) -> Result<i32>{
of_o.set_extension("o");
of_a.set_extension("nasm");
let file = fs::File::create(&of_a)?;
let mut writer = BufWriter::new(&file);
@ -72,7 +74,7 @@ pub fn compile(tokens: Vec<Operator>, args: Args) -> Result<i32>{
while ti < tokens.len() {
let token = &tokens[ti];
writeln!(writer, "addr_{}:", ti)?;
writeln!(writer, "addr_{ti}:")?;
match token.typ {
// stack
OpType::PushInt => {
@ -336,7 +338,7 @@ pub fn compile(tokens: Vec<Operator>, args: Args) -> Result<i32>{
}
OpType::End => {
writeln!(writer, " ;; -- end")?;
if ti + 1 != token.jmp as usize {
if ti + 1 != token.jmp {
writeln!(writer, " jmp addr_{}", token.jmp)?;
}
ti += 1;
@ -412,20 +414,18 @@ pub fn compile(tokens: Vec<Operator>, args: Args) -> Result<i32>{
writeln!(writer, " push rax")?;
ti += 1;
},
OpType::None => unreachable!(),
OpType::Macro => unreachable!(),
OpType::Include => unreachable!()
OpType::None | OpType::Macro | OpType::Include => unreachable!()
}
}
writeln!(writer, "addr_{}:", ti)?;
writeln!(writer, "addr_{ti}:")?;
writeln!(writer, " mov rax, 60")?;
writeln!(writer, " mov rdi, 0")?;
writeln!(writer, " syscall")?;
writeln!(writer, "segment .data")?;
for s in 0..strings.len() {
let s_chars = strings[s].chars().map(|c| (c as u32).to_string()).collect::<Vec<String>>();
for (_, s) in strings.iter().enumerate() {
let s_chars = s.chars().map(|c| (c as u32).to_string()).collect::<Vec<String>>();
let s_list = s_chars.join(",");
writeln!(writer, " str_{}: db {} ; {}", s, s_list, strings[s].escape_default())?;
writeln!(writer, " str_{}: db {} ; {}", s, s_list, s.escape_default())?;
}
writeln!(writer, "segment .bss")?;
@ -434,7 +434,7 @@ pub fn compile(tokens: Vec<Operator>, args: Args) -> Result<i32>{
writer.flush()?;
linux_x86_64_compile_and_link(&of_a, &of_o, &of_c, args.quiet)?;
if args.run {
let c = linux_x86_64_run(&of_c, vec![], args.quiet)?;
let c = linux_x86_64_run(&of_c, &[], args.quiet)?;
return Ok(c);
}

4
src/compile/mod.rs
View File

@ -1,5 +1,5 @@
pub mod linux_x86_64;
pub mod commands;
pub const MEM_SZ: u32 = 640 * 1000; // 4kb
pub const STRING_SZ: u32 = 640 * 1000; // 4kb
pub const MEM_SZ: usize = 640 * 1000; // 4kb
pub const STRING_SZ: usize = 640 * 1000; // 4kb

106
src/constants.rs
View File

@ -63,20 +63,20 @@ pub enum OpType {
#[derive(Debug, Clone)]
pub struct Operator {
pub typ: OpType,
pub value: i64,
pub value: usize,
pub text: String, //? only used for OpType::PushStr
pub addr: i64, //? only used for OpType::PushStr
pub jmp: i32,
pub loc: (String, u32, u32)
pub addr: Option<usize>, //? only used for OpType::PushStr
pub jmp: usize,
pub loc: (String, usize, usize)
}
impl Operator {
pub fn new(typ: OpType, value: i64, text: String, file: String, row: u32, col: u32) -> Self {
pub fn new(typ: OpType, value: usize, text: String, file: String, row: usize, col: usize) -> Self {
Self {
typ,
value,
jmp: 0,
addr: -1,
addr: None,
text,
loc: (file, row, col)
}
@ -86,48 +86,48 @@ impl Operator {
impl OpType {
pub fn human(&self) -> String {
match self {
&OpType::PushInt => "Number",
&OpType::PushStr => "String",
&OpType::Print => "print",
&OpType::Dup => "dup",
&OpType::Drop => "drop",
&OpType::Dup2 => "2dup",
&OpType::Rot => "rot",
&OpType::Over => "over",
&OpType::Swap => "swap",
&OpType::Plus => "+",
&OpType::Minus => "-",
&OpType::Equals => "=",
&OpType::Gt => ">",
&OpType::Lt => "<",
&OpType::NotEquals => "!=",
&OpType::Le => "<=",
&OpType::Ge => ">=",
&OpType::Band => "band",
&OpType::Bor => "bor",
&OpType::Shr => "shr",
&OpType::Shl => "shl",
&OpType::DivMod => "divmod",
&OpType::Mul => "*",
&OpType::If => "if",
&OpType::Else => "else",
&OpType::End => "end",
&OpType::While => "while",
&OpType::Do => "do",
&OpType::Macro => "macro",
&OpType::Include => "include",
&OpType::Mem => "mem",
&OpType::Load8 => "!8",
&OpType::Store8 => "@8",
&OpType::Syscall0 => "syscall0",
&OpType::Syscall1 => "syscall1",
&OpType::Syscall2 => "syscall2",
&OpType::Syscall3 => "syscall3",
&OpType::Syscall4 => "syscall4",
&OpType::Syscall5 => "syscall5",
&OpType::Syscall6 => "syscall6",
&OpType::None => "None"
match *self {
OpType::PushInt => "Number",
OpType::PushStr => "String",
OpType::Print => "print",
OpType::Dup => "dup",
OpType::Drop => "drop",
OpType::Dup2 => "2dup",
OpType::Rot => "rot",
OpType::Over => "over",
OpType::Swap => "swap",
OpType::Plus => "+",
OpType::Minus => "-",
OpType::Equals => "=",
OpType::Gt => ">",
OpType::Lt => "<",
OpType::NotEquals => "!=",
OpType::Le => "<=",
OpType::Ge => ">=",
OpType::Band => "band",
OpType::Bor => "bor",
OpType::Shr => "shr",
OpType::Shl => "shl",
OpType::DivMod => "divmod",
OpType::Mul => "*",
OpType::If => "if",
OpType::Else => "else",
OpType::End => "end",
OpType::While => "while",
OpType::Do => "do",
OpType::Macro => "macro",
OpType::Include => "include",
OpType::Mem => "mem",
OpType::Load8 => "!8",
OpType::Store8 => "@8",
OpType::Syscall0 => "syscall0",
OpType::Syscall1 => "syscall1",
OpType::Syscall2 => "syscall2",
OpType::Syscall3 => "syscall3",
OpType::Syscall4 => "syscall4",
OpType::Syscall5 => "syscall5",
OpType::Syscall6 => "syscall6",
OpType::None => "None"
}.to_string()
}
}
@ -135,13 +135,13 @@ impl OpType {
#[derive(Debug, Clone)]
pub struct Token {
pub file: String,
pub line: u32,
pub col: u32,
pub line: usize,
pub col: usize,
pub text: String,
pub typ: TokenType
}
#[derive(Debug, Clone, PartialEq)]
#[derive(Debug, Clone, PartialEq, Copy)]
pub enum TokenType {
Word,
Int,
@ -160,7 +160,7 @@ impl Token {
}
impl TokenType {
pub fn human(&self) -> String {
pub fn human(self) -> String {
match self {
TokenType::Word => "Word",
TokenType::Int => "Int",
@ -170,4 +170,4 @@ impl TokenType {
}
}
pub type Loc = (String, u32, u32);
pub type Loc = (String, usize, usize);

82
src/interpret/linux_x86_64/mod.rs
View File

@ -1,23 +1,20 @@
use crate::{constants::OpType, lerror, error};
use crate::{constants::{OpType, Loc}, lerror, error};
// use crate::util::logger;
use color_eyre::Result;
use eyre::eyre;
mod syscalls;
fn stack_pop(stack: &mut Vec<u64>, pos: &(String, u32, u32)) -> Result<u64> {
match stack.pop() {
Some(i) => Ok(i),
None => {
fn stack_pop(stack: &mut Vec<usize>, pos: &Loc) -> Result<usize> {
if let Some(i) = stack.pop() { Ok(i) } else {
lerror!(&pos.clone(), "Stack underflow");
Err(eyre!("Stack underflow"))
},
}
}
pub fn run(tokens: Vec<crate::constants::Operator>) -> Result<i32>{
let mut stack: Vec<u64> = Vec::new();
pub fn run(tokens: &[crate::constants::Operator]) -> Result<i32>{
let mut stack: Vec<usize> = Vec::new();
let mut ti = 0;
let mut mem: Vec<u8> = vec![0; crate::compile::MEM_SZ as usize + crate::compile::STRING_SZ as usize];
let mut mem: Vec<u8> = vec![0; crate::compile::MEM_SZ + crate::compile::STRING_SZ];
let mut string_idx = 0;
// for token in &tokens {
// println!("{{typ: \"{:?}\", val: {}, jmp: {}}}", token.typ, token.value, token.jmp);
@ -31,21 +28,23 @@ pub fn run(tokens: Vec<crate::constants::Operator>) -> Result<i32>{
// stack
OpType::PushInt => {
stack.push(token.value as u64);
stack.push(token.value);
ti += 1;
},
OpType::PushStr => {
if token.addr < 0 {
stack.push(token.text.len() as u64); // string len
stack.push(string_idx + crate::compile::MEM_SZ as u64);
if token.addr.is_none() {
stack.push(token.text.len()); // string len
stack.push(string_idx + crate::compile::MEM_SZ);
for c in token.text.bytes() {
mem[crate::compile::MEM_SZ as usize + string_idx as usize] = c;
mem[crate::compile::MEM_SZ + string_idx] = c;
string_idx += 1;
}
} else {
stack.push(token.text.len() as u64);
stack.push(token.addr as u64);
stack.push(token.text.len());
if let Some(addr) = token.addr {
stack.push(addr);
}
}
@ -109,15 +108,16 @@ pub fn run(tokens: Vec<crate::constants::Operator>) -> Result<i32>{
}
OpType::Load8 => {
let a = stack_pop(&mut stack, &pos)?;
let byte = mem[a as usize];
stack.push(byte as u64);
let byte = mem[a];
stack.push(byte as usize);
ti += 1;
}
#[allow(clippy::cast_possible_truncation)]
OpType::Store8 => {
let val = stack_pop(&mut stack, &pos)?;
let addr = stack_pop(&mut stack, &pos)?;
mem[addr as usize] = (val & 0xFF) as u8;
mem[addr] = (val & 0xFF) as u8;
ti += 1;
}
@ -137,79 +137,79 @@ pub fn run(tokens: Vec<crate::constants::Operator>) -> Result<i32>{
OpType::Equals => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((b == a) as u64);
stack.push(usize::from(b == a));
ti += 1;
},
OpType::Gt => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((b > a) as u64);
stack.push(usize::from(b > a));
ti += 1;
},
OpType::Lt => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((b < a) as u64);
stack.push(usize::from(b < a));
ti += 1;
},
OpType::NotEquals => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((b != a) as u64);
stack.push(usize::from(b != a));
ti += 1;
},
OpType::Ge => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((b >= a) as u64);
stack.push(usize::from(b >= a));
ti += 1;
},
OpType::Le => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((b <= a) as u64);
stack.push(usize::from(b <= a));
ti += 1;
},
OpType::Band => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((a & b) as u64);
stack.push(a & b);
ti += 1;
}
OpType::Bor => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((a | b) as u64);
stack.push(a | b);
ti += 1;
}
OpType::Shr => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((b >> a) as u64);
stack.push(b >> a);
ti += 1;
}
OpType::Shl => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((b << a) as u64);
stack.push(b << a);
ti += 1;
}
OpType::DivMod => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((b / a) as u64);
stack.push((b % a) as u64);
stack.push(b / a);
stack.push(b % a);
ti += 1;
}
OpType::Mul => {
let a = stack_pop(&mut stack, &pos)?;
let b = stack_pop(&mut stack, &pos)?;
stack.push((b * a) as u64);
stack.push(b * a);
ti += 1;
}
@ -219,18 +219,13 @@ pub fn run(tokens: Vec<crate::constants::Operator>) -> Result<i32>{
let a = stack_pop(&mut stack, &pos)?;
if a == 0 {
// println!("If({ti}) => t: {:?} j: {}", tokens[token.jmp as usize].typ, token.jmp);
ti = (token.jmp) as usize;
ti = token.jmp;
} else {
ti += 1;
}
},
OpType::Else => {
// println!("Else({ti}) => t: {:?} j: {}", tokens[token.jmp as usize].typ, token.jmp);
ti = (token.jmp) as usize;
},
OpType::End => {
// println!("End({ti}) => t: {:?} j: {}", tokens[token.jmp as usize].typ, token.jmp);
ti = (token.jmp) as usize;
OpType::Else | OpType::End => {
ti = token.jmp;
}
OpType::While => {
ti += 1;
@ -238,7 +233,7 @@ pub fn run(tokens: Vec<crate::constants::Operator>) -> Result<i32>{
OpType::Do => {
let a = stack.pop().unwrap();
if a == 0 {
ti = (token.jmp) as usize;
ti = token.jmp;
} else {
ti += 1;
}
@ -263,6 +258,7 @@ pub fn run(tokens: Vec<crate::constants::Operator>) -> Result<i32>{
// println!("yes");
let ret = match rax {
1 => syscalls::sys_write(rax, rdi, rsi, rdx, &mem),
0 => 0, //? temp, so clippy doesnt complain
_ => {
error!("Syscall(3) #{} is not implemented", rax);
return Err(eyre!("Syscall not implemented"));
@ -284,9 +280,7 @@ pub fn run(tokens: Vec<crate::constants::Operator>) -> Result<i32>{
todo!();
// ti += 1;
},
OpType::None => unreachable!(),
OpType::Macro => unreachable!(),
OpType::Include => unreachable!()
OpType::None | OpType::Macro | OpType::Include => unreachable!()
}
}

14
src/interpret/linux_x86_64/syscalls.rs
View File

@ -1,22 +1,20 @@
pub fn sys_write(sys_n: u64, fd: u64, buff: u64, count: u64, mem: &Vec<u8> ) -> u64 {
pub fn sys_write(sys_n: usize, fd: usize, buff: usize, count: usize, mem: &Vec<u8> ) -> usize {
let mem = (*mem).clone();
let buff = buff as usize;
let count = count as usize;
// println!("{:?}", &mem[buff..(buff + count)]);
// return 0 ;
let s = &mem[buff..(buff + count)].iter().map(|i| {
char::from_u32((*i) as u32).unwrap_or('_').to_string()
}).collect::<Vec<String>>().join("");
char::from_u32(u32::from(*i)).unwrap_or('_').to_string()
}).collect::<String>();
match fd {
1 => {
print!("{}", s);
print!("{s}");
},
2 => {
eprint!("{}", s);
eprint!("{s}");
},
_ => panic!("Unknown file {}", fd)
_ => panic!("Unknown file {fd}")
};
let _ = std::io::Write::flush(&mut std::io::stdout());
let _ = std::io::Write::flush(&mut std::io::stderr());

68
src/lexer.rs
View File

@ -5,45 +5,45 @@ use color_eyre::Result;
fn lex_word(s: String, tok_type: TokenType) -> (TokenType, String) {
match s {
s if s.parse::<u64>().is_ok() && tok_type == TokenType::Word => { // negative numbers not yet implemented
return (TokenType::Int, s);
(TokenType::Int, s)
},
s if tok_type == TokenType::Word => {
return (TokenType::Word, s);
(TokenType::Word, s)
},
s if tok_type == TokenType::String => {
return (TokenType::String, s);
(TokenType::String, s)
}
s if tok_type == TokenType::Char=> {
return (TokenType::Char, s);
s if tok_type == TokenType::Char => {
(TokenType::Char, s)
}
_ => unreachable!()
}
}
pub fn find_col<F>(text: String, mut col: u32, predicate: F) -> Result<u32> where F: Fn(char, char) -> bool {
pub fn find_col<F>(text: &str, mut col: usize, predicate: F) -> usize where F: Fn(char, char) -> bool {
let mut last = '\0';
while (col as usize) < text.len() && !predicate(text.chars().nth(col as usize).unwrap(), last) {
last = text.chars().nth(col as usize).unwrap();
while col < text.len() && !predicate(text.chars().nth(col).unwrap(), last) {
last = text.chars().nth(col).unwrap();
col += 1;
}
Ok(col)
col
}
// TODO: Implement multiline strings
fn lex_line(text: String) -> Result<Vec<(u32, String, TokenType)>> {
let mut tokens: Vec<(u32, String, TokenType)> = Vec::new();
fn lex_line(text: &str) -> Vec<(usize, String, TokenType)> {
let mut tokens: Vec<(usize, String, TokenType)> = Vec::new();
let mut col = find_col(text.clone(), 0, |x, _| !x.is_whitespace())?;
let mut col_end: u32 = 0;
while col_end < text.clone().len() as u32 {
if (text.len() - col as usize) < 1 {
return Ok(tokens);
let mut col = find_col(text, 0, |x, _| !x.is_whitespace());
let mut col_end: usize = 0;
while col_end < text.to_string().len() {
if (text.len() - col) < 1 {
return tokens;
}
if &text[(col as usize)..(col + 1) as usize] == "\"" {
col_end = find_col(text.clone(), col + 1, |x, x2| x == '"' && x2 != '\\')?;
let t = &text[((col + 1) as usize)..(col_end as usize)];
if &text[col..=col] == "\"" {
col_end = find_col(text, col + 1, |x, x2| x == '"' && x2 != '\\');
let t = &text[(col + 1)..col_end];
let t = t.replace("\\n", "\n")
.replace("\\t", "\t")
.replace("\\r", "\r")
@ -53,11 +53,11 @@ fn lex_line(text: String) -> Result<Vec<(u32, String, TokenType)>> {
if !t.is_empty() {
tokens.push((col, t.to_string(), TokenType::String));
}
col = find_col(text.clone(), col_end + 1, |x, _| !x.is_whitespace())?;
col = find_col(text, col_end + 1, |x, _| !x.is_whitespace());
} else if &text[(col as usize)..(col + 1) as usize] == "'"{
col_end = find_col(text.clone(), col + 1, |x, x2| x == '\'' && x2 != '\\')?;
let t = &text[((col + 1) as usize)..(col_end as usize)];
} else if &text[col..=col] == "'"{
col_end = find_col(text, col + 1, |x, x2| x == '\'' && x2 != '\\');
let t = &text[(col + 1)..col_end];
let t = t.replace("\\n", "\n")
.replace("\\t", "\t")
.replace("\\r", "\r")
@ -69,44 +69,44 @@ fn lex_line(text: String) -> Result<Vec<(u32, String, TokenType)>> {
if !t.is_empty() {
tokens.push((col, t.to_string(), TokenType::Char));
}
col = find_col(text.clone(), col_end + 1, |x, _| !x.is_whitespace())?;
col = find_col(text, col_end + 1, |x, _| !x.is_whitespace());
} else {
col_end = find_col(text.clone(), col, |x, _| x.is_whitespace())?;
let t = &text[(col as usize)..((col_end as usize))];
col_end = find_col(text, col, |x, _| x.is_whitespace());
let t = &text[col..col_end];
if t == "//" {
return Ok(tokens);
return tokens;
}
if !t.is_empty() {
tokens.push((col, t.to_string(), TokenType::Word));
}
col = find_col(text.clone(), col_end, |x, _| !x.is_whitespace())?;
col = find_col(text, col_end, |x, _| !x.is_whitespace());
}
}
Ok(tokens)
tokens
}
pub fn lex(code: String, file: &String, args: Args, preprocessing: bool) -> Result<Vec<Token>> {
pub fn lex<S: Into<String> + std::marker::Copy>(code: &str, file: S, args: &Args, preprocessing: bool) -> Result<Vec<Token>> {
let lines: Vec<(usize, &str)> = code
.split(['\n', '\r'])
.enumerate()
.collect();
let lines: Vec<(u32, String)> = lines.iter().map(|i| (i.0 as u32, i.1.to_string())).collect();
let lines: Vec<(usize, String)> = lines.iter().map(|i| (i.0, i.1.to_string())).collect();
let mut tokens: Vec<Token> = Vec::new();
for (row, line) in lines {
let lt = lex_line(line)?;
let lt = lex_line(&line);
for (col, tok, tok_type) in lt {
let (tok_type, tok) = lex_word(tok, tok_type);
let t = Token{
file: file.clone(),
file: file.into(),
line: row + 1,
col: col,
col,
text: tok,
typ: tok_type
};

35
src/main.rs
View File

@ -8,7 +8,6 @@ mod preprocessor;
use std::fs;
use color_eyre::Result;
use clap::Parser;
pub const DEFAULT_OUT_FILE: &str = "a.out";
@ -54,53 +53,39 @@ pub struct Args {
}
fn main() -> Result<()> {
fn main() {
let args = Args::parse();
let code = match fs::read_to_string(&args.in_file) {
Ok(t) => t,
Err(_) => {
let Ok(code) = fs::read_to_string(&args.in_file) else {
error!("Failed to read file {}, exiting!", &args.in_file);
return Ok(());
}
return;
};
let tokens = match lexer::lex(code, &args.in_file, args.clone(), true) {
Ok(t) => t,
Err(_) => {
let Ok(tokens) = lexer::lex(&code, &args.in_file, &args, true) else {
error!("Lexing failed, exiting!");
return Ok(());
}
return;
};
let mut parser = parser::Parser::new(tokens);
let tokens = match parser.parse() {
Ok(t) => t,
Err(_) => {
let Ok(tokens) = parser.parse() else {
error!("Parsing failed, exiting!");
return Ok(());
}
return;
};
let c = if args.compile && args.interpret {
error!("Cannot compile and interpret at the same time");
0
} else if args.interpret {
match interpret::linux_x86_64::run(tokens) {
Ok(c) => c,
Err(_) => {
if let Ok(c) = interpret::linux_x86_64::run(&tokens) { c } else {
error!("Interpretation failed, exiting!");
1
}
}
} else if args.compile {
match compile::linux_x86_64::compile(tokens, args) {
Ok(c) => c,
Err(_) => {
if let Ok(c) = compile::linux_x86_64::compile(&tokens, &args) { c } else {
error!("Compilation failed, exiting!");
1
}
}
} else {
error!("Did not choose to compile or to interpret, exiting");
0

142
src/parser.rs
View File

@ -1,61 +1,56 @@
use std::{collections::HashMap, ops::Deref};
use std::ops::Deref;
use crate::{constants::{Operator, OpType, Token, TokenType}, lerror};
use crate::{constants::{Operator, OpType, Token, TokenType, Loc}, lerror};
use color_eyre::Result;
use eyre::eyre;
pub fn cross_ref(mut program: Vec<Operator>) -> Result<Vec<Operator>> {
let mut stack: Vec<u32> = Vec::new();
let mut stack: Vec<usize> = Vec::new();
for ip in 0..program.len() {
let op = &program.clone()[ip];
match op.typ {
OpType::If => {
stack.push(ip as u32)
OpType::If | OpType::While => {
stack.push(ip);
}
OpType::Else => {
let if_ip = stack.pop().unwrap();
if program[if_ip as usize].typ != OpType::If {
if program[if_ip].typ != OpType::If {
lerror!(&op.clone().loc,"'end' can only close 'if' blocks");
return Err(eyre!("Bad block"));
}
// let mut if_og = &mut tokens[if_ip as usize];
// (*if_og).jmp = (ip + 1) as i32;
program[if_ip as usize].jmp = (ip + 1) as i32;
stack.push(ip as u32);
program[if_ip].jmp = ip + 1;
stack.push(ip);
},
OpType::End => {
let block_ip = stack.pop().unwrap();
// let mut block_og = &mut tokens[block_ip as usize].clone();
if program[block_ip as usize].typ == OpType::If ||
program[block_ip as usize].typ == OpType::Else {
program[block_ip as usize].jmp = ip as i32;
program[ip as usize].jmp = (ip + 1)as i32;
if program[block_ip].typ == OpType::If ||
program[block_ip].typ == OpType::Else {
} else if program[block_ip as usize].typ == OpType::Do {
program[ip].jmp = program[block_ip as usize].jmp;
program[block_ip as usize].jmp = (ip + 1) as i32;
program[block_ip].jmp = ip;
program[ip].jmp = ip + 1;
} else if program[block_ip].typ == OpType::Do {
program[ip].jmp = program[block_ip].jmp;
program[block_ip].jmp = ip + 1;
} else {
lerror!(&op.clone().loc,"'end' can only close 'if' blocks");
return Err(eyre!(""));
}
}
OpType::While => {
stack.push(ip as u32);
}
OpType::Do => {
let while_ip = stack.pop().unwrap();
program[ip as usize].jmp = while_ip as i32;
stack.push(ip as u32);
program[ip].jmp = while_ip;
stack.push(ip);
}
_ => ()
}
}
if stack.len() > 0 {
lerror!(&program[stack.pop().expect("Empy stack") as usize].clone().loc,"Unclosed block");
if !stack.is_empty() {
lerror!(&program[stack.pop().expect("Empy stack")].clone().loc,"Unclosed block");
return Err(eyre!("Unclosed block"));
}
@ -83,11 +78,11 @@ impl Parser {
let pos = (token.file.clone(), token.line, token.col);
match token.typ {
TokenType::Word => {
let word_type = lookup_word(token.text.clone(), &pos);
let word_type = lookup_word(&token.text, &pos);
tokens.push(Operator::new(word_type, 0, token.text.clone(), token.file.clone(), token.line, token.col));
},
TokenType::Int => {// negative numbers not yet implemented
tokens.push(Operator::new(OpType::PushInt, token.text.parse::<i64>()?, String::new(), token.file.clone(), token.line, token.col));
tokens.push(Operator::new(OpType::PushInt, token.text.parse::<usize>()?, String::new(), token.file.clone(), token.line, token.col));
},
TokenType::String => {
tokens.push(Operator::new(OpType::PushStr, 0, token.text.clone(), token.file.clone(), token.line, token.col));
@ -99,7 +94,7 @@ impl Parser {
return Err(eyre!(""));
}
tokens.push(Operator::new(OpType::PushInt, token.text.chars().next().unwrap() as i64, String::new(), token.file.clone(), token.line, token.col));
tokens.push(Operator::new(OpType::PushInt, token.text.chars().next().unwrap() as usize, String::new(), token.file.clone(), token.line, token.col));
}
};
@ -107,66 +102,61 @@ impl Parser {
//"print" => tokens.push(Operator::new(OpType::Print, 0, token.file.clone(), token.line, token.col)),
}
Ok(cross_ref(tokens)?)
cross_ref(tokens)
}
}
pub fn lookup_word<P: Deref<Target = (String, u32, u32)>>(s: String, _pos: P) -> OpType {
let lookup_table: HashMap<&str, OpType> = HashMap::from([
pub fn lookup_word<P: Deref<Target = Loc>>(s: &str, _pos: P) -> OpType {
match s {
//stack
("print", OpType::Print),
("dup", OpType::Dup),
("drop", OpType::Drop),
("2dup", OpType::Dup2),
("rot", OpType::Rot),
("over", OpType::Over),
("swap", OpType::Swap),
"print" => OpType::Print,
"dup" => OpType::Dup,
"drop" => OpType::Drop,
"2dup" => OpType::Dup2,
"rot" => OpType::Rot,
"over" => OpType::Over,
"swap" => OpType::Swap,
// comp and math
("+", OpType::Plus),
("-", OpType::Minus),
("=", OpType::Equals),
("!=", OpType::NotEquals),
(">", OpType::Gt),
("<", OpType::Lt),
(">=", OpType::Ge),
("<=", OpType::Le),
"+" => OpType::Plus,
"-" => OpType::Minus,
"=" => OpType::Equals,
"!=" => OpType::NotEquals,
">" => OpType::Gt,
"<" => OpType::Lt,
">=" => OpType::Ge,
"<=" => OpType::Le,
("band", OpType::Band),
("bor", OpType::Bor),
("shr", OpType::Shr),
("shl", OpType::Shl),
("divmod", OpType::DivMod),
("*", OpType::Mul),
"band" => OpType::Band,
"bor" => OpType::Bor,
"shr" => OpType::Shr,
"shl" => OpType::Shl,
"divmod" => OpType::DivMod,
"*" => OpType::Mul,
// block
("if", OpType::If),
("else", OpType::Else),
("end", OpType::End),
("while", OpType::While),
("do", OpType::Do),
("macro", OpType::Macro),
("include", OpType::Include), // technically not but it fits next to macros
"if" => OpType::If,
"else" => OpType::Else,
"end" => OpType::End,
"while" => OpType::While,
"do" => OpType::Do,
"macro" => OpType::Macro,
"include" => OpType::Include,
// mem
("mem", OpType::Mem),
("!8", OpType::Load8),
("@8", OpType::Store8),
"mem" => OpType::Mem,
"!8" => OpType::Load8,
"@8" => OpType::Store8,
("syscall0", OpType::Syscall0),
("syscall1", OpType::Syscall1),
("syscall2", OpType::Syscall2),
("syscall3", OpType::Syscall3),
("syscall4", OpType::Syscall4),
("syscall5", OpType::Syscall5),
("syscall6", OpType::Syscall6),
]);
"syscall0" => OpType::Syscall0,
"syscall1" => OpType::Syscall1,
"syscall2" => OpType::Syscall2,
"syscall3" => OpType::Syscall3,
"syscall4" => OpType::Syscall4,
"syscall5" => OpType::Syscall5,
"syscall6" => OpType::Syscall6,
_ => OpType::None
}
match lookup_table.get(s.as_str()) {
Some(v) => v.clone(),
None => {
OpType::None
}
}
}

47
src/preprocessor.rs
View File

@ -15,19 +15,19 @@ pub struct Macro {
pub tokens: Vec<Token>
}
pub fn preprocess(tokens: Vec<Token>, args: Args) -> Result<Vec<Token>>{
pub fn preprocess(tokens: Vec<Token>, args: &Args) -> Result<Vec<Token>>{
let mut program: Vec<Token> = Vec::new();
let mut macros: HashMap<String, Macro> = HashMap::new();
let mut rtokens = tokens.clone();
let mut rtokens = tokens;
rtokens.reverse();
while rtokens.len() > 0 {
while !rtokens.is_empty() {
let token = rtokens.pop().unwrap();
let op_type = lookup_word(token.text.clone(), &token.loc());
let op_type = lookup_word(&token.text, &token.loc());
match token.clone() {
_ if op_type == OpType::Macro => {
if rtokens.len() == 0 {
if rtokens.is_empty(){
lerror!(&token.loc(), "Macro name not found, expected {} but found nothing", TokenType::Word.human());
return Err(eyre!(""));
}
@ -37,41 +37,36 @@ pub fn preprocess(tokens: Vec<Token>, args: Args) -> Result<Vec<Token>>{
lerror!(&macro_name.loc(), "Bad macro name, expected {} but found {}", TokenType::Word.human(), macro_name.typ.human());
return Err(eyre!(""));
}
let word = lookup_word(macro_name.text.clone(), &macro_name.loc());
let word = lookup_word(&macro_name.text, &macro_name.loc());
if word != OpType::None {
lerror!(&macro_name.loc(), "Macro name cannot be a built in word, got '{}'", word.human());
return Err(eyre!(""));
}
if macros.get(&macro_name.text.clone()).is_some() {
if crate::constants::ALLOW_MACRO_REDEFINITION {
if macros.get(&macro_name.text.clone()).is_some() && crate::constants::ALLOW_MACRO_REDEFINITION {
lerror!(&macro_name.loc(), "Macro redefinition is not allowed");
lnote!(&macros.get(&macro_name.text.clone()).unwrap().loc, "First definition here");
lnote!(&macros.get(&macro_name.text).unwrap().loc, "First definition here");
return Err(eyre!(""));
} else {
//TODO: somehow warn about redefinition of only built in macros
}
}
let mut macr = Macro{ loc: macro_name.loc(), tokens: Vec::new() };
let mut depth = 0;
while rtokens.len() > 0 {
while !rtokens.is_empty() {
let t = rtokens.pop().unwrap();
let typ = lookup_word(t.text.clone(), &t.loc());
let typ = lookup_word(&t.text, &t.loc());
if typ == OpType::End && depth == 0 {
break;
} else if typ == OpType::End && depth != 0 {
depth -= 1;
macr.tokens.push(t);
} else {
if typ == OpType::If || typ == OpType::Do {
} else if typ == OpType::If || typ == OpType::Do {
macr.tokens.push(t);
depth += 1;
} else {
macr.tokens.push(t);
}
}
}
@ -81,7 +76,7 @@ pub fn preprocess(tokens: Vec<Token>, args: Args) -> Result<Vec<Token>>{
}
_ if op_type == OpType::Include => {
if rtokens.len() == 0 {
if rtokens.is_empty() {
lerror!(&token.loc(), "Include path not found, expected {} but found nothing", TokenType::String.human());
return Err(eyre!(""));
}
@ -94,7 +89,7 @@ pub fn preprocess(tokens: Vec<Token>, args: Args) -> Result<Vec<Token>>{
}
let mut in_paths = args.include.clone();
in_paths.append(&mut crate::DEFAULT_INCLUDES.to_vec().clone().iter().map(|f| f.to_string()).collect::<Vec<String>>());
in_paths.append(&mut crate::DEFAULT_INCLUDES.to_vec().clone().iter().map(|f| (*f).to_string()).collect::<Vec<String>>());
let mut include_code = String::new();
@ -113,7 +108,7 @@ pub fn preprocess(tokens: Vec<Token>, args: Args) -> Result<Vec<Token>>{
return Err(eyre!(""));
}
let mut code = lex(include_code, &include_path.text, args.clone(), false)?;
let mut code = lex(&include_code, &include_path.text, args, false)?;
code.reverse();
rtokens.append(&mut code);
@ -130,7 +125,7 @@ pub fn preprocess(tokens: Vec<Token>, args: Args) -> Result<Vec<Token>>{
let mut times = 0;
while program.iter().map(|f| {
if f.typ == TokenType::Word {
lookup_word(f.text.clone(), &f.loc())
lookup_word(&f.text, &f.loc())
} else {
OpType::PushInt // i hate myself, this is a randomly picked optype so its happy and works
}
@ -152,18 +147,20 @@ pub fn preprocess(tokens: Vec<Token>, args: Args) -> Result<Vec<Token>>{
pub fn expand_macros(tokens: Vec<Token>, macros: &HashMap<String, Macro>) -> Result<Vec<Token>> {
let mut program: Vec<Token> = Vec::new();
let mut rtokens = tokens.clone();
let mut rtokens = tokens;
rtokens.reverse();
while rtokens.len() > 0 {
while !rtokens.is_empty() {
let op = rtokens.pop().unwrap();
let op_type = lookup_word(op.text.clone(), &op.loc());
let op_type = lookup_word(&op.text, &op.loc());
if op.typ == TokenType::Word {
match op_type {
OpType::None => {
let m = macros.get(&op.text);
if m.is_some() {
program.append(&mut m.unwrap().tokens.clone())
if let Some(m) = m {
program.append(&mut m.tokens.clone());
}
} else {
lerror!(&op.loc(), "Unknown word '{}'", op.text.clone());
return Err(eyre!(""));