Squashed bugs, added some error reports

Cargo.lock

@@ -2,6 +2,12 @@
 # It is not intended for manual editing.
 version = 3
 
+[[package]]
+name = "anyhow"
+version = "1.0.79"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "080e9890a082662b09c1ad45f567faeeb47f22b5fb23895fbe1e651e718e25ca"
+
 [[package]]
 name = "codespan-reporting"
 version = "0.11.1"
@@ -76,6 +82,7 @@ dependencies = [
 name = "riscv_interpreter"
 version = "0.1.0"
 dependencies = [
+ "anyhow",
  "codespan-reporting",
  "itertools",
  "rayon",

Cargo.toml

@@ -4,6 +4,7 @@ version = "0.1.0"
 edition = "2021"
 
 [dependencies]
-codespan-reporting = { version = "0.11.1", features = ["ascii-only"] }
+anyhow = "1.0.79"
+codespan-reporting = "0.11.1"
 itertools = "0.12.0"
 rayon = "1.8.1"

src/env.rs

@@ -2,7 +2,7 @@ use std::collections::HashMap;
 
 #[derive(Debug)]
 pub struct Env {
-    register_alias: HashMap<String, usize>,
+    pub register_alias: HashMap<String, usize>,
     labels: HashMap<String, usize>,
     registers: [i64; 32],
     stack: Vec<i64>, // TODO: Find the size of the stack
@@ -68,6 +68,13 @@ impl Env {
     pub fn alias_to_register(&self, reg: &str) -> Option<usize> {
         self.register_alias.get(reg).copied()
     }
+    pub fn xn_to_register(&self, reg: &str) -> Option<usize> {
+        if reg.starts_with("x") {
+            reg[1..].parse::<usize>().ok()
+        } else {
+            None
+        }
+    }
 
     pub fn add_label(&mut self, label: &str, value: usize) {
         self.labels.insert(label.to_string(), value);

src/err.rs

@@ -7,6 +7,7 @@ pub enum SyntaxErr {
     UnmatchedParen(bool),
     UnexpectedChar,
     OutsideOp(String),
+    MemoryInvalidRegister,
 }
 
 impl Display for SyntaxErr {
@@ -16,6 +17,20 @@ impl Display for SyntaxErr {
             SyntaxErr::UnmatchedParen(_) => write!(f, "unmatched parenthesis"),
             SyntaxErr::UnexpectedChar => write!(f, "unexpected character"),
             SyntaxErr::OutsideOp(kind) => write!(f, "{kind} before opcode"),
+            SyntaxErr::MemoryInvalidRegister => write!(f, "invalid register"),
+        }
+    }
+}
+
+impl SyntaxErr {
+    pub fn note(&self) -> String {
+        match self {
+            SyntaxErr::TraillingComma => "remove the final comma".to_string(),
+            SyntaxErr::UnmatchedParen(false) => "add ')' after the register name".to_string(),
+            SyntaxErr::UnmatchedParen(true) => "add '(' before the register name".to_string(),
+            SyntaxErr::UnexpectedChar => "ensure the input is well-formed".to_string(),
+            SyntaxErr::OutsideOp(kind) => format!("add '{}'s only after an opcode", kind),
+            SyntaxErr::MemoryInvalidRegister => "valid registers are of the form xN, 0 <= N < 32, or the standard aliases".to_string(),
         }
     }
 }

src/main.rs

@@ -1,9 +1,50 @@
-use rayon::prelude::*;
-use riscv_interpreter::parser::{parse, Token};
+use codespan_reporting::{
+    diagnostic::{self, Diagnostic, Label},
+    files::SimpleFile,
+    term::{
+        self,
+        termcolor::{ColorChoice, StandardStream},
+        Config,
+    },
+};
+use riscv_interpreter::{env::Env, parser::parse};
 
-fn main() {
+fn main() -> anyhow::Result<()> {
+    let writer = StandardStream::stderr(ColorChoice::Always);
+    let config = Config::default();
     let input = std::fs::read_to_string("test.s").unwrap();
-    let tokens = parse(&input);
-    // println!("{:#?} -> {:#?}", input, tokens.into_par_iter().filter(|(token, _)| !matches!(token, Token::Spacing)).collect::<Vec<_>>());
-    println!("{:#?} -> {:#?}", input, tokens);
+
+    let file = SimpleFile::new("test.s", input.clone());
+
+    let env = Env::new();
+
+    match parse(&env, &input) {
+        Ok(tokens) => {
+            println!("{:#?} -> {:#?}", input, tokens);
+        }
+        Err(errs) => {
+            for err in errs {
+                let start = err.1.start;
+                let end = err.1.end;
+
+                let diagnostic = Diagnostic::error()
+                    .with_message("Syntax Error")
+                    .with_labels(vec![
+                        Label::primary((), start..end).with_message(err.0.to_string())
+                    ])
+                    .with_notes({
+                        let mut notes = Vec::new();
+                        if let Some(note) = err.3 {
+                            notes.push(note);
+                        }
+                        notes.push(err.0.note());
+                        notes
+                    });
+
+                term::emit(&mut writer.lock(), &config, &file, &diagnostic).unwrap();
+            }
+        }
+    };
+
+    Ok(())
 }

src/parser.rs

@@ -1,5 +1,5 @@
 /// TODO: Strings, Symbols
-use crate::err::SyntaxErr;
+use crate::{env::Env, err::SyntaxErr};
 use itertools::Itertools;
 use rayon::prelude::*;
 
@@ -58,7 +58,7 @@ pub struct Loc {
     pub end: usize,
 }
 
-fn parse_line(input: &str, line: usize) -> Result<Vec<(Token, Loc)>, ParseErr> {
+fn parse_line(env: &Env, input: &str, line: usize) -> Result<Vec<(Token, Loc)>, ParseErr> {
     let mut loc = Loc {
         line,
         col: 1,
@@ -104,6 +104,9 @@ fn parse_line(input: &str, line: usize) -> Result<Vec<(Token, Loc)>, ParseErr> {
                 Immediate(num.parse().unwrap())
             }
             '(' => {
+                let start = loc.start + 1;
+                let col = loc.col + 1;
+
                 let imm;
                 if let Some((Immediate(_), _)) = tokens.last() {
                     imm = Box::new(tokens.pop().unwrap());
@@ -123,6 +126,23 @@ fn parse_line(input: &str, line: usize) -> Result<Vec<(Token, Loc)>, ParseErr> {
                     chars.peek()
                 {
                     reg.push(chars.next().unwrap());
+                    loc.end += 1;
+                }
+                let end = loc.end + 1;
+
+                let reg = reg.trim();
+                if env.alias_to_register(reg).is_none() && env.xn_to_register(reg).is_none() {
+                    return Err((
+                        SyntaxErr::MemoryInvalidRegister,
+                        Loc {
+                            line,
+                            col,
+                            start,
+                            end,
+                        },
+                        tokens.clone(),
+                        None,
+                    ));
                 }
                 if chars.next() != Some(')') {
                     return Err((
@@ -179,7 +199,7 @@ fn parse_line(input: &str, line: usize) -> Result<Vec<(Token, Loc)>, ParseErr> {
         loc.start = loc.end;
     }
 
-    Ok(tokens
+    let tokens = tokens
         .into_iter()
         .filter(|(token, _)| !matches!(token, Token::Spacing))
         .group_by(|(token, _)| {
@@ -196,6 +216,8 @@ fn parse_line(input: &str, line: usize) -> Result<Vec<(Token, Loc)>, ParseErr> {
                 let (op, loc) = dbg!(dbg!(group[0].clone()));
                 let (op, mut args) = match op {
                     Op(op, args) => (op, args),
+                    // because any register/symbol/label def is interpreted as an Op by default, this only
+                    // partially works. This does trigger on immediate values and memory indexes
                     _ => {
                         return vec![(
                             Token::Error((
@@ -215,7 +237,14 @@ fn parse_line(input: &str, line: usize) -> Result<Vec<(Token, Loc)>, ParseErr> {
                 group.collect::<Vec<_>>()
             }
         })
-        .collect::<Vec<_>>())
+        .collect::<Vec<_>>();
+    if let Some((Token::Error(err), _)) =
+        tokens.iter().find(|line| matches!(line.0, Token::Error(_)))
+    {
+        Err(err.to_owned())
+    } else {
+        Ok(tokens)
+    }
 }
 
 /// Parse the input
@@ -223,12 +252,12 @@ fn parse_line(input: &str, line: usize) -> Result<Vec<(Token, Loc)>, ParseErr> {
 /// Returns a vector of tokens and their locations, if successful, or an error vector
 /// containing the error, the location of the error, the tokens parsed up to that point,
 /// and an optional message to display to the users
-pub fn parse(input: &str) -> Result<Vec<(Token, Loc)>, Vec<ParseErr>> {
+pub fn parse(env: &Env, input: &str) -> Result<Vec<(Token, Loc)>, Vec<ParseErr>> {
     let parsed_lines = input
         .lines()
         .enumerate()
         .par_bridge()
-        .map(|(i, line)| parse_line(line, i + 1))
+        .map(|(i, line)| parse_line(env, line, i + 1))
         .collect::<Vec<_>>();
 
     let (ok, err) = parsed_lines

test.s

@@ -1 +1 @@
-a:  lw a0 1(sp)
+    lw a0 -4(sp)