Removed makefile for a simpler sh script and finalized the compiler with loops

.gitignore

@@ -1 +1,4 @@
 /target
+
+# asm gen testing
+/asm

input/math.bf

@@ -1 +1 @@
-,++.
++++.--.

input/math2.bf

@@ -0,0 +1 @@
+,++.

input/move.bf

@@ -0,0 +1 @@
+>++++<+.>>+++<--.>.

src/asm_gen.rs

@@ -0,0 +1,90 @@
+use std::collections::VecDeque;
+
+use crate::tokenizer::{Token, TokenType};
+
+pub fn generate(tokens: Vec<Token>) -> String {
+    let mut output = String::new();
+
+    // reserving mem
+    output.push_str(
+        "section .bss\n\
+		data: resb 65536\n",
+    );
+
+    // global start label
+    output.push_str(
+        "global _start\n\
+		section .text\n\
+		_start:\n",
+    );
+
+    // create the pointer for our data
+    output.push_str("mov rdx, data\n");
+
+    let mut loops = 0;
+    let mut stack: VecDeque<usize> = VecDeque::new();
+    for token in tokens {
+        match token.r#type {
+            TokenType::Plus(x) => {
+                output.push_str(&format!("add byte[rdx], {x}\n"));
+            }
+            TokenType::Minus(x) => {
+                output.push_str(&format!("sub byte[rdx], {x}\n"));
+            }
+            TokenType::MoveRight(x) => {
+                output.push_str(&format!("add rdx, {x}\n"));
+            }
+            TokenType::MoveLeft(x) => {
+                output.push_str(&format!("sub rdx, {x}\n"));
+            }
+            TokenType::Output => {
+                output.push_str(
+                    "push rdx\n\
+					mov rax, 1\n\
+					mov rdi, 1\n\
+					mov rsi, rdx\n\
+					mov rdx, 1\n\
+					syscall\n\
+					pop rdx\n",
+                );
+            }
+            TokenType::Input => {
+                output.push_str(
+                    "push rdx\n\
+					mov rax, 0\n\
+					mov rdi, 0\n\
+					mov rsi, rdx\n\
+					mov rdx, 1\n\
+					syscall\n\
+					pop rdx\n",
+                );
+            }
+            TokenType::BracketOpen => {
+                output.push_str(&format!(
+                    ".loop_{loops}_start:\n\
+					cmp byte[rdx], 0\n\
+					je .loop_{loops}_end\n"
+                ));
+                stack.push_back(loops);
+                loops += 1;
+            }
+            TokenType::BracketClose => {
+                let current_loop = stack.pop_back().unwrap();
+                output.push_str(&format!(
+                    "cmp byte[rdx], 0\n\
+					jne .loop_{current_loop}_start\n\
+					.loop_{current_loop}_end:\n",
+                ));
+            }
+        }
+    }
+
+    // exit syscall
+    output.push_str(
+        "mov rax, 60\n\
+		mov rdi, 0\n\
+		syscall\n",
+    );
+
+    output
+}

src/compiler.rs

@@ -0,0 +1,12 @@
+use std::{fs, io::Write, path::PathBuf};
+
+use crate::{asm_gen, tokenizer};
+
+pub fn compile(program: Vec<u8>, name: String) {
+    let tokens = tokenizer::tokenize(program);
+    let asm = asm_gen::generate(tokens);
+
+    let path: PathBuf = [".", &format!("./{name}.asm")].iter().collect();
+    let mut file = fs::File::create(path).unwrap();
+    file.write_all(asm.as_bytes()).unwrap();
+}

src/interpreter.rs

@@ -0,0 +1,93 @@
+use std::{
+    collections::{HashMap, VecDeque},
+    io::{self, Write},
+};
+
+pub fn run(program: Vec<u8>) {
+    let mut tape: Vec<u8> = vec![0];
+    let mut cell_index: usize = 0;
+
+    let mut ip: usize = 0;
+    let mut icount: usize = 0;
+    let mut user_input: Vec<u8> = vec![];
+
+    let mut table: HashMap<usize, usize> = HashMap::new();
+    let mut stack: VecDeque<usize> = VecDeque::new();
+    for (ip, instruction) in program.iter().enumerate() {
+        if *instruction == b'[' {
+            stack.push_back(ip);
+        } else if *instruction == b']' {
+            let begin_index = stack.pop_back().unwrap();
+            table.insert(begin_index, ip);
+            table.insert(ip, begin_index);
+        }
+    }
+
+    while ip < program.len() {
+        let Some(instruction) = program.get(ip) else {
+            break;
+        };
+
+        match instruction {
+            b'+' => {
+                tape[cell_index] = tape[cell_index].wrapping_add(1);
+                icount += 1;
+            }
+            b'-' => {
+                tape[cell_index] = tape[cell_index].wrapping_sub(1);
+                icount += 1;
+            }
+            b'>' => {
+                cell_index = cell_index.wrapping_add(1);
+                if tape.len() <= cell_index {
+                    tape.push(0);
+                }
+                icount += 1;
+            }
+            b'<' => {
+                cell_index = cell_index.wrapping_sub(1);
+                icount += 1;
+            }
+            b'.' => {
+                print!("{}", tape[cell_index] as char);
+                icount += 1;
+            }
+            b',' => {
+                if user_input.is_empty() {
+                    user_input = input();
+                }
+                tape[cell_index] = user_input.remove(0);
+                icount += 1;
+            }
+            b'[' => {
+                if tape[cell_index] == 0 {
+                    ip = *table.get(&ip).unwrap();
+                }
+                icount += 1;
+            }
+            b']' => {
+                if tape[cell_index] != 0 {
+                    ip = *table.get(&ip).unwrap();
+                }
+                icount += 1;
+            }
+            _ => {}
+        }
+
+        ip += 1;
+    }
+
+    println!();
+    println!("Instructions ran: {icount}");
+    // println!("{tape:?}");
+}
+
+fn input() -> Vec<u8> {
+    let mut input = String::new();
+
+    print!("Input: ");
+    let _ = io::stdout().flush();
+    let _ = io::stdin().read_line(&mut input);
+
+    input.as_bytes().to_vec()
+}

src/main.rs

@@ -1,105 +1,34 @@
-use std::{
-    collections::{HashMap, VecDeque},
-    env, fs,
-    io::{self, Write},
-    path::PathBuf,
-};
+#![feature(let_chains)]
+
+use std::{env, fs, path::PathBuf};
+
+mod asm_gen;
+mod compiler;
+mod interpreter;
+mod tokenizer;
 
 fn main() {
-    let program: Vec<u8> = load_program();
-
-    let mut tape: Vec<u8> = vec![0];
-    let mut cell_index: usize = 0;
-
-    let mut ip: usize = 0;
-    let mut icount: usize = 0;
-    let mut user_input: Vec<u8> = vec![];
-
-    let mut table: HashMap<usize, usize> = HashMap::new();
-    let mut stack: VecDeque<usize> = VecDeque::new();
-    for (ip, instruction) in program.iter().enumerate() {
-        if *instruction == b'[' {
-            stack.push_back(ip);
-        } else if *instruction == b']' {
-            let begin_index = stack.pop_back().unwrap();
-            table.insert(begin_index, ip);
-            table.insert(ip, begin_index);
-        }
-    }
-
-    while ip < program.len() {
-        let Some(instruction) = program.get(ip) else {
-            break;
-        };
-
-        match instruction {
-            b'+' => {
-                tape[cell_index] = tape[cell_index].wrapping_add(1);
-                icount += 1;
-            }
-            b'-' => {
-                tape[cell_index] = tape[cell_index].wrapping_sub(1);
-                icount += 1;
-            }
-            b'>' => {
-                cell_index = cell_index.wrapping_add(1);
-                if tape.len() <= cell_index {
-                    tape.push(0);
-                }
-                icount += 1;
-            }
-            b'<' => {
-                cell_index = cell_index.wrapping_sub(1);
-                icount += 1;
-            }
-            b'.' => {
-                print!("{}", tape[cell_index] as char);
-                icount += 1;
-            }
-            b',' => {
-                if user_input.is_empty() {
-                    user_input = input();
-                }
-                tape[cell_index] = user_input.remove(0);
-                icount += 1;
-            }
-            b'[' => {
-                if tape[cell_index] == 0 {
-                    ip = *table.get(&ip).unwrap();
-                }
-                icount += 1;
-            }
-            b']' => {
-                if tape[cell_index] != 0 {
-                    ip = *table.get(&ip).unwrap();
-                }
-                icount += 1;
-            }
-            _ => {}
-        }
-
-        ip += 1;
-    }
-
-    println!();
-    println!("Instructions ran: {icount}");
-    // println!("{tape:?}");
-}
-
-fn input() -> Vec<u8> {
-    let mut input = String::new();
-
-    print!("Input: ");
-    let _ = io::stdout().flush();
-    let _ = io::stdin().read_line(&mut input);
-
-    input.as_bytes().to_vec()
-}
-
-fn load_program() -> Vec<u8> {
     let args: Vec<String> = env::args().collect();
-    let input = args.get(1).expect("malformed arguments");
+    let mode = args.get(1).expect("mode expected");
+    let input = args.get(2).expect("malformed arguments");
+    let program: Vec<u8> = load_program(input);
+    let name = args
+        .get(3)
+        .map(|s| s.to_string())
+        .unwrap_or_else(|| "program".to_string());
 
+    match mode {
+        x if x == "c" => {
+            compiler::compile(program, name);
+        }
+        x if x == "i" => {
+            interpreter::run(program);
+        }
+        _ => {}
+    }
+}
+
+fn load_program(input: &str) -> Vec<u8> {
     let path: PathBuf = input.into();
     let input = fs::read_to_string(path).expect("failed to read the program file");
     input.as_bytes().to_vec()

src/tokenizer.rs

@@ -0,0 +1,85 @@
+#[derive(Debug)]
+pub enum TokenType {
+    Plus(u32),
+    Minus(u32),
+    MoveRight(u32),
+    MoveLeft(u32),
+    Output,
+    Input,
+    BracketOpen,
+    BracketClose,
+}
+
+#[derive(Debug)]
+pub struct Token {
+    pub r#type: TokenType,
+}
+
+impl Token {
+    fn new(r#type: TokenType) -> Self {
+        Self { r#type }
+    }
+}
+
+pub fn tokenize(program: Vec<u8>) -> Vec<Token> {
+    let mut tokens = vec![];
+
+    let mut ip = 0;
+    while ip < program.len() {
+        let Some(instruction) = program.get(ip) else {
+            break;
+        };
+
+        ip += 1;
+
+        match instruction {
+            b'+' => {
+                let count = check_next(&program, &mut ip, instruction);
+                tokens.push(Token::new(TokenType::Plus(count)));
+            }
+            b'-' => {
+                let count = check_next(&program, &mut ip, instruction);
+                tokens.push(Token::new(TokenType::Minus(count)));
+            }
+            b'>' => {
+                let count = check_next(&program, &mut ip, instruction);
+                tokens.push(Token::new(TokenType::MoveRight(count)));
+            }
+            b'<' => {
+                let count = check_next(&program, &mut ip, instruction);
+                tokens.push(Token::new(TokenType::MoveLeft(count)));
+            }
+            b'.' => {
+                tokens.push(Token::new(TokenType::Output));
+            }
+            b',' => {
+                tokens.push(Token::new(TokenType::Input));
+            }
+            b'[' => {
+                tokens.push(Token::new(TokenType::BracketOpen));
+            }
+            b']' => {
+                tokens.push(Token::new(TokenType::BracketClose));
+            }
+            _ => {}
+        }
+    }
+
+    // dbg!(&tokens);
+
+    tokens
+}
+
+fn check_next(program: &[u8], ip: &mut usize, instruction: &u8) -> u32 {
+    let mut icount = 1;
+    let mut next_instruction = program.get(*ip);
+    while let Some(next) = next_instruction
+        && *next == *instruction
+    {
+        icount += 1;
+        *ip += 1;
+        next_instruction = program.get(*ip);
+    }
+
+    icount
+}

test.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+(
+mode=$1
+program=$2
+cargo run --release -- $mode $program
+if [ "$mode" == "c" ]; then
+	mv ./program.asm ./asm/program.asm \
+		&& nasm -felf64 ./asm/program.asm \
+		&& ld.lld ./asm/program.o -o ./asm/program \
+		&& ./asm/program
+fi
+)