From 8d479202b0c9dbe13bb95ad572a060b69642ec26 Mon Sep 17 00:00:00 2001
From: krolxon <krolyxon@tutanota.com>
Date: Mon, 5 Jan 2026 12:12:57 +0530
Subject: add labels, improve documentation, add step debug

---
 src/assembler.rs | 56 +++++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 55 insertions(+), 1 deletion(-)

(limited to 'src/assembler.rs')

diff --git a/src/assembler.rs b/src/assembler.rs
index 3cd2bf8..3eeec64 100644
--- a/src/assembler.rs
+++ b/src/assembler.rs
@@ -1,4 +1,5 @@
 use crate::instructions::Instruction;
+use std::collections::HashMap;
 
 fn tokenize(line: &str) -> Vec<String> {
     line.split(|c| c == ' ' || c == ',' || c == '\t')
@@ -17,14 +18,51 @@ fn parse_reg(s: &str) -> u8 {
     }
 }
 
+fn instr_size(tokens: &[String]) ->  u16 {
+    match tokens[0].as_str() {
+        "mov" | "add" | "sub" | "jmp" | "jz" | "jnz" => 3,
+        "hlt" => 1,
+        _ => panic!("Unknown instruction {}", tokens[0])
+
+    }
+}
+
+
+fn first_pass(source: &str) -> HashMap<String, u16> {
+    let mut symbols = HashMap::new();
+    let mut addr: u16 = 0;
+
+    for line in source.lines() {
+        let line = line.trim();
+
+        // Ignoring comments and empty lines
+        if line.is_empty() || line.starts_with(";") {
+            continue;
+        }
+
+        // Labels (ends with ":")
+        if line.ends_with(":") {
+            let label = line.trim_end_matches(":").to_string();
+            symbols.insert(label, addr);
+            continue;
+        }
+
+        let tokens = tokenize(line);
+        addr += instr_size(&tokens);
+    }
+
+    symbols
+}
+
 pub fn assembler(source: &str) -> Vec<u8> {
+    let symbols = first_pass(source);
     let mut bytes = Vec::new();
 
     for (line_no, line) in source.lines().enumerate() {
         let line = line.trim();
 
         // Comments in assembly start with ";"
-        if line.is_empty() || line.starts_with(';') {
+        if line.is_empty() || line.starts_with(';') || line.ends_with(":") {
             continue;
         }
 
@@ -61,6 +99,22 @@ pub fn assembler(source: &str) -> Vec<u8> {
                 bytes.push(r2);
             }
 
+            "jmp" | "jz" | "jnz" => {
+                let opcode = match tokens[0].as_str() {
+                    "jmp" => Instruction::JMP,
+                    "jz" => Instruction::JZ,
+                    "jnz" => Instruction::JNZ,
+                    _ => unreachable!()
+                };
+
+                let label = &tokens[1];
+                let addr = *symbols.get(label).expect("Uknown label");
+
+                bytes.push(opcode as u8);
+                bytes.push((addr & 0xFF) as u8); // low
+                bytes.push((addr >> 8) as u8); // high
+            }
+
             "hlt" => {
                 bytes.push(Instruction::HLT as u8);
             }
-- 
cgit v1.2.3