From 93f3dd1f416884f307e02efa529dcd650d54e0ff Mon Sep 17 00:00:00 2001
From: Camden Dixie O'Brien <cdo@wip.sh>
Date: Sat, 14 Mar 2026 13:10:48 +0000
Subject: [PATCH] Implement .utf directive

---
 asm.js | 22 ++++++++++++++++++++--
 1 file changed, 20 insertions(+), 2 deletions(-)

diff --git a/asm.js b/asm.js
index e79d62a..068e8eb 100644
--- a/asm.js
+++ b/asm.js
@@ -91,6 +91,7 @@ const State = Object.freeze({
     AT_ADDR:      19,
     BYTE:         20,
     WORD:         21,
+    UTF8:         22,
 });
 
 const Action = Object.freeze({
@@ -135,7 +136,8 @@ const const_opcodes = {
 };
 
 class Parser {
-	constructor() {
+	constructor(encoder) {
+        this.encoder = encoder;
 		this.tokens = [];
 		this.tokenizer = new Tokenizer();
         this.state = State.TOP;
@@ -151,6 +153,7 @@ class Parser {
             ".at":     State.AT_MEM,
             ".byte":   State.BYTE,
             ".word":   State.WORD,
+            ".utf8":   State.UTF8,
         };
         this.handlers = {
             [State.TOP]:          (token) => this.token_top(token),
@@ -175,6 +178,7 @@ class Parser {
             [State.AT_ADDR]:      (token) => this.token_at_addr(token),
             [State.BYTE]:         (token) => this.token_byte(token),
             [State.WORD]:         (token) => this.token_word(token),
+            [State.UTF8]:         (token) => this.token_utf8(token),
         };
 
         this.results = [];
@@ -499,6 +503,20 @@ class Parser {
         return action;
     }
 
+    token_utf8(token) {
+        if (token == LINE_END) {
+            this.state = State.TOP;
+            return;
+        } else if (token.string == undefined) {
+            console.error(
+                `ERROR: unexpected token ${token}, expected string`);
+            return;
+        }
+        const value = this.encoder.encode(token.string);
+        const action = { type: Action.DATA, size: value.length, value };
+        return action;
+    }
+
     mem_action() {
         const action = {
             type: Action.MEM,
@@ -539,7 +557,7 @@ export class Assembler {
 	constructor() {
         this.encoder = new TextEncoder("utf-8");
 		this.decoder = new TextDecoder("utf-8");
-		this.parser = new Parser();
+		this.parser = new Parser(this.encoder);
         this.handlers = {
             [Action.APPEND]: (action) => this.action_append(action),
             [Action.EXPORT]: (action) => this.action_export(action),