Fleshing out the 6502 Assembler.

src/MOS6502.js

@@ -71,6 +71,29 @@ function toHexString(v, l){
   return r;
 }
 
+function tokenize(l){
+  var m = l.match(/([^()]+)|([()])/g);
+  var ip = false;
+  t = [];
+  m.forEach((mi)=>{
+    if (mi === "("){
+      ip = true;
+    } else {
+      if (ip){
+        mi.split(/[\s,]+/).forEach((i)=>{
+          if (i !== "")
+            t.push("+" + i);
+        });
+      } else {
+        mi.split(/[\s,]+/).forEach((i)=>{
+          if (i !== "")
+            t.push(i);
+        });
+      }
+    }
+  });
+}
+
 class Assembler{
   constructor(initpc){
     // Labels that hold variable values.
@@ -82,31 +105,159 @@ class Assembler{
     this.__PC = (initpc >= 0) ? initpc : 0;
   }
 
+  function __AddrModeVal(tA, tB){
+    var mode = "";
+    var v = Number.NaN;
+    if (tA[0] === '#'){
+      mode = "i"; // Immediate
+      if (tA[1] === '$'){
+        v = parseInt(tA.substr(2), 16);
+      } else if (tA[1] === 'b'){
+        v = parseInt(tA.substr(2), 2);
+      } else {
+        v = parseInt(tA.substr(1));
+        if (isNaN(v)){
+          let lbl = tA.substr(1);
+          if (lbl in this.__varlabels){
+            if (this.__varlabels[lbl] < 256)
+              v = this.__varlabels[lbl]
+          }
+        }
+      }
+    } else if (tA[0] === '$'){
+      v = parseInt(tA.substr(1), 16);
+      if (tA.length === 3){
+        // Zero Page
+        if (tB !== null){
+          if (tB === "X")
+            mode = "zX";
+        } else {
+          mode = "z";
+        }
+      } else if (tA.length === 5){
+        // Absolute
+        if (tB !== null){
+          if (tB === "X"){
+            mode = "aX";
+          } else if (tB === "Y"){
+            mode = "aY";
+          }
+        } else {
+          mode = "a";
+        }
+      }
+    } else if (tA.startsWith("+")){
+      if (tB !== null){
+        if (tB === "+X"){ // The plus exists due to how tokenize() works.
+          mode = "nX";
+        } else if (tB === "Y"){
+          mode = "nY";
+        }
+
+        if (mode !== ""){
+          if (tA.startsWith("+$")){
+            v = parseInt(tA.substr(2));
+          } else {
+            let lbl = tA.substr(1);
+            if (lbl in this.__varlabels){
+              if (this.__varlabels[lbl] < 256)
+                v = vlb[lbl];
+            }
+          }
+        }
+      }
+    } else {
+      // We've been given a variable name. Need to look it up.
+      // if the var is 1 byte, assume "Zero Page". If var 2 bytes, assume "Absolute"
+      if (tA in this.__varlabel){
+        v = this.__varlabel[tA];
+        if (v < 256){
+          if (tB !== null){
+            if (tB === "X")
+              mode = "zX";
+          } else {
+            mode = "z";
+          }
+        } else {
+          if (tB !== null){
+            if (tB === "X"){
+              mode = "aX";
+            } else if (tB === "Y"){
+              mode = "aY";
+            }
+          } else {
+            mode = "a";
+          }
+        }
+      }
+    }
+
+    if (mode !== "" && ! isNaN(v))
+      return [mode, v];
+    throw new Error("Malformed op-code or value on program address " + toHexString(this.__PC));
+  }
+
   compile(src){
     var op = [];
     src.split("\n").forEach((line)=>{
       line = line.trim();
       if (line[0] !== ";"){ // Skip comment lines.
         line = line.split(";")[0].trim(); // Take out any trailing comments.
-        var tokens = line.split(/\s+/);
+        var tokens = tokenize(line);
 
         if (tokens[0] === 'define'){
           // Variable label!!
         } else if (tokens[0][tokens[0].length - 1] === ':'){
           // Jump label!!
-        } else if (tokens[0].length === 3){
+        } else if (tokens[0].length === 3){ 
+          let procFailed = false;
+          let mv = null;
           // Possible op code.
           switch(tokens[0].toLowerCase()){
+// --- ADC
             case 'adc':
-              break;
+              if (tokens.length >= 2 && tokens.length <= 3){
+                mv = addrModeVal(tokens[1]);
+                this.__PC += 2;
+                switch(mv[0]){
+                  case "i":
+                    op.push(0x69);
+                    op.push(mv[1]);
+                    break;
+                  case "z":
+                  case "zX":
+                    op.push((mv[0] === "z") ? 0x65 : 0x75);
+                    op.push(mv[1]);
+                    break;
+                  case "a":
+                  case "aX":
+                  case "aY":
+                    op.push((mv[0] === "a") ? 0x6D : ((mv[0] === "aX") ? 0x7D : 0x79));
+                    op.push(mv[1] & 0x000000FF);
+                    op.push((mv[1] & 0x0000FF00) >> 8);
+                    this.__PC += 1;
+                    break;
+                  case "nX":
+                  case "nY":
+                    op.push((mv[0] === "nX") ? 0x61 : 0x71);
+                    op.push(mv[1]);
+                    break;
+                }
+              } else { procFailed = true; }
+              break;
+// --- AND
             case 'and':
               break;
+// --- ASL
             case 'asl':
               break;
+// --- BCC
             case 'bcc':
               break;
+// --- BCS
             case 'bcs':
               break;
+// --- BEQ
             case 'beq':
               break;
             case 'bit':
@@ -117,19 +268,44 @@ class Assembler{
               break;
             case 'bpl':
               break;
+// --- BRK
             case 'brk':
+              if (tokens.length === 1){
+                op.push(0x00);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
             case 'bvc':
               break;
             case 'bvs':
               break;
+// --- CLC
             case 'clc':
+              if (tokens.length === 1){
+                op.push(0x18);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- CLD
             case 'cld':
+              if (tokens.length === 1){
+                op.push(0xD8);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- CLI
             case 'cli':
+              if (tokens.length === 1){
+                op.push(0x58);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- CLV
             case 'clv':
+              if (tokens.length === 1){
+                op.push(0xB8);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
             case 'cmp':
               break;
@@ -139,17 +315,37 @@ class Assembler{
               break;
             case 'dec':
               break;
+// --- DEX
             case 'dex':
+              if (tokens.length === 1){
+                op.push(0xCA);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- DEY
             case 'dey':
+              if (tokens.length === 1){
+                op.push(0x88);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
             case 'eor':
               break;
             case 'inc':
               break;
+// --- INX
             case 'inx':
+              if (tokens.length === 1){
+                op.push(0xE8);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- INY
             case 'iny':
+              if (tokens.length === 1){
+                op.push(0xC8);
+                this.__PC += 1
+              } else { procFailed = true; }
               break;
             case 'jmp':
               break;
@@ -163,7 +359,12 @@ class Assembler{
               break;
             case 'lsr':
               break;
+// --- NOP
             case 'nop':
+              if (tokens.length === 1){
+                op.push(0xEA);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
             case 'ora':
               break;
@@ -185,11 +386,24 @@ class Assembler{
               break;
             case 'sbc':
               break;
+// --- SEC
             case 'sec':
+              if (tokens.length === 1){
+                op.push(0x38);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- SED
             case 'sed':
+              if (tokens.length === 1){
+                op.push(0xF8);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- SEI
             case 'sei':
+              op.push(0x78);
+              this.__PC += 1;
               break;
             case 'sta':
               break;
@@ -197,21 +411,47 @@ class Assembler{
               break;
             case 'sty':
               break;
+// --- TAX
             case 'tax':
+              if (tokens.length === 1){
+                op.push(0xAA);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- TAY
             case 'tay':
+              if (tokens.length === 1){
+                op.push(0xA8);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- TSX
             case 'tsx':
               break;
+// --- TXA
             case 'txa':
+              if (tokens.length === 1){
+                op.push(0x8A);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- TXS
             case 'txs':
               break;
+// --- TYA
             case 'tya':
+              if (tokens.length === 1){
+                op.push(0x98);
+                this.__PC += 1;
+              } else { procFailed = true; }
               break;
+// --- ---
             default:
               throw new Error("Unknown op-code '" + tokens[0].toUpperCase() + "' at program address " + toHexString(this.__PC));
           }
+
+          if (procFailed)
+            throw new Error("Malformed op-code on program address " + toHexString(this.__PC));
         } else {
           throw new Error("Failed to compile line '" + line + "' at program address " + toHexString(this.__PC));
         }