etc/hello.tal

@@ -1,7 +1,5 @@
 ( init )
 
-|00 @System &vector $2 &expansion $2 &wst $1 &rst $1 &metadata $2 &r $2 &g $2 &b $2 &debug $1 &state $1
-
 %emit ( byte -- ) { #18 DEO }
 
 |0100 @program
@@ -25,4 +23,4 @@ BRK
 
 @program/extend BRK
 
-@hello-word "Hello 20 "World? 00
+@hello-word "Hello 20 "World! 00

src/uxnasm.c

@@ -11,28 +11,42 @@ THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 WITH REGARD TO THIS SOFTWARE.
 */
 
-#define PAGE 0x0100
+#define TRIM 0x0100
+#define LENGTH 0x10000
 
 typedef unsigned char Uint8;
 typedef signed char Sint8;
 typedef unsigned short Uint16;
 
 typedef struct {
-	char *name, rune, *content;
+	char *name, items[0x40][0x40];
-	Uint16 addr, refs;
+	Uint8 len;
-} Item;
+} Macro;
 
 typedef struct {
-	int line;
+	char *name;
-	char *path;
+	Uint16 addr, refs;
-} Context;
+} Label;
 
-static int ptr, length;
+typedef struct {
-static char token[0x40], scope[0x40], sublabel[0x80], lambda[0x05];
+	char name[0x40], rune;
-static char dict[0x10000], *dictnext = dict;
+	Uint16 addr;
-static Uint8 data[0x10000], lambda_stack[0x100], lambda_ptr, lambda_len;
+} Reference;
-static Uint16 label_len, refs_len, macro_len;
+
-static Item labels[0x400], refs[0x1000], macros[0x100];
+typedef struct {
+	int ptr, length;
+	Uint8 data[LENGTH];
+	Uint8 lambda_stack[0x100], lambda_ptr, lambda_len;
+	Uint16 line, label_len, macro_len, refs_len;
+	Label labels[0x400];
+	Macro macros[0x100];
+	Reference refs[0x1000];
+} Program;
+
+char source[0x40], token[0x40], scope[0x40], sublabel[0x40], lambda[0x05];
+char dict[0x10000], *storenext = dict;
+
+Program p;
 
 /* clang-format off */
 
@@ -51,30 +65,38 @@ static int   shex(char *s) { int n = 0; char c; while((c = *s++)) { n = n << 4,
 static int   scmp(char *a, char *b, int len) { int i = 0; while(a[i] == b[i]) if(!a[i] || ++i >= len) return 1; return 0; } /* str compare */
 static int   slen(char *s) { int i = 0; while(s[i]) i++; return i; } /* str length */
 static char *scpy(char *src, char *dst, int len) { int i = 0; while((dst[i] = src[i]) && i < len - 2) i++; dst[i + 1] = '\0'; return dst; } /* str copy */
-static char *scat(char *dst, char *src) { char *o = dst + slen(dst); while(*src) *o++ = *src++; *o = '\0'; return dst; } /* str concat */
+static char *scat(char *dst, char *src) { char *ptr = dst + slen(dst); while(*src) *ptr++ = *src++; *ptr = '\0'; return dst; } /* str cat */
-static char *push(char *s, char c) { char *o = dictnext; while((*dictnext++ = *s++) && *s); *dictnext++ = c; return o; } /* save str */
+static char *push(char *s) { char *ptr = storenext; while((*storenext++ = *s++)); *storenext++ = 0; return ptr; } /* save str */
 
 #define isopcode(x) (findopcode(x) || scmp(x, "BRK", 4))
-#define writeshort(x) (writebyte(x >> 8, ctx) && writebyte(x & 0xff, ctx))
+#define writeshort(x) (writebyte(x >> 8) && writebyte(x & 0xff))
-#define makesublabel(x) push(scat(scat(scpy(scope, sublabel, 0x40), "/"), x), 0)
+#define error_top(name, msg) !!fprintf(stderr, "%s: %s\n", name, msg)
-#define findlabel(x) finditem(x, labels, label_len)
+#define error_asm(name) !!fprintf(stderr, "%s: %s in @%s, %s:%d.\n", name, token, scope, source, p.line)
-#define findmacro(x) finditem(x, macros, macro_len)
-#define error_top(name, msg) !fprintf(stderr, "%s: %s\n", name, msg)
-#define error_asm(name) !fprintf(stderr, "%s: %s in @%s, %s:%d.\n", name, token, scope, ctx->path, ctx->line)
 
 /* clang-format on */
 
-static int parse(char *w, FILE *f, Context *ctx);
+static int parse(char *w, FILE *f);
+static char *makesublabel(char *src, char *name);
 
-static Item *
+static Macro *
-finditem(char *name, Item *list, int len)
+findmacro(char *name)
+{
+	int i;
+	for(i = 0; i < p.macro_len; i++)
+		if(scmp(p.macros[i].name, name, 0x40))
+			return &p.macros[i];
+	return NULL;
+}
+
+static Label *
+findlabel(char *name)
 {
 	int i;
 	if(name[0] == '&')
-		name = makesublabel(name + 1);
+		name = makesublabel(sublabel, name + 1);
-	for(i = 0; i < len; i++)
+	for(i = 0; i < p.label_len; i++)
-		if(scmp(list[i].name, name, 0x40))
+		if(scmp(p.labels[i].name, name, 0x40))
-			return &list[i];
+			return &p.labels[i];
 	return NULL;
 }
 
@@ -84,8 +106,10 @@ findopcode(char *s)
 	int i;
 	for(i = 0; i < 0x20; i++) {
 		int m = 3;
-		if(!scmp(ops[i], s, 3)) continue;
+		if(!scmp(ops[i], s, 3))
-		if(!i) i |= (1 << 7);
+			continue;
+		if(!i)
+			i |= (1 << 7);
 		while(s[m]) {
 			if(s[m] == '2')
 				i |= (1 << 5);
@@ -103,97 +127,65 @@ findopcode(char *s)
 }
 
 static int
-walkcomment(FILE *f, Context *ctx)
+walkcomment(char *w, FILE *f)
 {
-	char c;
 	int depth = 1;
-	while(f && fread(&c, 1, 1, f)) {
-		if(c == 0xa) ctx->line += 1;
-		if(c == '(') depth++;
-		if(c == ')' && --depth < 1) return 1;
-	}
-	return 0;
-}
-
-static int
-walkmacro(Item *m, Context *ctx)
-{
-	char c, *contentptr = m->content, *cptr = token;
-	while((c = *contentptr++)) {
-		if(c < 0x21) {
-			*cptr++ = 0x00;
-			if(token[0] && !parse(token, NULL, ctx)) return 0;
-			cptr = token;
-		} else
-			*cptr++ = c;
-	}
-	return 1;
-}
-
-static int
-walkfile(FILE *f, Context *ctx)
-{
-	char c, *cptr = token;
-	while(f && fread(&c, 1, 1, f)) {
-		if(c == 0xa) ctx->line += 1;
-		if(c < 0x21) {
-			*cptr++ = 0x00;
-			if(token[0] && !parse(token, f, ctx))
-				return 0;
-			cptr = token;
-		} else if(cptr - token < 0x3f)
-			*cptr++ = c;
-		else
-			return error_asm("Token too long");
-	}
-	return 1;
-}
-
-static int
-makemacro(char *name, FILE *f, Context *ctx)
-{
 	char c;
-	Item *m;
+	if(slen(w) == 1)
+		while(fread(&c, 1, 1, f)) {
+			if(c == '(')
+				depth++;
+			else if(c == ')' && --depth < 1)
+				return 1;
+		}
+	return 0;
+}
+
+static int
+makemacro(char *name, FILE *f)
+{
+	Macro *m;
+	char word[0x40];
 	if(!slen(name)) return error_asm("Macro is empty");
 	if(findmacro(name)) return error_asm("Macro is duplicate");
 	if(sihx(name)) return error_asm("Macro is hex number");
 	if(isopcode(name)) return error_asm("Macro is opcode");
-	if(macro_len == 0x100) return error_asm("Macros limit exceeded");
+	if(p.macro_len == 0x100) return error_asm("Macros limit exceeded");
-	m = &macros[macro_len++];
+	m = &p.macros[p.macro_len++];
-	m->name = push(name, 0);
+	m->name = push(name);
-	m->content = dictnext;
+	while(fscanf(f, "%63s", word) == 1) {
-	while(f && fread(&c, 1, 1, f) && c != '{')
+		if(word[0] == '{') continue;
-		if(c == 0xa) ctx->line += 1;
+		if(word[0] == '}') break;
-	while(f && fread(&c, 1, 1, f) && c != '}') {
+		if(word[0] == '%') return error_asm("Macro error");
-		if(c == 0xa) ctx->line += 1;
+		if(m->len >= 0x40) return error_asm("Macro size exceeded");
-		if(c == '%') return 0;
+		if(word[0] == '(') {
-		if(c == '(') walkcomment(f, ctx);
+			walkcomment(word, f);
-		*dictnext++ = c;
+			continue;
+		}
+		scpy(word, m->items[m->len++], 0x40);
 	}
-	*dictnext++ = 0;
 	return 1;
 }
 
 static int
-makelabel(char *name, int setscope, Context *ctx)
+makelabel(char *name, int setscope)
 {
-	Item *l;
+	Label *l;
 	if(name[0] == '&')
-		name = makesublabel(name + 1);
+		name = makesublabel(sublabel, name + 1);
 	if(!slen(name)) return error_asm("Label is empty");
 	if(findlabel(name)) return error_asm("Label is duplicate");
 	if(sihx(name)) return error_asm("Label is hex number");
 	if(isopcode(name)) return error_asm("Label is opcode");
 	if(cndx(runes, name[0]) >= 0) return error_asm("Label name is runic");
-	if(label_len == 0x400) return error_asm("Labels limit exceeded");
+	if(p.label_len == 0x400) return error_asm("Labels limit exceeded");
-	l = &labels[label_len++];
+	l = &p.labels[p.label_len++];
-	l->name = push(name, 0);
+	l->addr = p.ptr;
-	l->addr = ptr;
 	l->refs = 0;
+	l->name = push(name);
 	if(setscope) {
 		int i = 0;
-		while(name[i] != '/' && i < 0x3e && (scope[i] = name[i]))
+		while(name[i] != '/' && i < 0x3e && (scope[i] = name[i])) i++;
-			i++;
 		scope[i] = '\0';
 	}
 	return 1;
@@ -209,161 +201,205 @@ makelambda(int id)
 	return lambda;
 }
 
+static char *
+makesublabel(char *buf, char *name)
+{
+	if(slen(scope) + slen(name) >= 0x3f) {
+		(void)error_asm("Sublabel length too long");
+		return NULL;
+	}
+	return scat(scat(scpy(scope, buf, 0x40), "/"), name);
+}
+
+static int
+makepad(char *w)
+{
+	Label *l;
+	int rel = w[0] == '$' ? p.ptr : 0;
+	if(sihx(w + 1))
+		p.ptr = shex(w + 1) + rel;
+	else if((l = findlabel(w + 1)))
+		p.ptr = l->addr + rel;
+	else
+		return error_asm("Invalid padding");
+	return 1;
+}
+
 static int
 addref(char *label, char rune, Uint16 addr)
 {
-	Item *r;
+	Reference *r;
-	if(refs_len >= 0x1000)
+	if(p.refs_len >= 0x1000)
-		return error_top("References limit exceeded", label);
+		return error_asm("References limit exceeded");
-	r = &refs[refs_len++];
+	r = &p.refs[p.refs_len++];
 	if(label[0] == '{') {
-		lambda_stack[lambda_ptr++] = lambda_len;
+		p.lambda_stack[p.lambda_ptr++] = p.lambda_len;
-		r->name = push(makelambda(lambda_len++), 0);
+		scpy(makelambda(p.lambda_len++), r->name, 0x40);
 	} else if(label[0] == '&' || label[0] == '/') {
-		r->name = makesublabel(label + 1);
+		if(!makesublabel(r->name, label + 1))
+			return error_asm("Invalid sublabel");
 	} else
-		r->name = push(label, 0);
+		scpy(label, r->name, 0x40);
 	r->rune = rune;
 	r->addr = addr;
 	return 1;
 }
 
 static int
-writepad(char *w)
+writebyte(Uint8 b)
 {
-	Item *l;
+	if(p.ptr < TRIM)
-	int rel = w[0] == '$' ? ptr : 0;
-	if(sihx(w + 1)) {
-		ptr = shex(w + 1) + rel;
-		return 1;
-	}
-	if((l = findlabel(w + 1))) {
-		ptr = l->addr + rel;
-		return 1;
-	}
-	return 0;
-}
-
-static int
-writebyte(Uint8 b, Context *ctx)
-{
-	if(ptr < PAGE)
 		return error_asm("Writing in zero-page");
-	else if(ptr >= 0x10000)
+	else if(p.ptr >= 0x10000)
 		return error_asm("Writing outside memory");
-	else if(ptr < length)
+	else if(p.ptr < p.length)
 		return error_asm("Writing rewind");
-	data[ptr++] = b;
+	p.data[p.ptr++] = b;
-	if(b)
+	p.length = p.ptr;
-		length = ptr;
 	return 1;
 }
 
 static int
-writehex(char *w, Context *ctx)
+writehex(char *w)
 {
 	if(*w == '#')
-		writebyte(findopcode("LIT") | (slen(++w) > 2) << 5, ctx);
+		writebyte(findopcode("LIT") | (slen(++w) > 2) << 5);
 	if(slen(w) == 2)
-		return writebyte(shex(w), ctx);
+		return writebyte(shex(w));
 	else if(slen(w) == 4)
 		return writeshort(shex(w));
 	else
 		return 0;
 }
 
+static int
+tokenize(FILE *f)
+{
+	char c;
+	char *cptr = token;
+	while(fread(&c, 1, 1, f)) {
+		if(c < 0x21) {
+			*cptr++ = 0x00;
+			if(c == 0x0a)
+				p.line++;
+			if(token[0])
+				if(!parse(token, f))
+					return 0;
+			cptr = token;
+		} else if(cptr - token < 0x3f)
+			*cptr++ = c;
+		else
+			return error_asm("Token too long");
+	}
+	return 1;
+}
+
 static int
 makeinclude(char *filename)
 {
 	FILE *f;
 	int res = 0;
-	Context ctx;
-	ctx.line = 0;
-	ctx.path = push(filename, 0);
 	if(!(f = fopen(filename, "r")))
 		return error_top("Invalid source", filename);
-	res = walkfile(f, &ctx);
+	scpy(filename, source, 0x40);
+	p.line = 0;
+	res = tokenize(f);
 	fclose(f);
 	return res;
 }
 
 static int
-writestring(char *w, Context *ctx)
+parse(char *w, FILE *f)
 {
+	int i;
 	char c;
-	while((c = *(w++)))
+	Macro *m;
-		if(!writebyte(c, ctx)) return 0;
-	return 1;
-}
-
-static int
-parse(char *w, FILE *f, Context *ctx)
-{
-	Item *m;
 	switch(w[0]) {
-	case '(': return !walkcomment(f, ctx) ? error_asm("Invalid comment") : 1;
+	case '(': return !walkcomment(w, f) ? error_asm("Invalid comment") : 1;
 	case '~': return !makeinclude(w + 1) ? error_asm("Invalid include") : 1;
-	case '%': return !makemacro(w + 1, f, ctx) ? error_asm("Invalid macro") : 1;
+	case '%': return !makemacro(w + 1, f) ? error_asm("Invalid macro") : 1;
-	case '@': return !makelabel(w + 1, 1, ctx) ? error_asm("Invalid label") : 1;
+	case '@': return !makelabel(w + 1, 1) ? error_asm("Invalid label") : 1;
-	case '&': return !makelabel(w, 0, ctx) ? error_asm("Invalid sublabel") : 1;
+	case '&': return !makelabel(w, 0) ? error_asm("Invalid sublabel") : 1;
-	case '}': return !makelabel(makelambda(lambda_stack[--lambda_ptr]), 0, ctx) ? error_asm("Invalid label") : 1;
+	case '#': return !sihx(w + 1) || !writehex(w) ? error_asm("Invalid hexadecimal") : 1;
-	case '#': return !sihx(w + 1) || !writehex(w, ctx) ? error_asm("Invalid hexadecimal") : 1;
+	case '_': return addref(w + 1, w[0], p.ptr) && writebyte(0xff);
-	case '_': return addref(w + 1, w[0], ptr) && writebyte(0xff, ctx);
+	case ',': return addref(w + 1, w[0], p.ptr + 1) && writebyte(findopcode("LIT")) && writebyte(0xff);
-	case ',': return addref(w + 1, w[0], ptr + 1) && writebyte(findopcode("LIT"), ctx) && writebyte(0xff, ctx);
+	case '-': return addref(w + 1, w[0], p.ptr) && writebyte(0xff);
-	case '-': return addref(w + 1, w[0], ptr) && writebyte(0xff, ctx);
+	case '.': return addref(w + 1, w[0], p.ptr + 1) && writebyte(findopcode("LIT")) && writebyte(0xff);
-	case '.': return addref(w + 1, w[0], ptr + 1) && writebyte(findopcode("LIT"), ctx) && writebyte(0xff, ctx);
 	case ':': fprintf(stderr, "Deprecated rune %s, use =%s\n", w, w + 1); /* fall-through */
-	case '=': return addref(w + 1, w[0], ptr) && writeshort(0xffff);
+	case '=': return addref(w + 1, w[0], p.ptr) && writeshort(0xffff);
-	case ';': return addref(w + 1, w[0], ptr + 1) && writebyte(findopcode("LIT2"), ctx) && writeshort(0xffff);
+	case ';': return addref(w + 1, w[0], p.ptr + 1) && writebyte(findopcode("LIT2")) && writeshort(0xffff);
-	case '?': return addref(w + 1, w[0], ptr + 1) && writebyte(0x20, ctx) && writeshort(0xffff);
+	case '?': return addref(w + 1, w[0], p.ptr + 1) && writebyte(0x20) && writeshort(0xffff);
-	case '!': return addref(w + 1, w[0], ptr + 1) && writebyte(0x40, ctx) && writeshort(0xffff);
+	case '!': return addref(w + 1, w[0], p.ptr + 1) && writebyte(0x40) && writeshort(0xffff);
-	case '"': return !writestring(w + 1, ctx) ? error_asm("Invalid string") : 1;
+	case '}': return !makelabel(makelambda(p.lambda_stack[--p.lambda_ptr]), 0) ? error_asm("Invalid label") : 1;
 	case '$':
-	case '|': return !writepad(w) ? error_asm("Invalid padding") : 1;
+	case '|': return !makepad(w) ? error_asm("Invalid padding") : 1;
 	case '[':
-	case ']': return 1;
+	case ']':
-	}
+		if(slen(w) == 1) break; /* else fallthrough */
+	case '"':                   /* raw string */
+		while((c = *(++w)))
+			if(!writebyte(c)) return 0;
+		break;
+	default:
 		if(sihx(w))
-		return writehex(w, ctx);
+			return writehex(w);
 		else if(isopcode(w))
-		return writebyte(findopcode(w), ctx);
+			return writebyte(findopcode(w));
-	else if((m = findmacro(w)))
+		else if((m = findmacro(w))) {
-		return walkmacro(m, ctx);
+			for(i = 0; i < m->len; i++)
-	return addref(w, ' ', ptr + 1) && writebyte(0x60, ctx) && writeshort(0xffff);
+				if(!parse(m->items[i], f))
+					return 0;
+			return 1;
+		} else
+			return addref(w, ' ', p.ptr + 1) && writebyte(0x60) && writeshort(0xffff);
+	}
+	return 1;
 }
 
 static int
 resolve(void)
 {
-	int i, rel;
+	Label *l;
-	for(i = 0; i < refs_len; i++) {
+	int i;
-		Item *r = &refs[i], *l = findlabel(r->name);
+	Uint16 a;
-		Uint8 *rom = data + r->addr;
+	for(i = 0; i < p.refs_len; i++) {
-		if(!l) return 0;
+		Reference *r = &p.refs[i];
+		Uint8 *rom = p.data + r->addr;
 		switch(r->rune) {
 		case '_':
 		case ',':
-			*rom = rel = l->addr - r->addr - 2;
+			if(!(l = findlabel(r->name)))
-			if((Sint8)data[r->addr] != rel)
+				return error_top("Unknown relative reference", r->name);
+			*rom = (Sint8)(l->addr - r->addr - 2);
+			if((Sint8)p.data[r->addr] != (l->addr - r->addr - 2))
 				return error_top("Relative reference is too far", r->name);
+			l->refs++;
 			break;
 		case '-':
 		case '.':
+			if(!(l = findlabel(r->name)))
+				return error_top("Unknown zero-page reference", r->name);
 			*rom = l->addr;
+			l->refs++;
 			break;
 		case ':':
 		case '=':
 		case ';':
+			if(!(l = findlabel(r->name)))
+				return error_top("Unknown absolute reference", r->name);
 			*rom++ = l->addr >> 8, *rom = l->addr;
+			l->refs++;
 			break;
 		case '?':
 		case '!':
 		default:
-			rel = l->addr - r->addr - 2;
+			if(!(l = findlabel(r->name)))
-			*rom++ = rel >> 8, *rom = rel;
+				return error_top("Unknown subroutine reference", r->name);
+			a = l->addr - r->addr - 2;
+			*rom++ = a >> 8, *rom = a;
+			l->refs++;
 			break;
 		}
-		l->refs++;
 	}
 	return 1;
 }
@@ -372,16 +408,16 @@ static void
 review(char *filename)
 {
 	int i;
-	for(i = 0; i < label_len; i++)
+	for(i = 0; i < p.label_len; i++)
-		if(labels[i].name[0] - 'A' > 25 && !labels[i].refs)
+		if(p.labels[i].name[0] - 'A' > 25 && !p.labels[i].refs)
-			fprintf(stdout, "-- Unused label: %s\n", labels[i].name);
+			fprintf(stdout, "-- Unused label: %s\n", p.labels[i].name);
 	fprintf(stdout,
 		"Assembled %s in %d bytes(%.2f%% used), %d labels, %d macros.\n",
 		filename,
-		length - PAGE,
+		p.length - TRIM,
-		(length - PAGE) / 652.80,
+		(p.length - TRIM) / 652.80,
-		label_len,
+		p.label_len,
-		macro_len);
+		p.macro_len);
 }
 
 static void
@@ -394,11 +430,11 @@ writesym(char *filename)
 		return;
 	fp = fopen(scat(scpy(filename, symdst, slen(filename) + 1), ".sym"), "w");
 	if(fp != NULL) {
-		for(i = 0; i < label_len; i++) {
+		for(i = 0; i < p.label_len; i++) {
-			Uint8 hb = labels[i].addr >> 8, lb = labels[i].addr;
+			Uint8 hb = p.labels[i].addr >> 8, lb = p.labels[i].addr;
 			fwrite(&hb, 1, 1, fp);
 			fwrite(&lb, 1, 1, fp);
-			fwrite(labels[i].name, slen(labels[i].name) + 1, 1, fp);
+			fwrite(p.labels[i].name, slen(p.labels[i].name) + 1, 1, fp);
 		}
 	}
 	fclose(fp);
@@ -408,16 +444,15 @@ int
 main(int argc, char *argv[])
 {
 	FILE *dst;
-	ptr = PAGE;
+	p.ptr = 0x100;
 	scpy("on-reset", scope, 0x40);
 	if(argc == 1) return error_top("usage", "uxnasm [-v] input.tal output.rom");
-	if(scmp(argv[1], "-v", 2)) return !fprintf(stdout, "Uxnasm - Uxntal Assembler, 27 Mar 2024.\n");
+	if(scmp(argv[1], "-v", 2)) return !fprintf(stdout, "Uxnasm - Uxntal Assembler, 26 Mar 2024.\n");
-	if(!makeinclude(argv[1])) return !error_top("Assembly", "Failed to assemble rom.");
+	if(!makeinclude(argv[1]) || !resolve()) return !error_top("Assembly", "Failed to assemble rom.");
-	if(!resolve()) return !error_top("Assembly", "Failed to resolve symbols.");
 	if(!(dst = fopen(argv[2], "wb"))) return !error_top("Invalid Output", argv[2]);
-	if(length <= PAGE) return !error_top("Assembly", "Output rom is empty.");
+	if(p.length <= TRIM) return !error_top("Assembly", "Output rom is empty.");
 	review(argv[2]);
-	fwrite(data + PAGE, length - PAGE, 1, dst);
+	fwrite(p.data + TRIM, p.length - TRIM, 1, dst);
 	writesym(argv[2]);
 	return 0;
 }