A projects/06/POSSIBLE-BUGS.md

@@ -0,0 +1,8 @@ 
+
+assembler1:
+If a line contains a label larger than `(MAX_LINE_LEN - line_length)` (Note: technically beside the point here, but `MAX_LINE_LEN == 256`), the label may get truncated or something else will happen (I'm unsure...)
+
+Junk may be allowed after C-type instruction based on how it's parsed (parsing exits once expected characters are found in the right places, so the whole line isn't necessarily validated)
+
+It might be unsafe to assume `comp_start` and `jump_start` start at `&line[i+1]` after the '+' or ';' in `parse_c_type()` (unsure)
+

M projects/06/assembler1/assembler1.c → projects/06/assembler1/assembler1.c

@@ -10,6 +10,10 @@ #define error(...)              fprintf(stderr, __VA_ARGS__); \
                                 fprintf(stderr, "%lu | %s\n", \
                                         g_asm_line_number, g_asm_line);
 
+#define valid_symbol_char(c) (('0' <= c && c <= '9') || ('A' <= c && c <= 'Z') \
+                           || ('a' <= c && c <= 'z') || c == '_' || c == '.'   \
+		                                     || c == '$' || c == ':')
+
 #define MAX_LINE_LEN            256
 
 #define MAX_SYMBOL_STR_LEN      MAX_LINE_LEN - 2
@@ -42,6 +46,7 @@ // label and variable symbols will effectively be treated the same; the only
 // difference is in their use (in a hack program)
 struct symbol_t *g_symbol_list[MAX_SYMBOLS + RESERVED_LABEL_NUM];
 size_t g_symbol_list_len = 0;
+size_t g_current_variable_address = 16; // address for next variable symbol
 
 
 void print_binary_word16(uint16_t w)
@@ -181,15 +186,11 @@ size_t i;
 	struct symbol_t *s;
 
 	for (i = 0; i < RESERVED_LABEL_NUM; ++i) {
-		//DBGLOG("init_symbol_list: adding symbol {%s, %hu}\n",
-		//        g_reserved_symbol_strs[i], g_reserved_symbol_values[i]);
 		if((s = create_symbol(g_reserved_symbol_strs[i],
 		                      g_reserved_symbol_values[i])) == NULL) {
-			//DBGLOG("init_symbol_list: failed to create symbol\n");
 			return false;  // failed to create symbol
 		}
 		if (!add_symbol(s)) {
-			//DBGLOG("init_symbol_list: failed to add symbol\n");
 			return false;  // failed to add symbol
 		}
 	}
@@ -197,22 +198,17 @@ 
 	return true;
 }
 
-// assumes line[0] == '('
-bool parse_label(const char *line)
+// assumes line starts at symbol (_not_ at '(' character, if relevant)
+bool parse_symbol_str(const char *line, char *symbol_str, size_t *symbol_len)
 {
+	char c;
 	size_t i;
-	uint16_t value;
-	char c, labelstr[MAX_SYMBOL_STR_LEN + 1];
-	struct symbol_t *symbol;
 
-	//DBGLOG("Parsing label: line %lu | %s\n", g_asm_line_number, line);
+	//DBGLOG("Parsing symbol string: line %lu | %s\n", g_asm_line_number, line);
 
-	if (line[0] != '(') {  // just to be safe
-		return false;  // not a label; label's look like (THIS)
-	}
-
-	for (i = 1; (c = line[i]) != '\0'; ++i) {
-		if (i == 1 && ('0' <= c && c <= '9')) {
+	*symbol_len = 0;
+	for (i = 0; (c = line[i]) != '\0'; ++i) {
+		if (i == 0 && ('0' <= c && c <= '9')) {
 			error("syntax error: label starts with a number\n");
 			return false;
 		}
@@ -221,38 +217,57 @@ if (i > MAX_SYMBOL_STR_LEN) {
 			error("syntax error: label is too long (> %d chars)\n",
 			       MAX_SYMBOL_STR_LEN);
 			return false;
-		} else if (c == ')') {
+		} else if (valid_symbol_char(c)) {
+			symbol_str[i] = c;
+		} else {
 			if (i == 1) {
 				error("syntax error: empty label '()'\n");
 				return false;
 			}
 
-			labelstr[i-1] = '\0';
-
-			// create label, add to symbol list
-			value = (uint16_t)g_instruction_number; // safe. won't
-								// be > 65535
-			symbol = create_symbol(labelstr, value);
-			if (symbol == NULL)
-				return false;
-			return add_symbol(symbol);
 			break;
-		} else if (('0' <= c && c <= '9') || ('A' <= c && c <= 'Z')
-		        || ('a' <= c && c <= 'z') || c == '_') {
-			// TODO TODO: enforce label name NOT start with a number!
-			labelstr[i-1] = c;
-		} else {
-			error("syntax error: disallowed character '%hhu' found"
-			      " in label\n", c);
-			return false;
 		}
 	}
 
-	if (c == '\0') { // should_ be ')' after break if all goes as expected
+	symbol_str[i] = '\0';
+	*symbol_len = i;
+
+	return true;
+}
+
+// assumes line[0] == '('
+bool parse_label(const char *line)
+{
+	bool ret;
+	size_t label_len;
+	uint16_t value;
+	char label_str[MAX_SYMBOL_STR_LEN + 1];
+	struct symbol_t *symbol;
+
+	//DBGLOG("Parsing label: line %lu | %s\n", g_asm_line_number, line);
+
+	if (line[0] != '(') {  // just to be safe
+		return false;  // not a label; label's look like (THIS)
+	}
+
+	ret = parse_symbol_str(&line[1], label_str, &label_len);
+	if (!ret) {
+		return false;
+	}
+
+	if (line[label_len + 1] != ')') {
+		error("line[label_len = %lu] value is '%c'\n", label_len, line[label_len])
 		error("syntax error: no matching ')' found for label");
+		return false;
 	}
 
-	return false;
+	// create label, add to symbol list
+	value = (uint16_t)g_instruction_number; // safe. won't
+						// be > 65535
+	symbol = create_symbol(label_str, value);
+	if (symbol == NULL)
+		return false;
+	return add_symbol(symbol);
 }
 
 bool parse_line_for_label(const char *line)
@@ -283,9 +298,11 @@ }
 
 static bool parse_a_type(const char *line, uint16_t *instruction)
 {
-	char c, a_field_str[6];
+	bool ret;
+	char c, a_field_str[6], symbol_str[MAX_SYMBOL_STR_LEN + 1];
 	uint32_t a_field = 0;
-	size_t i, a = 0;
+	size_t i, a = 0, symbol_len;
+	struct symbol_t *symbol;
 
 	//DBGLOG("line: %s\n", line);
 
@@ -301,18 +318,24 @@ error("syntax error: A-type instruction empty after @\n");
 		return false;
 	}
 
-	if (('A' <= c && c <= 'Z') || ('a' <= c && c <= 'z')) {
-		// parse symbol/label, handle appropriately
-		// lookup symbol
-		// if (symbol already exists as label)
-		//     a_field = symbol value from table
-		// else
-		//     calc new variable value
-		//     put symbol in table w/ new variable value
-		//     a_field = new value
-		//     increment global variable value counter
-		DBGLOG("Found label/variable in A-instruction '%s'\n", line);
-		a_field = 65; // STUB!!!
+	if (valid_symbol_char(c) && !('0' <= c && c <= '9')) {
+		ret = parse_symbol_str(&line[1], symbol_str, &symbol_len);
+		if (!ret)
+			return ret;
+
+		symbol = lookup_symbol(symbol_str);
+		if (symbol == NULL) {  // symbol doesn't exist in table, is new
+			a_field = g_current_variable_address;
+			symbol = create_symbol(symbol_str, a_field);
+			if (symbol == NULL)
+				return false; // create failed
+			ret = add_symbol(symbol);
+			if (!ret)
+				return ret;
+			++g_current_variable_address;
+		} else {               // symbol exists in table, use its value
+			a_field = symbol->value;  // symbol value from table
+		}
 	} else {
 		for (i = 1; (c = line[i]) != '\0' && a < 6; ++i) {
 			if ('0' <= c && c <= '9') {
@@ -324,17 +347,22 @@ a_field_str[a] = c; // get number
 				a++;
 			} else if ((c == ' ' || c == '\t' || c == '/') && i > 1) {
 				break;
-			} else {       // any other character
+			} else {       // any other character encountered
 				error("syntax error: invalid char '%c' found "
 				      "after @\n", c);
 				return false;
 			}
 		}
 
+		if (a == 0) {
+			error("syntax error: A-type instruction empty "
+			      "after '@'\n");
+			return false;
+		}
+
 		a_field_str[a] = '\0';
-		// TODO: basic check on a_field_str length? at least 1?
 
-		a_field = myatoi(a_field_str);  // TODO: maybe negative number support?
+		a_field = myatoi(a_field_str);
 	}
 
 	if (a_field > 32767) {
@@ -352,6 +380,7 @@ */
 static bool parse_c_type_dest(const char *dest_line, uint8_t *dest)
 {
 	size_t len;
+
 	//DBGLOG("dest_line: %s\n", dest_line);
 
 	for (len = 0; dest_line[len] != '='; ++len) {  // read until '='
@@ -398,6 +427,7 @@ */
 static bool parse_c_type_comp(const char *comp_line, uint8_t *comp)
 {
 	size_t len;
+
 	//DBGLOG("comp_line: %s\n", comp_line);
 
 	for (len = 0; comp_line[len] == '0' || comp_line[len] == '1' ||
@@ -743,8 +773,6 @@ fprintf(stderr, "internal: failed to initialize symbol list\n");
 		exit(-1);
 	}
 
-	DBGLOG("Symbol initialization complete.\n");
-
 	file_line = 1;
 	// first pass to read labels and associate with values
 	while (fgets(in_line, MAX_LINE_LEN, fp) != NULL) {  // parse loop
@@ -762,9 +790,8 @@ 
 		g_asm_line = in_line;
 		g_asm_line_number = file_line;
 		result = parse_line_for_label(in_line);
-		if (result) {
-			DBGLOG("DEBUG: label found in line %s\n", in_line);
-		}
+		//if (result)
+		//	DBGLOG("DEBUG: label found in line %s\n", in_line);
 
 		++file_line;
 	}
@@ -773,6 +800,8 @@ 
 	if (fclose(fp)) {
 		fprintf(stderr, "Error closing file %s. Aborting...\n", in_file_path);
 	}
+
+	// TODO: skip second pass if error found in first pass
 
 	// TODO: keep file open, restart read from beginning
 	fp = fopen(in_file_path, "r");
@@ -807,7 +836,7 @@ 
 		++file_line;
 	}
 
-	debug_dump_all_symbols();
+	//debug_dump_all_symbols();
 
 	free_all_symbols();