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

@@ -6,24 +6,72 @@ #include <string.h>
 
 #include "../bindump.h"
 
-#define DEBUG(x) printf(x)
+#define DEBUG(...)                printf(__VA_ARGS__)
+
+#define die(err_msg, exit_val)  perror(err_msg); exit(exit_val)
+#define alert(...)              fprintf(stderr, __VA_ARGS__)
 
-void die(const char *err_msg, int exit_val)
-{
-	perror(err_msg);
-	exit(exit_val);
-}
 
-void alert(const char *err_msg)
+uint32_t myatoi(const char *a_field_str)
 {
-	fprintf(stderr, "%s", err_msg);
+	DEBUG("a_field_str = %s\n", a_field_str);
+	// I know, I could just use atoi(), but where's the fun in that?
+	// STUB
+	return 0xaff;
 }
 
 bool parse_a_type(const char *line, uint16_t *instruction)
 {
-	DEBUG("parse_a_type()\n");
-	// eventually, if error, die("error: TODO explain...\n", -1);
-	*instruction = 0x0000; // STUB, A-type MSB == 0 anyway
+	bool num_found = false;
+	char c, a_field_str[6];  // TODO: eventually factor out use of array
+	uint32_t a_field = 0;
+	size_t i, a = 0;
+
+	i = 0;
+	while ((c = line[i]) != '\n') {
+		if (c == ' ' || c == '\t' || c == '/') {
+			if (num_found) {
+				num_found = false;
+				if (a < 7) {
+					a_field_str[a] = '\0';
+					break;
+				} else {
+					die("fatal error: a-type value index "
+					    "%lu out of range\n", -1);
+				}
+			} else {
+				;  // skip whitespace
+			}
+		} else if (num_found) {
+			if ('0' <= c && c <= '9') {
+				if (a > 4) {
+					alert("error: @<number> too long\n");
+					return false;
+				}
+				a_field_str[a] = c; // get number
+			} else {
+				alert("syntax error: invalid char '%c' found "
+				      "after @\n", c);
+				return false;
+			}
+			++a;
+		} else if (c == '@') {
+			a = 0;
+			num_found = true;
+		} else {
+			alert("syntax error: invalid char '%c' before @\n", c);
+			return false;
+		}
+		++i;
+	}
+	// TODO: extension: support negative numbers
+	a_field = myatoi(a_field_str);
+	if (a_field > 32767) {
+		alert("error: %u > 32767, too large\n", a_field);
+		return false;
+	}
+
+	*instruction = 0x0000 | (uint16_t) a_field;
 	return true; // STUB, A-type MSB == 0 anyway
 }
 
@@ -33,7 +81,6 @@ DEBUG("parse_c_type()\n");
 	*instruction = 0x8000; // STUB, C-type MSB == 1 anyway
 	return true; // STUB, C-type MSB == 1 anyway
 }
-// */
 
 // does not care about line line length; exits at first newline or after
 // relevant portion parsed (allows for syntactically-incorrect lines, I know)
@@ -61,7 +108,6 @@ }
 
 	return ret;
 }
-// */
 
 bool parse_next_instruction(const char *line, uint16_t *instruction)
 {
@@ -93,13 +139,12 @@ if (slash_found) {
 				// second slash means this is a comment
 				return false;
 			}
-			slash_found= true;
+			slash_found = true;
 			continue;
 		} else if (slash_found) {
 			// this char not slash, but previous was: invalid syntax
 			// TODO: add line, column numbers
-			alert("\tsyntax error: found '/', comments "
-			      "need '//'\n");
+			alert("syntax error: found '/', comments need '//'\n");
 			return false;
 		} else {
 			// non-whitespace/slash char discovered
@@ -111,14 +156,17 @@ // comment not found, so attempting to parse instruction
 	return parse_next_instruction(line, instruction);
 }
 
-const char *test_lines[8] = {"// this is a comment\n",
-                             "@12345\n",
-                             "\n",
-                             "M+1\n",
-                             "@98\n",
-                             "// this is another comment\n",
-                             "/ this is a broken comment\n",
-                             "D=M;JNE\n"};
+const char *test_lines[] = {"// this is a comment\n",               // line 1
+			    "@12345\n",                             // line 2
+			    "\n",                                   // line 3
+                            "M+1\n",                                // line 4
+                            "@98 // test comment\n",                // line 5
+                            "// this is another comment\n",         // line 6
+                            "/ this is a broken comment\n",         // line 7
+                            "D=M;JNE\n",                            // line 8
+			    " @1337// immediate comment \n",        // line 9
+			    "\t@13371337 // number too long \n",    // line 10
+			    "    @0x1337 // invalid char in num\n"}; // line 11
 size_t num_test_lines = (sizeof(test_lines) / sizeof(test_lines[0]));
 
 int main()
@@ -133,10 +181,11 @@ // from a 'fgets()' loop or something
 
 	// for testing
 	for (size_t i = 0; i < num_test_lines; ++i) {
+		DEBUG("%lu|%s", i+1, test_lines[i]);
 		test_line_len = strlen(test_lines[i]);
 		result = parse_line(test_lines[i], test_line_len, &instruction);
 		if (result) {
-			printf("instruction: 0x%x  |  ", instruction);
+			DEBUG("instruction: 0x%x  |  ", instruction);
 			bindump_word16(instruction); // output instruction as binary
 			putchar('\n');
 		}