unify printing + add basic strings

ir.c

@@ -72,32 +72,11 @@ void irtoks_eat_irtoks(IRToks *v, IRToks *other, size_t jmp_offset) {
 	free(other->toks);
 }
 
-static void print_val(const Value *v);
 static void print_irparam(const IRParam *p);
 
-static void print_val(const Value *v) {
-	switch (v->type.kind) {
-		case TypeFloat:
-			printf("%f", v->Float);
-			break;
-		case TypeInt:
-			printf("%zd", v->Int);
-			break;
-		case TypeBool:
-			printf("%s", v->Bool ? "true" : "false");
-			break;
-		case TypeChar:
-			printf("'%c'", v->Char);
-			break;
-		default:
-			printf("(unknown type)");
-			break;
-	}
-}
-
 static void print_irparam(const IRParam *p) {
 	if (p->kind == IRParamLiteral) {
-		print_val(&p->Literal);
+		print_value(&p->Literal, false);
 	} else if (p->kind == IRParamAddr) {
 		printf("%%%zd", p->Addr);
 	}

lex.c

@@ -54,7 +54,7 @@ static char get_esc_char(char c) {
 	}
 }
 
-TokList lex(const char *s) {
+TokList lex(const char *s, Pool *static_vars) {
 	TokList toks;
 	toklist_init(&toks);
 	Pos pos = { .ln = 1, .col = 1 };
@@ -299,6 +299,52 @@ TokList lex(const char *s) {
 				emit(&toks, &pos, (Tok){ .kind = TokVal, .Val = { .type = { .kind = TypeChar, }, .Char = c, }, });
 				break;
 			}
+			case '"': {
+				consume(&pos, *(s++));
+				const char *start = s;
+				Pos start_pos = pos;
+				size_t size = 0;
+
+				/* count the string size before allocating */
+				while (s[0] != '"') {
+					if (!s[0]) {
+						set_err("Unexpected EOF in string literal");
+						return toks;
+					} else if (s[0] == '\\')
+						consume(&pos, *(s++));
+					consume(&pos, *(s++));
+					size++;
+				}
+
+				/* go through the actual string */
+				s = start;
+				pos = start_pos;
+				char *str = pool_alloc(static_vars, type_size[TypeChar] * size);
+				for (size_t i = 0; i < size; i++) {
+					char c = s[0];
+					if (c == '\\') {
+						consume(&pos, *(s++));
+						c = get_esc_char(s[0]);
+						if (!c) {
+							set_err("Unrecognized escape sequence: '\\%c'", c);
+							return toks;
+						}
+					}
+					consume(&pos, *(s++));
+					str[i] = c;
+				}
+				emit(&toks, &pos, (Tok){ .kind = TokVal, .Val = {
+						.type.kind = TypeArr,
+						.Arr = {
+							.is_string = true,
+							.type.kind = TypeChar,
+							.vals = str,
+							.len = size,
+							.cap = size,
+						},
+					},});
+				break;
+			}
 			default:
 				set_err("Unrecognized character: '%c'", s[0]);
 				return toks;

lex.h

@@ -3,6 +3,6 @@
 
 #include "tok.h"
 
-TokList lex(const char *s);
+TokList lex(const char *s, Pool *static_vars);
 
 #endif /* LEX_H */

main.c

@@ -34,15 +34,7 @@ static void die(const char *fmt, ...) {
 }
 
 static Value fn_put(Value *args) {
-	switch (args[0].type.kind) {
+	print_value(&args[0], true);
-		case TypeVoid:  printf("(void)");                              break;
-		case TypeFloat: printf("%f", args[0].Float);                   break;
-		case TypeInt:   printf("%zd", args[0].Int);                    break;
-		case TypeBool:  printf("%s", args[0].Bool ? "true" : "false"); break;
-		case TypeChar:  printf("%c", args[0].Char);                    break;
-		default:
-			ASSERT_UNREACHED();
-	}
 	return (Value){0};
 }
 
@@ -136,9 +128,11 @@ int main(int argc, const char **argv) {
 	}
 	fclose(fp);
 	/* lex source file */
-	TokList tokens = lex(file);
+	Pool *static_vars = pool_new(4096);
+	TokList tokens = lex(file, static_vars);
 	if (err) {
 		toklist_term(&tokens);
+		pool_term(static_vars);
 		free(file);
 		fprintf(stderr, C_IRED "Lexer error" C_RESET " in " C_CYAN "%s" C_RESET ":%zu:%zu: %s\n", filename, err_ln, err_col, errbuf);
 		return 1;
@@ -158,6 +152,7 @@ int main(int argc, const char **argv) {
 	if (err) {
 		irtoks_term(&ir);
 		toklist_term(&tokens);
+		pool_term(static_vars);
 		fprintf(stderr, C_IRED "Parser error" C_RESET " in " C_CYAN "%s" C_RESET ":%zu:%zu: %s\n", filename, err_ln, err_col, errbuf);
 		return 1;
 	}
@@ -169,9 +164,11 @@ int main(int argc, const char **argv) {
 		run(&ir, funcs);
 		if (err) {
 			irtoks_term(&ir);
+			pool_term(static_vars);
 			fprintf(stderr, C_IRED "Runtime error" C_RESET " in " C_CYAN "%s" C_RESET ":%zu:%zu: %s\n", filename, err_ln, err_col, errbuf);
 			return 1;
 		}
 	}
 	irtoks_term(&ir);
+	pool_term(static_vars);
 }

tok.c

@@ -5,6 +5,75 @@
 
 #include "util.h"
 
+size_t type_size[TypeEnumSize] = {
+	[TypeVoid]  = 0,
+	[TypeFloat] = sizeof(((Value*)NULL)->Float),
+	[TypeInt]   = sizeof(((Value*)NULL)->Int),
+	[TypeBool]  = sizeof(((Value*)NULL)->Bool),
+	[TypeChar]  = sizeof(((Value*)NULL)->Char),
+	[TypeArr]   = sizeof(((Value*)NULL)->Arr),
+};
+
+void print_value(const Value *v, bool raw) {
+	switch (v->type.kind) {
+		case TypeVoid:
+			printf("(void)");
+			break;
+		case TypeFloat:
+			printf("%f", v->Float);
+			break;
+		case TypeInt:
+			printf("%zd", v->Int);
+			break;
+		case TypeBool:
+			printf("%s", v->Bool ? "true" : "false");
+			break;
+		case TypeChar:
+			if (raw)
+				printf("%c", v->Char);
+			else {
+				const char *esc = unescape_char(v->Char);
+				if (esc) printf("'%s'", esc);
+				else     printf("'%c'", v->Char);
+			}
+			break;
+		case TypeArr:
+			if (v->Arr.is_string) {
+				if (v->Arr.type.kind != TypeChar)
+					ASSERT_UNREACHED();
+				char *str = v->Arr.vals;
+				if (!raw)
+					printf("\"");
+				for (size_t i = 0; i < v->Arr.len; i++) {
+					char c = str[i];
+					if (raw)
+						printf("%c", c);
+					else {
+						const char *esc = unescape_char(c);
+						if (esc) printf("%s", esc);
+						else     printf("%c", c);
+					}
+				}
+				if (!raw)
+					printf("\"");
+			} else {
+				printf("[");
+				for (size_t i = 0;; i++) {
+					size_t ty_sz = type_size[v->Arr.type.kind];
+					Value ty_val = { .type = v->Arr.type };
+					memcpy(&ty_val.Void, (uint8_t*)v->Arr.vals + ty_sz * i, ty_sz);
+					print_value(&ty_val, false);
+					if (i == v->Arr.len-1) break;
+					printf(", ");
+				}
+				printf("]");
+			}
+			break;
+		default:
+			ASSERT_UNREACHED();
+	}
+}
+
 int8_t op_prec[OperatorEnumSize] = {
 	[OpEOF]    = PREC_DELIM,
 	[OpNewLn]  = PREC_DELIM,
@@ -108,24 +177,9 @@ void print_toks(TokList *l) {
 				printf(": " C_ICYAN "%s" C_RESET, op_str[i->tok.Op]);
 				break;
 			case TokVal:
-				printf(C_IYELLOW "Val" C_RESET);
+				printf(C_IYELLOW "Val" C_RESET ": " C_ICYAN);
-				switch (i->tok.Val.type.kind) {
+				print_value(&i->tok.Val, false);
-					case TypeFloat:
+				printf(C_RESET);
-						printf(": " C_ICYAN "%f" C_RESET, i->tok.Val.Float);
-						break;
-					case TypeInt:
-						printf(": " C_ICYAN "%zd" C_RESET, i->tok.Val.Int);
-						break;
-					case TypeBool:
-						printf(": " C_ICYAN "%s" C_RESET, i->tok.Val.Bool ? "true" : "false");
-						break;
-					case TypeChar:
-						printf(": " C_ICYAN "'%c'" C_RESET, i->tok.Val.Char);
-						break;
-					default:
-						printf(" " C_ICYAN "(unknown type)" C_RESET);
-						break;
-				}
 				break;
 			case TokIdent:
 				printf(C_IYELLOW "Ident" C_RESET);

tok.h

@@ -13,23 +13,36 @@ typedef struct Type {
 		TypeInt,
 		TypeBool,
 		TypeChar,
+		TypeArr,
+		TypeEnumSize,
 	} kind;
 
 	/*union {
 	};*/
 } Type;
 
+extern size_t type_size[TypeEnumSize];
+
 typedef struct Value {
 	Type type;
 
 	union {
+		pseudo_void Void;
 		double Float;
 		ssize_t Int;
 		bool Bool;
 		char Char;
+		struct {
+			bool is_string : 1;
+			Type type;
+			void *vals;
+			size_t len, cap;
+		} Arr;
 	};
 } Value;
 
+void print_value(const Value *v, bool raw);
+
 enum Operator {
 	OpLCurl  = '{',
 	OpRCurl  = '}',

util.c

@@ -168,6 +168,23 @@ double stod(const char *s, size_t n, ssize_t *endpos) {
 	return res;
 }
 
+const char *unescape_char(char c) {
+	switch (c) {
+		case '\a':   return "\\a";
+		case '\b':   return "\\b";
+		case '\033': return "\\e";
+		case '\f':   return "\\f";
+		case '\n':   return "\\n";
+		case '\r':   return "\\r";
+		case '\t':   return "\\t";
+		case '\v':   return "\\v";
+		case '\\':   return "\\\\";
+		case '\'':   return "\\'";
+		case '"':    return "\\\"";
+		default:     return NULL;
+	}
+}
+
 char *mreadfile(FILE *fp) {
 	if (fseek(fp, 0l, SEEK_END) == -1)
 		return NULL;

util.h

@@ -8,6 +8,8 @@
 #include <string.h>
 #include <unistd.h>
 
+typedef uint8_t pseudo_void;
+
 /* some ANSI color codes */
 #define C_RED     "\x1b[31m"
 #define C_GREEN   "\x1b[32m"
@@ -67,6 +69,8 @@ char *psndup(Pool *p, const char *s, size_t n);
 intmax_t stoimax(const char *s, size_t n, size_t base, ssize_t *endpos /* -1 on success */);
 /* convert a non-null-terminated string to a double */
 double stod(const char *s, size_t n, ssize_t *endpos /* -1 on success */);
+/* return the escape sequence for a given character; return NULL if there is none */
+const char *unescape_char(char c);
 
 /* sets errno on failure */
 char *mreadfile(FILE *fp);