From b3e5203f6260cb400d7c73b56f91eddea6b24502 Mon Sep 17 00:00:00 2001
From: kdx <kikoodx@paranoici.org>
Date: Sat, 10 Jun 2023 19:06:23 +0200
Subject: string literal

---
 samples/helloworld.golem | 15 ++++++++++
 src/main.c               | 75 ++++++++++++++++++++++++++++++++++++++----------
 testing.sh               |  1 +
 3 files changed, 76 insertions(+), 15 deletions(-)

diff --git a/samples/helloworld.golem b/samples/helloworld.golem
index d1758ac..aac5202 100644
--- a/samples/helloworld.golem
+++ b/samples/helloworld.golem
@@ -3,6 +3,9 @@ global str[32];
 main() {
 	local i;
 
+	print("uwu");
+	wrt '\n';
+
 	strappend(str, 'H');
 	strappend(str, 'e');
 	strappend(str, 'L');
@@ -27,6 +30,18 @@ main() {
 	}
 }
 
+strcpy(dst, src) {
+	local rv;
+	rv = dst;
+	loop {
+		[dst] = [src];
+		src = src + 1;
+		if ([src] == 0)
+			return rv;
+		dst = dst + 1;
+	}
+}
+
 strappend(s, c) {
 	loop {
 		if ([s] == 0) {
diff --git a/src/main.c b/src/main.c
index ecb539a..cbddd10 100644
--- a/src/main.c
+++ b/src/main.c
@@ -10,6 +10,7 @@ typedef enum {
 	TOK_PUNCT,
 	TOK_NUM,
 	TOK_WORD,
+	TOK_STRING,
 	TOK_EOF,
 } TokenType;
 
@@ -17,6 +18,7 @@ static const char * str_tok[] = {
 	[TOK_PUNCT] = "TOK_PUNCT",
 	[TOK_NUM] = "TOK_NUM",
 	[TOK_WORD] = "TOK_WORD",
+	[TOK_STRING] = "TOK_STRING",
 	[TOK_EOF] = "TOK_EOF",
 };
 
@@ -109,6 +111,15 @@ tokenize(char *p)
 			cur->val = val;
 			continue;
 		}
+		
+		if (*p == '"') {
+			char *end = strchr(p + 1, '"');
+			if (end == NULL)
+				error("unclosed double quotes");
+			cur = cur->next = new_token(TOK_STRING, p + 1, end);
+			p = end + 1;
+			continue;
+		}
 
 		if (*p == '\'' && p[1] != '\0' && p[2] == '\'') {
 			cur = cur->next = new_token(TOK_NUM, p, p + 3);
@@ -159,6 +170,7 @@ typedef enum {
 	NOD_OR,  // |
 	NOD_XOR, // ^
 	NOD_NUM, // integer
+	NOD_STRING, // ""
 	NOD_DEREF, // [a]
 	NOD_BLOCK_STMT,
 	NOD_EXPR_STMT,
@@ -291,6 +303,8 @@ static Node *primary(Token **rest, Token *tok);
 
 static Node *locals;
 static int locals_size;
+static Node *strings;
+static int strings_size;
 
 static Node *
 function(Token **rest, Token *tok)
@@ -503,6 +517,14 @@ primary(Token **rest, Token *tok)
 		return node;
 	}
 
+	if (tok->type == TOK_STRING) {
+		Node *node = new_node(NOD_STRING);
+		node->loc = tok->loc;
+		node->len = tok->len;
+		*rest = tok->next;
+		return node;
+	}
+
 	if (tok->type == TOK_WORD && equal(tok->next, "(")) {
 		/* function call */
 		Node *node = new_word(tok);
@@ -676,29 +698,30 @@ static void gen_globaldec(Node *node);
 static void gen_fncall(Node *node);
 static void gen_variableget(Node *node);
 static void gen_deref(Node *node);
+static void gen_string(Node *node);
 
 static void
 gen_expr(Node *node)
 {
-	if (node->type == NOD_NUM) {
+	switch (node->type) {
+	case NOD_NUM:
 		printf("\tLIT %04x\n", node->val);
 		depth += 1;
 		return;
-	}
-
-	if (node->type == NOD_FNCALL) {
+	case NOD_FNCALL:
 		gen_fncall(node);
 		return;
-	}
-
-	if (node->type == NOD_WORD) {
+	case NOD_WORD:
 		gen_variableget(node);
 		return;
-	}
-
-	if (node->type == NOD_DEREF) {
+	case NOD_DEREF:
 		gen_deref(node);
 		return;
+	case NOD_STRING:
+		gen_string(node);
+		return;
+	default:
+		break;
 	}
 
 	gen_expr(node->lhs);
@@ -776,6 +799,18 @@ gen_deref(Node *node)
 	printf("\tLDA\n");
 }
 
+static void
+gen_string(Node *node)
+{
+	Node *store = new_node(0);
+	store->lhs = node;
+	store->next = strings;
+	strings = store;
+	strings_size += 1;
+	printf("\tLIT ,__str_%x\n", strings_size - 1);
+	depth += 1;
+}
+
 static void
 gen_assign_stmt(Node *node)
 {
@@ -925,6 +960,9 @@ gen_function(Node *node)
 static void
 codegen(Node *node)
 {
+	printf("\tJRT ,__fn_main\n");
+	printf("\tRET\n");
+
 	while (node != NULL) {
 		if (node->type == NOD_GLOBAL)
 			gen_globaldec(node);
@@ -932,6 +970,18 @@ codegen(Node *node)
 			gen_function(node);
 		node = node->next;
 	}
+
+	for (int i = 0; i < strings_size; i++) {
+		printf("@__str_%x\n", strings_size - i - 1);
+		for (int k = 0; k < strings->lhs->len; k++)
+			printf("\t%04x\n", (unsigned)strings->lhs->loc[k]);
+		printf("\t0000\n");
+		strings = strings->next;
+	}
+
+	printf("@__stack_ptr\n");
+	printf("\t,__stack\n");
+	printf("@__stack\n");
 }
 
 // program = stmt*
@@ -982,12 +1032,7 @@ main(int argc, char **argv)
 
 	Node *node = parse(tok);
 
-	printf("\tJRT ,__fn_main\n");
-	printf("\tRET\n");
 	codegen(node);
-	printf("@__stack_ptr\n");
-	printf("\t,__stack\n");
-	printf("@__stack\n");
 
 	assert(depth == 0);
 	return 0;
diff --git a/testing.sh b/testing.sh
index e1889da..a7bd81b 100755
--- a/testing.sh
+++ b/testing.sh
@@ -35,4 +35,5 @@ test "$1" "global a[5] = 4; main() { wrt([a + 4] + '0'); wrt '\n'; }"
 test "$1" "global a[5] = 4; main() { wrt([a + 5] + '0'); wrt '\n'; }"
 test "$1" "global a[3]; main() { [a] = 'U'; [a+1] = 'w'; [a+2] = '\n'; wrt[a]; wrt[a+1]; wrt[a]; wrt[a+2]; }"
 test "$1" "main() { wrt('0' + div(6, 3)); wrt('\n'); } div(a, b) { return a / b; }"
+test "$1" "main() { \"uwu\"; \"owo\"; }"
 rm -f build/tmp.*
-- 
cgit v1.2.3