From 4cedda2dbeca6bd81d2bd00032f7cff46e0a985e Mon Sep 17 00:00:00 2001
From: Rui Ueyama <ruiu@cs.stanford.edu>
Date: Wed, 7 Oct 2020 12:49:08 +0900
Subject: [PATCH] Add string literal

---
 chibicc.h  |  6 ++++++
 codegen.c  |  8 +++++++-
 parse.c    | 25 ++++++++++++++++++++++++-
 test.sh    |  9 +++++++++
 tokenize.c | 19 +++++++++++++++++++
 5 files changed, 65 insertions(+), 2 deletions(-)

diff --git a/chibicc.h b/chibicc.h
index 7448754..9ccc49f 100644
--- a/chibicc.h
+++ b/chibicc.h
@@ -19,6 +19,7 @@ typedef enum {
   TK_IDENT,   // Identifiers
   TK_PUNCT,   // Punctuators
   TK_KEYWORD, // Keywords
+  TK_STR,     // String literals
   TK_NUM,     // Numeric literals
   TK_EOF,     // End-of-file markers
 } TokenKind;
@@ -31,6 +32,8 @@ struct Token {
   int val;        // If kind is TK_NUM, its value
   char *loc;      // Token location
   int len;        // Token length
+  Type *ty;       // Used if TK_STR
+  char *str;      // String literal contents including terminating '\0'
 };
 
 void error(char *fmt, ...);
@@ -59,6 +62,9 @@ struct Obj {
   // Global variable or function
   bool is_function;
 
+  // Global variable
+  char *init_data;
+
   // Function
   Obj *params;
   Node *body;
diff --git a/codegen.c b/codegen.c
index b6eab83..7450d53 100644
--- a/codegen.c
+++ b/codegen.c
@@ -233,7 +233,13 @@ static void emit_data(Obj *prog) {
     printf("  .data\n");
     printf("  .globl %s\n", var->name);
     printf("%s:\n", var->name);
-    printf("  .zero %d\n", var->ty->size);
+
+    if (var->init_data) {
+      for (int i = 0; i < var->ty->size; i++)
+        printf("  .byte %d\n", var->init_data[i]);
+    } else {
+      printf("  .zero %d\n", var->ty->size);
+    }
   }
 }
 
diff --git a/parse.c b/parse.c
index 33eeab6..c6c1f37 100644
--- a/parse.c
+++ b/parse.c
@@ -106,6 +106,23 @@ static Obj *new_gvar(char *name, Type *ty) {
   return var;
 }
 
+static char *new_unique_name(void) {
+  static int id = 0;
+  char *buf = calloc(1, 20);
+  sprintf(buf, ".L..%d", id++);
+  return buf;
+}
+
+static Obj *new_anon_gvar(Type *ty) {
+  return new_gvar(new_unique_name(), ty);
+}
+
+static Obj *new_string_literal(char *p, Type *ty) {
+  Obj *var = new_anon_gvar(ty);
+  var->init_data = p;
+  return var;
+}
+
 static char *get_ident(Token *tok) {
   if (tok->kind != TK_IDENT)
     error_tok(tok, "expected an identifier");
@@ -530,7 +547,7 @@ static Node *funcall(Token **rest, Token *tok) {
   return node;
 }
 
-// primary = "(" expr ")" | "sizeof" unary | ident func-args? | num
+// primary = "(" expr ")" | "sizeof" unary | ident func-args? | str | num
 static Node *primary(Token **rest, Token *tok) {
   if (equal(tok, "(")) {
     Node *node = expr(&tok, tok->next);
@@ -557,6 +574,12 @@ static Node *primary(Token **rest, Token *tok) {
     return new_var_node(var, tok);
   }
 
+  if (tok->kind == TK_STR) {
+    Obj *var = new_string_literal(tok->str, tok->ty);
+    *rest = tok->next;
+    return new_var_node(var, tok);
+  }
+
   if (tok->kind == TK_NUM) {
     Node *node = new_num(tok->val, tok);
     *rest = tok->next;
diff --git a/test.sh b/test.sh
index 51d235d..fd75c46 100755
--- a/test.sh
+++ b/test.sh
@@ -171,4 +171,13 @@ assert 1 'int main() { char x; return sizeof(x); }'
 assert 10 'int main() { char x[10]; return sizeof(x); }'
 assert 1 'int main() { return sub_char(7, 3, 3); } int sub_char(char a, char b, char c) { return a-b-c; }'
 
+assert 0 'int main() { return ""[0]; }'
+assert 1 'int main() { return sizeof(""); }'
+
+assert 97 'int main() { return "abc"[0]; }'
+assert 98 'int main() { return "abc"[1]; }'
+assert 99 'int main() { return "abc"[2]; }'
+assert 0 'int main() { return "abc"[3]; }'
+assert 4 'int main() { return sizeof("abc"); }'
+
 echo OK
diff --git a/tokenize.c b/tokenize.c
index 7aa19ba..89a362f 100644
--- a/tokenize.c
+++ b/tokenize.c
@@ -99,6 +99,18 @@ static bool is_keyword(Token *tok) {
   return false;
 }
 
+static Token *read_string_literal(char *start) {
+  char *p = start + 1;
+  for (; *p != '"'; p++)
+    if (*p == '\n' || *p == '\0')
+      error_at(start, "unclosed string literal");
+
+  Token *tok = new_token(TK_STR, start, p + 1);
+  tok->ty = array_of(ty_char, p - start);
+  tok->str = strndup(start + 1, p - start - 1);
+  return tok;
+}
+
 static void convert_keywords(Token *tok) {
   for (Token *t = tok; t->kind != TK_EOF; t = t->next)
     if (is_keyword(t))
@@ -127,6 +139,13 @@ Token *tokenize(char *p) {
       continue;
     }
 
+    // String literal
+    if (*p == '"') {
+      cur = cur->next = read_string_literal(p);
+      p += cur->len;
+      continue;
+    }
+
     // Identifier or keyword
     if (is_ident1(*p)) {
       char *start = p;