Add U, L and LL suffixes

Deiolly · Dec 7, 2020 · aaf1045 · aaf1045
1 parent 34ab83b
commit aaf1045
Show file tree

Hide file tree

Showing 5 changed files with 122 additions and 4 deletions.
diff --git a/chibicc.h b/chibicc.h
@@ -46,7 +46,7 @@ struct Token {
   int64_t val;    // If kind is TK_NUM, its value
   char *loc;      // Token location
   int len;        // Token length
-  Type *ty;       // Used if TK_STR
+  Type *ty;       // Used if TK_NUM or TK_STR
   char *str;      // String literal contents including terminating '\0'
 
   int line_no;    // Line number

diff --git a/parse.c b/parse.c
@@ -2189,6 +2189,7 @@ static Node *primary(Token **rest, Token *tok) {
 
   if (tok->kind == TK_NUM) {
     Node *node = new_num(tok->val, tok);
+    node->ty = tok->ty;
     *rest = tok->next;
     return node;
   }

diff --git a/test/literal.c b/test/literal.c
@@ -17,6 +17,67 @@ int main() {
   ASSERT(47, 0b101111);
   ASSERT(47, 0B101111);
 
+  ASSERT(4, sizeof(0));
+  ASSERT(8, sizeof(0L));
+  ASSERT(8, sizeof(0LU));
+  ASSERT(8, sizeof(0UL));
+  ASSERT(8, sizeof(0LL));
+  ASSERT(8, sizeof(0LLU));
+  ASSERT(8, sizeof(0Ull));
+  ASSERT(8, sizeof(0l));
+  ASSERT(8, sizeof(0ll));
+  ASSERT(8, sizeof(0x0L));
+  ASSERT(8, sizeof(0b0L));
+  ASSERT(4, sizeof(2147483647));
+  ASSERT(8, sizeof(2147483648));
+  ASSERT(-1, 0xffffffffffffffff);
+  ASSERT(8, sizeof(0xffffffffffffffff));
+  ASSERT(4, sizeof(4294967295U));
+  ASSERT(8, sizeof(4294967296U));
+
+  ASSERT(3, -1U>>30);
+  ASSERT(3, -1Ul>>62);
+  ASSERT(3, -1ull>>62);
+
+  ASSERT(1, 0xffffffffffffffffl>>63);
+  ASSERT(1, 0xffffffffffffffffll>>63);
+
+  ASSERT(-1, 18446744073709551615);
+  ASSERT(8, sizeof(18446744073709551615));
+  ASSERT(-1, 18446744073709551615>>63);
+
+  ASSERT(-1, 0xffffffffffffffff);
+  ASSERT(8, sizeof(0xffffffffffffffff));
+  ASSERT(1, 0xffffffffffffffff>>63);
+
+  ASSERT(-1, 01777777777777777777777);
+  ASSERT(8, sizeof(01777777777777777777777));
+  ASSERT(1, 01777777777777777777777>>63);
+
+  ASSERT(-1, 0b1111111111111111111111111111111111111111111111111111111111111111);
+  ASSERT(8, sizeof(0b1111111111111111111111111111111111111111111111111111111111111111));
+  ASSERT(1, 0b1111111111111111111111111111111111111111111111111111111111111111>>63);
+
+  ASSERT(8, sizeof(2147483648));
+  ASSERT(4, sizeof(2147483647));
+
+  ASSERT(8, sizeof(0x1ffffffff));
+  ASSERT(4, sizeof(0xffffffff));
+  ASSERT(1, 0xffffffff>>31);
+
+  ASSERT(8, sizeof(040000000000));
+  ASSERT(4, sizeof(037777777777));
+  ASSERT(1, 037777777777>>31);
+
+  ASSERT(8, sizeof(0b111111111111111111111111111111111));
+  ASSERT(4, sizeof(0b11111111111111111111111111111111));
+  ASSERT(1, 0b11111111111111111111111111111111>>31);
+
+  ASSERT(-1, 1 << 31 >> 31);
+  ASSERT(-1, 01 << 31 >> 31);
+  ASSERT(-1, 0x1 << 31 >> 31);
+  ASSERT(-1, 0b1 << 31 >> 31);
+
   printf("OK\n");
   return 0;
 }
diff --git a/tokenize.c b/tokenize.c
@@ -243,29 +243,85 @@ static Token *read_char_literal(char *start) {
 
   Token *tok = new_token(TK_NUM, start, end + 1);
   tok->val = c;
+  tok->ty = ty_int;
   return tok;
 }
 
 static Token *read_int_literal(char *start) {
   char *p = start;
 
+  // Read a binary, octal, decimal or hexadecimal number.
   int base = 10;
-  if (!strncasecmp(p, "0x", 2) && isalnum(p[2])) {
+  if (!strncasecmp(p, "0x", 2) && isxdigit(p[2])) {
     p += 2;
     base = 16;
-  } else if (!strncasecmp(p, "0b", 2) && isalnum(p[2])) {
+  } else if (!strncasecmp(p, "0b", 2) && (p[2] == '0' || p[2] == '1')) {
     p += 2;
     base = 2;
   } else if (*p == '0') {
     base = 8;
   }
 
   int64_t val = strtoul(p, &p, base);
+
+  // Read U, L or LL suffixes.
+  bool l = false;
+  bool u = false;
+
+  if (startswith(p, "LLU") || startswith(p, "LLu") ||
+      startswith(p, "llU") || startswith(p, "llu") ||
+      startswith(p, "ULL") || startswith(p, "Ull") ||
+      startswith(p, "uLL") || startswith(p, "ull")) {
+    p += 3;
+    l = u = true;
+  } else if (!strncasecmp(p, "lu", 2) || !strncasecmp(p, "ul", 2)) {
+    p += 2;
+    l = u = true;
+  } else if (startswith(p, "LL") || startswith(p, "ll")) {
+    p += 2;
+    l = true;
+  } else if (*p == 'L' || *p == 'l') {
+    p++;
+    l = true;
+  } else if (*p == 'U' || *p == 'u') {
+    p++;
+    u = true;
+  }
+
   if (isalnum(*p))
     error_at(p, "invalid digit");
 
+  // Infer a type.
+  Type *ty;
+  if (base == 10) {
+    if (l && u)
+      ty = ty_ulong;
+    else if (l)
+      ty = ty_long;
+    else if (u)
+      ty = (val >> 32) ? ty_ulong : ty_uint;
+    else
+      ty = (val >> 31) ? ty_long : ty_int;
+  } else {
+    if (l && u)
+      ty = ty_ulong;
+    else if (l)
+      ty = (val >> 63) ? ty_ulong : ty_long;
+    else if (u)
+      ty = (val >> 32) ? ty_ulong : ty_uint;
+    else if (val >> 63)
+      ty = ty_ulong;
+    else if (val >> 32)
+      ty = ty_long;
+    else if (val >> 31)
+      ty = ty_uint;
+    else
+      ty = ty_int;
+  }
+
   Token *tok = new_token(TK_NUM, start, p);
   tok->val = val;
+  tok->ty = ty;
   return tok;
 }
 

diff --git a/type.c b/type.c
@@ -110,7 +110,7 @@ void add_type(Node *node) {
 
   switch (node->kind) {
   case ND_NUM:
-    node->ty = (node->val == (int)node->val) ? ty_int : ty_long;
+    node->ty = ty_int;
     return;
   case ND_ADD:
   case ND_SUB: