feat: Implement sizeof operator #11 (#25)

examples/inp5.c

@@ -1,25 +1,26 @@
 void printf(char *, ...);
 void *malloc(int sz);
+
 int main(int argc, char **argv)
 {
     for (int i = 0; i < argc; i += 1)
     {
         printf("Arg %u: %s\n", i, argv[i]);
     }
 
-    int *a = (int *)malloc(10 * 8);
+    int *a = (int *)malloc(10 * sizeof(int));
     for (int i = 0; i < 10; i = i + 1)
         a[i] = 2 * i;
     for (int i = 0; i < 10; i += 1)
         printf("%u\n", a[i]);
 
-    int **b = (int **)malloc(1024 * 8);
+    int **b = (int **)malloc(1024 * sizeof(int));
     int c = 123;
     b[0] = (&c);
     *(b[0]) = *(b[0]) * 2;
     printf("%u\n", c);
 
-    char *str = (char *)malloc(20);
+    char *str = (char *)malloc(20 * sizeof(char));
     str[0] = 'h';
     str[1] = 'i';
     str[2] = '!';

examples/inp6.c

@@ -9,7 +9,7 @@ struct MyStruct
 
 int main()
 {
-    struct MyStruct *m = (struct MyStruct *)malloc(16);
+    struct MyStruct *m = (struct MyStruct *)malloc(sizeof(struct MyStruct));
     m->a=10;
     m->b=20;
     printf("%u %u\n", m->a, m->b);

examples/inp_linked_list.c

@@ -14,13 +14,13 @@ struct linked_list
 
 struct linked_list *new_list()
 {
-    struct linked_list *ret = (struct linked_list *)malloc(8);
+    struct linked_list *ret = (struct linked_list *)malloc(sizeof(struct linked_list));
     return ret;
 }
 
 void add_val(struct linked_list *lst, int val)
 {
-    struct node *new = (struct node *)malloc(16);
+    struct node *new = (struct node *)malloc(sizeof(struct node));
     *((int *)(&new->next)) = 0;
     new->val = val;
     if (lst->first)

lexer.c

@@ -96,6 +96,8 @@ typed_token *next_keyword_or_identifier(char **inp_ptr)
             return new_simp_tkn(TKN_WHILE);
         else if (strcmp(val, "struct") == 0)
             return new_simp_tkn(TKN_STRUCT);
+        else if (strcmp(val, "sizeof") == 0)
+            return new_simp_tkn(TKN_SIZEOF);
         else
             return new_tkn(TKN_ID, val, ident_tkn_debug);
     }

lexer.h

@@ -22,6 +22,7 @@ typed_token *tokenize(char *inp);
 #define TKN_WHILE 8
 #define TKN_STRUCT 9
 #define TKN_ELSE 10
+#define TKN_SIZEOF 11
 
 // Single letter symbols
 #define TKN_L_PAREN 16

parser/expr/expr.c

@@ -320,6 +320,22 @@ apply_result *cast_apply(parser_node *node, context *ctx)
     return new_result(cast->val->apply(cast->val, ctx)->code, cast_type);
 }
 
+void sizeof_debug(int depth, parser_node *node)
+{
+    node_sizeof *sz = (node_sizeof *)node->data;
+    printtabs(depth);
+    printf("SizeOf:\n");
+    sz->type->debug(depth + 1, sz->type);
+}
+
+apply_result *sizeof_apply(parser_node *node, context *ctx)
+{
+    node_sizeof *size_of = (node_sizeof *)node->data;
+    general_type *type = ((node_type *)size_of->type->data)->type;
+    int sz = type->size(type, ctx);
+    return new_result(cc_asprintf("%u",sz), new_primitive_type("TKN_INT"));
+}
+
 parser_node *parse_paren(typed_token **tkns_ptr)
 {
     typed_token *tkn = *tkns_ptr;
@@ -346,6 +362,36 @@ parser_node *parse_paren(typed_token **tkns_ptr)
     return NULL;
 }
 
+parser_node *parse_sizeof(typed_token **tkns_ptr)
+{
+    typed_token *tkn = *tkns_ptr;
+    if (tkn->type_id == TKN_SIZEOF)
+    {
+        tkn = tkn->next;
+        if (tkn->type_id == TKN_L_PAREN)
+        {
+            tkn = tkn->next;
+            parser_node *type = parse_type(&tkn);
+            if (type)
+            {
+                if (tkn->type_id == TKN_R_PAREN)
+                {
+                    tkn = tkn->next;
+                    *tkns_ptr = tkn;
+                    parser_node *node = (parser_node *)malloc(sizeof(parser_node));
+                    node->data = (void *)malloc(sizeof(node_sizeof));
+                    node->debug = sizeof_debug;
+                    node->apply = sizeof_apply;
+                    node_sizeof *cast = (node_sizeof *)node->data;
+                    cast->type = type;
+                    return node;
+                }
+            }
+        }
+    }
+    return NULL;
+}
+
 parser_node *parse_cast(typed_token **tkns_ptr)
 {
     typed_token *tkn = *tkns_ptr;
@@ -383,6 +429,8 @@ parser_node *parse_terminal(typed_token **tkns_ptr)
 
     parser_node *curr = NULL;
 
+    if (!curr)
+        curr = parse_sizeof(&tkn);
     if (!curr)
         curr = parse_cast(&tkn);
     if (!curr)

parser/expr/expr.h

@@ -14,6 +14,11 @@ typedef struct
     parser_node *right;
 } node_binary_op;
 
+typedef struct
+{
+    parser_node *type;
+} node_sizeof;
+
 typedef struct
 {
     int op;

tests/output/inp5.c_asm_output.txt

@@ -5,7 +5,7 @@ __temp_str_2 db `Arg %u: %s\n`, 0
 __temp_str_3 db `%u\n`, 0
 __temp_str_4 db `%u\n`, 0
 __temp_str_5 db `%s`, 0
-__main_size: equ 768
+__main_size: equ 784
 section .text
 extern printf
 extern malloc
@@ -306,114 +306,118 @@ mov [rsp+272], rax
 call printf
 mov [rsp+280], rax
 mov rax, 20
+mov rbx, 1
+mul rbx
 mov [rsp+288], rax
-mov rdi, [rsp+288]
+mov rax, [rsp+288]
+mov [rsp+296], rax
+mov rdi, [rsp+296]
 mov rax, rsp
 add rax, 0
-mov [rsp+296], rax
-call malloc
 mov [rsp+304], rax
-mov rax, [rsp+304]
+call malloc
+mov [rsp+312], rax
+mov rax, [rsp+312]
 mov [rsp+280], rax
 mov rax, rsp
 add rax, 280
-mov [rsp+312], rax
+mov [rsp+320], rax
 mov rax, 0
 mov rbx, 1
 mul rbx
 mov rbx, [rsp+280]
 add rbx, rax
-mov [rsp+320], rbx
+mov [rsp+328], rbx
 mov rax, [rbx]
-mov [rsp+328], rax
-mov rax, [rsp+328]
+mov [rsp+336], rax
+mov rax, [rsp+336]
 mov rbx, 104
 mov rax, 104
-mov rbx, [rsp+320]
+mov rbx, [rsp+328]
 mov [rbx], rax
-mov [rsp+329], rax
+mov [rsp+337], rax
 mov rax, rsp
 add rax, 280
-mov [rsp+330], rax
+mov [rsp+338], rax
 mov rax, 1
 mov rbx, 1
 mul rbx
 mov rbx, [rsp+280]
 add rbx, rax
-mov [rsp+338], rbx
+mov [rsp+346], rbx
 mov rax, [rbx]
-mov [rsp+346], rax
-mov rax, [rsp+346]
+mov [rsp+354], rax
+mov rax, [rsp+354]
 mov rbx, 105
 mov rax, 105
-mov rbx, [rsp+338]
+mov rbx, [rsp+346]
 mov [rbx], rax
-mov [rsp+347], rax
+mov [rsp+355], rax
 mov rax, rsp
 add rax, 280
-mov [rsp+348], rax
+mov [rsp+356], rax
 mov rax, 2
 mov rbx, 1
 mul rbx
 mov rbx, [rsp+280]
 add rbx, rax
-mov [rsp+356], rbx
+mov [rsp+364], rbx
 mov rax, [rbx]
-mov [rsp+364], rax
-mov rax, [rsp+364]
+mov [rsp+372], rax
+mov rax, [rsp+372]
 mov rbx, 33
 mov rax, 33
-mov rbx, [rsp+356]
+mov rbx, [rsp+364]
 mov [rbx], rax
-mov [rsp+365], rax
+mov [rsp+373], rax
 mov rax, rsp
 add rax, 280
-mov [rsp+366], rax
+mov [rsp+374], rax
 mov rax, 3
 mov rbx, 1
 mul rbx
 mov rbx, [rsp+280]
 add rbx, rax
-mov [rsp+374], rbx
+mov [rsp+382], rbx
 mov rax, [rbx]
-mov [rsp+382], rax
-mov rax, [rsp+382]
+mov [rsp+390], rax
+mov rax, [rsp+390]
 mov rbx, 10
 mov rax, 10
-mov rbx, [rsp+374]
+mov rbx, [rsp+382]
 mov [rbx], rax
-mov [rsp+383], rax
+mov [rsp+391], rax
 mov rax, rsp
 add rax, 280
-mov [rsp+384], rax
+mov [rsp+392], rax
 mov rax, 4
 mov rbx, 1
 mul rbx
 mov rbx, [rsp+280]
 add rbx, rax
-mov [rsp+392], rbx
+mov [rsp+400], rbx
 mov rax, [rbx]
-mov [rsp+400], rax
-mov rax, [rsp+400]
+mov [rsp+408], rax
+mov rax, [rsp+408]
 mov rbx, 0
 mov rax, 0
-mov rbx, [rsp+392]
+mov rbx, [rsp+400]
 mov [rbx], rax
-mov [rsp+401], rax
+mov [rsp+409], rax
 mov rax, __temp_str_5
-mov [rsp+402], rax
+mov [rsp+410], rax
 mov rax, rsp
 add rax, 280
-mov [rsp+410], rax
-mov rax, [rsp+280]
 mov [rsp+418], rax
-mov rdi, [rsp+402]
-mov rsi, [rsp+418]
+mov rax, [rsp+280]
+mov [rsp+426], rax
+mov rdi, [rsp+410]
+mov rsi, [rsp+426]
 mov rax, rsp
 add rax, 0
-mov [rsp+426], rax
-call printf
 mov [rsp+434], rax
+call printf
+mov [rsp+442], rax
 mov rsp, rbp
 pop rbp
 ret

tests/output/inp5.c_tree_output.txt

@@ -76,7 +76,8 @@ Program(
                     Left:
                       Literal(Type: 34, Value: 10)
                     Right:
-                      Literal(Type: 34, Value: 8)
+                      SizeOf:
+                        TKN_INT
             Type:
               Pointer of:
                 TKN_INT
@@ -161,7 +162,8 @@ Program(
                     Left:
                       Literal(Type: 34, Value: 1024)
                     Right:
-                      Literal(Type: 34, Value: 8)
+                      SizeOf:
+                        TKN_INT
             Type:
               Pointer of:
                 Pointer of:
@@ -216,7 +218,12 @@ Program(
                 Function:
                   Variable(malloc)
                 Args:
-                  Literal(Type: 34, Value: 20)
+                  BinaryOp(Op: 78)
+                    Left:
+                      Literal(Type: 34, Value: 20)
+                    Right:
+                      SizeOf:
+                        TKN_CHAR
             Type:
               Pointer of:
                 TKN_CHAR

tests/output/inp6.c_lex_output.txt

@@ -43,7 +43,11 @@ TKN_STAR
 TKN_R_PAREN
 TKN_ID(malloc)
 TKN_L_PAREN
-TKN_LIT_INT(16)
+TKN_SIZEOF
+TKN_L_PAREN
+TKN_STRUCT
+TKN_ID(MyStruct)
+TKN_R_PAREN
 TKN_R_PAREN
 TKN_SEMICOLON
 TKN_ID(m)

tests/output/inp6.c_tree_output.txt

@@ -34,7 +34,8 @@ Program(
                 Function:
                   Variable(malloc)
                 Args:
-                  Literal(Type: 34, Value: 16)
+                  SizeOf:
+                    struct MyStruct
             Type:
               Pointer of:
                 struct MyStruct