convert BASIC LEN() into a keyword operator, so we can compile time optimize when the argument is constant

Author: totalspectrum

ast.c

@@ -1013,6 +1013,9 @@ static void doASTDump(AST *ast, int indent)
             case K_ASC:
                 opString = "asc";
                 break;
+            case K_LEN:
+                opString = "strlen";
+                break;
             case K_POWER:
                 opString = "power";
                 break;

doc/basic.md

@@ -255,6 +255,7 @@ if
 import
 input
 integer
+len
 let
 lib
 line
@@ -367,7 +368,6 @@ instrrev
 lcase$
 kill
 left$
-len
 _lockclr
 _locknew
 _lockrel

frontends/basic/basic.y

@@ -461,6 +461,7 @@ AdjustParamForByVal(AST *param)
 %token BAS_INPUT      "input"
 %token BAS_CAST_INT   "int"
 %token BAS_INTEGER_KW "integer"
+%token BAS_LEN        "len"
 %token BAS_LET        "let"
 %token BAS_LIB        "lib"
 %token BAS_LINE       "line"
@@ -1640,6 +1641,8 @@ unary_op:
     { $$ = AstOperator(K_ABS, NULL, NULL); }
   | BAS_ASC
     { $$ = AstOperator(K_ASC, NULL, NULL); }
+  | BAS_LEN
+    { $$ = AstOperator(K_LEN, NULL, NULL); }
   | BAS_SQRT
     { $$ = AstOperator(K_SQRT, NULL, NULL); }
 ;

frontends/lexer.c

@@ -2318,6 +2318,7 @@ struct reservedword basic_keywords[] = {
     { "input", BAS_INPUT },
     { "int", BAS_CAST_INT },
     { "integer", BAS_INTEGER_KW },
+    { "len", BAS_LEN },
     { "let", BAS_LET },
     { "lib", BAS_LIB },
     { "line", BAS_LINE },

frontends/types.c

@@ -107,6 +107,7 @@ static AST *struct_copy;
 static AST *struct_memset;
 static AST *string_cmp;
 static AST *string_concat;
+static AST *string_length;
 
 static AST *string_tointeger;
 static AST *string_frominteger;
@@ -1127,6 +1128,26 @@ AST *CoerceOperatorTypes(AST *ast, AST *lefttype, AST *righttype)
             }
         }
         return ast_type_long;
+    case K_LEN:
+        if (!CompatibleTypes(righttype, ast_type_string)) {
+            if (curfunc && IsBasicLang(curfunc->language)) {
+                ERROR(ast, "expected string argument to LEN");
+            } else {
+                ERROR(ast, "expected string argument to __builtin_strlen");
+            }
+        } else {
+            AST *newast;
+            AST *sexpr = ast->right;
+            
+            if (sexpr && sexpr->kind == AST_STRINGPTR) {
+                /* the -1 is to remove the trailing 0 */
+                newast = AstInteger(AstStringLen(sexpr) - 1);
+            } else {
+                newast = MakeOperatorCall(string_length, sexpr, NULL, NULL);
+            }
+            *ast = *newast;
+        }
+        return ast_type_long;
     case K_BOOL_NOT:
     case K_BOOL_AND:
     case K_BOOL_OR:
@@ -2169,6 +2190,7 @@ InitGlobalFuncs(void)
 
         string_cmp = getBasicPrimitive("_string_cmp");
         string_concat = getBasicPrimitive("_string_concat");
+        string_length = getBasicPrimitive("__builtin_strlen");
         string_tointeger = getBasicPrimitive("__builtin_atoi");
         string_frominteger = getBasicPrimitive("strint$");
         gc_alloc_managed = getBasicPrimitive("_gc_alloc_managed");

optokens.h

@@ -41,7 +41,10 @@ enum OpToken {
     K_LEU,
     K_GTU,
     K_GEU,
-    K_ASC,
+    
+    K_ASC,  /* BASIC ASC operator */
+    K_LEN,  /* BASIC LEN operator */
+    
     K_LIMITMIN_UNS,
     K_LIMITMAX_UNS,
     K_POWER,