Merge pull request #445 from Daniel-Cortez/__addrof

Operator '__addressof'

source/compiler/sc.h

@@ -237,6 +237,8 @@ typedef struct s_symbol {
 #define uMISSING    0x080
 #define uFORWARD    0x100
 #define uNODESTRUCT 0x200 /* "no destruct(or)", not "node struct" */
+/* symbol is referenced "globally", e.g. via "__emit" or "#emit" used outside functions */
+#define uGLOBALREF  0x400
 /* uRETNONE is not stored in the "usage" field of a symbol. It is
  * used during parsing a function, to detect a mix of "return;" and
  * "return value;" in a few special cases.
@@ -354,6 +356,7 @@ enum {
   tMIDDLE = tDBLDOT, /* value of last multi-character operator */
 
   /* reserved words (statements) */
+  t__ADDRESSOF,
   tASSERT,
   tBEGIN,
   tBREAK,
@@ -520,15 +523,28 @@ enum {  /* identifier types */
   estAUTOMATON,
   estSTATE
 };
-enum {  /* symbol type flags */
+enum {  /* search types for error_suggest() when the identifier type is "estSYMBOL" */
-  esfLABEL      = 1 << 0,
+/* symbol type flags */
+  esfLABEL      = 1 << 0, /* label */
   esfCONST      = 1 << 1, /* named constant */
   esfVARIABLE   = 1 << 2, /* single variable */
   esfARRAY      = 1 << 3, /* array */
-  esfFUNCTION   = 1 << 4, /* Pawn or native function */
+  esfPAWNFUNC   = 1 << 4, /* Pawn function */
+  esfNATIVE     = 1 << 5, /* native function */
+
+/* composite search types */
+  /* find symbols of any type (used only to define other search types) */
+  esfANY        = esfLABEL | esfCONST | esfVARIABLE | esfARRAY | esfPAWNFUNC | esfNATIVE,
+
+  /* any function */
+  esfFUNCTION   = esfPAWNFUNC | esfNATIVE,
 
   /* find symbols of any type but labels */
-  esfNONLABEL   = esfCONST | esfVARIABLE | esfARRAY | esfFUNCTION,
+  esfNONLABEL   = esfANY & ~esfLABEL,
+
+  /* find symbols of any type except constants and native functions
+   * (for the "__addressof" operator) */
+  esfADDRESSOF  = esfANY & ~(esfCONST | esfNATIVE),
 
   /* find an array, a single variable, or a named constant */
   esfVARCONST   = esfCONST | esfVARIABLE | esfARRAY

source/compiler/sc1.c

@@ -4276,15 +4276,13 @@ static void doarg(char *name,int ident,int offset,int tags[],int numtags,
   } /* if */
 }
 
-static int count_referrers(symbol *entry)
+static int has_referrers(symbol *entry)
 {
-  int i,count;
+  int i;
-
-  count=0;
   for (i=0; i<entry->numrefers; i++)
     if (entry->refer[i]!=NULL)
-      count++;
+      return TRUE;
-  return count;
+  return ((entry->usage & uGLOBALREF)!=0);
 }
 
 #if !defined SC_LIGHT
@@ -4724,7 +4722,7 @@ static void reduce_referrers(symbol *root)
       if (sym->ident==iFUNCTN
           && (sym->usage & uNATIVE)==0
           && (sym->usage & uPUBLIC)==0 && strcmp(sym->name,uMAINFUNC)!=0 && strcmp(sym->name,uENTRYFUNC)!=0
-          && count_referrers(sym)==0)
+          && !has_referrers(sym))
       {
         sym->usage&=~(uREAD | uWRITTEN);  /* erase usage bits if there is no referrer */
         /* find all symbols that are referred by this symbol */
@@ -4743,7 +4741,7 @@ static void reduce_referrers(symbol *root)
       } else if ((sym->ident==iVARIABLE || sym->ident==iARRAY)
                  && (sym->usage & uPUBLIC)==0
                  && sym->parent==NULL
-                 && count_referrers(sym)==0)
+                 && !has_referrers(sym))
       {
         sym->usage&=~(uREAD | uWRITTEN);  /* erase usage bits if there is no referrer */
       } /* if */

source/compiler/sc2.c

@@ -2170,9 +2170,9 @@ char *sc_tokens[] = {
          "*=", "/=", "%=", "+=", "-=", "<<=", ">>>=", ">>=", "&=", "^=", "|=",
          "||", "&&", "==", "!=", "<=", ">=", "<<", ">>>", ">>", "++", "--",
          "...", "..",
-         "assert", "*begin", "break", "case", "char", "const", "continue", "default",
+         "__addressof", "assert", "*begin", "break", "case", "char", "const", "continue",
-         "defined", "do", "else", "__emit", "*end", "enum", "exit", "for", "forward",
+         "default", "defined", "do", "else", "__emit", "*end", "enum", "exit", "for",
-         "goto", "if", "native", "new", "operator", "public", "return", "sizeof",
+         "forward", "goto", "if", "native", "new", "operator", "public", "return", "sizeof",
          "sleep", "state", "static", "stock", "switch", "tagof", "*then", "while",
          "#assert", "#define", "#else", "#elseif", "#emit", "#endif", "#endinput",
          "#endscript", "#error", "#file", "#if", "#include", "#line", "#pragma",
@@ -3163,12 +3163,10 @@ SC_FUNC void markusage(symbol *sym,int usage)
   if ((usage & (uREAD | uWRITTEN))!=0) {
     /* only do this for global symbols */
     if (sym->vclass==sGLOBAL) {
-      /* "curfunc" should always be valid, since statements may not occurs
-       * outside functions; in the case of syntax errors, however, the
-       * compiler may arrive through this function
-       */
       if (curfunc!=NULL)
         refer_symbol(sym,curfunc);
+      else
+        sym->usage |= uGLOBALREF;
     } /* if */
   } /* if */
 }

source/compiler/sc3.c

@@ -1315,6 +1315,103 @@ static int hier2(value *lval)
     lvalue=hier2(lval);
     lval->tag=tag;
     return lvalue;
+  case t__ADDRESSOF: {
+    extern char *sc_tokens[];
+    static const char allowed_sym_types[]="-variable, array, array cell, label or function-";
+    paranthese=0;
+    while (matchtoken('('))
+      paranthese++;
+    tok=lex(&val,&st);
+    if (tok!=tSYMBOL)
+      return error_suggest(20,st,NULL,estNONSYMBOL,tok);    /* invalid symbol name */
+    sym=findloc(st);
+    if (sym==NULL)
+      sym=findglb(st,sSTATEVAR);
+    if (sym==NULL) {
+      return error_suggest(17,st,NULL,estSYMBOL,esfADDRESSOF);  /* undefined symbol */
+    } else if ((sym->usage & uDEFINE)==0) {
+      /* the symbol is defined after the point of its use,
+         so the compiler can't know its address yet
+       */
+      return error(17,st);      /* undefined symbol (don't suggest other symbols) */
+    } /* if */
+    /* Mark the symbol as read, so the compiler won't throw it away
+     * if it's only indirectly used from "__addressof" */
+    markusage(sym,uREAD);
+    clear_value(lval);
+    lval->ident=iCONSTEXPR;
+    switch (sym->ident) {
+    case iVARIABLE:
+    case iREFERENCE:
+    case iARRAY:
+    case iREFARRAY:
+      if (sym->vclass==sLOCAL) {
+        lval->ident=iEXPRESSION;
+        address(sym,sPRI);
+        break;
+      } /* if */
+      /* fallthrough */
+    case iFUNCTN:
+    case iLABEL:
+      lval->constval=sym->addr;
+      if (sym->ident==iFUNCTN) {
+        if ((sym->usage & uNATIVE)!=0)
+          error(1,allowed_sym_types,sc_tokens[teNATIVE-tFIRST]);
+        break;
+      } else if (sym->ident==iARRAY || sym->ident==iREFARRAY) {
+        cell arrayidx=0,numoffsets=0,offsmul=1;
+        int level;
+        symbol *subsym=sym;
+        for (level=0; matchtoken('['); level++) {
+          if (subsym!=NULL) {
+            if (level==subsym->dim.array.level && matchtoken(tSYMBOL)) {
+              char *idxname;
+              int cmptag=subsym->x.tags.index;
+              tokeninfo(&val,&idxname);
+              if (findconst(idxname,&cmptag)==NULL)
+                error_suggest(80,idxname,NULL,estSYMBOL,esfCONST);  /* unknown symbol, or non-constant */
+              else if (cmptag>1)
+                error(91,idxname);                                  /* ambiguous constant */
+            } else {
+              int index,ident;
+              cell cidx;
+              stgget(&index,&cidx);     /* mark position in code generator */
+              ident=expression(&val,&tag,NULL,TRUE);
+              stgdel(index,cidx);       /* scratch generated code */
+              if (ident!=iCONSTEXPR)
+                error(8);               /* must be constant expression */
+            } /* if */
+            numoffsets+=offsmul;
+            offsmul*=subsym->dim.array.length;
+            arrayidx=(arrayidx*subsym->dim.array.length)+val;
+            subsym=finddepend(subsym);
+          }
+          needtoken(']');
+        } /* for */
+        if (level>sym->dim.array.level+1)
+          error(28,sym->name);    /* invalid subscript */
+        while (subsym!=NULL) {
+          numoffsets+=offsmul;
+          offsmul*=subsym->dim.array.length;
+          arrayidx*=arrayidx*subsym->dim.array.length;
+          subsym=finddepend(subsym);
+        } /* if */
+        lval->constval+=(numoffsets-1+arrayidx)*(cell)sizeof(cell);
+        ldconst(lval->constval,sPRI);
+      } /* if */
+      break;
+    case iCONSTEXPR:
+      error(1,allowed_sym_types,sc_tokens[teNUMERIC-tFIRST]);
+      break;
+    default:
+      assert(0);
+    } /* switch */
+    while (paranthese--)
+      needtoken(')');
+    if (strchr((char*)pline,PREPROC_TERM)!=NULL)
+      return error(93); /* "__addressof" operator is invalid in preprocessor directives */
+    return FALSE;
+  } /* case */
   case tDEFINED:
     paranthese=0;
     while (matchtoken('('))

source/compiler/sc5.c

@@ -131,7 +131,8 @@ static char *errmsg[] = {
 /*089*/  "state variables may not be initialized (symbol \"%s\")\n",
 /*090*/  "public functions may not return arrays (symbol \"%s\")\n",
 /*091*/  "ambiguous constant; tag override is required (symbol \"%s\")\n",
-/*092*/  "functions may not return arrays of unknown size (symbol \"%s\")\n"
+/*092*/  "functions may not return arrays of unknown size (symbol \"%s\")\n",
+/*093*/  "\"__addressof\" operator is invalid in preprocessor expressions\n"
 };
 
 static char *fatalmsg[] = {
@@ -534,7 +535,7 @@ static int find_closest_symbol_table(const char *name,const symbol *root,int sym
       break;
     case iFUNCTN:
     case iREFFUNC:
-      if ((symboltype & esfFUNCTION)==0)
+      if ((symboltype & (((sym->usage & uNATIVE)!=0) ? esfNATIVE : esfFUNCTION))==0)
         continue;
       break;
     default:

source/compiler/tests/__addressof.meta

@@ -0,0 +1,9 @@
+{
+  'test_type': 'output_check',
+  'errors': """
+__addressof.pwn(3) : error 017: undefined symbol "Func1"
+__addressof.pwn(16) : error 093: "__addressof" operator is invalid in preprocessor expressions
+__addressof.pwn(21) : error 001: expected token: "-variable, array, array cell, label or function-", but found "-numeric value-"
+__addressof.pwn(24) : error 001: expected token: "-variable, array, array cell, label or function-", but found "-native function-"
+  """
+}

source/compiler/tests/__addressof.pwn

@@ -0,0 +1,52 @@
+#include <console>
+
+const func1_addr = __addressof(Func1); // error
+
+Func1() {}
+Func2() {}
+Func3() {}
+
+new const functions[3] =
+{
+	__addressof(Func1),
+	__addressof(Func2),
+	__addressof(Func3)
+};
+
+#if __addressof Func1 > 0 // error
+#endif
+
+main()
+{
+	const a = __addressof func1_addr; // error
+	#pragma unused a
+
+	const b = __addressof printf; // error
+	#pragma unused b
+
+lbl:
+	const c = __addressof lbl;
+	#pragma unused c
+
+	const d = __addressof functions;
+	#pragma unused d
+
+	const e = __addressof functions[1];
+	#pragma unused e
+
+	static x = 0;
+	const f = __addressof x;
+	#pragma unused f
+
+	static arr1[3][2] = { { 0, ... }, ... };
+	const g = __addressof arr1[2][1];
+	#pragma unused g
+
+	new y = 0;
+	new const h = __addressof y;
+	#pragma unused h
+
+	new arr2[3][2] = { { 0, ... }, ... };
+	new const i = __addressof arr2;
+	#pragma unused i
+}