Unicode literals

Examples/test-suite/csharp_features.i

@@ -1,4 +1,5 @@
 %module csharp_features
+%include "wchar.i"
 
 // SWIG gets the method modifiers wrong occasionally, like with private inheritance, %csmethodmodifiers can fix this
 %csmethodmodifiers Derived::VirtualMethod() "public virtual"
@@ -19,6 +20,9 @@ public:
 class MoreDerived : public Derived {
 public:
   int variable;
+  // test wide char literals support for C# module
+  void methodWithDefault1(const wchar_t* s = L"literal with escapes \x1234"){}
+  void methodWithDefault2(wchar_t c = L'\x1234'){}
 };
 %}
 

Source/CParse/cscanner.c

@@ -383,6 +383,10 @@ static int yylook(void) {
     case SWIG_TOKEN_STRING:
       yylval.id = Swig_copy_string(Char(Scanner_text(scan)));
       return STRING;
+
+    case SWIG_TOKEN_WSTRING:
+      yylval.id = Swig_copy_string(Char(Scanner_text(scan)));
+      return WSTRING;
       
     case SWIG_TOKEN_CHAR:
       yylval.str = NewString(Scanner_text(scan));
@@ -391,7 +395,15 @@ static int yylook(void) {
 	Printf(stdout,"%d\n", Len(Scanner_text(scan)));
       }
       return CHARCONST;
-      
+
+    case SWIG_TOKEN_WCHAR:
+      yylval.str = NewString(Scanner_text(scan));
+      if (Len(yylval.str) == 0) {
+	Swig_error(cparse_file, cparse_line, "Empty character constant\n");
+	Printf(stdout,"%d\n", Len(Scanner_text(scan)));
+      }
+      return WCHARCONST;
+
       /* Numbers */
       
     case SWIG_TOKEN_INT:

Source/CParse/parser.y

@@ -13,8 +13,8 @@
  * some point.  Beware.
  * ----------------------------------------------------------------------------- */
 
+%expect 6
 %{
-
 #define yylex yylex
 
 char cvsroot_parser_y[] = "$Id$";
@@ -1635,9 +1635,9 @@ static void tag_nodes(Node *n, const_String_or_char_ptr attrname, DOH *value) {
 %token <id> ID
 %token <str> HBLOCK
 %token <id> POUND 
-%token <id> STRING
+%token <id> STRING WSTRING
 %token <loc> INCLUDE IMPORT INSERT
-%token <str> CHARCONST 
+%token <str> CHARCONST WCHARCONST
 %token <dtype> NUM_INT NUM_FLOAT NUM_UNSIGNED NUM_LONG NUM_ULONG NUM_LONGLONG NUM_ULONGLONG NUM_BOOL
 %token <ivalue> TYPEDEF
 %token <type> TYPE_INT TYPE_UNSIGNED TYPE_SHORT TYPE_LONG TYPE_FLOAT TYPE_DOUBLE TYPE_CHAR TYPE_WCHAR TYPE_VOID TYPE_SIGNED TYPE_BOOL TYPE_COMPLEX TYPE_TYPEDEF TYPE_RAW TYPE_NON_ISO_INT8 TYPE_NON_ISO_INT16 TYPE_NON_ISO_INT32 TYPE_NON_ISO_INT64
@@ -1728,7 +1728,7 @@ static void tag_nodes(Node *n, const_String_or_char_ptr attrname, DOH *value) {
 %type <decl>     abstract_declarator direct_abstract_declarator ctor_end;
 %type <tmap>     typemap_type;
 %type <str>      idcolon idcolontail idcolonnt idcolontailnt idtemplate stringbrace stringbracesemi;
-%type <id>       string stringnum ;
+%type <id>       string stringnum wstring;
 %type <tparms>   template_parms;
 %type <dtype>    cpp_end cpp_vend;
 %type <ivalue>   rename_namewarn;
@@ -6001,7 +6001,7 @@ definetype     : { /* scanner_check_typedef(); */ } expr {
                    $$ = $2;
 		   if ($$.type == T_STRING) {
 		     $$.rawval = NewStringf("\"%(escape)s\"",$$.val);
-		   } else if ($$.type != T_CHAR) {
+		   } else if ($$.type != T_CHAR && $$.type != T_WSTRING && $$.type != T_WCHAR) {
 		     $$.rawval = 0;
 		   }
 		   $$.bitfield = 0;
@@ -6127,6 +6127,11 @@ valexpr        : exprnum { $$ = $1; }
 		  $$.type = T_ULONG;
                }
                | exprcompound { $$ = $1; }
+	       | wstring {
+		    $$.val = NewString($1);
+		    $$.rawval = NewStringf("L\"%s\"", $$.val);
+                    $$.type = T_WSTRING;
+	       }
                | CHARCONST {
 		  $$.val = NewString($1);
 		  if (Len($$.val)) {
@@ -6139,6 +6144,18 @@ valexpr        : exprnum { $$ = $1; }
 		  $$.throws = 0;
 		  $$.throwf = 0;
 	       }
+               | WCHARCONST {
+		  $$.val = NewString($1);
+		  if (Len($$.val)) {
+		    $$.rawval = NewStringf("L\'%s\'", $$.val);
+		  } else {
+		    $$.rawval = NewString("L'\\0'");
+		  }
+		  $$.type = T_WCHAR;
+		  $$.bitfield = 0;
+		  $$.throws = 0;
+		  $$.throwf = 0;
+	       }
 
 /* grouping */
                |  LPAREN expr RPAREN %prec CAST {
@@ -6661,6 +6678,21 @@ string         : string STRING {
                }
                | STRING { $$ = $1;}
                ; 
+/* Concatenated wide strings: L"str1" L"str2" */
+wstring         : wstring WSTRING {
+                   $$ = (char *) malloc(strlen($1)+strlen($2)+1);
+                   strcpy($$,$1);
+                   strcat($$,$2);
+               }
+/* Concatenated wide string and normal string literal: L"str1" "str2" */
+/*not all the compilers support this concatenation mode, so perhaps better to postpone it*/
+               /*| wstring STRING { here $2 comes unescaped, we have to escape it back first via NewStringf("%(escape)s)"
+                   $$ = (char *) malloc(strlen($1)+strlen($2)+1);
+                   strcpy($$,$1);
+                   strcat($$,$2);
+	       }*/
+               | WSTRING { $$ = $1;}
+               ;
 
 stringbrace    : string {
 		 $$ = NewString($1);

Source/Swig/cwrap.c

@@ -274,7 +274,7 @@ int Swig_cargs(Wrapper *w, ParmList *p) {
 	  Delete(defname);
 	  Delete(defvalue);
 	}
-      } else if (!pvalue && ((tycode == T_POINTER) || (tycode == T_STRING))) {
+      } else if (!pvalue && ((tycode == T_POINTER) || (tycode == T_STRING) || (tycode == T_WSTRING))) {
 	pvalue = (String *) "0";
       }
       if (!altty) {

Source/Swig/scanner.c

@@ -509,7 +509,6 @@ static int look(Scanner * s) {
 	state = 4;		/* Possibly a SWIG directive */
       
       /* Look for possible identifiers or unicode/delimiter strings */
-
       else if ((isalpha(c)) || (c == '_') ||
 	       (s->idstart && strchr(s->idstart, c))) {
 	state = 7;
@@ -867,11 +866,15 @@ static int look(Scanner * s) {
       break;
     
     case 7:			/* Identifier or true/false or unicode/custom delimiter string */
-      if (c=='R') { /* Possibly CUSTOM DELIMITER string */
+      if (c == 'R') { /* Possibly CUSTOM DELIMITER string */
 	state = 72;
 	break;
       }
-      else if (c!='u' && c!='U' && c!='L') { /* Definitely an identifier */
+      else if (c == 'L') { /* Probably identifier but may be a wide string literal */
+	state = 77;
+	break;
+      }
+      else if (c != 'u' && c != 'U') { /* Definitely an identifier */
 	state = 70;
 	break;
       }
@@ -879,14 +882,14 @@ static int look(Scanner * s) {
       if ((c = nextchar(s)) == 0) {
 	state = 76;
       }
-      else if (c=='\"') { /* Definitely u, U or L string */
+      else if (c == '\"') { /* Definitely u, U or L string */
 	retract(s, 1);
 	state = 1000;
       }
-      else if (c=='R') { /* Possibly CUSTOM DELIMITER u, U, L string */
+      else if (c == 'R') { /* Possibly CUSTOM DELIMITER u, U, L string */
 	state = 73;
       }
-      else if (c=='8') { /* Possibly u8 string */
+      else if (c == '8') { /* Possibly u8 string */
 	state = 71;
       }
       else {
@@ -950,6 +953,59 @@ static int look(Scanner * s) {
       
       break;
 
+      case 77: /*identifier or wide string literal*/
+	if ((c = nextchar(s)) == 0)
+	  return SWIG_TOKEN_ID;
+	else if (c == '\"') {
+	  s->start_line = s->line;
+	  Clear(s->text);
+	  state = 78;
+	}
+	else if (c == '\'') {
+	  s->start_line = s->line;
+	  Clear(s->text);
+	  state = 79;
+	}
+	else if (isalnum(c) || (c == '_') || (c == '$'))
+	  state = 7;
+	else {
+	  retract(s, 1);
+	  return SWIG_TOKEN_ID;
+	}
+      break;
+
+      case 78:			/* Processing a wide string literal*/
+	if ((c = nextchar(s)) == 0) {
+	  Swig_error(cparse_file, cparse_start_line, "Unterminated wide string\n");
+	  return SWIG_TOKEN_ERROR;
+	}
+	if (c == '\"') {
+	  Delitem(s->text, DOH_END);
+	  return SWIG_TOKEN_WSTRING;
+	} else if (c == '\\') {
+	  if ((c = nextchar(s)) == 0) {
+	    Swig_error(cparse_file, cparse_start_line, "Unterminated wide string\n");
+	    return SWIG_TOKEN_ERROR;
+	  }
+	}
+	break;
+
+      case 79:			/* Processing a wide char literal */
+	if ((c = nextchar(s)) == 0) {
+	  Swig_error(cparse_file, cparse_start_line, "Unterminated character constant\n");
+	  return SWIG_TOKEN_ERROR;
+	}
+	if (c == '\'') {
+	  Delitem(s->text, DOH_END);
+	  return (SWIG_TOKEN_WCHAR);
+	} else if (c == '\\') {
+	  if ((c = nextchar(s)) == 0) {
+	    Swig_error(cparse_file, cparse_start_line, "Unterminated wide char literal\n");
+	    return SWIG_TOKEN_ERROR;
+	  }
+	}
+	break;
+
     case 75:			/* Special identifier $ */
       if ((c = nextchar(s)) == 0)
 	return SWIG_TOKEN_DOLLAR;

Source/Swig/stype.c

@@ -131,12 +131,21 @@ SwigType *NewSwigType(int t) {
   case T_UCHAR:
     return NewString("unsigned char");
     break;
-  case T_STRING:{
+  case T_STRING: {
       SwigType *t = NewString("char");
       SwigType_add_pointer(t);
       return t;
       break;
     }
+  case T_WCHAR:
+    return NewString("wchar_t");
+    break;
+  case T_WSTRING: {
+    SwigType *t = NewString("wchar_t");
+    SwigType_add_pointer(t);
+    return t;
+    break;
+  }
   case T_LONGLONG:
     return NewString("long long");
     break;

Source/Swig/swig.h

@@ -100,6 +100,9 @@ extern "C" {
 #define   T_SYMBOL     98
 #define   T_ERROR      99
 
+/* wide string literal, may contain escaped wide chars like \x1234 as well as normal escape sequences */
+#define   T_WSTRING    39
+
 
 /* --- File interface --- */
 

Source/Swig/swigscan.h

@@ -67,6 +67,9 @@ extern void         Scanner_freeze_line(Scanner *s, int val);
 #define   SWIG_TOKEN_QUESTION     30       /* ? */
 #define   SWIG_TOKEN_COMMENT      31       /* C or C++ comment */
 #define   SWIG_TOKEN_BOOL         32       /* true or false */
+#define   SWIG_TOKEN_WSTRING      33       /* L"str" */
+#define   SWIG_TOKEN_WCHAR        34       /* L'c' */
+
 #define   SWIG_TOKEN_ILLEGAL      99
 #define   SWIG_TOKEN_ERROR        -1
 

Source/Swig/typesys.c

@@ -1194,6 +1194,8 @@ int SwigType_type(SwigType *t) {
   if (strncmp(c, "p.", 2) == 0) {
     if (SwigType_type(c + 2) == T_CHAR)
       return T_STRING;
+    else if (SwigType_type(c + 2) == T_WCHAR)
+      return T_WSTRING;
     else
       return T_POINTER;
   }
@@ -1236,6 +1238,8 @@ int SwigType_type(SwigType *t) {
     return T_SCHAR;
   if (strcmp(c, "unsigned char") == 0)
     return T_UCHAR;
+  if (strcmp(c, "wchar_t") == 0)
+    return T_WCHAR;
   if (strcmp(c, "float") == 0)
     return T_FLOAT;
   if (strcmp(c, "double") == 0)