Improve alternative operator names support

Handle alternative operator names in C++ preprocessor expressions.
Handle full set of alternative operator names in C++ expressions
(previously only "and", "or" and "not" were understood).

Fixes #2914

CHANGES.current

@@ -7,6 +7,12 @@ the issue number to the end of the URL: https://github.com/swig/swig/issues/
 Version 4.3.0 (in progress)
 ===========================
 
+2024-05-30: olly
+	    #2914 Handle alternative operator names in C++ preprocessor
+	    expressions.  Handle full set of alternative operator names in
+	    C++ expressions (previously only "and", "or" and "not" were
+	    understood).
+
 2024-05-15: olly
 	    #2868 Support C++17 fold expressions.
 

Examples/test-suite/cpp11_decltype.i

@@ -74,6 +74,9 @@
     // so this would end up wrapped as int.
     decltype(!0) should_be_bool;
 
+    // Test alternative operator names work in this context.
+    decltype(((compl 42) and (not 1)) or (2 xor 4)) should_be_bool2;
+
     decltype(E1) should_be_enum;
 
     auto get_number_sum(decltype(i+j) a) -> decltype(i+j) {

Examples/test-suite/expressions.i

@@ -9,5 +9,9 @@ struct A
     // Regression test for preprocessor bug with handling a slash immediately
     // followed by a single quote, fixed in 4.2.0.  (#2630)
     int f(int i = 64/' ') { return i; }
+
+    // Regression test for alternative operator names - this failed to parse in
+    // SWIG 4.2.0 and earlier.
+    int g(bool b = (compl 1 or not 2) xor (3 and 4) xor (3 bitand 6) xor (3 bitor 5) xor (2 + 2 not_eq 5)) { return (int)b; }
 };
 %}

Examples/test-suite/php/cpp11_decltype_runme.php

@@ -52,6 +52,7 @@ check::equal(gettype($b->should_be_int9), "integer");
 check::equal(gettype($b->should_be_int10), "integer");
 
 check::equal(gettype($b->should_be_bool), "boolean");
+check::equal(gettype($b->should_be_bool2), "boolean");
 
 check::equal(gettype($b::should_be_char), "string");
 check::equal($b::should_be_char, "\0");

Examples/test-suite/preproc_cpp.i

@@ -1,7 +1,7 @@
 %module preproc_cpp
 
 
-// booleans start
+// Test boolean literals (https://github.com/swig/swig/pull/2394):
 #if false
 # error "boolean preproc check fail false"
 #else
@@ -31,7 +31,43 @@
 #else
 # error "boolean preproc check fail (0 || true)"
 #endif
-// booleans end
+
+
+// Test alternative operator names (https://github.com/swig/swig/issues/2914):
+#if true or 0
+#else
+# error "boolean preproc check fail true or 0"
+#endif
+
+#if true and 1
+#else
+# error "boolean preproc check fail true and 1"
+#endif
+
+#if true xor false
+#else
+# error "boolean preproc check fail true xor false"
+#endif
+
+#if 1 bitor 2
+#else
+# error "boolean preproc check fail 1 bitor 2"
+#endif
+
+#if not(1 bitand 2)
+#else
+# error "boolean preproc check fail not(1 bitand 2)"
+#endif
+
+#if compl 0
+#else
+# error "boolean preproc check fail compl 0"
+#endif
+
+#if (1 bitor 2) not_eq 1
+#else
+# error "boolean preproc check fail (1 bitor 2) not_eq 1"
+#endif
 
 
 // Regression tests for bugs handling a comma in a comment in a macro:

Source/CParse/cscanner.c

@@ -719,12 +719,6 @@ num_common:
 
       /* C++ keywords */
       if (cparse_cplusplus) {
-	if (strcmp(yytext, "and") == 0)
-	  return (LAND);
-	if (strcmp(yytext, "or") == 0)
-	  return (LOR);
-	if (strcmp(yytext, "not") == 0)
-	  return (LNOT);
 	if (strcmp(yytext, "class") == 0)
 	  return (CLASS);
 	if (strcmp(yytext, "private") == 0)
@@ -802,9 +796,12 @@ num_common:
 	    yylval.str = s;
 	    return OPERATOR;
 	  } else if (nexttok == SWIG_TOKEN_ID) {
-	    /* We have an identifier.  This could be any number of things. It could be a named version of
-               an operator (e.g., 'and_eq') or it could be a conversion operator.   To deal with this, we're
-               going to read tokens until we encounter a ( or ;.  Some care is needed for formatting. */
+	    /* We have an identifier.  It could be "new" or "delete",
+	     * potentially followed by "[]", or it could be a conversion
+	     * operator (it can't be "and_eq" or similar as those are returned
+	     * as SWIG_TOKEN_ANDEQUAL, etc by Scanner_token()).  To deal with
+	     * this we read tokens until we encounter a suitable terminating
+	     * token.  Some care is needed for formatting. */
 	    int needspace = 1;
 	    int termtoken = 0;
 	    const char *termvalue = 0;
@@ -856,17 +853,6 @@ num_common:
 		    || (strcmp(t, "delete") == 0)
 		    || (strcmp(t, "new[]") == 0)
 		    || (strcmp(t, "delete[]") == 0)
-		    || (strcmp(t, "and") == 0)
-		    || (strcmp(t, "and_eq") == 0)
-		    || (strcmp(t, "bitand") == 0)
-		    || (strcmp(t, "bitor") == 0)
-		    || (strcmp(t, "compl") == 0)
-		    || (strcmp(t, "not") == 0)
-		    || (strcmp(t, "not_eq") == 0)
-		    || (strcmp(t, "or") == 0)
-		    || (strcmp(t, "or_eq") == 0)
-		    || (strcmp(t, "xor") == 0)
-		    || (strcmp(t, "xor_eq") == 0)
 		    )) {
 		/*              retract(strlen(t)); */
 

Source/Swig/scanner.c

@@ -1025,12 +1025,35 @@ static int look(Scanner *s) {
       }
       break;
 
-    case 76:			/* Identifier or true/false */
+    case 76:			/* Identifier, true/false or alternative token */
       if (cparse_cplusplus) {
 	if (Strcmp(s->text, "true") == 0)
 	  return SWIG_TOKEN_BOOL;
-	else if (Strcmp(s->text, "false") == 0)
+	if (Strcmp(s->text, "false") == 0)
 	  return SWIG_TOKEN_BOOL;
+
+	if (Strcmp(s->text, "and") == 0)
+	  return SWIG_TOKEN_LAND;
+	if (Strcmp(s->text, "and_eq") == 0)
+	  return SWIG_TOKEN_ANDEQUAL;
+	if (Strcmp(s->text, "bitand") == 0)
+	  return SWIG_TOKEN_AND;
+	if (Strcmp(s->text, "bitor") == 0)
+	  return SWIG_TOKEN_OR;
+	if (Strcmp(s->text, "compl") == 0)
+	  return SWIG_TOKEN_NOT;
+	if (Strcmp(s->text, "not") == 0)
+	  return SWIG_TOKEN_LNOT;
+	if (Strcmp(s->text, "not_eq") == 0)
+	  return SWIG_TOKEN_NOTEQUAL;
+	if (Strcmp(s->text, "or") == 0)
+	  return SWIG_TOKEN_LOR;
+	if (Strcmp(s->text, "or_eq") == 0)
+	  return SWIG_TOKEN_OREQUAL;
+	if (Strcmp(s->text, "xor") == 0)
+	  return SWIG_TOKEN_XOR;
+	if (Strcmp(s->text, "xor_eq") == 0)
+	  return SWIG_TOKEN_XOREQUAL;
       }
       return SWIG_TOKEN_ID;