Disable EQUS expansion for raw symbols (by parsing them as strings) (#1648)

Author: Rangi42

man/rgbasm.5

@@ -1125,7 +1125,9 @@ Additionally, label names can contain up to a single dot
 .Ql \&. ,
 which may not be the first character.
 .Pp
-A symbol cannot have the same name as a reserved keyword, unless it is prefixed by a hash
+A symbol cannot have the same name as a reserved keyword, unless its name is a
+.Dq raw identifier
+prefixed by a hash
 .Sq # .
 For example,
 .Ql #load
@@ -1300,7 +1302,7 @@ it at the same time.
 below).
 .Ss Numeric constants
 .Ic EQU
-is used to define immutable numeric symbols.
+is used to define numeric constant symbols.
 Unlike
 .Sq =
 above, constants defined this way cannot be redefined.
@@ -1408,6 +1410,8 @@ This expansion is disabled in a few contexts:
 and
 .Ql MACRO name
 will not expand string constants in their names.
+Expansion is also disabled if the string constant's name is a raw identifier prefixed by a hash
+.Sq # .
 .Bd -literal -offset indent
 DEF COUNTREG EQUS "[hl+]"
     ld a, COUNTREG
@@ -1873,7 +1877,7 @@ being the second, and so on. Since there are only nine digits, you can only use
 To use the rest, you put the argument number in angle brackets, like
 .Ic \e<10> .
 .Pp
-This bracketed syntax supports decimal numbers and numeric constant symbols.
+This bracketed syntax supports decimal numbers and numeric symbols.
 For example,
 .Ql \e<_NARG>
 will get the last argument.

src/asm/lexer.cpp

@@ -1915,8 +1915,8 @@ static Token yylex_NORMAL() {
 				// `token` is either a `SYMBOL` or a `LOCAL`, and both have a `std::string` value.
 				assume(token.value.holds<std::string>());
 
-				// Local symbols cannot be string expansions
-				if (token.type == T_(SYMBOL) && lexerState->expandStrings) {
+				// Raw symbols and local symbols cannot be string expansions
+				if (!raw && token.type == T_(SYMBOL) && lexerState->expandStrings) {
 					// Attempt string expansion
 					Symbol const *sym = sym_FindExactSymbol(token.value.get<std::string>());
 

src/asm/parser.y

@@ -80,6 +80,17 @@
 	static void failAssert(AssertionType type);
 	static void failAssertMsg(AssertionType type, std::string const &message);
 
+	template <typename N, typename S>
+	static auto handleSymbolByType(std::string const &symName, N numCallback, S strCallback) {
+		if (Symbol *sym = sym_FindScopedSymbol(symName); sym && sym->type == SYM_EQUS) {
+			return strCallback(*sym->getEqus());
+		} else {
+			Expression expr;
+			expr.makeSymbol(symName);
+			return numCallback(expr);
+		}
+	}
+
 	// The CPU encodes instructions in a logical way, so most instructions actually follow patterns.
 	// These enums thus help with bit twiddling to compute opcodes.
 	enum { REG_B, REG_C, REG_D, REG_E, REG_H, REG_L, REG_HL_IND, REG_A };
@@ -343,6 +354,7 @@
 
 // Strings
 %type <std::string> string
+%type <std::string> string_literal
 %type <std::string> strcat_args
 // Strings used for identifiers
 %type <std::string> def_id
@@ -1210,10 +1222,17 @@ print_expr:
 	relocexpr_no_str {
 		printf("$%" PRIX32, $1.getConstVal());
 	}
-	| string {
+	| string_literal {
 		// Allow printing NUL characters
 		fwrite($1.data(), 1, $1.length(), stdout);
 	}
+	| scoped_sym {
+		handleSymbolByType(
+		    $1,
+		    [](Expression const &expr) { printf("$%" PRIX32, expr.getConstVal()); },
+		    [](std::string const &str) { fwrite(str.data(), 1, str.length(), stdout); }
+		);
+	}
 ;
 
 reloc_3bit:
@@ -1233,10 +1252,23 @@ constlist_8bit_entry:
 		$1.checkNBit(8);
 		sect_RelByte($1, 0);
 	}
-	| string {
+	| string_literal {
 		std::vector<int32_t> output = charmap_Convert($1);
 		sect_ByteString(output);
 	}
+	| scoped_sym {
+		handleSymbolByType(
+		    $1,
+		    [](Expression const &expr) {
+			    expr.checkNBit(8);
+			    sect_RelByte(expr, 0);
+		    },
+		    [](std::string const &str) {
+			    std::vector<int32_t> output = charmap_Convert(str);
+			    sect_ByteString(output);
+		    }
+		);
+	}
 ;
 
 constlist_16bit:
@@ -1249,10 +1281,23 @@ constlist_16bit_entry:
 		$1.checkNBit(16);
 		sect_RelWord($1, 0);
 	}
-	| string {
+	| string_literal {
 		std::vector<int32_t> output = charmap_Convert($1);
 		sect_WordString(output);
 	}
+	| scoped_sym {
+		handleSymbolByType(
+		    $1,
+		    [](Expression const &expr) {
+			    expr.checkNBit(16);
+			    sect_RelWord(expr, 0);
+		    },
+		    [](std::string const &str) {
+			    std::vector<int32_t> output = charmap_Convert(str);
+			    sect_WordString(output);
+		    }
+		);
+	}
 ;
 
 constlist_32bit:
@@ -1264,10 +1309,20 @@ constlist_32bit_entry:
 	relocexpr_no_str {
 		sect_RelLong($1, 0);
 	}
-	| string {
+	| string_literal {
 		std::vector<int32_t> output = charmap_Convert($1);
 		sect_LongString(output);
 	}
+	| scoped_sym {
+		handleSymbolByType(
+		    $1,
+		    [](Expression const &expr) { sect_RelLong(expr, 0); },
+		    [](std::string const &str) {
+			    std::vector<int32_t> output = charmap_Convert(str);
+			    sect_LongString(output);
+		    }
+		);
+	}
 ;
 
 reloc_8bit:
@@ -1299,17 +1354,26 @@ relocexpr:
 	relocexpr_no_str {
 		$$ = std::move($1);
 	}
-	| string {
+	| string_literal {
 		std::vector<int32_t> output = charmap_Convert($1);
 		$$.makeNumber(strToNum(output));
 	}
+	| scoped_sym {
+		$$ = handleSymbolByType(
+		    $1,
+		    [](Expression const &expr) { return expr; },
+		    [](std::string const &str) {
+			    std::vector<int32_t> output = charmap_Convert(str);
+			    Expression expr;
+			    expr.makeNumber(strToNum(output));
+			    return expr;
+		    }
+		);
+	}
 ;
 
 relocexpr_no_str:
-	scoped_sym {
-		$$.makeSymbol($1);
-	}
-	| NUMBER {
+	NUMBER {
 		$$.makeNumber($1);
 	}
 	| OP_LOGICNOT relocexpr %prec NEG {
@@ -1403,7 +1467,7 @@ relocexpr_no_str:
 		// '@' is also a SYMBOL; it is handled here
 		$$.makeBankSymbol($3);
 	}
-	| OP_BANK LPAREN string RPAREN {
+	| OP_BANK LPAREN string_literal RPAREN {
 		$$.makeBankSection($3);
 	}
 	| OP_SIZEOF LPAREN string RPAREN {
@@ -1540,7 +1604,7 @@ precision_arg:
 	}
 ;
 
-string:
+string_literal:
 	STRING {
 		$$ = std::move($1);
 	}
@@ -1625,6 +1689,19 @@ string:
 	}
 ;
 
+string:
+	string_literal {
+		$$ = std::move($1);
+	}
+	| scoped_sym {
+		if (Symbol *sym = sym_FindScopedSymbol($1); sym && sym->type == SYM_EQUS) {
+			$$ = *sym->getEqus();
+		} else {
+			::error("'%s' is not a string symbol\n", $1.c_str());
+		}
+	}
+;
+
 strcat_args:
 	string {
 		$$ = std::move($1);
@@ -1649,10 +1726,20 @@ strfmt_va_args:
 		$$ = std::move($1);
 		$$.args.push_back(static_cast<uint32_t>($3.getConstVal()));
 	}
-	| strfmt_va_args COMMA string {
+	| strfmt_va_args COMMA string_literal {
 		$$ = std::move($1);
 		$$.args.push_back(std::move($3));
 	}
+	| strfmt_va_args COMMA scoped_sym {
+		$$ = std::move($1);
+		handleSymbolByType(
+		    $3,
+		    [&](Expression const &expr) {
+			    $$.args.push_back(static_cast<uint32_t>(expr.getConstVal()));
+		    },
+		    [&](std::string const &str) { $$.args.push_back(str); }
+		);
+	}
 ;
 
 section:

test/asm/raw-string-symbol-errors.asm

@@ -0,0 +1,29 @@
+def n equ 42
+def s equs "hello"
+macro m
+endm
+
+assert (#n) == 42
+assert (#s) == $656c6c6f
+assert (#m) == 0
+assert (#u) == 0
+
+assert strlen(#n) == 0
+assert strlen(#s) == 5
+assert strlen(#m) == 0
+assert strlen(#u) == 0
+
+def d_n = (#n)
+def d_s = (#s)
+def d_m = (#m)
+def d_u = (#u)
+
+def s_n equs #n
+def s_s equs #s
+def s_m equs #m
+def s_u equs #u
+
+purge #s
+purge #s
+assert (#s) == 0
+assert strlen(#s) == 0

test/asm/raw-string-symbol-errors.err

@@ -0,0 +1,27 @@
+warning: raw-string-symbol-errors.asm(7): [-Wobsolete]
+    Treating multi-unit strings as numbers is deprecated
+error: raw-string-symbol-errors.asm(8):
+    'm' is not a numeric symbol
+error: raw-string-symbol-errors.asm(11):
+    'n' is not a string symbol
+error: raw-string-symbol-errors.asm(13):
+    'm' is not a string symbol
+error: raw-string-symbol-errors.asm(14):
+    'u' is not a string symbol
+warning: raw-string-symbol-errors.asm(17): [-Wobsolete]
+    Treating multi-unit strings as numbers is deprecated
+error: raw-string-symbol-errors.asm(18):
+    'm' is not a numeric symbol
+error: raw-string-symbol-errors.asm(19):
+    Expected constant expression: 'u' is not constant at assembly time
+error: raw-string-symbol-errors.asm(21):
+    'n' is not a string symbol
+error: raw-string-symbol-errors.asm(23):
+    'm' is not a string symbol
+error: raw-string-symbol-errors.asm(24):
+    'u' is not a string symbol
+error: raw-string-symbol-errors.asm(27):
+    's' was already purged
+error: raw-string-symbol-errors.asm(29):
+    's' is not a string symbol
+error: Assembly aborted (11 errors)!

test/asm/raw-string-symbols.asm

@@ -0,0 +1,34 @@
+opt Wno-unmapped-char
+
+def hello equs "world"
+def name equs "hello"
+println "{name}"
+println #name
+assert !strcmp(strsub(#name, 1, 4), "hell")
+assert strlen(#hello) == charlen(#hello)
+assert strlen("{hello}") == 5
+
+def multi equs """the	quick
+brown	fox"""
+println #multi
+
+def char equs "A"
+def n = #char
+println n
+def n = (#char)
+println n
+def n = 1 + #char
+println n
+assert #char == $41
+
+def fmt equs "%s %s %d"
+println strfmt(#fmt, #name, #hello, (#char))
+
+purge #name
+assert !def(name) && !def(#name) && def(hello)
+
+section "test", rom0
+#label:
+db #hello
+dw #hello
+dw BANK(#label), #label

test/asm/raw-string-symbols.out

@@ -0,0 +1,8 @@
+hello
+hello
+the	quick
+brown	fox
+$41
+$41
+$42
+hello world 65