Implement 'character' literals (#1747)

src/asm/lexer.cpp

@@ -1428,6 +1428,7 @@ static void appendExpandedString(std::string &str, std::string const &expanded)
 			break;
 		case '\\':
 		case '"':
+		case '\'':
 		case '{':
 			str += '\\';
 			[[fallthrough]];
@@ -1448,6 +1449,7 @@ static void appendCharInLiteral(std::string &str, int c) {
 		// Character escape
 		case '\\':
 		case '"':
+		case '\'':
 		case '{':
 		case '}':
 			if (rawMode) {
@@ -1518,7 +1520,7 @@ static void appendCharInLiteral(std::string &str, int c) {
 		break;
 
 	case '{': // Symbol interpolation
-		// We'll be exiting the string scope, so re-enable expansions
+		// We'll be exiting the string/character scope, so re-enable expansions
 		// (Not interpolations, since they're handled by the function itself...)
 		lexerState->disableMacroArgs = false;
 		if (auto interpolation = readInterpolation(0); interpolation) {
@@ -1614,6 +1616,42 @@ static void readString(std::string &str, bool rawString) {
 	}
 }
 
+static void readCharacter(std::string &str) {
+	// This is essentially a simplified `readString`
+	Defer reenableExpansions = scopedDisableExpansions();
+
+	bool rawMode = lexerState->mode == LEXER_RAW;
+
+	// We reach this function after reading a single quote
+	if (rawMode) {
+		str += '\'';
+	}
+
+	for (;;) {
+		int c = peek();
+
+		// '\r', '\n' or EOF ends a character early
+		if (c == EOF || c == '\r' || c == '\n') {
+			error("Unterminated character");
+			return;
+		}
+
+		// We'll be staying in the character, so we can safely consume the char
+		shiftChar();
+
+		// Close the character and return if it's terminated
+		if (c == '\'') {
+			if (rawMode) {
+				str += c;
+			}
+			return;
+		}
+
+		// Append the character or handle special ones
+		appendCharInLiteral(str, c);
+	}
+}
+
 // Lexer core
 
 static Token yylex_SKIP_TO_ENDC(); // forward declaration for yylex_NORMAL
@@ -1896,7 +1934,7 @@ static Token yylex_NORMAL() {
 		case '`': // Gfx constant
 			return Token(T_(NUMBER), readGfxConstant());
 
-			// Handle strings
+			// Handle string and character literals
 
 		case '"': {
 			std::string str;
@@ -1904,6 +1942,12 @@ static Token yylex_NORMAL() {
 			return Token(T_(STRING), str);
 		}
 
+		case '\'': {
+			std::string chr;
+			readCharacter(chr);
+			return Token(T_(CHARACTER), chr);
+		}
+
 			// Handle newlines and EOF
 
 		case '\r':
@@ -2036,6 +2080,11 @@ static Token yylex_RAW() {
 			readString(str, false);
 			break;
 
+		case '\'': // Character literals inside macro args
+			shiftChar();
+			readCharacter(str);
+			break;
+
 		case '#': // Raw string literals inside macro args
 			str += c;
 			shiftChar();
@@ -2093,6 +2142,7 @@ backslash:
 			case ')':
 			case '\\': // Escapes shared with string literals
 			case '"':
+			case '\'':
 			case '{':
 			case '}':
 				break;

src/asm/parser.y

@@ -339,6 +339,7 @@
 // Literals
 %token <int32_t> NUMBER "number"
 %token <std::string> STRING "string"
+%token <std::string> CHARACTER "character"
 %token <std::string> SYMBOL "symbol"
 %token <std::string> LABEL "label"
 %token <std::string> LOCAL "local label"
@@ -1415,6 +1416,15 @@ relocexpr_no_str:
 	NUMBER {
 		$$.makeNumber($1);
 	}
+	| CHARACTER {
+		std::vector<int32_t> output = charmap_Convert($1);
+		if (output.size() == 1) {
+			$$.makeNumber(static_cast<uint32_t>(output[0]));
+		} else {
+			::error("Character literals must be a single charmap unit");
+			$$.makeNumber(0);
+		}
+	}
 	| OP_LOGICNOT relocexpr %prec NEG {
 		$$.makeUnaryOp(RPN_LOGNOT, std::move($2));
 	}

src/link/script.y

@@ -293,7 +293,7 @@ yy::parser::symbol_type yylex() {
 					c = '\r';
 				} else if (c == 't') {
 					c = '\t';
-				} else if (c != '\\' && c != '"') {
+				} else if (c != '\\' && c != '"' && c != '\'') {
 					scriptError(context, "Cannot escape character %s", printChar(c));
 				}
 				context.file.sbumpc();