X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=lexer.c;h=14526b95694a642b9c3557c2656fd770d72beab1;hb=ef26133e70bcbc508d2991bafef373f5ca667ff2;hp=d9bd2c4d446ff72e48f958d43aa3e67945fa4451;hpb=22e1e31672ac88705fe8621657f10a48fe661118;p=cparser

diff --git a/lexer.c b/lexer.c
index d9bd2c4..14526b9 100644
--- a/lexer.c
+++ b/lexer.c
@@ -108,16 +108,17 @@ static inline void put_back(utf32 const pc)
 
 static inline void next_char(void);
 
-#define MATCH_NEWLINE(code)  \
-	case '\r':               \
-		next_char();         \
-		if (c == '\n') {     \
-	case '\n':               \
-			next_char();     \
-		}                    \
-		lexer_pos.lineno++;  \
+#define NEWLINE  \
+	'\r': \
+		next_char(); \
+		if (c == '\n') { \
+	case '\n': \
+			next_char(); \
+		} \
+		lexer_pos.lineno++; \
 		lexer_pos.colno = 1; \
-		code
+		goto newline; \
+		newline // Let it look like an ordinary case label.
 
 #define eat(c_type) (assert(c == c_type), next_char())
 
@@ -126,7 +127,8 @@ static void maybe_concat_lines(void)
 	eat('\\');
 
 	switch (c) {
-	MATCH_NEWLINE(return;)
+	case NEWLINE:
+		return;
 
 	default:
 		break;
@@ -247,15 +249,103 @@ static inline void next_char(void)
 	case '8':         \
 	case '9':
 
+static bool is_universal_char_valid(utf32 const v)
+{
+	/* C11 §6.4.3:2 */
+	if (v < 0xA0U && v != 0x24 && v != 0x40 && v != 0x60)
+		return false;
+	if (0xD800 <= v && v <= 0xDFFF)
+		return false;
+	return true;
+}
+
+static int digit_value(utf32 digit);
+
+static utf32 parse_universal_char(unsigned const n_digits)
+{
+	utf32 v = 0;
+	for (unsigned k = n_digits; k != 0; --k) {
+		if (isxdigit(c)) {
+			v = 16 * v + digit_value(c);
+			next_char();
+		} else {
+			errorf(&lexer_pos, "short universal character name, expected %u more digits", k);
+			break;
+		}
+	}
+	if (!is_universal_char_valid(v)) {
+		errorf(&lexer_pos, "\\%c%0*X is not a valid universal character name", n_digits == 4 ? 'u' : 'U', (int)n_digits, v);
+	}
+	return v;
+}
+
+static bool is_universal_char_valid_identifier(utf32 const v)
+{
+	/* C11 Annex D.1 */
+	if (                v == 0x000A8) return true;
+	if (                v == 0x000AA) return true;
+	if (                v == 0x000AD) return true;
+	if (                v == 0x000AF) return true;
+	if (0x000B2 <= v && v <= 0x000B5) return true;
+	if (0x000B7 <= v && v <= 0x000BA) return true;
+	if (0x000BC <= v && v <= 0x000BE) return true;
+	if (0x000C0 <= v && v <= 0x000D6) return true;
+	if (0x000D8 <= v && v <= 0x000F6) return true;
+	if (0x000F8 <= v && v <= 0x000FF) return true;
+	if (0x00100 <= v && v <= 0x0167F) return true;
+	if (0x01681 <= v && v <= 0x0180D) return true;
+	if (0x0180F <= v && v <= 0x01FFF) return true;
+	if (0x0200B <= v && v <= 0x0200D) return true;
+	if (0x0202A <= v && v <= 0x0202E) return true;
+	if (0x0203F <= v && v <= 0x02040) return true;
+	if (                v == 0x02054) return true;
+	if (0x02060 <= v && v <= 0x0206F) return true;
+	if (0x02070 <= v && v <= 0x0218F) return true;
+	if (0x02460 <= v && v <= 0x024FF) return true;
+	if (0x02776 <= v && v <= 0x02793) return true;
+	if (0x02C00 <= v && v <= 0x02DFF) return true;
+	if (0x02E80 <= v && v <= 0x02FFF) return true;
+	if (0x03004 <= v && v <= 0x03007) return true;
+	if (0x03021 <= v && v <= 0x0302F) return true;
+	if (0x03031 <= v && v <= 0x0303F) return true;
+	if (0x03040 <= v && v <= 0x0D7FF) return true;
+	if (0x0F900 <= v && v <= 0x0FD3D) return true;
+	if (0x0FD40 <= v && v <= 0x0FDCF) return true;
+	if (0x0FDF0 <= v && v <= 0x0FE44) return true;
+	if (0x0FE47 <= v && v <= 0x0FFFD) return true;
+	if (0x10000 <= v && v <= 0x1FFFD) return true;
+	if (0x20000 <= v && v <= 0x2FFFD) return true;
+	if (0x30000 <= v && v <= 0x3FFFD) return true;
+	if (0x40000 <= v && v <= 0x4FFFD) return true;
+	if (0x50000 <= v && v <= 0x5FFFD) return true;
+	if (0x60000 <= v && v <= 0x6FFFD) return true;
+	if (0x70000 <= v && v <= 0x7FFFD) return true;
+	if (0x80000 <= v && v <= 0x8FFFD) return true;
+	if (0x90000 <= v && v <= 0x9FFFD) return true;
+	if (0xA0000 <= v && v <= 0xAFFFD) return true;
+	if (0xB0000 <= v && v <= 0xBFFFD) return true;
+	if (0xC0000 <= v && v <= 0xCFFFD) return true;
+	if (0xD0000 <= v && v <= 0xDFFFD) return true;
+	if (0xE0000 <= v && v <= 0xEFFFD) return true;
+	return false;
+}
+
+static bool is_universal_char_valid_identifier_start(utf32 const v)
+{
+	/* C11 Annex D.2 */
+	if (0x0300 <= v && v <= 0x036F) return false;
+	if (0x1DC0 <= v && v <= 0x1DFF) return false;
+	if (0x20D0 <= v && v <= 0x20FF) return false;
+	if (0xFE20 <= v && v <= 0xFE2F) return false;
+	return true;
+}
+
 /**
  * Read a symbol from the input and build
  * the lexer_token.
  */
 static void parse_symbol(void)
 {
-	obstack_1grow(&symbol_obstack, (char) c);
-	next_char();
-
 	while (true) {
 		switch (c) {
 		DIGITS
@@ -264,6 +354,34 @@ static void parse_symbol(void)
 			next_char();
 			break;
 
+		case '\\':
+			next_char();
+			switch (c) {
+			{
+				unsigned n;
+			case 'U': n = 8; goto universal;
+			case 'u': n = 4; goto universal;
+universal:
+				next_char();
+				utf32 const v = parse_universal_char(n);
+				if (!is_universal_char_valid_identifier(v)) {
+					if (is_universal_char_valid(v)) {
+						errorf(&lexer_pos, "universal character \\%c%0*X is not valid in an identifier", n == 4 ? 'u' : 'U', (int)n, v);
+					}
+				} else if (obstack_object_size(&symbol_obstack) == 0 && !is_universal_char_valid_identifier_start(v)) {
+					errorf(&lexer_pos, "universal character \\%c%0*X is not valid as start of an identifier", n == 4 ? 'u' : 'U', (int)n, v);
+				} else {
+					obstack_grow_symbol(&symbol_obstack, v);
+				}
+				break;
+			}
+
+			default:
+				put_back(c);
+				c = '\\';
+				goto end_symbol;
+			}
+
 		default:
 dollar_sign:
 			goto end_symbol;
@@ -620,10 +738,10 @@ static utf32 parse_escape_sequence(void)
 		if (c_mode & _GNUC)
 			return 27;   /* hopefully 27 is ALWAYS the code for ESCAPE */
 		break;
-	case 'u':
-	case 'U':
-		parse_error("universal character parsing not implemented yet");
-		return EOF;
+
+	case 'U': return parse_universal_char(8);
+	case 'u': return parse_universal_char(4);
+
 	default:
 		break;
 	}
@@ -632,21 +750,6 @@ static utf32 parse_escape_sequence(void)
 	return EOF;
 }
 
-/**
- * Concatenate two strings.
- */
-string_t concat_strings(const string_t *const s1, const string_t *const s2)
-{
-	const size_t len1 = s1->size - 1;
-	const size_t len2 = s2->size - 1;
-
-	char *const concat = obstack_alloc(&symbol_obstack, len1 + len2 + 1);
-	memcpy(concat, s1->begin, len1);
-	memcpy(concat + len1, s2->begin, len2 + 1);
-
-	return identify_string(concat, len1 + len2 + 1);
-}
-
 string_t make_string(const char *string)
 {
 	size_t      len   = strlen(string) + 1;
@@ -656,158 +759,70 @@ string_t make_string(const char *string)
 	return identify_string(space, len);
 }
 
-/**
- * Parse a string literal and set lexer_token.
- */
-static void parse_string_literal(void)
+static void parse_string(utf32 const delim, token_kind_t const kind, string_encoding_t const enc, char const *const context)
 {
-	eat('"');
+	eat(delim);
 
 	while (true) {
 		switch (c) {
 		case '\\': {
 			utf32 const tc = parse_escape_sequence();
-			if (tc >= 0x100) {
-				warningf(WARN_OTHER, &lexer_pos, "escape sequence out of range");
+			if (enc == STRING_ENCODING_CHAR) {
+				if (tc >= 0x100) {
+					warningf(WARN_OTHER, &lexer_pos, "escape sequence out of range");
+				}
+				obstack_1grow(&symbol_obstack, tc);
+			} else {
+				obstack_grow_symbol(&symbol_obstack, tc);
 			}
-			obstack_1grow(&symbol_obstack, tc);
-			break;
-		}
-
-		case EOF:
-			errorf(&lexer_token.base.source_position, "string has no end");
-			goto end_of_string;
-
-		case '"':
-			next_char();
-			goto end_of_string;
-
-		default:
-			obstack_grow_symbol(&symbol_obstack, c);
-			next_char();
-			break;
-		}
-	}
-
-end_of_string:
-
-	/* TODO: concatenate multiple strings separated by whitespace... */
-
-	/* add finishing 0 to the string */
-	obstack_1grow(&symbol_obstack, '\0');
-	const size_t  size   = (size_t)obstack_object_size(&symbol_obstack);
-	char         *string = obstack_finish(&symbol_obstack);
-
-	lexer_token.kind          = T_STRING_LITERAL;
-	lexer_token.string.string = identify_string(string, size);
-}
-
-/**
- * Parse a wide character constant and set lexer_token.
- */
-static void parse_wide_character_constant(void)
-{
-	eat('\'');
-
-	while (true) {
-		switch (c) {
-		case '\\': {
-			const utf32 tc = parse_escape_sequence();
-			obstack_grow_symbol(&symbol_obstack, tc);
 			break;
 		}
 
-		MATCH_NEWLINE(
-			parse_error("newline while parsing character constant");
+		case NEWLINE:
+			errorf(&lexer_pos, "newline while parsing %s", context);
 			break;
-		)
-
-		case '\'':
-			next_char();
-			goto end_of_wide_char_constant;
 
 		case EOF:
-			errorf(&lexer_token.base.source_position, "EOF while parsing character constant");
-			goto end_of_wide_char_constant;
+			errorf(&lexer_token.base.source_position, "EOF while parsing %s", context);
+			goto end_of_string;
 
 		default:
-			obstack_grow_symbol(&symbol_obstack, c);
-			next_char();
-			break;
+			if (c == delim) {
+				next_char();
+				goto end_of_string;
+			} else {
+				obstack_grow_symbol(&symbol_obstack, c);
+				next_char();
+				break;
+			}
 		}
 	}
 
-end_of_wide_char_constant:;
+end_of_string:
 	obstack_1grow(&symbol_obstack, '\0');
-	size_t  size   = (size_t) obstack_object_size(&symbol_obstack) - 1;
-	char   *string = obstack_finish(&symbol_obstack);
+	size_t const size   = obstack_object_size(&symbol_obstack) - 1;
+	char  *const string = obstack_finish(&symbol_obstack);
 
-	lexer_token.kind          = T_WIDE_CHARACTER_CONSTANT;
-	lexer_token.string.string = identify_string(string, size);
-
-	if (size == 0) {
-		errorf(&lexer_token.base.source_position, "empty character constant");
-	}
+	lexer_token.kind            = kind;
+	lexer_token.string.encoding = enc;
+	lexer_token.string.string   = identify_string(string, size);
 }
 
 /**
- * Parse a wide string literal and set lexer_token.
+ * Parse a string literal and set lexer_token.
  */
-static void parse_wide_string_literal(void)
+static void parse_string_literal(string_encoding_t const enc)
 {
-	parse_string_literal();
-	if (lexer_token.kind == T_STRING_LITERAL)
-		lexer_token.kind = T_WIDE_STRING_LITERAL;
+	parse_string('"', T_STRING_LITERAL, enc, "string literal");
 }
 
 /**
  * Parse a character constant and set lexer_token.
  */
-static void parse_character_constant(void)
+static void parse_character_constant(string_encoding_t const enc)
 {
-	eat('\'');
-
-	while (true) {
-		switch (c) {
-		case '\\': {
-			utf32 const tc = parse_escape_sequence();
-			if (tc >= 0x100) {
-				warningf(WARN_OTHER, &lexer_pos, "escape sequence out of range");
-			}
-			obstack_1grow(&symbol_obstack, tc);
-			break;
-		}
-
-		MATCH_NEWLINE(
-			parse_error("newline while parsing character constant");
-			break;
-		)
-
-		case '\'':
-			next_char();
-			goto end_of_char_constant;
-
-		case EOF:
-			errorf(&lexer_token.base.source_position, "EOF while parsing character constant");
-			goto end_of_char_constant;
-
-		default:
-			obstack_grow_symbol(&symbol_obstack, c);
-			next_char();
-			break;
-
-		}
-	}
-
-end_of_char_constant:;
-	obstack_1grow(&symbol_obstack, '\0');
-	const size_t        size   = (size_t)obstack_object_size(&symbol_obstack)-1;
-	char         *const string = obstack_finish(&symbol_obstack);
-
-	lexer_token.kind          = T_CHARACTER_CONSTANT;
-	lexer_token.string.string = identify_string(string, size);
-
-	if (size == 0) {
+	parse_string('\'', T_CHARACTER_CONSTANT, enc, "character constant");
+	if (lexer_token.string.string.size == 0) {
 		errorf(&lexer_token.base.source_position, "empty character constant");
 	}
 }
@@ -834,7 +849,8 @@ static void skip_multiline_comment(void)
 			}
 			break;
 
-		MATCH_NEWLINE(break;)
+		case NEWLINE:
+			break;
 
 		case EOF: {
 			errorf(&lexer_token.base.source_position,
@@ -912,7 +928,7 @@ static void parse_line_directive(void)
 		lexer_pos.lineno = atoi(pp_token.number.number.begin) - 1;
 		next_pp_token();
 	}
-	if (pp_token.kind == T_STRING_LITERAL) {
+	if (pp_token.kind == T_STRING_LITERAL && pp_token.string.encoding == STRING_ENCODING_CHAR) {
 		lexer_pos.input_name = pp_token.string.string.begin;
 		lexer_pos.is_system_header = false;
 		next_pp_token();
@@ -1093,32 +1109,33 @@ void lexer_next_preprocessing_token(void)
 			next_char();
 			break;
 
-		MATCH_NEWLINE(
+		case NEWLINE:
 			lexer_token.kind = '\n';
 			return;
-		)
 
-		SYMBOL_CHARS
+		SYMBOL_CHARS {
 			parse_symbol();
 			/* might be a wide string ( L"string" ) */
+			string_encoding_t const enc = STRING_ENCODING_WIDE;
 			if (lexer_token.base.symbol == symbol_L) {
 				switch (c) {
-					case '"':  parse_wide_string_literal();     break;
-					case '\'': parse_wide_character_constant(); break;
+				case '"':  parse_string_literal(enc);     break;
+				case '\'': parse_character_constant(enc); break;
 				}
 			}
 			return;
+		}
 
 		DIGITS
 			parse_number();
 			return;
 
 		case '"':
-			parse_string_literal();
+			parse_string_literal(STRING_ENCODING_CHAR);
 			return;
 
 		case '\'':
-			parse_character_constant();
+			parse_character_constant(STRING_ENCODING_CHAR);
 			return;
 
 		case '.':
@@ -1233,6 +1250,17 @@ void lexer_next_preprocessing_token(void)
 			MAYBE('#', T_HASHHASH)
 			ELSE('#')
 
+		case '\\':
+			next_char();
+			if (c == 'U' || c == 'u') {
+				put_back(c);
+				c = '\\';
+				parse_symbol();
+			} else {
+				lexer_token.kind = '\\';
+			}
+			return;
+
 		case '?':
 		case '[':
 		case ']':
@@ -1243,7 +1271,6 @@ void lexer_next_preprocessing_token(void)
 		case '~':
 		case ';':
 		case ',':
-		case '\\':
 			lexer_token.kind = c;
 			next_char();
 			return;