X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=lexer.c;h=1f0c23d46b004cb0cea3ed84a7689f3c9c55c5df;hb=34a560b81b8899995ff67421650edcad53efb4d4;hp=2177a303bf79b74b63de8d97b5662f9f583a8d14;hpb=d95d714c02991fdf212f1ff56fab9c98746a8f73;p=cparser

diff --git a/lexer.c b/lexer.c
index 2177a30..1f0c23d 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;
@@ -179,13 +181,12 @@ static inline void next_char(void)
 	}
 }
 
-#define SYMBOL_CHARS  \
-	case '$': if (!allow_dollar_in_symbol) goto dollar_sign; \
+#define SYMBOL_CHARS_WITHOUT_E_P \
+	     '$': if (!allow_dollar_in_symbol) goto dollar_sign; \
 	case 'a':         \
 	case 'b':         \
 	case 'c':         \
 	case 'd':         \
-	case 'e':         \
 	case 'f':         \
 	case 'g':         \
 	case 'h':         \
@@ -196,7 +197,6 @@ static inline void next_char(void)
 	case 'm':         \
 	case 'n':         \
 	case 'o':         \
-	case 'p':         \
 	case 'q':         \
 	case 'r':         \
 	case 's':         \
@@ -211,7 +211,6 @@ static inline void next_char(void)
 	case 'B':         \
 	case 'C':         \
 	case 'D':         \
-	case 'E':         \
 	case 'F':         \
 	case 'G':         \
 	case 'H':         \
@@ -222,7 +221,6 @@ static inline void next_char(void)
 	case 'M':         \
 	case 'N':         \
 	case 'O':         \
-	case 'P':         \
 	case 'Q':         \
 	case 'R':         \
 	case 'S':         \
@@ -233,10 +231,20 @@ static inline void next_char(void)
 	case 'X':         \
 	case 'Y':         \
 	case 'Z':         \
-	case '_':
+	case '_'
+
+#define SYMBOL_CHARS_E_P \
+	     'E': \
+	case 'P': \
+	case 'e': \
+	case 'p'
+
+#define SYMBOL_CHARS  \
+	     SYMBOL_CHARS_WITHOUT_E_P: \
+	case SYMBOL_CHARS_E_P
 
 #define DIGITS        \
-	case '0':         \
+	     '0':         \
 	case '1':         \
 	case '2':         \
 	case '3':         \
@@ -245,7 +253,98 @@ static inline void next_char(void)
 	case '6':         \
 	case '7':         \
 	case '8':         \
-	case '9':
+	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
@@ -253,17 +352,42 @@ static inline void next_char(void)
  */
 static void parse_symbol(void)
 {
-	obstack_1grow(&symbol_obstack, (char) c);
-	next_char();
-
 	while (true) {
 		switch (c) {
-		DIGITS
-		SYMBOL_CHARS
+		case DIGITS:
+		case SYMBOL_CHARS:
 			obstack_1grow(&symbol_obstack, (char) c);
 			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_utf8(&symbol_obstack, v);
+				}
+				break;
+			}
+
+			default:
+				put_back(c);
+				c = '\\';
+				goto end_symbol;
+			}
+
 		default:
 dollar_sign:
 			goto end_symbol;
@@ -276,16 +400,20 @@ end_symbol:
 	char     *string = obstack_finish(&symbol_obstack);
 	symbol_t *symbol = symbol_table_insert(string);
 
-	lexer_token.kind              = symbol->ID;
-	lexer_token.identifier.symbol = symbol;
+	lexer_token.kind        = symbol->ID;
+	lexer_token.base.symbol = symbol;
 
 	if (symbol->string != string) {
 		obstack_free(&symbol_obstack, string);
 	}
 }
 
-static string_t identify_string(char *string, size_t len)
+static string_t sym_make_string(string_encoding_t const enc)
 {
+	obstack_1grow(&symbol_obstack, '\0');
+	size_t const len    = obstack_object_size(&symbol_obstack) - 1;
+	char  *const string = obstack_finish(&symbol_obstack);
+
 	/* TODO hash */
 #if 0
 	const char *result = strset_insert(&stringset, concat);
@@ -295,110 +423,32 @@ static string_t identify_string(char *string, size_t len)
 #else
 	const char *result = string;
 #endif
-	return (string_t) {result, len};
+	return (string_t){ result, len, enc };
 }
 
-/**
- * parse suffixes like 'LU' or 'f' after numbers
- */
-static void parse_number_suffix(void)
+static void parse_pp_number(void)
 {
-	assert(obstack_object_size(&symbol_obstack) == 0);
-	while (true) {
+	for (;;) {
 		switch (c) {
-		SYMBOL_CHARS
-			obstack_1grow(&symbol_obstack, (char) c);
-			next_char();
-			break;
-		default:
-		dollar_sign:
-			goto finish_suffix;
-		}
-	}
-finish_suffix:
-	if (obstack_object_size(&symbol_obstack) == 0) {
-		lexer_token.number.suffix.begin = NULL;
-		lexer_token.number.suffix.size  = 0;
-		return;
-	}
-
-	obstack_1grow(&symbol_obstack, '\0');
-	size_t size   = obstack_object_size(&symbol_obstack) - 1;
-	char  *string = obstack_finish(&symbol_obstack);
-
-	lexer_token.number.suffix = identify_string(string, size);
-}
-
-static void parse_exponent(void)
-{
-	if (c == '-' || c == '+') {
-		obstack_1grow(&symbol_obstack, (char)c);
-		next_char();
-	}
-
-	if (isdigit(c)) {
-		do {
+		case SYMBOL_CHARS_E_P:
 			obstack_1grow(&symbol_obstack, (char)c);
 			next_char();
-		} while (isdigit(c));
-	} else {
-		errorf(&lexer_token.base.source_position, "exponent has no digits");
-	}
-}
-
-/**
- * Parses a hex number including hex floats and set the
- * lexer_token.
- */
-static void parse_number_hex(void)
-{
-	bool is_float   = false;
-	bool has_digits = false;
-
-	assert(obstack_object_size(&symbol_obstack) == 0);
-	while (isxdigit(c)) {
-		has_digits = true;
-		obstack_1grow(&symbol_obstack, (char) c);
-		next_char();
-	}
-
-	if (c == '.') {
-		is_float = true;
-		obstack_1grow(&symbol_obstack, (char) c);
-		next_char();
+			if (c == '+' || c == '-') {
+		case '.':
+		case DIGITS:
+		case SYMBOL_CHARS_WITHOUT_E_P:
+				obstack_1grow(&symbol_obstack, (char)c);
+				next_char();
+			}
+			break;
 
-		while (isxdigit(c)) {
-			has_digits = true;
-			obstack_1grow(&symbol_obstack, (char) c);
-			next_char();
+		default:
+dollar_sign:
+			lexer_token.kind           = T_NUMBER;
+			lexer_token.literal.string = sym_make_string(STRING_ENCODING_CHAR);
+			return;
 		}
 	}
-	if (c == 'p' || c == 'P') {
-		is_float = true;
-		obstack_1grow(&symbol_obstack, (char) c);
-		next_char();
-		parse_exponent();
-	} else if (is_float) {
-		errorf(&lexer_token.base.source_position,
-		       "hexadecimal floatingpoint constant requires an exponent");
-	}
-	obstack_1grow(&symbol_obstack, '\0');
-
-	size_t  size   = obstack_object_size(&symbol_obstack) - 1;
-	char   *string = obstack_finish(&symbol_obstack);
-	lexer_token.number.number = identify_string(string, size);
-
-	lexer_token.kind    =
-		is_float ? T_FLOATINGPOINT_HEXADECIMAL : T_INTEGER_HEXADECIMAL;
-
-	if (!has_digits) {
-		errorf(&lexer_token.base.source_position,
-		       "invalid number literal '0x%S'", &lexer_token.number.number);
-		lexer_token.number.number.begin = "0";
-		lexer_token.number.number.size  = 1;
-	}
-
-	parse_number_suffix();
 }
 
 /**
@@ -411,81 +461,6 @@ static bool is_octal_digit(utf32 chr)
 	return '0' <= chr && chr <= '7';
 }
 
-/**
- * Parses a number and sets the lexer_token.
- */
-static void parse_number(void)
-{
-	bool is_float   = false;
-	bool has_digits = false;
-
-	assert(obstack_object_size(&symbol_obstack) == 0);
-	if (c == '0') {
-		next_char();
-		if (c == 'x' || c == 'X') {
-			next_char();
-			parse_number_hex();
-			return;
-		} else {
-			has_digits = true;
-		}
-		obstack_1grow(&symbol_obstack, '0');
-	}
-
-	while (isdigit(c)) {
-		has_digits = true;
-		obstack_1grow(&symbol_obstack, (char) c);
-		next_char();
-	}
-
-	if (c == '.') {
-		is_float = true;
-		obstack_1grow(&symbol_obstack, '.');
-		next_char();
-
-		while (isdigit(c)) {
-			has_digits = true;
-			obstack_1grow(&symbol_obstack, (char) c);
-			next_char();
-		}
-	}
-	if (c == 'e' || c == 'E') {
-		is_float = true;
-		obstack_1grow(&symbol_obstack, 'e');
-		next_char();
-		parse_exponent();
-	}
-
-	obstack_1grow(&symbol_obstack, '\0');
-	size_t  size   = obstack_object_size(&symbol_obstack) - 1;
-	char   *string = obstack_finish(&symbol_obstack);
-	lexer_token.number.number = identify_string(string, size);
-
-	/* is it an octal number? */
-	if (is_float) {
-		lexer_token.kind = T_FLOATINGPOINT;
-	} else if (string[0] == '0') {
-		lexer_token.kind = T_INTEGER_OCTAL;
-
-		/* check for invalid octal digits */
-		for (size_t i= 0; i < size; ++i) {
-			char t = string[i];
-			if (t >= '8')
-				errorf(&lexer_token.base.source_position,
-				       "invalid digit '%c' in octal number", t);
-		}
-	} else {
-		lexer_token.kind = T_INTEGER;
-	}
-
-	if (!has_digits) {
-		errorf(&lexer_token.base.source_position, "invalid number literal '%S'",
-		       &lexer_token.number.number);
-	}
-
-	parse_number_suffix();
-}
-
 /**
  * Returns the value of a digit.
  * The only portable way to do it ...
@@ -594,10 +569,10 @@ 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;
 	}
@@ -606,182 +581,71 @@ 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;
-	char *const space = obstack_alloc(&symbol_obstack, len);
-	memcpy(space, string, len);
-
-	return identify_string(space, len);
+	obstack_grow(&symbol_obstack, string, strlen(string));
+	return sym_make_string(STRING_ENCODING_CHAR);
 }
 
-/**
- * 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_utf8(&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();
+		case NEWLINE:
+			errorf(&lexer_pos, "newline while parsing %s", context);
 			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");
-			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_utf8(&symbol_obstack, c);
+				next_char();
+				break;
+			}
 		}
 	}
 
-end_of_wide_char_constant:;
-	obstack_1grow(&symbol_obstack, '\0');
-	size_t  size   = (size_t) obstack_object_size(&symbol_obstack) - 1;
-	char   *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");
-	}
+end_of_string:
+	lexer_token.kind           = kind;
+	lexer_token.literal.string = sym_make_string(enc);
 }
 
 /**
- * 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.literal.string.size == 0) {
 		errorf(&lexer_token.base.source_position, "empty character constant");
 	}
 }
@@ -808,7 +672,8 @@ static void skip_multiline_comment(void)
 			}
 			break;
 
-		MATCH_NEWLINE(break;)
+		case NEWLINE:
+			break;
 
 		case EOF: {
 			errorf(&lexer_token.base.source_position,
@@ -879,20 +744,26 @@ static void eat_until_newline(void)
  */
 static void parse_line_directive(void)
 {
-	if (pp_token.kind != T_INTEGER) {
+	if (pp_token.kind != T_NUMBER) {
 		parse_error("expected integer");
 	} else {
 		/* use offset -1 as this is about the next line */
-		lexer_pos.lineno = atoi(pp_token.number.number.begin) - 1;
+		char      *end;
+		long const line = strtol(pp_token.literal.string.begin, &end, 0);
+		if (*end == '\0') {
+			lexer_pos.lineno = line - 1;
+		} else {
+			errorf(&lexer_pos, "'%S' is not a valid line number", &pp_token.literal.string);
+		}
 		next_pp_token();
 	}
-	if (pp_token.kind == T_STRING_LITERAL) {
-		lexer_pos.input_name = pp_token.string.string.begin;
+	if (pp_token.kind == T_STRING_LITERAL && pp_token.literal.string.encoding == STRING_ENCODING_CHAR) {
+		lexer_pos.input_name       = pp_token.literal.string.begin;
 		lexer_pos.is_system_header = false;
 		next_pp_token();
 
 		/* attempt to parse numeric flags as outputted by gcc preprocessor */
-		while (pp_token.kind == T_INTEGER) {
+		while (pp_token.kind == T_NUMBER) {
 			/* flags:
 			 * 1 - indicates start of a new file
 			 * 2 - indicates return from a file
@@ -901,7 +772,7 @@ static void parse_line_directive(void)
 			 *
 			 * currently we're only interested in "3"
 			 */
-			if (streq(pp_token.number.number.begin, "3")) {
+			if (streq(pp_token.literal.string.begin, "3")) {
 				lexer_pos.is_system_header = true;
 			}
 			next_pp_token();
@@ -936,8 +807,6 @@ typedef enum stdc_pragma_value_kind_t {
  */
 static void parse_pragma(void)
 {
-	bool unknown_pragma = true;
-
 	next_pp_token();
 	if (pp_token.kind != T_IDENTIFIER) {
 		warningf(WARN_UNKNOWN_PRAGMAS, &pp_token.base.source_position,
@@ -946,55 +815,34 @@ static void parse_pragma(void)
 		return;
 	}
 
-	symbol_t *symbol = pp_token.identifier.symbol;
-	if (symbol->pp_ID == TP_STDC) {
-		stdc_pragma_kind_t kind = STDC_UNKNOWN;
+	stdc_pragma_kind_t kind = STDC_UNKNOWN;
+	if (pp_token.base.symbol->pp_ID == TP_STDC && c_mode & _C99) {
 		/* a STDC pragma */
-		if (c_mode & _C99) {
-			next_pp_token();
+		next_pp_token();
 
-			switch (pp_token.identifier.symbol->pp_ID) {
-			case TP_FP_CONTRACT:
-				kind = STDC_FP_CONTRACT;
-				break;
-			case TP_FENV_ACCESS:
-				kind = STDC_FENV_ACCESS;
-				break;
-			case TP_CX_LIMITED_RANGE:
-				kind = STDC_CX_LIMITED_RANGE;
-				break;
-			default:
-				break;
+		switch (pp_token.base.symbol->pp_ID) {
+		case TP_FP_CONTRACT:      kind = STDC_FP_CONTRACT;      break;
+		case TP_FENV_ACCESS:      kind = STDC_FENV_ACCESS;      break;
+		case TP_CX_LIMITED_RANGE: kind = STDC_CX_LIMITED_RANGE; break;
+		default:                  break;
+		}
+		if (kind != STDC_UNKNOWN) {
+			next_pp_token();
+			stdc_pragma_value_kind_t value;
+			switch (pp_token.base.symbol->pp_ID) {
+			case TP_ON:      value = STDC_VALUE_ON;      break;
+			case TP_OFF:     value = STDC_VALUE_OFF;     break;
+			case TP_DEFAULT: value = STDC_VALUE_DEFAULT; break;
+			default:         value = STDC_VALUE_UNKNOWN; break;
 			}
-			if (kind != STDC_UNKNOWN) {
-				stdc_pragma_value_kind_t value = STDC_VALUE_UNKNOWN;
-				next_pp_token();
-				switch (pp_token.identifier.symbol->pp_ID) {
-				case TP_ON:
-					value = STDC_VALUE_ON;
-					break;
-				case TP_OFF:
-					value = STDC_VALUE_OFF;
-					break;
-				case TP_DEFAULT:
-					value = STDC_VALUE_DEFAULT;
-					break;
-				default:
-					break;
-				}
-				if (value != STDC_VALUE_UNKNOWN) {
-					unknown_pragma = false;
-				} else {
-					errorf(&pp_token.base.source_position,
-					       "bad STDC pragma argument");
-				}
+			if (value == STDC_VALUE_UNKNOWN) {
+				kind = STDC_UNKNOWN;
+				errorf(&pp_token.base.source_position, "bad STDC pragma argument");
 			}
 		}
-	} else {
-		unknown_pragma = true;
 	}
 	eat_until_newline();
-	if (unknown_pragma) {
+	if (kind == STDC_UNKNOWN) {
 		warningf(WARN_UNKNOWN_PRAGMAS, &pp_token.base.source_position,
 		         "encountered unknown #pragma");
 	}
@@ -1006,9 +854,7 @@ static void parse_pragma(void)
 static void parse_preprocessor_identifier(void)
 {
 	assert(pp_token.kind == T_IDENTIFIER);
-	symbol_t *symbol = pp_token.identifier.symbol;
-
-	switch (symbol->pp_ID) {
+	switch (pp_token.base.symbol->pp_ID) {
 	case TP_line:
 		next_pp_token();
 		parse_line_directive();
@@ -1034,7 +880,7 @@ static void parse_preprocessor_directive(void)
 	case T_IDENTIFIER:
 		parse_preprocessor_identifier();
 		break;
-	case T_INTEGER:
+	case T_NUMBER:
 		parse_line_directive();
 		break;
 	case '\n':
@@ -1084,6 +930,7 @@ void lexer_next_preprocessing_token(void)
 {
 	while (true) {
 		lexer_token.base.source_position = lexer_pos;
+		lexer_token.base.symbol          = NULL;
 
 		switch (c) {
 		case ' ':
@@ -1091,40 +938,41 @@ void lexer_next_preprocessing_token(void)
 			next_char();
 			break;
 
-		MATCH_NEWLINE(
+		case NEWLINE:
 			lexer_token.kind = '\n';
 			return;
-		)
 
-		SYMBOL_CHARS
+		case SYMBOL_CHARS: {
 			parse_symbol();
 			/* might be a wide string ( L"string" ) */
-			if (lexer_token.identifier.symbol == symbol_L) {
+			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();
+		case DIGITS:
+			parse_pp_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 '.':
 			MAYBE_PROLOG
-				DIGITS
+				case DIGITS:
 					put_back(c);
 					c = '.';
-					parse_number();
+					parse_pp_number();
 					return;
 
 				case '.':
@@ -1231,6 +1079,17 @@ 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 ']':
@@ -1241,7 +1100,6 @@ void lexer_next_preprocessing_token(void)
 		case '~':
 		case ';':
 		case ',':
-		case '\\':
 			lexer_token.kind = c;
 			next_char();
 			return;
@@ -1252,7 +1110,7 @@ void lexer_next_preprocessing_token(void)
 
 		default:
 dollar_sign:
-			errorf(&lexer_pos, "unknown character '%c' found", c);
+			errorf(&lexer_pos, "unknown character '%lc' found", c);
 			next_char();
 			break;
 		}