X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=lexer.c;h=daeda8a492c11c85ceb2652ed833cffe71fc3205;hb=f619e0f8119ae30ecd0e4a56505b99fdf0de1ee4;hp=4ff847dd21a06cb5f448445694a549ecbd3df601;hpb=4878641d50359e25f113833bf1b9fe0102eef407;p=cparser

diff --git a/lexer.c b/lexer.c
index 4ff847d..daeda8a 100644
--- a/lexer.c
+++ b/lexer.c
@@ -19,6 +19,8 @@
  */
 #include <config.h>
 
+#include "adt/strutil.h"
+#include "input.h"
 #include "diagnostic.h"
 #include "lexer.h"
 #include "symbol_t.h"
@@ -44,24 +46,19 @@
 #include <strings.h>
 #endif
 
-//#define DEBUG_CHARS
-#define MAX_PUTBACK 3
+#define MAX_PUTBACK 16    // 3 would be enough, but 16 gives a nicer alignment
 #define BUF_SIZE    1024
 
-#if defined(_WIN32) || defined(__CYGWIN__)
-/* No strtold on windows and no replacement yet */
-#define strtold(s, e) strtod(s, e)
-#endif
-
-static utf32        c;
-token_t             lexer_token;
-symbol_t           *symbol_L;
-static FILE        *input;
-static utf32        buf[BUF_SIZE + MAX_PUTBACK];
-static const utf32 *bufend;
-static const utf32 *bufpos;
-static strset_t     stringset;
-bool                allow_dollar_in_symbol = true;
+static input_t           *input;
+static utf32              input_buf[BUF_SIZE + MAX_PUTBACK];
+static const utf32       *bufpos;
+static const utf32       *bufend;
+static utf32              c;
+static source_position_t  lexer_pos;
+token_t                   lexer_token;
+static symbol_t          *symbol_L;
+static strset_t           stringset;
+bool                      allow_dollar_in_symbol = true;
 
 /**
  * Prints a parse error message at the current token.
@@ -70,7 +67,7 @@ bool                allow_dollar_in_symbol = true;
  */
 static void parse_error(const char *msg)
 {
-	errorf(&lexer_token.source_position, "%s", msg);
+	errorf(&lexer_pos, "%s", msg);
 }
 
 /**
@@ -80,281 +77,23 @@ static void parse_error(const char *msg)
  */
 static NORETURN internal_error(const char *msg)
 {
-	internal_errorf(&lexer_token.source_position, "%s", msg);
-}
-
-static size_t read_block(unsigned char *const read_buf, size_t const n)
-{
-	size_t const s = fread(read_buf, 1, n, input);
-	if (s == 0) {
-		if (ferror(input))
-			parse_error("read from input failed");
-		buf[MAX_PUTBACK] = EOF;
-		bufpos           = buf + MAX_PUTBACK;
-		bufend           = buf + MAX_PUTBACK + 1;
-	}
-	return s;
-}
-
-static void decode_iso_8859_1(void)
-{
-	unsigned char read_buf[BUF_SIZE];
-	size_t const s = read_block(read_buf, sizeof(read_buf));
-	if (s == 0)
-		return;
-
-	unsigned char const *src = read_buf;
-	unsigned char const *end = read_buf + s;
-	utf32               *dst = buf + MAX_PUTBACK;
-	while (src != end)
-		*dst++ = *src++;
-
-	bufpos = buf + MAX_PUTBACK;
-	bufend = dst;
-}
-
-static void decode_iso_8859_15(void)
-{
-	unsigned char read_buf[BUF_SIZE];
-	size_t const s = read_block(read_buf, sizeof(read_buf));
-	if (s == 0)
-		return;
-
-	unsigned char const *src = read_buf;
-	unsigned char const *end = read_buf + s;
-	utf32               *dst = buf + MAX_PUTBACK;
-	while (src != end) {
-		utf32 tc = *src++;
-		switch (tc) {
-			case 0xA4: tc = 0x20AC; break; // €
-			case 0xA6: tc = 0x0160; break; // Å 
-			case 0xA8: tc = 0x0161; break; // Å¡
-			case 0xB4: tc = 0x017D; break; // Ž
-			case 0xB8: tc = 0x017E; break; // ž
-			case 0xBC: tc = 0x0152; break; // Œ
-			case 0xBD: tc = 0x0153; break; // œ
-			case 0xBE: tc = 0x0178; break; // Ÿ
-		}
-		*dst++ = tc;
-	}
-
-	bufpos = buf + MAX_PUTBACK;
-	bufend = dst;
-}
-
-static void decode_utf8(void)
-{
-	static utf32  part_decoded_min_code;
-	static utf32  part_decoded_char;
-	static size_t part_decoded_rest_len;
-
-	do {
-		unsigned char read_buf[BUF_SIZE];
-		size_t const s = read_block(read_buf, sizeof(read_buf));
-		if (s == 0) {
-			if (part_decoded_rest_len > 0)
-				parse_error("incomplete input char at end of input");
-			return;
-		}
-
-		unsigned char const *src = read_buf;
-		unsigned char const *end = read_buf + s;
-		utf32               *dst = buf + MAX_PUTBACK;
-		utf32                decoded;
-		utf32                min_code;
-
-		if (part_decoded_rest_len != 0) {
-			min_code              = part_decoded_min_code;
-			decoded               = part_decoded_char;
-			size_t const rest_len = part_decoded_rest_len;
-			part_decoded_rest_len = 0;
-			switch (rest_len) {
-				case 4:  goto realign;
-				case 3:  goto three_more;
-				case 2:  goto two_more;
-				default: goto one_more;
-			}
-		}
-
-		while (src != end) {
-			if ((*src & 0x80) == 0) {
-				decoded = *src++;
-			} else if ((*src & 0xE0) == 0xC0) {
-				min_code = 0x80;
-				decoded  = *src++ & 0x1F;
-one_more:
-				if (src == end) {
-					part_decoded_min_code = min_code;
-					part_decoded_char     = decoded;
-					part_decoded_rest_len = 1;
-					break;
-				}
-				if ((*src & 0xC0) == 0x80) {
-					decoded = (decoded << 6) | (*src++ & 0x3F);
-				} else {
-					goto invalid_char;
-				}
-				if (decoded < min_code                      ||
-						decoded > 0x10FFFF                      ||
-						(0xD800 <= decoded && decoded < 0xE000) || // high/low surrogates
-						(0xFDD0 <= decoded && decoded < 0xFDF0) || // noncharacters
-						(decoded & 0xFFFE) == 0xFFFE) {            // noncharacters
-					parse_error("invalid byte sequence in input");
-				}
-			} else if ((*src & 0xF0) == 0xE0) {
-				min_code = 0x800;
-				decoded  = *src++ & 0x0F;
-two_more:
-				if (src == end) {
-					part_decoded_min_code = min_code;
-					part_decoded_char     = decoded;
-					part_decoded_rest_len = 2;
-					break;
-				}
-				if ((*src & 0xC0) == 0x80) {
-					decoded = (decoded << 6) | (*src++ & 0x3F);
-				} else {
-					goto invalid_char;
-				}
-				goto one_more;
-			} else if ((*src & 0xF8) == 0xF0) {
-				min_code = 0x10000;
-				decoded  = *src++ & 0x07;
-three_more:
-				if (src == end) {
-					part_decoded_min_code = min_code;
-					part_decoded_char     = decoded;
-					part_decoded_rest_len = 3;
-					break;
-				}
-				if ((*src & 0xC0) == 0x80) {
-					decoded = (decoded << 6) | (*src++ & 0x3F);
-				} else {
-					goto invalid_char;
-				}
-				goto two_more;
-			} else {
-invalid_char:
-				parse_error("invalid byte sequence in input");
-realign:
-				do {
-					++src;
-					if (src == end) {
-						part_decoded_rest_len = 4;
-						break;
-					}
-				} while ((*src & 0xC0) == 0x80 || (*src & 0xF8) == 0xF8);
-				continue;
-			}
-			*dst++ = decoded;
-		}
-
-		bufpos = buf + MAX_PUTBACK;
-		bufend = dst;
-	} while (bufpos == bufend);
-}
-
-static void decode_windows_1252(void)
-{
-	unsigned char read_buf[BUF_SIZE];
-	size_t const s = read_block(read_buf, sizeof(read_buf));
-	if (s == 0)
-		return;
-
-	unsigned char const *src = read_buf;
-	unsigned char const *end = read_buf + s;
-	utf32               *dst = buf + MAX_PUTBACK;
-	while (src != end) {
-		utf32 tc = *src++;
-		switch (tc) {
-			case 0x80: tc = 0x20AC; break; // €
-			case 0x82: tc = 0x201A; break; // ‚
-			case 0x83: tc = 0x0192; break; // ƒ
-			case 0x84: tc = 0x201E; break; // „
-			case 0x85: tc = 0x2026; break; // …
-			case 0x86: tc = 0x2020; break; // †
-			case 0x87: tc = 0x2021; break; // ‡
-			case 0x88: tc = 0x02C6; break; // ˆ
-			case 0x89: tc = 0x2030; break; // ‰
-			case 0x8A: tc = 0x0160; break; // Å 
-			case 0x8B: tc = 0x2039; break; // ‹
-			case 0x8C: tc = 0x0152; break; // Œ
-			case 0x8E: tc = 0x017D; break; // Ž
-			case 0x91: tc = 0x2018; break; // ‘
-			case 0x92: tc = 0x2019; break; // ’
-			case 0x93: tc = 0x201C; break; // “
-			case 0x94: tc = 0x201D; break; // ”
-			case 0x95: tc = 0x2022; break; // •
-			case 0x96: tc = 0x2013; break; // –
-			case 0x97: tc = 0x2014; break; // —
-			case 0x98: tc = 0x02DC; break; // ˜
-			case 0x99: tc = 0x2122; break; // ™
-			case 0x9A: tc = 0x0161; break; // Å¡
-			case 0x9B: tc = 0x203A; break; // ›
-			case 0x9C: tc = 0x0153; break; // œ
-			case 0x9E: tc = 0x017E; break; // ž
-			case 0x9F: tc = 0x0178; break; // Ÿ
-		}
-		*dst++ = tc;
-	}
-
-	bufpos = buf + MAX_PUTBACK;
-	bufend = dst;
-}
-
-typedef void (*decoder_t)(void);
-
-static decoder_t decoder = decode_utf8;
-
-typedef struct named_decoder_t {
-	char const *name;
-	decoder_t   decoder;
-} named_decoder_t;
-
-static named_decoder_t const decoders[] = {
-	{ "CP819",           decode_iso_8859_1   }, // offical alias
-	{ "IBM819",          decode_iso_8859_1   }, // offical alias
-	{ "ISO-8859-1",      decode_iso_8859_1   }, // offical alias
-	{ "ISO-8859-15",     decode_iso_8859_15  }, // offical name
-	{ "ISO8859-1",       decode_iso_8859_1   },
-	{ "ISO8859-15",      decode_iso_8859_15  },
-	{ "ISO_8859-1",      decode_iso_8859_1   }, // offical alias
-	{ "ISO_8859-15",     decode_iso_8859_15  }, // offical alias
-	{ "ISO_8859-1:1987", decode_iso_8859_1   }, // offical name
-	{ "Latin-9",         decode_iso_8859_15  }, // offical alias
-	{ "UTF-8",           decode_utf8         }, // offical name
-	{ "csISOLatin1",     decode_iso_8859_1   }, // offical alias
-	{ "cp1252",          decode_windows_1252 },
-	{ "iso-ir-100",      decode_iso_8859_1   }, // offical alias
-	{ "l1",              decode_iso_8859_1   }, // offical alias
-	{ "latin1",          decode_iso_8859_1   }, // offical alias
-	{ "windows-1252",    decode_windows_1252 }, // official name
-
-	{ NULL,              NULL                }
-};
-
-void select_input_encoding(char const* const encoding)
-{
-	for (named_decoder_t const *i = decoders; i->name != NULL; ++i) {
-		if (strcasecmp(encoding, i->name) != 0)
-			continue;
-		decoder = i->decoder;
-		return;
-	}
-	fprintf(stderr, "error: input encoding \"%s\" not supported\n", encoding);
+	internal_errorf(&lexer_pos, "%s", msg);
 }
 
 static inline void next_real_char(void)
 {
 	assert(bufpos <= bufend);
 	if (bufpos >= bufend) {
-		if (input == NULL) {
+		size_t n = decode(input, input_buf+MAX_PUTBACK, BUF_SIZE);
+		if (n == 0) {
 			c = EOF;
 			return;
 		}
-		decoder();
+		bufpos = input_buf + MAX_PUTBACK;
+		bufend = bufpos + n;
 	}
 	c = *bufpos++;
+	++lexer_pos.colno;
 }
 
 /**
@@ -364,30 +103,24 @@ static inline void next_real_char(void)
  */
 static inline void put_back(utf32 const pc)
 {
-	assert(bufpos > buf);
-	*(--bufpos - buf + buf) = pc;
-
-#ifdef DEBUG_CHARS
-	printf("putback '%lc'\n", pc);
-#endif
+	*(--bufpos - input_buf + input_buf) = pc;
+	--lexer_pos.colno;
 }
 
 static inline void next_char(void);
 
-#define MATCH_NEWLINE(code)                   \
-	case '\r':                                \
-		next_char();                          \
-		if (c == '\n') {                      \
-			next_char();                      \
-		}                                     \
-		lexer_token.source_position.linenr++; \
-		code                                  \
-	case '\n':                                \
-		next_char();                          \
-		lexer_token.source_position.linenr++; \
+#define MATCH_NEWLINE(code)  \
+	case '\r':               \
+		next_char();         \
+		if (c == '\n') {     \
+	case '\n':               \
+			next_char();     \
+		}                    \
+		lexer_pos.lineno++;  \
+		lexer_pos.colno = 1; \
 		code
 
-#define eat(c_type)  do { assert(c == c_type); next_char(); } while (0)
+#define eat(c_type) (assert(c == c_type), next_char())
 
 static void maybe_concat_lines(void)
 {
@@ -415,17 +148,17 @@ static inline void next_char(void)
 	/* filter trigraphs */
 	if (UNLIKELY(c == '\\')) {
 		maybe_concat_lines();
-		goto end_of_next_char;
+		return;
 	}
 
 	if (LIKELY(c != '?'))
-		goto end_of_next_char;
+		return;
 
 	next_real_char();
 	if (LIKELY(c != '?')) {
 		put_back(c);
 		c = '?';
-		goto end_of_next_char;
+		return;
 	}
 
 	next_real_char();
@@ -445,11 +178,6 @@ static inline void next_char(void)
 		c = '?';
 		break;
 	}
-
-end_of_next_char:;
-#ifdef DEBUG_CHARS
-	printf("nchar '%c'\n", c);
-#endif
 }
 
 #define SYMBOL_CHARS  \
@@ -549,14 +277,28 @@ end_symbol:
 	char     *string = obstack_finish(&symbol_obstack);
 	symbol_t *symbol = symbol_table_insert(string);
 
-	lexer_token.type   = symbol->ID;
-	lexer_token.symbol = symbol;
+	lexer_token.kind              = symbol->ID;
+	lexer_token.identifier.symbol = symbol;
 
 	if (symbol->string != string) {
 		obstack_free(&symbol_obstack, string);
 	}
 }
 
+static string_t identify_string(char *string, size_t len)
+{
+	/* TODO hash */
+#if 0
+	const char *result = strset_insert(&stringset, concat);
+	if (result != concat) {
+		obstack_free(&symbol_obstack, concat);
+	}
+#else
+	const char *result = string;
+#endif
+	return (string_t) {result, len};
+}
+
 /**
  * parse suffixes like 'LU' or 'f' after numbers
  */
@@ -576,32 +318,16 @@ static void parse_number_suffix(void)
 	}
 finish_suffix:
 	if (obstack_object_size(&symbol_obstack) == 0) {
-		lexer_token.symbol = NULL;
+		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);
 	char     *string = obstack_finish(&symbol_obstack);
-	symbol_t *symbol = symbol_table_insert(string);
-
-	if (symbol->string != string) {
-		obstack_free(&symbol_obstack, string);
-	}
-	lexer_token.symbol = symbol;
-}
 
-static string_t identify_string(char *string, size_t len)
-{
-	/* TODO hash */
-#if 0
-	const char *result = strset_insert(&stringset, concat);
-	if (result != concat) {
-		obstack_free(&symbol_obstack, concat);
-	}
-#else
-	const char *result = string;
-#endif
-	return (string_t) {result, len};
+	lexer_token.number.suffix = identify_string(string, size);
 }
 
 /**
@@ -646,23 +372,23 @@ static void parse_number_hex(void)
 			next_char();
 		}
 	} else if (is_float) {
-		errorf(&lexer_token.source_position,
+		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.literal = identify_string(string, size);
+	lexer_token.number.number = identify_string(string, size);
 
-	lexer_token.type    =
+	lexer_token.kind    =
 		is_float ? T_FLOATINGPOINT_HEXADECIMAL : T_INTEGER_HEXADECIMAL;
 
 	if (!has_digits) {
-		errorf(&lexer_token.source_position, "invalid number literal '0x%S'",
-		       &lexer_token.literal);
-		lexer_token.literal.begin = "0";
-		lexer_token.literal.size  = 1;
+		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();
@@ -675,19 +401,7 @@ static void parse_number_hex(void)
  */
 static bool is_octal_digit(utf32 chr)
 {
-	switch (chr) {
-	case '0':
-	case '1':
-	case '2':
-	case '3':
-	case '4':
-	case '5':
-	case '6':
-	case '7':
-		return true;
-	default:
-		return false;
-	}
+	return '0' <= chr && chr <= '7';
 }
 
 /**
@@ -747,28 +461,28 @@ static void parse_number(void)
 	obstack_1grow(&symbol_obstack, '\0');
 	size_t  size   = obstack_object_size(&symbol_obstack) - 1;
 	char   *string = obstack_finish(&symbol_obstack);
-	lexer_token.literal = identify_string(string, size);
+	lexer_token.number.number = identify_string(string, size);
 
 	/* is it an octal number? */
 	if (is_float) {
-		lexer_token.type = T_FLOATINGPOINT;
+		lexer_token.kind = T_FLOATINGPOINT;
 	} else if (string[0] == '0') {
-		lexer_token.type = T_INTEGER_OCTAL;
+		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' || t == '9')
-				errorf(&lexer_token.source_position,
+			if (t >= '8')
+				errorf(&lexer_token.base.source_position,
 				       "invalid digit '%c' in octal number", t);
 		}
 	} else {
-		lexer_token.type = T_INTEGER;
+		lexer_token.kind = T_INTEGER;
 	}
 
 	if (!has_digits) {
-		errorf(&lexer_token.source_position, "invalid number literal '%S'",
-		       &lexer_token.literal);
+		errorf(&lexer_token.base.source_position, "invalid number literal '%S'",
+		       &lexer_token.number.number);
 	}
 
 	parse_number_suffix();
@@ -918,33 +632,11 @@ string_t make_string(const char *string)
 	return identify_string(space, len);
 }
 
-static void grow_symbol(utf32 const tc)
-{
-	struct obstack *const o  = &symbol_obstack;
-	if (tc < 0x80U) {
-		obstack_1grow(o, tc);
-	} else if (tc < 0x800) {
-		obstack_1grow(o, 0xC0 | (tc >> 6));
-		obstack_1grow(o, 0x80 | (tc & 0x3F));
-	} else if (tc < 0x10000) {
-		obstack_1grow(o, 0xE0 | ( tc >> 12));
-		obstack_1grow(o, 0x80 | ((tc >>  6) & 0x3F));
-		obstack_1grow(o, 0x80 | ( tc        & 0x3F));
-	} else {
-		obstack_1grow(o, 0xF0 | ( tc >> 18));
-		obstack_1grow(o, 0x80 | ((tc >> 12) & 0x3F));
-		obstack_1grow(o, 0x80 | ((tc >>  6) & 0x3F));
-		obstack_1grow(o, 0x80 | ( tc        & 0x3F));
-	}
-}
-
 /**
  * Parse a string literal and set lexer_token.
  */
 static void parse_string_literal(void)
 {
-	const unsigned start_linenr = lexer_token.source_position.linenr;
-
 	eat('"');
 
 	while (true) {
@@ -952,19 +644,15 @@ static void parse_string_literal(void)
 		case '\\': {
 			utf32 const tc = parse_escape_sequence();
 			if (tc >= 0x100) {
-				warningf(&lexer_token.source_position,
-						"escape sequence out of range");
+				warningf(WARN_OTHER, &lexer_pos, "escape sequence out of range");
 			}
 			obstack_1grow(&symbol_obstack, tc);
 			break;
 		}
 
 		case EOF: {
-			source_position_t source_position;
-			source_position.input_name = lexer_token.source_position.input_name;
-			source_position.linenr     = start_linenr;
-			errorf(&source_position, "string has no end");
-			lexer_token.type = T_ERROR;
+			errorf(&lexer_token.base.source_position, "string has no end");
+			lexer_token.kind = T_ERROR;
 			return;
 		}
 
@@ -973,7 +661,7 @@ static void parse_string_literal(void)
 			goto end_of_string;
 
 		default:
-			grow_symbol(c);
+			obstack_grow_symbol(&symbol_obstack, c);
 			next_char();
 			break;
 		}
@@ -988,8 +676,8 @@ end_of_string:
 	const size_t  size   = (size_t)obstack_object_size(&symbol_obstack);
 	char         *string = obstack_finish(&symbol_obstack);
 
-	lexer_token.type    = T_STRING_LITERAL;
-	lexer_token.literal = identify_string(string, size);
+	lexer_token.kind          = T_STRING_LITERAL;
+	lexer_token.string.string = identify_string(string, size);
 }
 
 /**
@@ -997,15 +685,13 @@ end_of_string:
  */
 static void parse_wide_character_constant(void)
 {
-	const unsigned start_linenr = lexer_token.source_position.linenr;
-
 	eat('\'');
 
 	while (true) {
 		switch (c) {
 		case '\\': {
 			const utf32 tc = parse_escape_sequence();
-			grow_symbol(tc);
+			obstack_grow_symbol(&symbol_obstack, tc);
 			break;
 		}
 
@@ -1019,15 +705,14 @@ static void parse_wide_character_constant(void)
 			goto end_of_wide_char_constant;
 
 		case EOF: {
-			source_position_t source_position = lexer_token.source_position;
-			source_position.linenr = start_linenr;
-			errorf(&source_position, "EOF while parsing character constant");
-			lexer_token.type = T_ERROR;
+			errorf(&lexer_token.base.source_position,
+			       "EOF while parsing character constant");
+			lexer_token.kind = T_ERROR;
 			return;
 		}
 
 		default:
-			grow_symbol(c);
+			obstack_grow_symbol(&symbol_obstack, c);
 			next_char();
 			break;
 		}
@@ -1038,8 +723,12 @@ end_of_wide_char_constant:;
 	size_t  size   = (size_t) obstack_object_size(&symbol_obstack) - 1;
 	char   *string = obstack_finish(&symbol_obstack);
 
-	lexer_token.type     = T_WIDE_CHARACTER_CONSTANT;
-	lexer_token.literal  = identify_string(string, size);
+	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");
+	}
 }
 
 /**
@@ -1048,8 +737,8 @@ end_of_wide_char_constant:;
 static void parse_wide_string_literal(void)
 {
 	parse_string_literal();
-	if (lexer_token.type == T_STRING_LITERAL)
-		lexer_token.type = T_WIDE_STRING_LITERAL;
+	if (lexer_token.kind == T_STRING_LITERAL)
+		lexer_token.kind = T_WIDE_STRING_LITERAL;
 }
 
 /**
@@ -1057,8 +746,6 @@ static void parse_wide_string_literal(void)
  */
 static void parse_character_constant(void)
 {
-	const unsigned start_linenr = lexer_token.source_position.linenr;
-
 	eat('\'');
 
 	while (true) {
@@ -1066,8 +753,7 @@ static void parse_character_constant(void)
 		case '\\': {
 			utf32 const tc = parse_escape_sequence();
 			if (tc >= 0x100) {
-				warningf(&lexer_token.source_position,
-						"escape sequence out of range");
+				warningf(WARN_OTHER, &lexer_pos, "escape sequence out of range");
 			}
 			obstack_1grow(&symbol_obstack, tc);
 			break;
@@ -1083,16 +769,14 @@ static void parse_character_constant(void)
 			goto end_of_char_constant;
 
 		case EOF: {
-			source_position_t source_position;
-			source_position.input_name = lexer_token.source_position.input_name;
-			source_position.linenr     = start_linenr;
-			errorf(&source_position, "EOF while parsing character constant");
-			lexer_token.type = T_ERROR;
+			errorf(&lexer_token.base.source_position,
+			       "EOF while parsing character constant");
+			lexer_token.kind = T_ERROR;
 			return;
 		}
 
 		default:
-			grow_symbol(c);
+			obstack_grow_symbol(&symbol_obstack, c);
 			next_char();
 			break;
 
@@ -1104,8 +788,12 @@ end_of_char_constant:;
 	const size_t        size   = (size_t)obstack_object_size(&symbol_obstack)-1;
 	char         *const string = obstack_finish(&symbol_obstack);
 
-	lexer_token.type    = T_CHARACTER_CONSTANT;
-	lexer_token.literal = identify_string(string, size);
+	lexer_token.kind          = T_CHARACTER_CONSTANT;
+	lexer_token.string.string = identify_string(string, size);
+
+	if (size == 0) {
+		errorf(&lexer_token.base.source_position, "empty character constant");
+	}
 }
 
 /**
@@ -1113,17 +801,13 @@ end_of_char_constant:;
  */
 static void skip_multiline_comment(void)
 {
-	unsigned start_linenr = lexer_token.source_position.linenr;
-
 	while (true) {
 		switch (c) {
 		case '/':
 			next_char();
 			if (c == '*') {
 				/* nested comment, warn here */
-				if (warning.comment) {
-					warningf(&lexer_token.source_position, "'/*' within comment");
-				}
+				warningf(WARN_COMMENT, &lexer_pos, "'/*' within comment");
 			}
 			break;
 		case '*':
@@ -1137,10 +821,8 @@ static void skip_multiline_comment(void)
 		MATCH_NEWLINE(break;)
 
 		case EOF: {
-			source_position_t source_position;
-			source_position.input_name = lexer_token.source_position.input_name;
-			source_position.linenr     = start_linenr;
-			errorf(&source_position, "at end of file while looking for comment end");
+			errorf(&lexer_token.base.source_position,
+			       "at end of file while looking for comment end");
 			return;
 		}
 
@@ -1168,8 +850,7 @@ static void skip_line_comment(void)
 		case '\\':
 			next_char();
 			if (c == '\n' || c == '\r') {
-				if (warning.comment)
-					warningf(&lexer_token.source_position, "multi-line comment");
+				warningf(WARN_COMMENT, &lexer_pos, "multi-line comment");
 				return;
 			}
 			break;
@@ -1198,56 +879,43 @@ static inline void next_pp_token(void)
  */
 static void eat_until_newline(void)
 {
-	while (pp_token.type != '\n' && pp_token.type != T_EOF) {
+	while (pp_token.kind != '\n' && pp_token.kind != T_EOF) {
 		next_pp_token();
 	}
 }
 
-/**
- * Handle the define directive.
- */
-static void define_directive(void)
-{
-	lexer_next_preprocessing_token();
-	if (lexer_token.type != T_IDENTIFIER) {
-		parse_error("expected identifier after #define\n");
-		eat_until_newline();
-	}
-}
-
-/**
- * Handle the ifdef directive.
- */
-static void ifdef_directive(int is_ifndef)
-{
-	(void) is_ifndef;
-	lexer_next_preprocessing_token();
-	//expect_identifier();
-	//extect_newline();
-}
-
-/**
- * Handle the endif directive.
- */
-static void endif_directive(void)
-{
-	//expect_newline();
-}
-
 /**
  * Parse the line directive.
  */
 static void parse_line_directive(void)
 {
-	if (pp_token.type != T_INTEGER) {
+	if (pp_token.kind != T_INTEGER) {
 		parse_error("expected integer");
 	} else {
-		lexer_token.source_position.linenr = atoi(pp_token.literal.begin);
+		/* use offset -1 as this is about the next line */
+		lexer_pos.lineno = atoi(pp_token.number.number.begin) - 1;
 		next_pp_token();
 	}
-	if (pp_token.type == T_STRING_LITERAL) {
-		lexer_token.source_position.input_name = pp_token.literal.begin;
+	if (pp_token.kind == T_STRING_LITERAL) {
+		lexer_pos.input_name = pp_token.string.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) {
+			/* flags:
+			 * 1 - indicates start of a new file
+			 * 2 - indicates return from a file
+			 * 3 - indicates system header
+			 * 4 - indicates implicit extern "C" in C++ mode
+			 *
+			 * currently we're only interested in "3"
+			 */
+			if (streq(pp_token.number.number.begin, "3")) {
+				lexer_pos.is_system_header = true;
+			}
+			next_pp_token();
+		}
 	}
 
 	eat_until_newline();
@@ -1281,13 +949,21 @@ static void parse_pragma(void)
 	bool unknown_pragma = true;
 
 	next_pp_token();
-	if (pp_token.symbol->pp_ID == TP_STDC) {
+	if (pp_token.kind != T_IDENTIFIER) {
+		warningf(WARN_UNKNOWN_PRAGMAS, &pp_token.base.source_position,
+		         "expected identifier after #pragma");
+		eat_until_newline();
+		return;
+	}
+
+	symbol_t *symbol = pp_token.identifier.symbol;
+	if (symbol->pp_ID == TP_STDC) {
 		stdc_pragma_kind_t kind = STDC_UNKNOWN;
 		/* a STDC pragma */
 		if (c_mode & _C99) {
 			next_pp_token();
 
-			switch (pp_token.symbol->pp_ID) {
+			switch (pp_token.identifier.symbol->pp_ID) {
 			case TP_FP_CONTRACT:
 				kind = STDC_FP_CONTRACT;
 				break;
@@ -1303,7 +979,7 @@ static void parse_pragma(void)
 			if (kind != STDC_UNKNOWN) {
 				stdc_pragma_value_kind_t value = STDC_VALUE_UNKNOWN;
 				next_pp_token();
-				switch (pp_token.symbol->pp_ID) {
+				switch (pp_token.identifier.symbol->pp_ID) {
 				case TP_ON:
 					value = STDC_VALUE_ON;
 					break;
@@ -1319,7 +995,8 @@ static void parse_pragma(void)
 				if (value != STDC_VALUE_UNKNOWN) {
 					unknown_pragma = false;
 				} else {
-					errorf(&pp_token.source_position, "bad STDC pragma argument");
+					errorf(&pp_token.base.source_position,
+					       "bad STDC pragma argument");
 				}
 			}
 		}
@@ -1327,8 +1004,9 @@ static void parse_pragma(void)
 		unknown_pragma = true;
 	}
 	eat_until_newline();
-	if (unknown_pragma && warning.unknown_pragmas) {
-		warningf(&pp_token.source_position, "encountered unknown #pragma");
+	if (unknown_pragma) {
+		warningf(WARN_UNKNOWN_PRAGMAS, &pp_token.base.source_position,
+		         "encountered unknown #pragma");
 	}
 }
 
@@ -1337,40 +1015,21 @@ static void parse_pragma(void)
  */
 static void parse_preprocessor_identifier(void)
 {
-	assert(pp_token.type == T_IDENTIFIER);
-	symbol_t *symbol = pp_token.symbol;
+	assert(pp_token.kind == T_IDENTIFIER);
+	symbol_t *symbol = pp_token.identifier.symbol;
 
 	switch (symbol->pp_ID) {
-	case TP_include:
-		printf("include - enable header name parsing!\n");
-		break;
-	case TP_define:
-		define_directive();
-		break;
-	case TP_ifdef:
-		ifdef_directive(0);
-		break;
-	case TP_ifndef:
-		ifdef_directive(1);
-		break;
-	case TP_endif:
-		endif_directive();
-		break;
 	case TP_line:
 		next_pp_token();
 		parse_line_directive();
 		break;
-	case TP_if:
-	case TP_else:
-	case TP_elif:
-	case TP_undef:
-	case TP_error:
-		/* TODO; output the rest of the line */
-		parse_error("#error directive: ");
-		break;
 	case TP_pragma:
 		parse_pragma();
 		break;
+	case TP_error:
+		/* TODO; output the rest of the line */
+		parse_error("#error directive");
+		break;
 	}
 }
 
@@ -1381,7 +1040,7 @@ static void parse_preprocessor_directive(void)
 {
 	next_pp_token();
 
-	switch (pp_token.type) {
+	switch (pp_token.kind) {
 	case T_IDENTIFIER:
 		parse_preprocessor_identifier();
 		break;
@@ -1406,7 +1065,7 @@ static void parse_preprocessor_directive(void)
 #define MAYBE(ch, set_type)                                \
 				case ch:                                   \
 					next_char();                           \
-					lexer_token.type = set_type;           \
+					lexer_token.kind = set_type;           \
 					return;
 
 /* must use this as last thing */
@@ -1414,7 +1073,7 @@ static void parse_preprocessor_directive(void)
 				case ch:                                   \
 					if (c_mode & mode) {                   \
 						next_char();                       \
-						lexer_token.type = set_type;       \
+						lexer_token.kind = set_type;       \
 						return;                            \
 					}                                      \
 					/* fallthrough */
@@ -1422,19 +1081,20 @@ static void parse_preprocessor_directive(void)
 #define ELSE_CODE(code)                                    \
 				default:                                   \
 					code                                   \
+					return;                                \
 				}                                          \
 			} /* end of while (true) */                    \
-			break;
 
 #define ELSE(set_type)                                     \
 		ELSE_CODE(                                         \
-			lexer_token.type = set_type;                   \
-			return;                                        \
+			lexer_token.kind = set_type;                   \
 		)
 
 void lexer_next_preprocessing_token(void)
 {
 	while (true) {
+		lexer_token.base.source_position = lexer_pos;
+
 		switch (c) {
 		case ' ':
 		case '\t':
@@ -1442,14 +1102,14 @@ void lexer_next_preprocessing_token(void)
 			break;
 
 		MATCH_NEWLINE(
-			lexer_token.type = '\n';
+			lexer_token.kind = '\n';
 			return;
 		)
 
 		SYMBOL_CHARS
 			parse_symbol();
 			/* might be a wide string ( L"string" ) */
-			if (lexer_token.symbol == symbol_L) {
+			if (lexer_token.identifier.symbol == symbol_L) {
 				switch (c) {
 					case '"':  parse_wide_string_literal();     break;
 					case '\'': parse_wide_character_constant(); break;
@@ -1483,8 +1143,7 @@ void lexer_next_preprocessing_token(void)
 					ELSE_CODE(
 						put_back(c);
 						c = '.';
-						lexer_token.type = '.';
-						return;
+						lexer_token.kind = '.';
 					)
 			ELSE('.')
 		case '&':
@@ -1537,8 +1196,7 @@ void lexer_next_preprocessing_token(void)
 							ELSE_CODE(
 								put_back(c);
 								c = '%';
-								lexer_token.type = '#';
-								return;
+								lexer_token.kind = '#';
 							)
 					ELSE('#')
 			ELSE('%')
@@ -1594,19 +1252,19 @@ void lexer_next_preprocessing_token(void)
 		case ';':
 		case ',':
 		case '\\':
-			lexer_token.type = c;
+			lexer_token.kind = c;
 			next_char();
 			return;
 
 		case EOF:
-			lexer_token.type = T_EOF;
+			lexer_token.kind = T_EOF;
 			return;
 
 		default:
 dollar_sign:
-			errorf(&lexer_token.source_position, "unknown character '%c' found", c);
+			errorf(&lexer_pos, "unknown character '%c' found", c);
 			next_char();
-			lexer_token.type = T_ERROR;
+			lexer_token.kind = T_ERROR;
 			return;
 		}
 	}
@@ -1616,12 +1274,12 @@ void lexer_next_token(void)
 {
 	lexer_next_preprocessing_token();
 
-	while (lexer_token.type == '\n') {
+	while (lexer_token.kind == '\n') {
 newline_found:
 		lexer_next_preprocessing_token();
 	}
 
-	if (lexer_token.type == '#') {
+	if (lexer_token.kind == '#') {
 		parse_preprocessor_directive();
 		goto newline_found;
 	}
@@ -1633,34 +1291,24 @@ void init_lexer(void)
 	symbol_L = symbol_table_insert("L");
 }
 
-void lexer_open_stream(FILE *stream, const char *input_name)
+static void input_error(unsigned delta_lines, unsigned delta_cols,
+                        const char *message)
 {
-	input                                  = stream;
-	lexer_token.source_position.linenr     = 0;
-	lexer_token.source_position.input_name = input_name;
-
-	bufpos = NULL;
-	bufend = NULL;
-
-	/* place a virtual \n at the beginning so the lexer knows that we're
-	 * at the beginning of a line */
-	c = '\n';
+	lexer_pos.lineno += delta_lines;
+	lexer_pos.colno  += delta_cols;
+	errorf(&lexer_pos, "%s", message);
 }
 
-void lexer_open_buffer(const char *buffer, size_t len, const char *input_name)
+void lexer_switch_input(input_t *new_input, const char *input_name)
 {
-	input                                  = NULL;
-	lexer_token.source_position.linenr     = 0;
-	lexer_token.source_position.input_name = input_name;
+	lexer_pos.lineno     = 0;
+	lexer_pos.colno      = 0;
+	lexer_pos.input_name = input_name;
 
-#if 0 // TODO
-	bufpos = buffer;
-	bufend = buffer + len;
-#else
-	(void)buffer;
-	(void)len;
-	panic("builtin lexing not done yet");
-#endif
+	set_input_error_callback(input_error);
+	input  = new_input;
+	bufpos = NULL;
+	bufend = NULL;
 
 	/* place a virtual \n at the beginning so the lexer knows that we're
 	 * at the beginning of a line */
@@ -1675,7 +1323,7 @@ void exit_lexer(void)
 static __attribute__((unused))
 void dbg_pos(const source_position_t source_position)
 {
-	fprintf(stdout, "%s:%u\n", source_position.input_name,
-	        source_position.linenr);
+	fprintf(stdout, "%s:%u:%u\n", source_position.input_name,
+	        source_position.lineno, (unsigned)source_position.colno);
 	fflush(stdout);
 }