X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=lexer.c;h=d368aed847b9bd0ed00f36e7ea301d6230d4b923;hb=2fb66fd8bd2a5956ab2cad26978ccfb7e105d45f;hp=d7b280978c3aaead5661de210be1dccf536a7d93;hpb=f2dbd6ca8767e8ac0692b9a4f2ad9be4df7dffd6;p=cparser diff --git a/lexer.c b/lexer.c index d7b2809..d368aed 100644 --- a/lexer.c +++ b/lexer.c @@ -1,7 +1,29 @@ +/* + * This file is part of cparser. + * Copyright (C) 2007-2009 Matthias Braun + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA + * 02111-1307, USA. + */ #include +#include "adt/strutil.h" +#include "input.h" #include "diagnostic.h" #include "lexer.h" +#include "symbol_t.h" #include "token_t.h" #include "symbol_table_t.h" #include "adt/error.h" @@ -9,9 +31,9 @@ #include "adt/util.h" #include "types.h" #include "type_t.h" -#include "target_architecture.h" #include "parser.h" #include "warning.h" +#include "lang_features.h" #include #include @@ -19,111 +41,94 @@ #include #include -//#define DEBUG_CHARS -#define MAX_PUTBACK 3 - -#ifdef _WIN32 -/* No strtold on windows and no replacement yet */ -#define strtold(s, e) strtod(s, e) +#ifndef _WIN32 +#include #endif -#if defined HAS_SIGNED_CHAR -typedef signed char char_type; -#elif defined HAS_UNSIGNED_CHAR -typedef unsigned char char_type; -#else -# error signedness of char not determined -#endif +#define MAX_PUTBACK 16 // 3 would be enough, but 16 gives a nicer alignment +#define BUF_SIZE 1024 -static int c; -token_t lexer_token; -symbol_t *symbol_L; -static FILE *input; -static char buf[1024 + MAX_PUTBACK]; -static const char *bufend; -static const char *bufpos; -static strset_t stringset; +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; /** - * Print an error prefix at the given coordinates. + * Prints a parse error message at the current token. * - * @param input_name the input file name - * @param linenr the line number - */ -static void error_prefix_at(const char *input_name, unsigned linenr) -{ - fprintf(stderr, "%s:%u: Error: ", input_name, linenr); -} - -/** - * Print an error prefix at the current token coordinates. + * @param msg the error message */ -static void error_prefix(void) +static void parse_error(const char *msg) { - error_prefix_at(lexer_token.source_position.input_name, - lexer_token.source_position.linenr); + errorf(&lexer_pos, "%s", msg); } /** - * Prints a parse error message at the current token. + * Prints an internal error message at the current token. * * @param msg the error message */ -static void parse_error(const char *msg) +static NORETURN internal_error(const char *msg) { - error_prefix(); - fprintf(stderr, "%s\n", msg); + internal_errorf(&lexer_pos, "%s", msg); } static inline void next_real_char(void) { assert(bufpos <= bufend); if (bufpos >= bufend) { - size_t s = fread(buf + MAX_PUTBACK, 1, sizeof(buf) - MAX_PUTBACK, - input); - if(s == 0) { + size_t n = decode(input, input_buf+MAX_PUTBACK, BUF_SIZE); + if (n == 0) { c = EOF; return; } - bufpos = buf + MAX_PUTBACK; - bufend = buf + MAX_PUTBACK + s; + bufpos = input_buf + MAX_PUTBACK; + bufend = bufpos + n; } c = *bufpos++; + ++lexer_pos.colno; } -static inline void put_back(int pc) +/** + * Put a character back into the buffer. + * + * @param pc the character to put back + */ +static inline void put_back(utf32 const pc) { - assert(bufpos > buf); - *(--bufpos - buf + buf) = (char) pc; - -#ifdef DEBUG_CHARS - printf("putback '%c'\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++; \ - code - -#define eat(c_type) do { assert(c == c_type); next_char(); } while(0) +#define NEWLINE \ + '\r': \ + next_char(); \ + if (c == '\n') { \ + case '\n': \ + next_char(); \ + } \ + lexer_pos.lineno++; \ + lexer_pos.colno = 1; \ + goto newline; \ + newline // Let it look like an ordinary case label. + +#define eat(c_type) (assert(c == c_type), next_char()) static void maybe_concat_lines(void) { eat('\\'); - switch(c) { - MATCH_NEWLINE(return;) + switch (c) { + case NEWLINE: + return; default: break; @@ -133,28 +138,32 @@ static void maybe_concat_lines(void) c = '\\'; } +/** + * Set c to the next input character, ie. + * after expanding trigraphs. + */ static inline void next_char(void) { next_real_char(); /* filter trigraphs */ - if(UNLIKELY(c == '\\')) { + if (UNLIKELY(c == '\\')) { maybe_concat_lines(); - goto end_of_next_char; + return; } - if(LIKELY(c != '?')) - goto end_of_next_char; + if (LIKELY(c != '?')) + return; next_real_char(); - if(LIKELY(c != '?')) { + if (LIKELY(c != '?')) { put_back(c); c = '?'; - goto end_of_next_char; + return; } next_real_char(); - switch(c) { + switch (c) { case '=': c = '#'; break; case '(': c = '['; break; case '/': c = '\\'; maybe_concat_lines(); break; @@ -170,14 +179,10 @@ static inline void next_char(void) c = '?'; break; } - -end_of_next_char:; -#ifdef DEBUG_CHARS - printf("nchar '%c'\n", c); -#endif } #define SYMBOL_CHARS \ + case '$': if (!allow_dollar_in_symbol) goto dollar_sign; \ case 'a': \ case 'b': \ case 'c': \ @@ -244,23 +249,141 @@ end_of_next_char:; case '8': \ case '9': -static void parse_symbol(void) +static bool is_universal_char_valid(utf32 const v) { - symbol_t *symbol; - char *string; + /* 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; +} - obstack_1grow(&symbol_obstack, (char) c); - next_char(); +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; +} - while(1) { - switch(c) { +/** + * Read a symbol from the input and build + * the lexer_token. + */ +static void parse_symbol(void) +{ + while (true) { + switch (c) { DIGITS 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_symbol(&symbol_obstack, v); + } + break; + } + + default: + put_back(c); + c = '\\'; + goto end_symbol; + } + default: +dollar_sign: goto end_symbol; } } @@ -268,392 +391,312 @@ static void parse_symbol(void) end_symbol: obstack_1grow(&symbol_obstack, '\0'); - string = obstack_finish(&symbol_obstack); - symbol = symbol_table_insert(string); + char *string = obstack_finish(&symbol_obstack); + symbol_t *symbol = symbol_table_insert(string); - lexer_token.type = symbol->ID; - lexer_token.v.symbol = symbol; + lexer_token.kind = symbol->ID; + lexer_token.base.symbol = symbol; - if(symbol->string != string) { + if (symbol->string != string) { obstack_free(&symbol_obstack, string); } } -static void parse_integer_suffix(bool is_oct_hex) +static string_t sym_make_string(string_encoding_t const enc) { - bool is_unsigned = false; - bool min_long = false; - bool min_longlong = false; + obstack_1grow(&symbol_obstack, '\0'); + size_t const len = obstack_object_size(&symbol_obstack) - 1; + char *const string = obstack_finish(&symbol_obstack); - if(c == 'U' || c == 'u') { - is_unsigned = true; - next_char(); - if(c == 'L' || c == 'l') { - min_long = true; - next_char(); - if(c == 'L' || c == 'l') { - min_longlong = true; - next_char(); - } - } - } else if(c == 'l' || c == 'L') { - min_long = true; - next_char(); - if(c == 'l' || c == 'L') { - min_longlong = true; - next_char(); - if(c == 'u' || c == 'U') { - is_unsigned = true; - next_char(); - } - } else if(c == 'u' || c == 'U') { - is_unsigned = true; - next_char(); - lexer_token.datatype = type_unsigned_long; - } + /* 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, enc }; +} - if(!is_unsigned) { - long long v = lexer_token.v.intvalue; - if(!min_long) { - if(v >= TARGET_INT_MIN && v <= TARGET_INT_MAX) { - lexer_token.datatype = type_int; - return; - } else if(is_oct_hex && v >= 0 && v <= TARGET_UINT_MAX) { - lexer_token.datatype = type_unsigned_int; - return; - } - } - if(!min_longlong) { - if(v >= TARGET_LONG_MIN && v <= TARGET_LONG_MAX) { - lexer_token.datatype = type_long; - return; - } else if(is_oct_hex && v >= 0 && v <= TARGET_ULONG_MAX) { - lexer_token.datatype = type_unsigned_long; - return; - } - } - unsigned long long uv = (unsigned long long) v; - if(is_oct_hex && uv > (unsigned long long) TARGET_LONGLONG_MAX) { - lexer_token.datatype = type_unsigned_long_long; - return; - } - - lexer_token.datatype = type_long_long; - } else { - unsigned long long v = (unsigned long long) lexer_token.v.intvalue; - if(!min_long && v <= TARGET_UINT_MAX) { - lexer_token.datatype = type_unsigned_int; - return; - } - if(!min_longlong && v <= TARGET_ULONG_MAX) { - lexer_token.datatype = type_unsigned_long; - return; +/** + * parse suffixes like 'LU' or 'f' after numbers + */ +static void parse_number_suffix(void) +{ + assert(obstack_object_size(&symbol_obstack) == 0); + while (true) { + switch (c) { + SYMBOL_CHARS + obstack_1grow(&symbol_obstack, (char) c); + next_char(); + break; + default: + dollar_sign: + goto finish_suffix; } - lexer_token.datatype = type_unsigned_long_long; } +finish_suffix: + if (obstack_object_size(&symbol_obstack) == 0) { + lexer_token.number.suffix.begin = NULL; + lexer_token.number.suffix.size = 0; + return; + } + + lexer_token.number.suffix = sym_make_string(STRING_ENCODING_CHAR); } -static void parse_floating_suffix(void) +static void parse_exponent(void) { - switch(c) { - /* TODO: do something usefull with the suffixes... */ - case 'f': - case 'F': - next_char(); - lexer_token.datatype = type_float; - break; - case 'l': - case 'L': + if (c == '-' || c == '+') { + obstack_1grow(&symbol_obstack, (char)c); next_char(); - lexer_token.datatype = type_long_double; - break; - default: - lexer_token.datatype = type_double; - break; } -} -/** - * A replacement for strtoull. Only those parts needed for - * our parser are implemented. - */ -static unsigned long long parse_int_string(const char *s, const char **endptr, int base) { - unsigned long long v = 0; - - switch (base) { - case 16: - for (;; ++s) { - /* check for overrun */ - if (v >= 0x1000000000000000ULL) - break; - switch (tolower(*s)) { - case '0': v <<= 4; break; - case '1': v <<= 4; v |= 0x1; break; - case '2': v <<= 4; v |= 0x2; break; - case '3': v <<= 4; v |= 0x3; break; - case '4': v <<= 4; v |= 0x4; break; - case '5': v <<= 4; v |= 0x5; break; - case '6': v <<= 4; v |= 0x6; break; - case '7': v <<= 4; v |= 0x7; break; - case '8': v <<= 4; v |= 0x8; break; - case '9': v <<= 4; v |= 0x9; break; - case 'a': v <<= 4; v |= 0xa; break; - case 'b': v <<= 4; v |= 0xb; break; - case 'c': v <<= 4; v |= 0xc; break; - case 'd': v <<= 4; v |= 0xd; break; - case 'e': v <<= 4; v |= 0xe; break; - case 'f': v <<= 4; v |= 0xf; break; - default: - goto end; - } - } - break; - case 8: - for (;; ++s) { - /* check for overrun */ - if (v >= 0x2000000000000000ULL) - break; - switch (tolower(*s)) { - case '0': v <<= 3; break; - case '1': v <<= 3; v |= 1; break; - case '2': v <<= 3; v |= 2; break; - case '3': v <<= 3; v |= 3; break; - case '4': v <<= 3; v |= 4; break; - case '5': v <<= 3; v |= 5; break; - case '6': v <<= 3; v |= 6; break; - case '7': v <<= 3; v |= 7; break; - default: - goto end; - } - } - break; - case 10: - for (;; ++s) { - /* check for overrun */ - if (v > 0x1999999999999999ULL) - break; - switch (tolower(*s)) { - case '0': v *= 10; break; - case '1': v *= 10; v += 1; break; - case '2': v *= 10; v += 2; break; - case '3': v *= 10; v += 3; break; - case '4': v *= 10; v += 4; break; - case '5': v *= 10; v += 5; break; - case '6': v *= 10; v += 6; break; - case '7': v *= 10; v += 7; break; - case '8': v *= 10; v += 8; break; - case '9': v *= 10; v += 9; break; - default: - goto end; - } - } - break; - default: - assert(0); - break; + if (isdigit(c)) { + do { + obstack_1grow(&symbol_obstack, (char)c); + next_char(); + } while (isdigit(c)); + } else { + errorf(&lexer_token.base.source_position, "exponent has no digits"); } -end: - *endptr = s; - return v; } +/** + * Parses a hex number including hex floats and set the + * lexer_token. + */ static void parse_number_hex(void) { - assert(c == 'x' || c == 'X'); - next_char(); + bool is_float = false; + bool has_digits = false; - while(isxdigit(c)) { + while (isxdigit(c)) { + has_digits = true; obstack_1grow(&symbol_obstack, (char) c); next_char(); } - obstack_1grow(&symbol_obstack, '\0'); - char *string = obstack_finish(&symbol_obstack); - if(c == '.' || c == 'p' || c == 'P') { + if (c == '.') { + is_float = true; + obstack_1grow(&symbol_obstack, (char) c); next_char(); - panic("Hex floating point numbers not implemented yet"); + + while (isxdigit(c)) { + has_digits = true; + obstack_1grow(&symbol_obstack, (char) c); + next_char(); + } } - if(*string == '\0') { - parse_error("invalid hex number"); - lexer_token.type = T_ERROR; + 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"); } - const char *endptr; - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = parse_int_string(string, &endptr, 16); - if(*endptr != '\0') { - parse_error("hex number literal too long"); + lexer_token.number.number = sym_make_string(STRING_ENCODING_CHAR); + + lexer_token.kind = is_float ? T_FLOATINGPOINT : T_INTEGER; + + if (!has_digits) { + errorf(&lexer_token.base.source_position, "invalid number literal '%S'", &lexer_token.number.number); + lexer_token.number.number.begin = "0"; + lexer_token.number.number.size = 1; } - obstack_free(&symbol_obstack, string); - parse_integer_suffix(true); + parse_number_suffix(); } -static inline bool is_octal_digit(int chr) +static void parse_number_bin(void) { - return '0' <= chr && chr <= '7'; -} + bool has_digits = false; -static void parse_number_oct(void) -{ - while(is_octal_digit(c)) { - obstack_1grow(&symbol_obstack, (char) c); + while (c == '0' || c == '1') { + has_digits = true; + obstack_1grow(&symbol_obstack, (char)c); next_char(); } - obstack_1grow(&symbol_obstack, '\0'); - char *string = obstack_finish(&symbol_obstack); - const char *endptr; - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = parse_int_string(string, &endptr, 8); - if(*endptr != '\0') { - parse_error("octal number literal too long"); + lexer_token.number.number = sym_make_string(STRING_ENCODING_CHAR); + lexer_token.kind = T_INTEGER; + + if (!has_digits) { + errorf(&lexer_token.base.source_position, "invalid number literal '%S'", &lexer_token.number.number); + lexer_token.number.number.begin = "0"; + lexer_token.number.number.size = 1; } - obstack_free(&symbol_obstack, string); - parse_integer_suffix(true); + parse_number_suffix(); } -static void parse_number_dec(void) +/** + * Returns true if the given char is a octal digit. + * + * @param char the character to check + */ +static bool is_octal_digit(utf32 chr) { - bool is_float = false; - while(isdigit(c)) { + 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') { + obstack_1grow(&symbol_obstack, (char)c); + next_char(); + if (c == 'x' || c == 'X') { + obstack_1grow(&symbol_obstack, (char)c); + next_char(); + parse_number_hex(); + return; + } else if (c == 'b' || c == 'B') { + /* GCC extension: binary constant 0x[bB][01]+. */ + obstack_1grow(&symbol_obstack, (char)c); + next_char(); + parse_number_bin(); + return; + } + has_digits = true; + } + + while (isdigit(c)) { + has_digits = true; obstack_1grow(&symbol_obstack, (char) c); next_char(); } - if(c == '.') { + if (c == '.') { + is_float = true; obstack_1grow(&symbol_obstack, '.'); next_char(); - while(isdigit(c)) { + while (isdigit(c)) { + has_digits = true; obstack_1grow(&symbol_obstack, (char) c); next_char(); } - is_float = true; } - if(c == 'e' || c == 'E') { + if (c == 'e' || c == 'E') { + is_float = true; obstack_1grow(&symbol_obstack, 'e'); next_char(); - - if(c == '-' || c == '+') { - obstack_1grow(&symbol_obstack, (char) c); - next_char(); - } - - while(isdigit(c)) { - obstack_1grow(&symbol_obstack, (char) c); - next_char(); - } - is_float = true; + parse_exponent(); } - obstack_1grow(&symbol_obstack, '\0'); - char *string = obstack_finish(&symbol_obstack); - - if(is_float) { - char *endptr; - lexer_token.type = T_FLOATINGPOINT; - lexer_token.v.floatvalue = strtold(string, &endptr); - - if(*endptr != '\0') { - parse_error("invalid number literal"); - } + lexer_token.number.number = sym_make_string(STRING_ENCODING_CHAR); - parse_floating_suffix(); + if (is_float) { + lexer_token.kind = T_FLOATINGPOINT; } else { - const char *endptr; - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = parse_int_string(string, &endptr, 10); - - if(*endptr != '\0') { - parse_error("invalid number literal"); + lexer_token.kind = T_INTEGER; + + if (lexer_token.number.number.begin[0] == '0') { + /* check for invalid octal digits */ + for (size_t i= 0; i < lexer_token.number.number.size; ++i) { + char t = lexer_token.number.number.begin[i]; + if (t >= '8') + errorf(&lexer_token.base.source_position, "invalid digit '%c' in octal number", t); + } } + } - parse_integer_suffix(false); + if (!has_digits) { + errorf(&lexer_token.base.source_position, "invalid number literal '%S'", + &lexer_token.number.number); } - obstack_free(&symbol_obstack, string); + + parse_number_suffix(); } -static void parse_number(void) +/** + * Returns the value of a digit. + * The only portable way to do it ... + */ +static int digit_value(utf32 const digit) { - if (c == '0') { - next_char(); - switch (c) { - case 'X': - case 'x': - parse_number_hex(); - break; - case '0': - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - parse_number_oct(); - break; - case '8': - case '9': - next_char(); - parse_error("invalid octal number"); - lexer_token.type = T_ERROR; - return; - case '.': - case 'e': - case 'E': - default: - obstack_1grow(&symbol_obstack, '0'); - parse_number_dec(); - return; - } - } else { - parse_number_dec(); + switch (digit) { + case '0': return 0; + case '1': return 1; + case '2': return 2; + case '3': return 3; + case '4': return 4; + case '5': return 5; + case '6': return 6; + case '7': return 7; + case '8': return 8; + case '9': return 9; + case 'a': + case 'A': return 10; + case 'b': + case 'B': return 11; + case 'c': + case 'C': return 12; + case 'd': + case 'D': return 13; + case 'e': + case 'E': return 14; + case 'f': + case 'F': return 15; + default: + internal_error("wrong character given"); } } -static int parse_octal_sequence(const int first_digit) +/** + * Parses an octal character sequence. + * + * @param first_digit the already read first digit + */ +static utf32 parse_octal_sequence(utf32 const first_digit) { assert(is_octal_digit(first_digit)); - int value = first_digit - '0'; + utf32 value = digit_value(first_digit); if (!is_octal_digit(c)) return value; - value = 8 * value + c - '0'; + value = 8 * value + digit_value(c); next_char(); if (!is_octal_digit(c)) return value; - value = 8 * value + c - '0'; + value = 8 * value + digit_value(c); next_char(); - return (char_type)value; + return value; } -static int parse_hex_sequence(void) +/** + * Parses a hex character sequence. + */ +static utf32 parse_hex_sequence(void) { - int value = 0; - while(1) { - if (c >= '0' && c <= '9') { - value = 16 * value + c - '0'; - } else if ('A' <= c && c <= 'F') { - value = 16 * value + c - 'A' + 10; - } else if ('a' <= c && c <= 'f') { - value = 16 * value + c - 'a' + 10; - } else { - break; - } + utf32 value = 0; + while (isxdigit(c)) { + value = 16 * value + digit_value(c); next_char(); } - - return (char_type)value; + return value; } -static int parse_escape_sequence(void) +/** + * Parse an escape sequence. + */ +static utf32 parse_escape_sequence(void) { eat('\\'); - int ec = c; + utf32 const ec = c; next_char(); - switch(ec) { + switch (ec) { case '"': return '"'; case '\'': return '\''; case '\\': return '\\'; @@ -679,259 +722,124 @@ static int parse_escape_sequence(void) case EOF: parse_error("reached end of file while parsing escape sequence"); return EOF; - default: - parse_error("unknown escape sequence"); - return EOF; - } -} - -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; + /* \E is not documented, but handled, by GCC. It is acceptable according + * to §6.11.4, whereas \e is not. */ + case 'E': + case 'e': + if (c_mode & _GNUC) + return 27; /* hopefully 27 is ALWAYS the code for ESCAPE */ + break; - char *const concat = obstack_alloc(&symbol_obstack, len1 + len2 + 1); - memcpy(concat, s1->begin, len1); - memcpy(concat + len1, s2->begin, len2 + 1); + case 'U': return parse_universal_char(8); + case 'u': return parse_universal_char(4); -#if 0 /* TODO hash */ - const char *result = strset_insert(&stringset, concat); - if(result != concat) { - obstack_free(&symbol_obstack, concat); + default: + break; } - - return result; -#else - return (string_t){ concat, len1 + len2 + 1 }; -#endif + /* §6.4.4.4:8 footnote 64 */ + parse_error("unknown escape sequence"); + return EOF; } -static void parse_string_literal(void) +string_t make_string(const char *string) { - const unsigned start_linenr = lexer_token.source_position.linenr; - - assert(c == '"'); - next_char(); - - int tc; - while(1) { - switch(c) { - case '\\': - tc = parse_escape_sequence(); - obstack_1grow(&symbol_obstack, (char) tc); - break; - - case EOF: - error_prefix_at(lexer_token.source_position.input_name, - start_linenr); - fprintf(stderr, "string has no end\n"); - lexer_token.type = T_ERROR; - return; - - case '"': - next_char(); - goto end_of_string; - - default: - obstack_1grow(&symbol_obstack, (char) 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); - const char *const string = obstack_finish(&symbol_obstack); - -#if 0 /* TODO hash */ - /* check if there is already a copy of the string */ - result = strset_insert(&stringset, string); - if(result != string) { - obstack_free(&symbol_obstack, string); - } -#else - const char *const result = string; -#endif - - lexer_token.type = T_STRING_LITERAL; - lexer_token.v.string.begin = result; - lexer_token.v.string.size = size; + obstack_grow(&symbol_obstack, string, strlen(string)); + return sym_make_string(STRING_ENCODING_CHAR); } -static void parse_wide_character_constant(void) +static void parse_string(utf32 const delim, token_kind_t const kind, string_encoding_t const enc, char const *const context) { - eat('\''); - - int found_char = 0; - while(1) { - switch(c) { - case '\\': - found_char = parse_escape_sequence(); - break; + eat(delim); - MATCH_NEWLINE( - parse_error("newline while parsing character constant"); - break; - ) - - case '\'': - next_char(); - goto end_of_wide_char_constant; - - case EOF: - parse_error("EOF while parsing character constant"); - lexer_token.type = T_ERROR; - return; - - default: - if(found_char != 0) { - parse_error("more than 1 characters in character " - "constant"); - goto end_of_wide_char_constant; + while (true) { + switch (c) { + case '\\': { + utf32 const tc = parse_escape_sequence(); + if (enc == STRING_ENCODING_CHAR) { + if (tc >= 0x100) { + warningf(WARN_OTHER, &lexer_pos, "escape sequence out of range"); + } + obstack_1grow(&symbol_obstack, tc); } else { - found_char = c; - next_char(); + obstack_grow_symbol(&symbol_obstack, tc); } break; } - } - -end_of_wide_char_constant: - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = found_char; - lexer_token.datatype = type_wchar_t; -} - -static void parse_wide_string_literal(void) -{ - const unsigned start_linenr = lexer_token.source_position.linenr; - - assert(c == '"'); - next_char(); - while(1) { - switch(c) { - case '\\': { - wchar_rep_t tc = parse_escape_sequence(); - obstack_grow(&symbol_obstack, &tc, sizeof(tc)); - break; - } + case NEWLINE: + errorf(&lexer_pos, "newline while parsing %s", context); + break; - case EOF: - error_prefix_at(lexer_token.source_position.input_name, - start_linenr); - fprintf(stderr, "string has no end\n"); - lexer_token.type = T_ERROR; - return; + case EOF: + errorf(&lexer_token.base.source_position, "EOF while parsing %s", context); + goto end_of_string; - case '"': + default: + if (c == delim) { next_char(); goto end_of_string; - - default: { - wchar_rep_t tc = c; - obstack_grow(&symbol_obstack, &tc, sizeof(tc)); + } else { + 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 */ - wchar_rep_t nul = L'\0'; - obstack_grow(&symbol_obstack, &nul, sizeof(nul)); - const size_t size = (size_t)obstack_object_size(&symbol_obstack) / sizeof(wchar_rep_t); - const wchar_rep_t *const string = obstack_finish(&symbol_obstack); - -#if 0 /* TODO hash */ - /* check if there is already a copy of the string */ - const wchar_rep_t *const result = strset_insert(&stringset, string); - if(result != string) { - obstack_free(&symbol_obstack, string); - } -#else - const wchar_rep_t *const result = string; -#endif - - lexer_token.type = T_WIDE_STRING_LITERAL; - lexer_token.v.wide_string.begin = result; - lexer_token.v.wide_string.size = size; +end_of_string: + lexer_token.kind = kind; + lexer_token.string.string = sym_make_string(enc); } -static void parse_character_constant(void) +/** + * Parse a string literal and set lexer_token. + */ +static void parse_string_literal(string_encoding_t const enc) { - eat('\''); - - int found_char = 0; - while(1) { - switch(c) { - case '\\': - found_char = parse_escape_sequence(); - break; - - MATCH_NEWLINE( - parse_error("newline while parsing character constant"); - break; - ) - - case '\'': - next_char(); - goto end_of_char_constant; - - case EOF: - parse_error("EOF while parsing character constant"); - lexer_token.type = T_ERROR; - return; + parse_string('"', T_STRING_LITERAL, enc, "string literal"); +} - default: - if(found_char != 0) { - parse_error("more than 1 characters in character " - "constant"); - goto end_of_char_constant; - } else { - found_char = c; - next_char(); - } - break; - } +/** + * Parse a character constant and set lexer_token. + */ +static void parse_character_constant(string_encoding_t const enc) +{ + parse_string('\'', T_CHARACTER_CONSTANT, enc, "character constant"); + if (lexer_token.string.string.size == 0) { + errorf(&lexer_token.base.source_position, "empty character constant"); } - -end_of_char_constant: - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = found_char; - lexer_token.datatype = type_int; } +/** + * Skip a multiline comment. + */ static void skip_multiline_comment(void) { - unsigned start_linenr = lexer_token.source_position.linenr; - - while(1) { - switch(c) { + while (true) { + switch (c) { + case '/': + next_char(); + if (c == '*') { + /* nested comment, warn here */ + warningf(WARN_COMMENT, &lexer_pos, "'/*' within comment"); + } + break; case '*': next_char(); - if(c == '/') { + if (c == '/') { next_char(); return; } break; - MATCH_NEWLINE(break;) + case NEWLINE: + break; - case EOF: - error_prefix_at(lexer_token.source_position.input_name, - start_linenr); - fprintf(stderr, "at end of file while looking for comment end\n"); + case EOF: { + errorf(&lexer_token.base.source_position, + "at end of file while looking for comment end"); return; + } default: next_char(); @@ -940,10 +848,13 @@ static void skip_multiline_comment(void) } } +/** + * Skip a single line comment. + */ static void skip_line_comment(void) { - while(1) { - switch(c) { + while (true) { + switch (c) { case EOF: return; @@ -951,6 +862,14 @@ static void skip_line_comment(void) case '\r': return; + case '\\': + next_char(); + if (c == '\n' || c == '\r') { + warningf(WARN_COMMENT, &lexer_pos, "multi-line comment"); + return; + } + break; + default: next_char(); break; @@ -958,119 +877,169 @@ static void skip_line_comment(void) } } +/** The current preprocessor token. */ static token_t pp_token; +/** + * Read the next preprocessor token. + */ static inline void next_pp_token(void) { lexer_next_preprocessing_token(); pp_token = lexer_token; } +/** + * Eat all preprocessor tokens until newline. + */ 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(); } } -static void error_directive(void) +/** + * Parse the line directive. + */ +static void parse_line_directive(void) { - error_prefix(); - fprintf(stderr, "#error directive: \n"); - - /* parse pp-tokens until new-line */ -} + if (pp_token.kind != T_INTEGER) { + 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; + next_pp_token(); + } + if (pp_token.kind == T_STRING_LITERAL && pp_token.string.string.encoding == STRING_ENCODING_CHAR) { + lexer_pos.input_name = pp_token.string.string.begin; + lexer_pos.is_system_header = false; + next_pp_token(); -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(); + /* 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(); + } } -} -static void ifdef_directive(int is_ifndef) -{ - (void) is_ifndef; - lexer_next_preprocessing_token(); - //expect_identifier(); - //extect_newline(); + eat_until_newline(); } -static void endif_directive(void) -{ - //expect_newline(); -} +/** + * STDC pragmas. + */ +typedef enum stdc_pragma_kind_t { + STDC_UNKNOWN, + STDC_FP_CONTRACT, + STDC_FENV_ACCESS, + STDC_CX_LIMITED_RANGE +} stdc_pragma_kind_t; -static void parse_line_directive(void) +/** + * STDC pragma values. + */ +typedef enum stdc_pragma_value_kind_t { + STDC_VALUE_UNKNOWN, + STDC_VALUE_ON, + STDC_VALUE_OFF, + STDC_VALUE_DEFAULT +} stdc_pragma_value_kind_t; + +/** + * Parse a pragma directive. + */ +static void parse_pragma(void) { - if(pp_token.type != T_INTEGER) { - parse_error("expected integer"); - } else { - lexer_token.source_position.linenr = (unsigned int)(pp_token.v.intvalue - 1); - next_pp_token(); + next_pp_token(); + if (pp_token.kind != T_IDENTIFIER) { + warningf(WARN_UNKNOWN_PRAGMAS, &pp_token.base.source_position, + "expected identifier after #pragma"); + eat_until_newline(); + return; } - if(pp_token.type == T_STRING_LITERAL) { - lexer_token.source_position.input_name = pp_token.v.string.begin; + + stdc_pragma_kind_t kind = STDC_UNKNOWN; + if (pp_token.base.symbol->pp_ID == TP_STDC && c_mode & _C99) { + /* a STDC pragma */ next_pp_token(); - } + 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 (value == STDC_VALUE_UNKNOWN) { + kind = STDC_UNKNOWN; + errorf(&pp_token.base.source_position, "bad STDC pragma argument"); + } + } + } eat_until_newline(); + if (kind == STDC_UNKNOWN) { + warningf(WARN_UNKNOWN_PRAGMAS, &pp_token.base.source_position, + "encountered unknown #pragma"); + } } +/** + * Parse a preprocessor non-null directive. + */ static void parse_preprocessor_identifier(void) { - assert(pp_token.type == T_IDENTIFIER); - symbol_t *symbol = pp_token.v.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; + assert(pp_token.kind == T_IDENTIFIER); + switch (pp_token.base.symbol->pp_ID) { 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: - error_directive(); - break; case TP_pragma: - if (warning.unknown_pragmas) { - warningf(lexer_token.source_position, "encountered unknown #pragma"); - } - eat_until_newline(); + parse_pragma(); + break; + case TP_error: + /* TODO; output the rest of the line */ + parse_error("#error directive"); break; } } +/** + * Parse a preprocessor directive. + */ static void parse_preprocessor_directive(void) { next_pp_token(); - switch(pp_token.type) { + switch (pp_token.kind) { case T_IDENTIFIER: parse_preprocessor_identifier(); break; case T_INTEGER: parse_line_directive(); break; + case '\n': + /* NULL directive, see §6.10.7 */ + break; default: parse_error("invalid preprocessor directive"); eat_until_newline(); @@ -1080,82 +1049,84 @@ static void parse_preprocessor_directive(void) #define MAYBE_PROLOG \ next_char(); \ - while(1) { \ - switch(c) { + while (true) { \ + switch (c) { #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 */ +#define MAYBE_MODE(ch, set_type, mode) \ + case ch: \ + if (c_mode & mode) { \ + next_char(); \ + lexer_token.kind = set_type; \ + return; \ + } \ + /* fallthrough */ + #define ELSE_CODE(code) \ default: \ - code; \ + code \ + return; \ } \ - } /* end of while(1) */ \ - break; + } /* end of while (true) */ \ #define ELSE(set_type) \ ELSE_CODE( \ - lexer_token.type = set_type; \ - return; \ + lexer_token.kind = set_type; \ ) void lexer_next_preprocessing_token(void) { - while(1) { - switch(c) { + while (true) { + lexer_token.base.source_position = lexer_pos; + lexer_token.base.symbol = NULL; + + switch (c) { case ' ': case '\t': next_char(); break; - MATCH_NEWLINE( - lexer_token.type = '\n'; + case NEWLINE: + lexer_token.kind = '\n'; return; - ) - SYMBOL_CHARS + SYMBOL_CHARS { parse_symbol(); /* might be a wide string ( L"string" ) */ - if(lexer_token.type == T_IDENTIFIER && - lexer_token.v.symbol == symbol_L) { - if(c == '"') { - parse_wide_string_literal(); - } else if(c == '\'') { - parse_wide_character_constant(); + string_encoding_t const enc = STRING_ENCODING_WIDE; + if (lexer_token.base.symbol == symbol_L) { + switch (c) { + 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 '.': MAYBE_PROLOG - case '0': - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - case '8': - case '9': + DIGITS put_back(c); c = '.'; - parse_number_dec(); + parse_number(); return; case '.': @@ -1164,8 +1135,7 @@ void lexer_next_preprocessing_token(void) ELSE_CODE( put_back(c); c = '.'; - lexer_token.type = '.'; - return; + lexer_token.kind = '.'; ) ELSE('.') case '&': @@ -1208,25 +1178,24 @@ void lexer_next_preprocessing_token(void) ELSE('/') case '%': MAYBE_PROLOG - MAYBE('>', T_PERCENTGREATER) + MAYBE('>', '}') MAYBE('=', T_PERCENTEQUAL) case ':': MAYBE_PROLOG case '%': MAYBE_PROLOG - MAYBE(':', T_PERCENTCOLONPERCENTCOLON) + MAYBE(':', T_HASHHASH) ELSE_CODE( put_back(c); c = '%'; - lexer_token.type = T_PERCENTCOLON; - return; + lexer_token.kind = '#'; ) - ELSE(T_PERCENTCOLON) + ELSE('#') ELSE('%') case '<': MAYBE_PROLOG - MAYBE(':', T_LESSCOLON) - MAYBE('%', T_LESSPERCENT) + MAYBE(':', '[') + MAYBE('%', '{') MAYBE('=', T_LESSEQUAL) case '<': MAYBE_PROLOG @@ -1252,7 +1221,8 @@ void lexer_next_preprocessing_token(void) ELSE('|') case ':': MAYBE_PROLOG - MAYBE('>', T_COLONGREATER) + MAYBE('>', ']') + MAYBE_MODE(':', T_COLONCOLON, _CXX) ELSE(':') case '=': MAYBE_PROLOG @@ -1263,6 +1233,17 @@ 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 ']': @@ -1273,21 +1254,19 @@ void lexer_next_preprocessing_token(void) case '~': 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_pos, "unknown character '%c' found", c); next_char(); - error_prefix(); - fprintf(stderr, "unknown character '%c' found\n", c); - lexer_token.type = T_ERROR; - return; + break; } } } @@ -1295,15 +1274,13 @@ void lexer_next_preprocessing_token(void) void lexer_next_token(void) { lexer_next_preprocessing_token(); - if(lexer_token.type != '\n') - return; + while (lexer_token.kind == '\n') { newline_found: - do { lexer_next_preprocessing_token(); - } while(lexer_token.type == '\n'); + } - if(lexer_token.type == '#') { + if (lexer_token.kind == '#') { parse_preprocessor_directive(); goto newline_found; } @@ -1312,15 +1289,25 @@ newline_found: void init_lexer(void) { strset_init(&stringset); + symbol_L = symbol_table_insert("L"); +} + +static void input_error(unsigned delta_lines, unsigned delta_cols, + const char *message) +{ + lexer_pos.lineno += delta_lines; + lexer_pos.colno += delta_cols; + errorf(&lexer_pos, "%s", message); } -void lexer_open_stream(FILE *stream, const char *input_name) +void lexer_switch_input(input_t *new_input, const char *input_name) { - input = stream; - 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; - symbol_L = symbol_table_insert("L"); + set_input_error_callback(input_error); + input = new_input; bufpos = NULL; bufend = NULL; @@ -1337,7 +1324,7 @@ 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); }