X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=lexer.c;h=3a2a748cb8b5ca66734a4a7eba6b676b3c05f691;hb=f7f1fcc7021a3c622316b6415c16f83d0a64ae4a;hp=9643ec5953fcc5b62b2d830f64463032e21fe98e;hpb=5c804a26fbf22e58afedba1551d31f63d81d4d84;p=cparser diff --git a/lexer.c b/lexer.c index 9643ec5..3a2a748 100644 --- a/lexer.c +++ b/lexer.c @@ -1,6 +1,6 @@ /* * This file is part of cparser. - * Copyright (C) 2007-2008 Matthias Braun + * 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 @@ -19,6 +19,7 @@ */ #include +#include "input.h" #include "diagnostic.h" #include "lexer.h" #include "symbol_t.h" @@ -40,22 +41,23 @@ #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 -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; +#define MAX_PUTBACK 16 // 3 would be enough, but 16 gives a nicer alignment +#define BUF_SIZE 1024 + +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. @@ -64,7 +66,7 @@ static strset_t stringset; */ static void parse_error(const char *msg) { - errorf(lexer_token.source_position, "%s", msg); + errorf(&lexer_pos, "%s", msg); } /** @@ -74,23 +76,23 @@ static void parse_error(const char *msg) */ static NORETURN internal_error(const char *msg) { - internal_errorf(lexer_token.source_position, "%s", 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; } /** @@ -98,38 +100,32 @@ static inline void next_real_char(void) * * @param pc the character to put back */ -static inline void put_back(int pc) +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++; \ +#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) { eat('\\'); - switch(c) { + switch (c) { MATCH_NEWLINE(return;) default: @@ -149,23 +145,23 @@ 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; @@ -181,14 +177,10 @@ 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': \ @@ -261,14 +253,11 @@ end_of_next_char:; */ static void parse_symbol(void) { - symbol_t *symbol; - char *string; - obstack_1grow(&symbol_obstack, (char) c); next_char(); - while(1) { - switch(c) { + while (true) { + switch (c) { DIGITS SYMBOL_CHARS obstack_1grow(&symbol_obstack, (char) c); @@ -276,6 +265,7 @@ static void parse_symbol(void) break; default: +dollar_sign: goto end_symbol; } } @@ -283,194 +273,60 @@ 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.identifier.symbol = symbol; - if(symbol->string != string) { + if (symbol->string != string) { obstack_free(&symbol_obstack, string); } } -static void parse_integer_suffix(bool is_oct_hex) -{ - bool is_unsigned = false; - bool min_long = false; - bool min_longlong = false; - - 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; - } - } - - 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; - } - lexer_token.datatype = type_unsigned_long_long; - } -} - -static void parse_floating_suffix(void) +static string_t identify_string(char *string, size_t len) { - switch(c) { - /* TODO: do something useful with the suffixes... */ - case 'f': - case 'F': - next_char(); - lexer_token.datatype = type_float; - break; - case 'l': - case 'L': - next_char(); - lexer_token.datatype = type_long_double; - break; - default: - lexer_token.datatype = type_double; - break; + /* 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}; } /** - * A replacement for strtoull. Only those parts needed for - * our parser are implemented. + * parse suffixes like 'LU' or 'f' after numbers */ -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; - } +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; } - break; - default: - assert(0); - break; } -end: - *endptr = s; - return v; +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); + char *string = obstack_finish(&symbol_obstack); + + lexer_token.number.suffix = identify_string(string, size); } /** @@ -479,34 +335,62 @@ end: */ static void parse_number_hex(void) { - assert(c == 'x' || c == 'X'); - next_char(); + bool is_float = false; + bool has_digits = false; - while(isxdigit(c)) { + assert(obstack_object_size(&symbol_obstack) == 0); + 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(); - internal_error("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(); + + if (c == '-' || c == '+') { + obstack_1grow(&symbol_obstack, (char) c); + next_char(); + } + + while (isxdigit(c)) { + obstack_1grow(&symbol_obstack, (char) c); + next_char(); + } + } 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); - 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.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; } - obstack_free(&symbol_obstack, string); - parse_integer_suffix(true); + parse_number_suffix(); } /** @@ -514,157 +398,101 @@ static void parse_number_hex(void) * * @param char the character to check */ -static inline bool is_octal_digit(int chr) +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'; } /** - * Parses a octal number and set the lexer_token. + * Parses a number and sets the lexer_token. */ -static void parse_number_oct(void) +static void parse_number(void) { - while(is_octal_digit(c)) { - obstack_1grow(&symbol_obstack, (char) c); - next_char(); - } - obstack_1grow(&symbol_obstack, '\0'); - char *string = obstack_finish(&symbol_obstack); + bool is_float = false; + bool has_digits = false; - 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"); + 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'); } - obstack_free(&symbol_obstack, string); - parse_integer_suffix(true); -} - -/** - * Parses a decimal including float number and set the - * lexer_token. - */ -static void parse_number_dec(void) -{ - bool is_float = false; - while(isdigit(c)) { + 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 == '+') { + if (c == '-' || c == '+') { obstack_1grow(&symbol_obstack, (char) c); next_char(); } - while(isdigit(c)) { + while (isdigit(c)) { obstack_1grow(&symbol_obstack, (char) c); next_char(); } - is_float = true; } 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"); + 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); } - - parse_floating_suffix(); } 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"); - } - - parse_integer_suffix(false); + lexer_token.kind = T_INTEGER; } - obstack_free(&symbol_obstack, string); -} -/** - * Parses a number and sets the lexer_token. - */ -static void parse_number(void) -{ - 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(); + 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 ... */ -static int digit_value(int digit) { +static int digit_value(utf32 const digit) +{ switch (digit) { case '0': return 0; case '1': return 1; @@ -698,53 +526,43 @@ static int digit_value(int digit) { * * @param first_digit the already read first digit */ -static int parse_octal_sequence(const int first_digit) +static utf32 parse_octal_sequence(utf32 const first_digit) { assert(is_octal_digit(first_digit)); - int value = digit_value(first_digit); + utf32 value = digit_value(first_digit); if (!is_octal_digit(c)) return value; value = 8 * value + digit_value(c); next_char(); if (!is_octal_digit(c)) return value; value = 8 * value + digit_value(c); next_char(); - - if(char_is_signed) { - return (signed char) value; - } else { - return (unsigned char) value; - } + return value; } /** * Parses a hex character sequence. */ -static int parse_hex_sequence(void) +static utf32 parse_hex_sequence(void) { - int value = 0; - while(isxdigit(c)) { + utf32 value = 0; + while (isxdigit(c)) { value = 16 * value + digit_value(c); next_char(); } - - if(char_is_signed) { - return (signed char) value; - } else { - return (unsigned char) value; - } + return value; } /** * Parse an escape sequence. */ -static int parse_escape_sequence(void) +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 '\\'; @@ -770,10 +588,23 @@ 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"); + /* \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; + case 'u': + case 'U': + parse_error("universal character parsing not implemented yet"); return EOF; + default: + break; } + /* §6.4.4.4:8 footnote 64 */ + parse_error("unknown escape sequence"); + return EOF; } /** @@ -788,67 +619,16 @@ string_t concat_strings(const string_t *const s1, const string_t *const s2) memcpy(concat, s1->begin, len1); memcpy(concat + len1, s2->begin, len2 + 1); -#if 0 /* TODO hash */ - const char *result = strset_insert(&stringset, concat); - if(result != concat) { - obstack_free(&symbol_obstack, concat); - } - - return result; -#else - return (string_t){ concat, len1 + len2 + 1 }; -#endif -} - -/** - * Concatenate a string and a wide string. - */ -wide_string_t concat_string_wide_string(const string_t *const s1, const wide_string_t *const s2) -{ - const size_t len1 = s1->size - 1; - const size_t len2 = s2->size - 1; - - wchar_rep_t *const concat = obstack_alloc(&symbol_obstack, (len1 + len2 + 1) * sizeof(*concat)); - const char *const src = s1->begin; - for (size_t i = 0; i != len1; ++i) { - concat[i] = src[i]; - } - memcpy(concat + len1, s2->begin, (len2 + 1) * sizeof(*concat)); - - return (wide_string_t){ concat, len1 + len2 + 1 }; -} - -/** - * Concatenate two wide strings. - */ -wide_string_t concat_wide_strings(const wide_string_t *const s1, const wide_string_t *const s2) -{ - const size_t len1 = s1->size - 1; - const size_t len2 = s2->size - 1; - - wchar_rep_t *const concat = obstack_alloc(&symbol_obstack, (len1 + len2 + 1) * sizeof(*concat)); - memcpy(concat, s1->begin, len1 * sizeof(*concat)); - memcpy(concat + len1, s2->begin, (len2 + 1) * sizeof(*concat)); - - return (wide_string_t){ concat, len1 + len2 + 1 }; + return identify_string(concat, len1 + len2 + 1); } -/** - * Concatenate a wide string and a string. - */ -wide_string_t concat_wide_string_string(const wide_string_t *const s1, const string_t *const s2) +string_t make_string(const char *string) { - const size_t len1 = s1->size - 1; - const size_t len2 = s2->size - 1; + size_t len = strlen(string) + 1; + char *const space = obstack_alloc(&symbol_obstack, len); + memcpy(space, string, len); - wchar_rep_t *const concat = obstack_alloc(&symbol_obstack, (len1 + len2 + 1) * sizeof(*concat)); - memcpy(concat, s1->begin, len1 * sizeof(*concat)); - const char *const src = s2->begin; - for (size_t i = 0; i != len2 + 1; ++i) { - concat[i] = src[i]; - } - - return (wide_string_t){ concat, len1 + len2 + 1 }; + return identify_string(space, len); } /** @@ -856,24 +636,22 @@ wide_string_t concat_wide_string_string(const wide_string_t *const s1, const str */ static void parse_string_literal(void) { - const unsigned start_linenr = lexer_token.source_position.linenr; - eat('"'); - int tc; - while(1) { - switch(c) { - case '\\': - tc = parse_escape_sequence(); - obstack_1grow(&symbol_obstack, (char) tc); + 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; + } 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; } @@ -882,7 +660,7 @@ static void parse_string_literal(void) goto end_of_string; default: - obstack_1grow(&symbol_obstack, (char) c); + obstack_grow_symbol(&symbol_obstack, c); next_char(); break; } @@ -894,22 +672,11 @@ end_of_string: /* 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 + 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.v.string.begin = result; - lexer_token.v.string.size = size; + lexer_token.kind = T_STRING_LITERAL; + lexer_token.string.string = identify_string(string, size); } /** @@ -917,15 +684,13 @@ end_of_string: */ static void parse_wide_character_constant(void) { - const unsigned start_linenr = lexer_token.source_position.linenr; - eat('\''); - while(1) { - switch(c) { + while (true) { + switch (c) { case '\\': { - wchar_rep_t tc = parse_escape_sequence(); - obstack_grow(&symbol_obstack, &tc, sizeof(tc)); + const utf32 tc = parse_escape_sequence(); + obstack_grow_symbol(&symbol_obstack, tc); break; } @@ -939,30 +704,30 @@ 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: { - wchar_rep_t tc = (wchar_rep_t) c; - obstack_grow(&symbol_obstack, &tc, sizeof(tc)); + default: + obstack_grow_symbol(&symbol_obstack, c); next_char(); break; } - } } end_of_wide_char_constant:; - size_t size = (size_t) obstack_object_size(&symbol_obstack); - const wchar_rep_t *string = obstack_finish(&symbol_obstack); + 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); - lexer_token.type = T_WIDE_CHARACTER_CONSTANT; - lexer_token.v.wide_string.begin = string; - lexer_token.v.wide_string.size = size; - lexer_token.datatype = type_wchar_t; + if (size == 0) { + errorf(&lexer_token.base.source_position, "empty character constant"); + } } /** @@ -970,64 +735,9 @@ end_of_wide_char_constant:; */ 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 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; - return; - } - - case '"': - next_char(); - goto end_of_string; - - default: { - wchar_rep_t tc = c; - obstack_grow(&symbol_obstack, &tc, sizeof(tc)); - 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; + parse_string_literal(); + if (lexer_token.kind == T_STRING_LITERAL) + lexer_token.kind = T_WIDE_STRING_LITERAL; } /** @@ -1035,15 +745,16 @@ end_of_string:; */ static void parse_character_constant(void) { - const unsigned start_linenr = lexer_token.source_position.linenr; - eat('\''); - while(1) { - switch(c) { + while (true) { + switch (c) { case '\\': { - int tc = parse_escape_sequence(); - obstack_1grow(&symbol_obstack, (char) tc); + 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; } @@ -1057,16 +768,14 @@ 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: - obstack_1grow(&symbol_obstack, (char) c); + obstack_grow_symbol(&symbol_obstack, c); next_char(); break; @@ -1074,13 +783,16 @@ static void parse_character_constant(void) } end_of_char_constant:; - const size_t size = (size_t)obstack_object_size(&symbol_obstack); - const char *const string = obstack_finish(&symbol_obstack); + 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.type = T_CHARACTER_CONSTANT; - lexer_token.v.string.begin = string; - lexer_token.v.string.size = size; - lexer_token.datatype = type_int; + 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"); + } } /** @@ -1088,19 +800,18 @@ end_of_char_constant:; */ 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 == '*') { - /* TODO: nested comment, warn here */ + /* nested comment, warn here */ + warningf(WARN_COMMENT, &lexer_pos, "'/*' within comment"); } break; case '*': next_char(); - if(c == '/') { + if (c == '/') { next_char(); return; } @@ -1109,10 +820,8 @@ 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; } @@ -1128,8 +837,8 @@ static void skip_multiline_comment(void) */ static void skip_line_comment(void) { - while(1) { - switch(c) { + while (true) { + switch (c) { case EOF: return; @@ -1137,6 +846,14 @@ 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; @@ -1161,56 +878,43 @@ 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 = (unsigned int)(pp_token.v.intvalue - 1); + /* 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.v.string.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 (strcmp(pp_token.number.number.begin, "3") == 0) { + lexer_pos.is_system_header = true; + } + next_pp_token(); + } } eat_until_newline(); @@ -1219,7 +923,7 @@ static void parse_line_directive(void) /** * STDC pragmas. */ -typedef enum { +typedef enum stdc_pragma_kind_t { STDC_UNKNOWN, STDC_FP_CONTRACT, STDC_FENV_ACCESS, @@ -1229,7 +933,7 @@ typedef enum { /** * STDC pragma values. */ -typedef enum { +typedef enum stdc_pragma_value_kind_t { STDC_VALUE_UNKNOWN, STDC_VALUE_ON, STDC_VALUE_OFF, @@ -1239,17 +943,26 @@ typedef enum { /** * Parse a pragma directive. */ -static void parse_pragma(void) { +static void parse_pragma(void) +{ bool unknown_pragma = true; next_pp_token(); - if (pp_token.v.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.v.symbol->pp_ID) { + switch (pp_token.identifier.symbol->pp_ID) { case TP_FP_CONTRACT: kind = STDC_FP_CONTRACT; break; @@ -1265,7 +978,7 @@ static void parse_pragma(void) { if (kind != STDC_UNKNOWN) { stdc_pragma_value_kind_t value = STDC_VALUE_UNKNOWN; next_pp_token(); - switch (pp_token.v.symbol->pp_ID) { + switch (pp_token.identifier.symbol->pp_ID) { case TP_ON: value = STDC_VALUE_ON; break; @@ -1281,7 +994,8 @@ 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"); } } } @@ -1289,8 +1003,9 @@ 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"); } } @@ -1299,40 +1014,21 @@ static void parse_pragma(void) { */ static void parse_preprocessor_identifier(void) { - assert(pp_token.type == T_IDENTIFIER); - symbol_t *symbol = pp_token.v.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; + switch (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: - /* 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; } } @@ -1343,7 +1039,7 @@ static void parse_preprocessor_directive(void) { next_pp_token(); - switch(pp_token.type) { + switch (pp_token.kind) { case T_IDENTIFIER: parse_preprocessor_identifier(); break; @@ -1351,7 +1047,7 @@ static void parse_preprocessor_directive(void) parse_line_directive(); break; case '\n': - /* NULL directive, see § 6.10.7 */ + /* NULL directive, see §6.10.7 */ break; default: parse_error("invalid preprocessor directive"); @@ -1362,51 +1058,60 @@ 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; + + switch (c) { case ' ': case '\t': next_char(); 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.type == T_IDENTIFIER && - lexer_token.v.symbol == symbol_L) { - if(c == '"') { - parse_wide_string_literal(); - } else if(c == '\'') { - parse_wide_character_constant(); + if (lexer_token.identifier.symbol == symbol_L) { + switch (c) { + case '"': parse_wide_string_literal(); break; + case '\'': parse_wide_character_constant(); break; } } return; @@ -1425,19 +1130,10 @@ void lexer_next_preprocessing_token(void) 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 '.': @@ -1446,8 +1142,7 @@ void lexer_next_preprocessing_token(void) ELSE_CODE( put_back(c); c = '.'; - lexer_token.type = '.'; - return; + lexer_token.kind = '.'; ) ELSE('.') case '&': @@ -1500,8 +1195,7 @@ void lexer_next_preprocessing_token(void) ELSE_CODE( put_back(c); c = '%'; - lexer_token.type = '#'; - return; + lexer_token.kind = '#'; ) ELSE('#') ELSE('%') @@ -1535,6 +1229,7 @@ void lexer_next_preprocessing_token(void) case ':': MAYBE_PROLOG MAYBE('>', ']') + MAYBE_MODE(':', T_COLONCOLON, _CXX) ELSE(':') case '=': MAYBE_PROLOG @@ -1556,18 +1251,19 @@ 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_pos, "unknown character '%c' found", c); next_char(); - errorf(lexer_token.source_position, "unknown character '%c' found\n", c); - lexer_token.type = T_ERROR; + lexer_token.kind = T_ERROR; return; } } @@ -1576,15 +1272,13 @@ 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; } @@ -1593,15 +1287,25 @@ newline_found: void init_lexer(void) { strset_init(&stringset); + 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; + lexer_pos.lineno += delta_lines; + lexer_pos.colno += delta_cols; + errorf(&lexer_pos, "%s", message); +} - symbol_L = symbol_table_insert("L"); +void lexer_switch_input(input_t *new_input, const char *input_name) +{ + lexer_pos.lineno = 0; + lexer_pos.colno = 0; + lexer_pos.input_name = input_name; + + set_input_error_callback(input_error); + input = new_input; bufpos = NULL; bufend = NULL; @@ -1618,7 +1322,7 @@ 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); }