*/
#include <config.h>
+#include "adt/strutil.h"
#include "input.h"
#include "diagnostic.h"
#include "lexer.h"
#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"
token_t lexer_token;
static symbol_t *symbol_L;
static strset_t stringset;
-static char *encoding;
bool allow_dollar_in_symbol = true;
/**
char *string = obstack_finish(&symbol_obstack);
symbol_t *symbol = symbol_table_insert(string);
- lexer_token.kind = symbol->ID;
- lexer_token.identifier.symbol = symbol;
+ lexer_token.kind = symbol->ID;
+ lexer_token.base.symbol = symbol;
if (symbol->string != string) {
obstack_free(&symbol_obstack, string);
}
obstack_1grow(&symbol_obstack, '\0');
- size_t size = obstack_object_size(&symbol_obstack);
- char *string = obstack_finish(&symbol_obstack);
+ size_t size = obstack_object_size(&symbol_obstack) - 1;
+ char *string = obstack_finish(&symbol_obstack);
lexer_token.number.suffix = identify_string(string, size);
}
+static void parse_exponent(void)
+{
+ if (c == '-' || c == '+') {
+ obstack_1grow(&symbol_obstack, (char)c);
+ next_char();
+ }
+
+ if (isdigit(c)) {
+ do {
+ obstack_1grow(&symbol_obstack, (char)c);
+ next_char();
+ } while (isdigit(c));
+ } else {
+ errorf(&lexer_token.base.source_position, "exponent has no digits");
+ }
+}
+
/**
* Parses a hex number including hex floats and set the
* lexer_token.
bool is_float = false;
bool has_digits = false;
- assert(obstack_object_size(&symbol_obstack) == 0);
while (isxdigit(c)) {
has_digits = true;
obstack_1grow(&symbol_obstack, (char) c);
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();
- }
+ parse_exponent();
} else if (is_float) {
errorf(&lexer_token.base.source_position,
"hexadecimal floatingpoint constant requires an exponent");
char *string = obstack_finish(&symbol_obstack);
lexer_token.number.number = identify_string(string, size);
- lexer_token.kind =
- is_float ? T_FLOATINGPOINT_HEXADECIMAL : T_INTEGER_HEXADECIMAL;
+ lexer_token.kind = is_float ? T_FLOATINGPOINT : T_INTEGER;
if (!has_digits) {
- errorf(&lexer_token.base.source_position,
- "invalid number literal '0x%S'", &lexer_token.number.number);
+ 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;
+ }
+
+ parse_number_suffix();
+}
+
+static void parse_number_bin(void)
+{
+ bool has_digits = false;
+
+ while (c == '0' || c == '1') {
+ has_digits = true;
+ obstack_1grow(&symbol_obstack, (char)c);
+ next_char();
+ }
+ obstack_1grow(&symbol_obstack, '\0');
+
+ size_t const size = obstack_object_size(&symbol_obstack) - 1;
+ char *const string = obstack_finish(&symbol_obstack);
+ lexer_token.number.number = identify_string(string, size);
+ 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;
}
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 {
- has_digits = true;
+ } 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;
}
- obstack_1grow(&symbol_obstack, '0');
+ has_digits = true;
}
while (isdigit(c)) {
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();
- }
+ parse_exponent();
}
obstack_1grow(&symbol_obstack, '\0');
char *string = obstack_finish(&symbol_obstack);
lexer_token.number.number = identify_string(string, size);
- /* is it an octal number? */
if (is_float) {
lexer_token.kind = T_FLOATINGPOINT;
- } else if (string[0] == '0') {
- lexer_token.kind = T_INTEGER_OCTAL;
-
- /* check for invalid octal digits */
- for (size_t i= 0; i < size; ++i) {
- char t = string[i];
- if (t >= '8')
- errorf(&lexer_token.base.source_position,
- "invalid digit '%c' in octal number", t);
- }
} else {
lexer_token.kind = T_INTEGER;
+
+ if (string[0] == '0') {
+ /* 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);
+ }
+ }
}
if (!has_digits) {
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.
*/
break;
}
- case EOF: {
+ case EOF:
errorf(&lexer_token.base.source_position, "string has no end");
- lexer_token.kind = T_ERROR;
- return;
- }
+ goto end_of_string;
case '"':
next_char();
goto end_of_string;
default:
- grow_symbol(c);
+ obstack_grow_symbol(&symbol_obstack, c);
next_char();
break;
}
switch (c) {
case '\\': {
const utf32 tc = parse_escape_sequence();
- grow_symbol(tc);
+ obstack_grow_symbol(&symbol_obstack, tc);
break;
}
next_char();
goto end_of_wide_char_constant;
- case EOF: {
- errorf(&lexer_token.base.source_position,
- "EOF while parsing character constant");
- lexer_token.kind = T_ERROR;
- return;
- }
+ case EOF:
+ errorf(&lexer_token.base.source_position, "EOF while parsing character constant");
+ goto end_of_wide_char_constant;
default:
- grow_symbol(c);
+ obstack_grow_symbol(&symbol_obstack, c);
next_char();
break;
}
next_char();
goto end_of_char_constant;
- case EOF: {
- errorf(&lexer_token.base.source_position,
- "EOF while parsing character constant");
- lexer_token.kind = T_ERROR;
- return;
- }
+ case EOF:
+ errorf(&lexer_token.base.source_position, "EOF while parsing character constant");
+ goto end_of_char_constant;
default:
- grow_symbol(c);
+ obstack_grow_symbol(&symbol_obstack, c);
next_char();
break;
}
}
-/**
- * Handle the define directive.
- */
-static void define_directive(void)
-{
- lexer_next_preprocessing_token();
- if (lexer_token.kind != 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.
*/
}
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();
*/
static void parse_pragma(void)
{
- bool unknown_pragma = true;
-
next_pp_token();
if (pp_token.kind != T_IDENTIFIER) {
warningf(WARN_UNKNOWN_PRAGMAS, &pp_token.base.source_position,
return;
}
- symbol_t *symbol = pp_token.identifier.symbol;
- if (symbol->pp_ID == TP_STDC) {
- stdc_pragma_kind_t kind = STDC_UNKNOWN;
+ stdc_pragma_kind_t kind = STDC_UNKNOWN;
+ if (pp_token.base.symbol->pp_ID == TP_STDC && c_mode & _C99) {
/* a STDC pragma */
- if (c_mode & _C99) {
- next_pp_token();
+ next_pp_token();
- switch (pp_token.identifier.symbol->pp_ID) {
- case TP_FP_CONTRACT:
- kind = STDC_FP_CONTRACT;
- break;
- case TP_FENV_ACCESS:
- kind = STDC_FENV_ACCESS;
- break;
- case TP_CX_LIMITED_RANGE:
- kind = STDC_CX_LIMITED_RANGE;
- break;
- default:
- break;
+ switch (pp_token.base.symbol->pp_ID) {
+ case TP_FP_CONTRACT: kind = STDC_FP_CONTRACT; break;
+ case TP_FENV_ACCESS: kind = STDC_FENV_ACCESS; break;
+ case TP_CX_LIMITED_RANGE: kind = STDC_CX_LIMITED_RANGE; break;
+ default: break;
+ }
+ if (kind != STDC_UNKNOWN) {
+ next_pp_token();
+ stdc_pragma_value_kind_t value;
+ switch (pp_token.base.symbol->pp_ID) {
+ case TP_ON: value = STDC_VALUE_ON; break;
+ case TP_OFF: value = STDC_VALUE_OFF; break;
+ case TP_DEFAULT: value = STDC_VALUE_DEFAULT; break;
+ default: value = STDC_VALUE_UNKNOWN; break;
}
- if (kind != STDC_UNKNOWN) {
- stdc_pragma_value_kind_t value = STDC_VALUE_UNKNOWN;
- next_pp_token();
- switch (pp_token.identifier.symbol->pp_ID) {
- case TP_ON:
- value = STDC_VALUE_ON;
- break;
- case TP_OFF:
- value = STDC_VALUE_OFF;
- break;
- case TP_DEFAULT:
- value = STDC_VALUE_DEFAULT;
- break;
- default:
- break;
- }
- if (value != STDC_VALUE_UNKNOWN) {
- unknown_pragma = false;
- } else {
- errorf(&pp_token.base.source_position,
- "bad STDC pragma argument");
- }
+ if (value == STDC_VALUE_UNKNOWN) {
+ kind = STDC_UNKNOWN;
+ errorf(&pp_token.base.source_position, "bad STDC pragma argument");
}
}
- } else {
- unknown_pragma = true;
}
eat_until_newline();
- if (unknown_pragma) {
+ if (kind == STDC_UNKNOWN) {
warningf(WARN_UNKNOWN_PRAGMAS, &pp_token.base.source_position,
"encountered unknown #pragma");
}
static void parse_preprocessor_identifier(void)
{
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 (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:
- /* 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;
}
}
{
while (true) {
lexer_token.base.source_position = lexer_pos;
+ lexer_token.base.symbol = NULL;
switch (c) {
case ' ':
SYMBOL_CHARS
parse_symbol();
/* might be a wide string ( L"string" ) */
- if (lexer_token.identifier.symbol == symbol_L) {
+ if (lexer_token.base.symbol == symbol_L) {
switch (c) {
case '"': parse_wide_string_literal(); break;
case '\'': parse_wide_character_constant(); break;
dollar_sign:
errorf(&lexer_pos, "unknown character '%c' found", c);
next_char();
- lexer_token.kind = T_ERROR;
- return;
+ break;
}
}
}
errorf(&lexer_pos, "%s", message);
}
-void select_input_encoding(char const* new_encoding)
-{
- if (encoding != NULL)
- xfree(encoding);
- encoding = xstrdup(new_encoding);
-}
-
-void lexer_open_stream(FILE *stream, const char *input_name)
+void lexer_switch_input(input_t *new_input, const char *input_name)
{
- if (input != NULL) {
- input_free(input);
- input = NULL;
- }
-
lexer_pos.lineno = 0;
lexer_pos.colno = 0;
lexer_pos.input_name = input_name;
set_input_error_callback(input_error);
- input = input_from_stream(stream, encoding);
+ input = new_input;
bufpos = NULL;
bufend = NULL;
void exit_lexer(void)
{
- if (input != NULL) {
- input_free(input);
- input = NULL;
- }
strset_destroy(&stringset);
}
void dbg_pos(const source_position_t source_position)
{
fprintf(stdout, "%s:%u:%u\n", source_position.input_name,
- source_position.lineno, source_position.colno);
+ source_position.lineno, (unsigned)source_position.colno);
fflush(stdout);
}