/* CPP Library - lexical analysis.
- Copyright (C) 2000 Free Software Foundation, Inc.
+ Copyright (C) 2000, 2001, 2002 Free Software Foundation, Inc.
Contributed by Per Bothner, 1994-95.
Based on CCCP program by Paul Rubin, June 1986
Adapted to ANSI C, Richard Stallman, Jan 1987
along with this program; if not, write to the Free Software
Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
-/* This lexer works with a single pass of the file. Recently I
- re-wrote it to minimize the places where we step backwards in the
- input stream, to make future changes to support multi-byte
- character sets fairly straight-forward.
-
- There is now only one routine where we do step backwards:
- skip_escaped_newlines. This routine could probably also be changed
- so that it doesn't need to step back. One possibility is to use a
- trick similar to that used in lex_period and lex_percent. Two
- extra characters might be needed, but skip_escaped_newlines itself
- would probably be the only place that needs to be aware of that,
- and changes to the remaining routines would probably only be needed
- if they process a backslash. */
-
#include "config.h"
#include "system.h"
#include "cpplib.h"
SPELL_OPERATOR = 0,
SPELL_CHAR,
SPELL_IDENT,
+ SPELL_NUMBER,
SPELL_STRING,
SPELL_NONE
};
const unsigned char *name;
};
-const unsigned char *digraph_spellings [] = {U"%:", U"%:%:", U"<:",
- U":>", U"<%", U"%>"};
+static const unsigned char *const digraph_spellings[] =
+{ U"%:", U"%:%:", U"<:", U":>", U"<%", U"%>" };
#define OP(e, s) { SPELL_OPERATOR, U s },
#define TK(e, s) { s, U STRINGX (e) },
-const struct token_spelling token_spellings [N_TTYPES] = {TTYPE_TABLE };
+static const struct token_spelling token_spellings[N_TTYPES] = { TTYPE_TABLE };
#undef OP
#undef TK
#define TOKEN_SPELL(token) (token_spellings[(token)->type].category)
#define TOKEN_NAME(token) (token_spellings[(token)->type].name)
+#define BACKUP() do {buffer->cur = buffer->backup_to;} while (0)
-static cppchar_t handle_newline PARAMS ((cpp_reader *, cppchar_t));
-static cppchar_t skip_escaped_newlines PARAMS ((cpp_reader *, cppchar_t));
+static void handle_newline PARAMS ((cpp_reader *));
+static cppchar_t skip_escaped_newlines PARAMS ((cpp_reader *));
static cppchar_t get_effective_char PARAMS ((cpp_reader *));
static int skip_block_comment PARAMS ((cpp_reader *));
static int skip_line_comment PARAMS ((cpp_reader *));
static void adjust_column PARAMS ((cpp_reader *));
-static void skip_whitespace PARAMS ((cpp_reader *, cppchar_t));
+static int skip_whitespace PARAMS ((cpp_reader *, cppchar_t));
static cpp_hashnode *parse_identifier PARAMS ((cpp_reader *));
static cpp_hashnode *parse_identifier_slow PARAMS ((cpp_reader *,
const U_CHAR *));
static int unescaped_terminator_p PARAMS ((cpp_reader *, const U_CHAR *));
static void parse_string PARAMS ((cpp_reader *, cpp_token *, cppchar_t));
static void unterminated PARAMS ((cpp_reader *, int));
-static int trigraph_ok PARAMS ((cpp_reader *, cppchar_t));
+static bool trigraph_p PARAMS ((cpp_reader *));
static void save_comment PARAMS ((cpp_reader *, cpp_token *, const U_CHAR *));
-static void lex_percent PARAMS ((cpp_reader *, cpp_token *));
-static void lex_dot PARAMS ((cpp_reader *, cpp_token *));
static int name_p PARAMS ((cpp_reader *, const cpp_string *));
static int maybe_read_ucs PARAMS ((cpp_reader *, const unsigned char **,
const unsigned char *, unsigned int *));
static tokenrun *next_tokenrun PARAMS ((tokenrun *));
-static cpp_chunk *new_chunk PARAMS ((unsigned int));
-static int chunk_suitable PARAMS ((cpp_chunk *, unsigned int));
static unsigned int hex_digit_value PARAMS ((unsigned int));
-static _cpp_buff *new_buff PARAMS ((unsigned int));
+static _cpp_buff *new_buff PARAMS ((size_t));
/* Utility routine:
Compares, the token TOKEN to the NUL-terminated string STRING.
TOKEN must be a CPP_NAME. Returns 1 for equal, 0 for unequal. */
-
int
cpp_ideq (token, string)
const cpp_token *token;
return !ustrcmp (NODE_NAME (token->val.node), (const U_CHAR *) string);
}
-/* Call when meeting a newline. Returns the character after the newline
- (or carriage-return newline combination), or EOF. */
-static cppchar_t
-handle_newline (pfile, newline_char)
+/* Call when meeting a newline, assumed to be in buffer->cur[-1].
+ Returns with buffer->cur pointing to the character immediately
+ following the newline (combination). */
+static void
+handle_newline (pfile)
cpp_reader *pfile;
- cppchar_t newline_char;
{
- cpp_buffer *buffer;
- cppchar_t next = EOF;
-
- pfile->line++;
- buffer = pfile->buffer;
- buffer->col_adjust = 0;
- buffer->line_base = buffer->cur;
+ cpp_buffer *buffer = pfile->buffer;
- /* Handle CR-LF and LF-CR combinations, get the next character. */
- if (buffer->cur < buffer->rlimit)
- {
- next = *buffer->cur++;
- if (next + newline_char == '\r' + '\n')
- {
- buffer->line_base = buffer->cur;
- if (buffer->cur < buffer->rlimit)
- next = *buffer->cur++;
- else
- next = EOF;
- }
- }
+ /* Handle CR-LF and LF-CR. Most other implementations (e.g. java)
+ only accept CR-LF; maybe we should fall back to that behaviour? */
+ if (buffer->cur[-1] + buffer->cur[0] == '\r' + '\n')
+ buffer->cur++;
- buffer->read_ahead = next;
- return next;
+ buffer->line_base = buffer->cur;
+ buffer->col_adjust = 0;
+ pfile->line++;
}
-/* Subroutine of skip_escaped_newlines; called when a trigraph is
- encountered. It warns if necessary, and returns true if the
- trigraph should be honoured. FROM_CHAR is the third character of a
- trigraph, and presumed to be the previous character for position
- reporting. */
-static int
-trigraph_ok (pfile, from_char)
+/* Subroutine of skip_escaped_newlines; called when a 3-character
+ sequence beginning with "??" is encountered. buffer->cur points to
+ the second '?'.
+
+ Warn if necessary, and returns true if the sequence forms a
+ trigraph and the trigraph should be honoured. */
+static bool
+trigraph_p (pfile)
cpp_reader *pfile;
- cppchar_t from_char;
{
- int accept = CPP_OPTION (pfile, trigraphs);
-
+ cpp_buffer *buffer = pfile->buffer;
+ cppchar_t from_char = buffer->cur[1];
+ bool accept;
+
+ if (!_cpp_trigraph_map[from_char])
+ return false;
+
+ accept = CPP_OPTION (pfile, trigraphs);
+
/* Don't warn about trigraphs in comments. */
if (CPP_OPTION (pfile, warn_trigraphs) && !pfile->state.lexing_comment)
{
- cpp_buffer *buffer = pfile->buffer;
-
if (accept)
- cpp_warning_with_line (pfile, pfile->line, CPP_BUF_COL (buffer) - 2,
+ cpp_warning_with_line (pfile, pfile->line, CPP_BUF_COL (buffer) - 1,
"trigraph ??%c converted to %c",
(int) from_char,
(int) _cpp_trigraph_map[from_char]);
{
buffer->last_Wtrigraphs = buffer->cur;
cpp_warning_with_line (pfile, pfile->line,
- CPP_BUF_COL (buffer) - 2,
+ CPP_BUF_COL (buffer) - 1,
"trigraph ??%c ignored", (int) from_char);
}
}
return accept;
}
-/* Assumes local variables buffer and result. */
-#define ACCEPT_CHAR(t) \
- do { result->type = t; buffer->read_ahead = EOF; } while (0)
-
-/* When we move to multibyte character sets, add to these something
- that saves and restores the state of the multibyte conversion
- library. This probably involves saving and restoring a "cookie".
- In the case of glibc it is an 8-byte structure, so is not a high
- overhead operation. In any case, it's out of the fast path. */
-#define SAVE_STATE() do { saved_cur = buffer->cur; } while (0)
-#define RESTORE_STATE() do { buffer->cur = saved_cur; } while (0)
-
-/* Skips any escaped newlines introduced by NEXT, which is either a
- '?' or a '\\'. Returns the next character, which will also have
- been placed in buffer->read_ahead. This routine performs
- preprocessing stages 1 and 2 of the ISO C standard. */
+/* Skips any escaped newlines introduced by '?' or a '\\', assumed to
+ lie in buffer->cur[-1]. Returns the next byte, which will be in
+ buffer->cur[-1]. This routine performs preprocessing stages 1 and
+ 2 of the ISO C standard. */
static cppchar_t
-skip_escaped_newlines (pfile, next)
+skip_escaped_newlines (pfile)
cpp_reader *pfile;
- cppchar_t next;
{
cpp_buffer *buffer = pfile->buffer;
+ cppchar_t next = buffer->cur[-1];
/* Only do this if we apply stages 1 and 2. */
if (!buffer->from_stage3)
{
- cppchar_t next1;
const unsigned char *saved_cur;
- int space;
+ cppchar_t next1;
do
{
- if (buffer->cur == buffer->rlimit)
- break;
-
- SAVE_STATE ();
if (next == '?')
{
- next1 = *buffer->cur++;
- if (next1 != '?' || buffer->cur == buffer->rlimit)
- {
- RESTORE_STATE ();
- break;
- }
-
- next1 = *buffer->cur++;
- if (!_cpp_trigraph_map[next1]
- || !trigraph_ok (pfile, next1))
- {
- RESTORE_STATE ();
- break;
- }
-
- /* We have a full trigraph here. */
- next = _cpp_trigraph_map[next1];
- if (next != '\\' || buffer->cur == buffer->rlimit)
+ if (buffer->cur[0] != '?' || !trigraph_p (pfile))
break;
- SAVE_STATE ();
- }
- /* We have a backslash, and room for at least one more character. */
- space = 0;
- do
- {
- next1 = *buffer->cur++;
- if (!is_nvspace (next1))
+ /* Translate the trigraph. */
+ next = _cpp_trigraph_map[buffer->cur[1]];
+ buffer->cur += 2;
+ if (next != '\\')
break;
- space = 1;
}
- while (buffer->cur < buffer->rlimit);
+
+ if (buffer->cur == buffer->rlimit)
+ break;
+
+ /* We have a backslash, and room for at least one more
+ character. Skip horizontal whitespace. */
+ saved_cur = buffer->cur;
+ do
+ next1 = *buffer->cur++;
+ while (is_nvspace (next1) && buffer->cur < buffer->rlimit);
if (!is_vspace (next1))
{
- RESTORE_STATE ();
+ buffer->cur = saved_cur;
break;
}
- if (space && !pfile->state.lexing_comment)
+ if (saved_cur != buffer->cur - 1
+ && !pfile->state.lexing_comment)
cpp_warning (pfile, "backslash and newline separated by space");
- next = handle_newline (pfile, next1);
- if (next == EOF)
- cpp_pedwarn (pfile, "backslash-newline at end of file");
+ handle_newline (pfile);
+ buffer->backup_to = buffer->cur;
+ if (buffer->cur == buffer->rlimit)
+ {
+ cpp_pedwarn (pfile, "backslash-newline at end of file");
+ next = EOF;
+ }
+ else
+ next = *buffer->cur++;
}
while (next == '\\' || next == '?');
}
- buffer->read_ahead = next;
return next;
}
/* Obtain the next character, after trigraph conversion and skipping
- an arbitrary string of escaped newlines. The common case of no
- trigraphs or escaped newlines falls through quickly. */
+ an arbitrarily long string of escaped newlines. The common case of
+ no trigraphs or escaped newlines falls through quickly. On return,
+ buffer->backup_to points to where to return to if the character is
+ not to be processed. */
static cppchar_t
get_effective_char (pfile)
cpp_reader *pfile;
{
+ cppchar_t next;
cpp_buffer *buffer = pfile->buffer;
- cppchar_t next = EOF;
- if (buffer->cur < buffer->rlimit)
- {
- next = *buffer->cur++;
-
- /* '?' can introduce trigraphs (and therefore backslash); '\\'
- can introduce escaped newlines, which we want to skip, or
- UCNs, which, depending upon lexer state, we will handle in
- the future. */
- if (next == '?' || next == '\\')
- next = skip_escaped_newlines (pfile, next);
- }
+ buffer->backup_to = buffer->cur;
+ next = *buffer->cur++;
+ if (__builtin_expect (next == '?' || next == '\\', 0))
+ next = skip_escaped_newlines (pfile);
- buffer->read_ahead = next;
- return next;
+ return next;
}
/* Skip a C-style block comment. We find the end of the comment by
{
prevc = c, c = *buffer->cur++;
- next_char:
/* FIXME: For speed, create a new character class of characters
of interest inside block comments. */
if (c == '?' || c == '\\')
- c = skip_escaped_newlines (pfile, c);
+ c = skip_escaped_newlines (pfile);
/* People like decorating comments with '*', so check for '/'
instead for efficiency. */
break;
/* Warn about potential nested comments, but not if the '/'
- comes immediately before the true comment delimeter.
+ comes immediately before the true comment delimiter.
Don't bother to get it right across escaped newlines. */
if (CPP_OPTION (pfile, warn_comments)
- && buffer->cur != buffer->rlimit)
- {
- prevc = c, c = *buffer->cur++;
- if (c == '*' && buffer->cur != buffer->rlimit)
- {
- prevc = c, c = *buffer->cur++;
- if (c != '/')
- cpp_warning_with_line (pfile, pfile->line,
- CPP_BUF_COL (buffer) - 2,
- "\"/*\" within comment");
- }
- goto next_char;
- }
+ && buffer->cur[0] == '*' && buffer->cur[1] != '/')
+ cpp_warning_with_line (pfile,
+ pfile->line, CPP_BUF_COL (buffer),
+ "\"/*\" within comment");
}
else if (is_vspace (c))
- {
- prevc = c, c = handle_newline (pfile, c);
- goto next_char;
- }
+ handle_newline (pfile);
else if (c == '\t')
adjust_column (pfile);
}
pfile->state.lexing_comment = 0;
- buffer->read_ahead = EOF;
return c != '/' || prevc != '*';
}
-/* Skip a C++ line comment. Handles escaped newlines. Returns
- non-zero if a multiline comment. The following new line, if any,
- is left in buffer->read_ahead. */
+/* Skip a C++ line comment, leaving buffer->cur pointing to the
+ terminating newline. Handles escaped newlines. Returns non-zero
+ if a multiline comment. */
static int
skip_line_comment (pfile)
cpp_reader *pfile;
pfile->state.lexing_comment = 1;
do
{
- c = EOF;
if (buffer->cur == buffer->rlimit)
- break;
+ goto at_eof;
c = *buffer->cur++;
if (c == '?' || c == '\\')
- c = skip_escaped_newlines (pfile, c);
+ c = skip_escaped_newlines (pfile);
}
while (!is_vspace (c));
+ /* Step back over the newline, except at EOF. */
+ buffer->cur--;
+ at_eof:
+
pfile->state.lexing_comment = 0;
- buffer->read_ahead = c; /* Leave any newline for caller. */
return orig_line != pfile->line;
}
/* Skips whitespace, saving the next non-whitespace character.
Adjusts pfile->col_adjust to account for tabs. Without this,
tokens might be assigned an incorrect column. */
-static void
+static int
skip_whitespace (pfile, c)
cpp_reader *pfile;
cppchar_t c;
/* Just \f \v or \0 left. */
else if (c == '\0')
{
+ if (buffer->cur - 1 == buffer->rlimit)
+ return 0;
if (!warned)
{
cpp_warning (pfile, "null character(s) ignored");
"%s in preprocessing directive",
c == '\f' ? "form feed" : "vertical tab");
- c = EOF;
- if (buffer->cur == buffer->rlimit)
- break;
c = *buffer->cur++;
}
/* We only want non-vertical space, i.e. ' ' \t \f \v \0. */
while (is_nvspace (c));
- /* Remember the next character. */
- buffer->read_ahead = c;
+ buffer->cur--;
+ return 1;
}
/* See if the characters of a number token are valid in a name (no
Poisson-like). Second most common case is a new identifier, not
split and no dollar sign. The other possibilities are rare and
have been relegated to parse_identifier_slow. */
-
static cpp_hashnode *
parse_identifier (pfile)
cpp_reader *pfile;
{
cpp_hashnode *result;
- const U_CHAR *cur, *rlimit;
+ const U_CHAR *cur;
/* Fast-path loop. Skim over a normal identifier.
N.B. ISIDNUM does not include $. */
- cur = pfile->buffer->cur - 1;
- rlimit = pfile->buffer->rlimit;
- do
+ cur = pfile->buffer->cur;
+ while (ISIDNUM (*cur))
cur++;
- while (cur < rlimit && ISIDNUM (*cur));
/* Check for slow-path cases. */
- if (cur < rlimit && (*cur == '?' || *cur == '\\' || *cur == '$'))
+ if (*cur == '?' || *cur == '\\' || *cur == '$')
result = parse_identifier_slow (pfile, cur);
else
{
if (c == '$')
saw_dollar++;
- c = EOF;
- if (buffer->cur == buffer->rlimit)
- break;
-
c = *buffer->cur++;
}
/* Potential escaped newline? */
+ buffer->backup_to = buffer->cur - 1;
if (c != '?' && c != '\\')
break;
- c = skip_escaped_newlines (pfile, c);
+ c = skip_escaped_newlines (pfile);
}
while (is_idchar (c));
- /* Remember the next character. */
- buffer->read_ahead = c;
+ /* Step back over the unwanted char. */
+ BACKUP ();
- /* $ is not a identifier character in the standard, but is commonly
+ /* $ is not an identifier character in the standard, but is commonly
accepted as an extension. Don't warn about it in skipped
conditional blocks. */
if (saw_dollar && CPP_PEDANTIC (pfile) && ! pfile->state.skipping)
ht_lookup (pfile->hash_table, obstack_finish (stack), len, HT_ALLOCED);
}
-/* Parse a number, skipping embedded backslash-newlines. */
+/* Parse a number, beginning with character C, skipping embedded
+ backslash-newlines. LEADING_PERIOD is non-zero if there was a "."
+ before C. Place the result in NUMBER. */
static void
parse_number (pfile, number, c, leading_period)
cpp_reader *pfile;
int leading_period;
{
cpp_buffer *buffer = pfile->buffer;
- cpp_pool *pool = &pfile->ident_pool;
unsigned char *dest, *limit;
- dest = POOL_FRONT (pool);
- limit = POOL_LIMIT (pool);
+ dest = BUFF_FRONT (pfile->u_buff);
+ limit = BUFF_LIMIT (pfile->u_buff);
/* Place a leading period. */
if (leading_period)
{
- if (dest >= limit)
- limit = _cpp_next_chunk (pool, 0, &dest);
+ if (dest == limit)
+ {
+ _cpp_extend_buff (pfile, &pfile->u_buff, 1);
+ dest = BUFF_FRONT (pfile->u_buff);
+ limit = BUFF_LIMIT (pfile->u_buff);
+ }
*dest++ = '.';
}
do
{
/* Need room for terminating null. */
- if (dest + 1 >= limit)
- limit = _cpp_next_chunk (pool, 0, &dest);
+ if ((size_t) (limit - dest) < 2)
+ {
+ size_t len_so_far = dest - BUFF_FRONT (pfile->u_buff);
+ _cpp_extend_buff (pfile, &pfile->u_buff, 2);
+ dest = BUFF_FRONT (pfile->u_buff) + len_so_far;
+ limit = BUFF_LIMIT (pfile->u_buff);
+ }
*dest++ = c;
- c = EOF;
- if (buffer->cur == buffer->rlimit)
- break;
-
c = *buffer->cur++;
}
while (is_numchar (c) || c == '.' || VALID_SIGN (c, dest[-1]));
/* Potential escaped newline? */
+ buffer->backup_to = buffer->cur - 1;
if (c != '?' && c != '\\')
break;
- c = skip_escaped_newlines (pfile, c);
+ c = skip_escaped_newlines (pfile);
}
while (is_numchar (c) || c == '.' || VALID_SIGN (c, dest[-1]));
- /* Remember the next character. */
- buffer->read_ahead = c;
+ /* Step back over the unwanted char. */
+ BACKUP ();
/* Null-terminate the number. */
*dest = '\0';
- number->text = POOL_FRONT (pool);
+ number->text = BUFF_FRONT (pfile->u_buff);
number->len = dest - number->text;
- POOL_COMMIT (pool, number->len + 1);
+ BUFF_FRONT (pfile->u_buff) = dest + 1;
}
/* Subroutine of parse_string. Emits error for unterminated strings. */
if (pfile->state.angled_headers)
return 1;
- start = POOL_FRONT (&pfile->ident_pool);
+ start = BUFF_FRONT (pfile->u_buff);
/* An odd number of consecutive backslashes represents an escaped
terminator. */
name. Handles embedded trigraphs and escaped newlines. The stored
string is guaranteed NUL-terminated, but it is not guaranteed that
this is the first NUL since embedded NULs are preserved.
+ Multi-line strings are allowed, but they are deprecated.
- Multi-line strings are allowed, but they are deprecated. */
+ When this function returns, buffer->cur points to the next
+ character to be processed. */
static void
parse_string (pfile, token, terminator)
cpp_reader *pfile;
cppchar_t terminator;
{
cpp_buffer *buffer = pfile->buffer;
- cpp_pool *pool = &pfile->ident_pool;
unsigned char *dest, *limit;
cppchar_t c;
bool warned_nulls = false, warned_multi = false;
- dest = POOL_FRONT (pool);
- limit = POOL_LIMIT (pool);
+ dest = BUFF_FRONT (pfile->u_buff);
+ limit = BUFF_LIMIT (pfile->u_buff);
for (;;)
{
- if (buffer->cur == buffer->rlimit)
- c = EOF;
- else
- c = *buffer->cur++;
-
- have_char:
- /* We need space for the terminating NUL. */
- if (dest >= limit)
- limit = _cpp_next_chunk (pool, 0, &dest);
-
- if (c == EOF)
+ /* We need room for another char, possibly the terminating NUL. */
+ if ((size_t) (limit - dest) < 1)
{
- unterminated (pfile, terminator);
- break;
+ size_t len_so_far = dest - BUFF_FRONT (pfile->u_buff);
+ _cpp_extend_buff (pfile, &pfile->u_buff, 2);
+ dest = BUFF_FRONT (pfile->u_buff) + len_so_far;
+ limit = BUFF_LIMIT (pfile->u_buff);
}
/* Handle trigraphs, escaped newlines etc. */
+ c = *buffer->cur++;
if (c == '?' || c == '\\')
- c = skip_escaped_newlines (pfile, c);
+ c = skip_escaped_newlines (pfile);
- if (c == terminator && unescaped_terminator_p (pfile, dest))
+ if (c == terminator)
{
- c = EOF;
- break;
+ if (unescaped_terminator_p (pfile, dest))
+ break;
}
else if (is_vspace (c))
{
character literals at end of line. This is a kludge
around not knowing where comments are. */
if (CPP_OPTION (pfile, lang) == CLK_ASM && terminator != '>')
- break;
+ {
+ buffer->cur--;
+ break;
+ }
/* Character constants and header names may not extend over
multiple lines. In Standard C, neither may strings.
if (terminator != '"' || pfile->state.angled_headers)
{
unterminated (pfile, terminator);
+ buffer->cur--;
break;
}
pfile->mls_col = token->col;
}
- c = handle_newline (pfile, c);
- *dest++ = '\n';
- goto have_char;
+ handle_newline (pfile);
+ c = '\n';
}
- else if (c == '\0' && !warned_nulls)
+ else if (c == '\0')
{
- warned_nulls = true;
- cpp_warning (pfile, "null character(s) preserved in literal");
+ if (buffer->cur - 1 == buffer->rlimit)
+ {
+ unterminated (pfile, terminator);
+ buffer->cur--;
+ break;
+ }
+ if (!warned_nulls)
+ {
+ warned_nulls = true;
+ cpp_warning (pfile, "null character(s) preserved in literal");
+ }
}
*dest++ = c;
}
- /* Remember the next character. */
- buffer->read_ahead = c;
*dest = '\0';
- token->val.str.text = POOL_FRONT (pool);
- token->val.str.len = dest - token->val.str.text;
- POOL_COMMIT (pool, token->val.str.len + 1);
+ token->val.str.text = BUFF_FRONT (pfile->u_buff);
+ token->val.str.len = dest - BUFF_FRONT (pfile->u_buff);
+ BUFF_FRONT (pfile->u_buff) = dest + 1;
}
/* The stored comment includes the comment start and any terminator. */
unsigned int len;
len = pfile->buffer->cur - from + 1; /* + 1 for the initial '/'. */
+
/* C++ comments probably (not definitely) have moved past a new
line, which we don't want to save in the comment. */
- if (pfile->buffer->read_ahead != EOF)
+ if (is_vspace (pfile->buffer->cur[-1]))
len--;
- buffer = _cpp_pool_alloc (&pfile->ident_pool, len);
+ buffer = _cpp_unaligned_alloc (pfile, len);
token->type = CPP_COMMENT;
token->val.str.len = len;
memcpy (buffer + 1, from, len - 1);
}
-/* Subroutine of _cpp_lex_direct to handle '%'. A little tricky, since we
- want to avoid stepping back when lexing %:%X. */
-static void
-lex_percent (pfile, result)
- cpp_reader *pfile;
- cpp_token *result;
-{
- cpp_buffer *buffer= pfile->buffer;
- cppchar_t c;
-
- result->type = CPP_MOD;
- /* Parsing %:%X could leave an extra character. */
- if (buffer->extra_char == EOF)
- c = get_effective_char (pfile);
- else
- {
- c = buffer->read_ahead = buffer->extra_char;
- buffer->extra_char = EOF;
- }
-
- if (c == '=')
- ACCEPT_CHAR (CPP_MOD_EQ);
- else if (CPP_OPTION (pfile, digraphs))
- {
- if (c == ':')
- {
- result->flags |= DIGRAPH;
- ACCEPT_CHAR (CPP_HASH);
- if (get_effective_char (pfile) == '%')
- {
- buffer->extra_char = get_effective_char (pfile);
- if (buffer->extra_char == ':')
- {
- buffer->extra_char = EOF;
- ACCEPT_CHAR (CPP_PASTE);
- }
- else
- /* We'll catch the extra_char when we're called back. */
- buffer->read_ahead = '%';
- }
- }
- else if (c == '>')
- {
- result->flags |= DIGRAPH;
- ACCEPT_CHAR (CPP_CLOSE_BRACE);
- }
- }
-}
-
-/* Subroutine of _cpp_lex_direct to handle '.'. This is tricky, since we
- want to avoid stepping back when lexing '...' or '.123'. In the
- latter case we should also set a flag for parse_number. */
-static void
-lex_dot (pfile, result)
- cpp_reader *pfile;
- cpp_token *result;
-{
- cpp_buffer *buffer = pfile->buffer;
- cppchar_t c;
-
- /* Parsing ..X could leave an extra character. */
- if (buffer->extra_char == EOF)
- c = get_effective_char (pfile);
- else
- {
- c = buffer->read_ahead = buffer->extra_char;
- buffer->extra_char = EOF;
- }
-
- /* All known character sets have 0...9 contiguous. */
- if (c >= '0' && c <= '9')
- {
- result->type = CPP_NUMBER;
- parse_number (pfile, &result->val.str, c, 1);
- }
- else
- {
- result->type = CPP_DOT;
- if (c == '.')
- {
- buffer->extra_char = get_effective_char (pfile);
- if (buffer->extra_char == '.')
- {
- buffer->extra_char = EOF;
- ACCEPT_CHAR (CPP_ELLIPSIS);
- }
- else
- /* We'll catch the extra_char when we're called back. */
- buffer->read_ahead = '.';
- }
- else if (c == '*' && CPP_OPTION (pfile, cplusplus))
- ACCEPT_CHAR (CPP_DOT_STAR);
- }
-}
-
/* Allocate COUNT tokens for RUN. */
void
_cpp_init_tokenrun (run, count)
/* Lex a token into RESULT (external interface). Takes care of issues
like directive handling, token lookahead, multiple include
- opimisation and skipping. */
+ optimization and skipping. */
const cpp_token *
_cpp_lex_token (pfile)
cpp_reader *pfile;
/* Is this a directive. If _cpp_handle_directive returns
false, it is an assembler #. */
if (result->type == CPP_HASH
- && !pfile->state.parsing_args
+ /* 6.10.3 p 11: Directives in a list of macro arguments
+ gives undefined behavior. This implementation
+ handles the directive as normal. */
+ && pfile->state.parsing_args != 1
&& _cpp_handle_directive (pfile, result->flags & PREV_WHITE))
continue;
if (pfile->cb.line_change && !pfile->state.skipping)
return result;
}
+#define IF_NEXT_IS(CHAR, THEN_TYPE, ELSE_TYPE) \
+ do { \
+ if (get_effective_char (pfile) == CHAR) \
+ result->type = THEN_TYPE; \
+ else \
+ { \
+ BACKUP (); \
+ result->type = ELSE_TYPE; \
+ } \
+ } while (0)
+
/* Lex a token into pfile->cur_token, which is also incremented, to
get diagnostics pointing to the correct location.
result->line = pfile->line;
skipped_white:
- c = buffer->read_ahead;
- if (c == EOF && buffer->cur < buffer->rlimit)
- c = *buffer->cur++;
+ c = *buffer->cur++;
result->col = CPP_BUF_COLUMN (buffer, buffer->cur);
- buffer->read_ahead = EOF;
trigraph:
switch (c)
{
- case EOF:
+ case ' ': case '\t': case '\f': case '\v': case '\0':
+ result->flags |= PREV_WHITE;
+ if (skip_whitespace (pfile, c))
+ goto skipped_white;
+
+ /* EOF. */
+ buffer->cur--;
buffer->saved_flags = BOL;
if (!pfile->state.parsing_args && !pfile->state.in_directive)
{
for command line and _Pragma buffers. */
if (!buffer->from_stage3)
cpp_pedwarn (pfile, "no newline at end of file");
- handle_newline (pfile, '\n');
+ handle_newline (pfile);
}
/* Don't pop the last buffer. */
result->type = CPP_EOF;
break;
- case ' ': case '\t': case '\f': case '\v': case '\0':
- skip_whitespace (pfile, c);
- result->flags |= PREV_WHITE;
- goto skipped_white;
-
case '\n': case '\r':
- handle_newline (pfile, c);
+ handle_newline (pfile);
buffer->saved_flags = BOL;
if (! pfile->state.in_directive)
{
{
unsigned int line = pfile->line;
- c = skip_escaped_newlines (pfile, c);
+ c = skip_escaped_newlines (pfile);
if (line != pfile->line)
- /* We had at least one escaped newline of some sort, and the
- next character is in buffer->read_ahead. Update the
- token's line and column. */
+ {
+ buffer->cur--;
+ /* We had at least one escaped newline of some sort.
+ Update the token's line and column. */
goto update_tokens_line;
+ }
+ }
- /* We are either the original '?' or '\\', or a trigraph. */
+ /* We are either the original '?' or '\\', or a trigraph. */
+ if (c == '?')
result->type = CPP_QUERY;
- buffer->read_ahead = EOF;
- if (c == '\\')
- goto random_char;
- else if (c != '?')
- goto trigraph;
- }
+ else if (c == '\\')
+ goto random_char;
+ else
+ goto trigraph;
break;
case '0': case '1': case '2': case '3': case '4':
parse_number (pfile, &result->val.str, c, 0);
break;
- case '$':
- if (!CPP_OPTION (pfile, dollars_in_ident))
- goto random_char;
- /* Fall through... */
+ case 'L':
+ /* 'L' may introduce wide characters or strings. */
+ {
+ const unsigned char *pos = buffer->cur;
+
+ c = get_effective_char (pfile);
+ if (c == '\'' || c == '"')
+ {
+ result->type = (c == '"' ? CPP_WSTRING: CPP_WCHAR);
+ parse_string (pfile, result, c);
+ break;
+ }
+ buffer->cur = pos;
+ }
+ /* Fall through. */
+ start_ident:
case '_':
case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
case 's': case 't': case 'u': case 'v': case 'w': case 'x':
case 'y': case 'z':
case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
- case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
+ case 'G': case 'H': case 'I': case 'J': case 'K':
case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
case 'Y': case 'Z':
result->type = CPP_NAME;
result->val.node = parse_identifier (pfile);
- /* 'L' may introduce wide characters or strings. */
- if (result->val.node == pfile->spec_nodes.n_L)
- {
- c = buffer->read_ahead;
- if (c == EOF && buffer->cur < buffer->rlimit)
- c = *buffer->cur;
- if (c == '\'' || c == '"')
- {
- buffer->cur++;
- ACCEPT_CHAR (c == '"' ? CPP_WSTRING: CPP_WCHAR);
- goto make_string;
- }
- }
/* Convert named operators to their proper types. */
- else if (result->val.node->flags & NODE_OPERATOR)
+ if (result->val.node->flags & NODE_OPERATOR)
{
result->flags |= NAMED_OP;
result->type = result->val.node->value.operator;
case '\'':
case '"':
result->type = c == '"' ? CPP_STRING: CPP_CHAR;
- make_string:
parse_string (pfile, result, c);
break;
case '/':
/* A potential block or line comment. */
comment_start = buffer->cur;
- result->type = CPP_DIV;
c = get_effective_char (pfile);
- if (c == '=')
- ACCEPT_CHAR (CPP_DIV_EQ);
- if (c != '/' && c != '*')
- break;
-
+
if (c == '*')
{
if (skip_block_comment (pfile))
cpp_error (pfile, "unterminated comment");
}
- else
+ else if (c == '/' && (CPP_OPTION (pfile, cplusplus_comments)
+ || CPP_IN_SYSTEM_HEADER (pfile)))
{
- if (!CPP_OPTION (pfile, cplusplus_comments)
- && !CPP_IN_SYSTEM_HEADER (pfile))
- break;
-
/* Warn about comments only if pedantically GNUC89, and not
in system headers. */
if (CPP_OPTION (pfile, lang) == CLK_GNUC89 && CPP_PEDANTIC (pfile)
buffer->warned_cplusplus_comments = 1;
}
- /* Skip_line_comment updates buffer->read_ahead. */
if (skip_line_comment (pfile) && CPP_OPTION (pfile, warn_comments))
cpp_warning (pfile, "multi-line comment");
}
+ else if (c == '=')
+ {
+ result->type = CPP_DIV_EQ;
+ break;
+ }
+ else
+ {
+ BACKUP ();
+ result->type = CPP_DIV;
+ break;
+ }
- /* Skipping the comment has updated buffer->read_ahead. */
if (!pfile->state.save_comments)
{
result->flags |= PREV_WHITE;
if (pfile->state.angled_headers)
{
result->type = CPP_HEADER_NAME;
- c = '>'; /* terminator. */
- goto make_string;
+ parse_string (pfile, result, '>');
+ break;
}
- result->type = CPP_LESS;
c = get_effective_char (pfile);
if (c == '=')
- ACCEPT_CHAR (CPP_LESS_EQ);
+ result->type = CPP_LESS_EQ;
else if (c == '<')
- {
- ACCEPT_CHAR (CPP_LSHIFT);
- if (get_effective_char (pfile) == '=')
- ACCEPT_CHAR (CPP_LSHIFT_EQ);
- }
+ IF_NEXT_IS ('=', CPP_LSHIFT_EQ, CPP_LSHIFT);
else if (c == '?' && CPP_OPTION (pfile, cplusplus))
- {
- ACCEPT_CHAR (CPP_MIN);
- if (get_effective_char (pfile) == '=')
- ACCEPT_CHAR (CPP_MIN_EQ);
- }
+ IF_NEXT_IS ('=', CPP_MIN_EQ, CPP_MIN);
else if (c == ':' && CPP_OPTION (pfile, digraphs))
{
- ACCEPT_CHAR (CPP_OPEN_SQUARE);
+ result->type = CPP_OPEN_SQUARE;
result->flags |= DIGRAPH;
}
else if (c == '%' && CPP_OPTION (pfile, digraphs))
{
- ACCEPT_CHAR (CPP_OPEN_BRACE);
+ result->type = CPP_OPEN_BRACE;
result->flags |= DIGRAPH;
}
+ else
+ {
+ BACKUP ();
+ result->type = CPP_LESS;
+ }
break;
case '>':
- result->type = CPP_GREATER;
c = get_effective_char (pfile);
if (c == '=')
- ACCEPT_CHAR (CPP_GREATER_EQ);
+ result->type = CPP_GREATER_EQ;
else if (c == '>')
- {
- ACCEPT_CHAR (CPP_RSHIFT);
- if (get_effective_char (pfile) == '=')
- ACCEPT_CHAR (CPP_RSHIFT_EQ);
- }
+ IF_NEXT_IS ('=', CPP_RSHIFT_EQ, CPP_RSHIFT);
else if (c == '?' && CPP_OPTION (pfile, cplusplus))
+ IF_NEXT_IS ('=', CPP_MAX_EQ, CPP_MAX);
+ else
{
- ACCEPT_CHAR (CPP_MAX);
- if (get_effective_char (pfile) == '=')
- ACCEPT_CHAR (CPP_MAX_EQ);
+ BACKUP ();
+ result->type = CPP_GREATER;
}
break;
case '%':
- lex_percent (pfile, result);
+ c = get_effective_char (pfile);
+ if (c == '=')
+ result->type = CPP_MOD_EQ;
+ else if (CPP_OPTION (pfile, digraphs) && c == ':')
+ {
+ result->flags |= DIGRAPH;
+ result->type = CPP_HASH;
+ if (get_effective_char (pfile) == '%')
+ {
+ const unsigned char *pos = buffer->cur;
+
+ if (get_effective_char (pfile) == ':')
+ result->type = CPP_PASTE;
+ else
+ buffer->cur = pos - 1;
+ }
+ else
+ BACKUP ();
+ }
+ else if (CPP_OPTION (pfile, digraphs) && c == '>')
+ {
+ result->flags |= DIGRAPH;
+ result->type = CPP_CLOSE_BRACE;
+ }
+ else
+ {
+ BACKUP ();
+ result->type = CPP_MOD;
+ }
break;
case '.':
- lex_dot (pfile, result);
+ result->type = CPP_DOT;
+ c = get_effective_char (pfile);
+ if (c == '.')
+ {
+ const unsigned char *pos = buffer->cur;
+
+ if (get_effective_char (pfile) == '.')
+ result->type = CPP_ELLIPSIS;
+ else
+ buffer->cur = pos - 1;
+ }
+ /* All known character sets have 0...9 contiguous. */
+ else if (ISDIGIT (c))
+ {
+ result->type = CPP_NUMBER;
+ parse_number (pfile, &result->val.str, c, 1);
+ }
+ else if (c == '*' && CPP_OPTION (pfile, cplusplus))
+ result->type = CPP_DOT_STAR;
+ else
+ BACKUP ();
break;
case '+':
- result->type = CPP_PLUS;
c = get_effective_char (pfile);
- if (c == '=')
- ACCEPT_CHAR (CPP_PLUS_EQ);
- else if (c == '+')
- ACCEPT_CHAR (CPP_PLUS_PLUS);
+ if (c == '+')
+ result->type = CPP_PLUS_PLUS;
+ else if (c == '=')
+ result->type = CPP_PLUS_EQ;
+ else
+ {
+ BACKUP ();
+ result->type = CPP_PLUS;
+ }
break;
case '-':
- result->type = CPP_MINUS;
c = get_effective_char (pfile);
if (c == '>')
{
- ACCEPT_CHAR (CPP_DEREF);
- if (CPP_OPTION (pfile, cplusplus)
- && get_effective_char (pfile) == '*')
- ACCEPT_CHAR (CPP_DEREF_STAR);
+ result->type = CPP_DEREF;
+ if (CPP_OPTION (pfile, cplusplus))
+ {
+ if (get_effective_char (pfile) == '*')
+ result->type = CPP_DEREF_STAR;
+ else
+ BACKUP ();
+ }
}
- else if (c == '=')
- ACCEPT_CHAR (CPP_MINUS_EQ);
else if (c == '-')
- ACCEPT_CHAR (CPP_MINUS_MINUS);
- break;
-
- case '*':
- result->type = CPP_MULT;
- if (get_effective_char (pfile) == '=')
- ACCEPT_CHAR (CPP_MULT_EQ);
- break;
-
- case '=':
- result->type = CPP_EQ;
- if (get_effective_char (pfile) == '=')
- ACCEPT_CHAR (CPP_EQ_EQ);
- break;
-
- case '!':
- result->type = CPP_NOT;
- if (get_effective_char (pfile) == '=')
- ACCEPT_CHAR (CPP_NOT_EQ);
+ result->type = CPP_MINUS_MINUS;
+ else if (c == '=')
+ result->type = CPP_MINUS_EQ;
+ else
+ {
+ BACKUP ();
+ result->type = CPP_MINUS;
+ }
break;
case '&':
- result->type = CPP_AND;
c = get_effective_char (pfile);
- if (c == '=')
- ACCEPT_CHAR (CPP_AND_EQ);
- else if (c == '&')
- ACCEPT_CHAR (CPP_AND_AND);
+ if (c == '&')
+ result->type = CPP_AND_AND;
+ else if (c == '=')
+ result->type = CPP_AND_EQ;
+ else
+ {
+ BACKUP ();
+ result->type = CPP_AND;
+ }
break;
- case '#':
- result->type = CPP_HASH;
- if (get_effective_char (pfile) == '#')
- ACCEPT_CHAR (CPP_PASTE);
- break;
-
case '|':
- result->type = CPP_OR;
c = get_effective_char (pfile);
- if (c == '=')
- ACCEPT_CHAR (CPP_OR_EQ);
- else if (c == '|')
- ACCEPT_CHAR (CPP_OR_OR);
- break;
-
- case '^':
- result->type = CPP_XOR;
- if (get_effective_char (pfile) == '=')
- ACCEPT_CHAR (CPP_XOR_EQ);
+ if (c == '|')
+ result->type = CPP_OR_OR;
+ else if (c == '=')
+ result->type = CPP_OR_EQ;
+ else
+ {
+ BACKUP ();
+ result->type = CPP_OR;
+ }
break;
case ':':
- result->type = CPP_COLON;
c = get_effective_char (pfile);
if (c == ':' && CPP_OPTION (pfile, cplusplus))
- ACCEPT_CHAR (CPP_SCOPE);
+ result->type = CPP_SCOPE;
else if (c == '>' && CPP_OPTION (pfile, digraphs))
{
result->flags |= DIGRAPH;
- ACCEPT_CHAR (CPP_CLOSE_SQUARE);
+ result->type = CPP_CLOSE_SQUARE;
+ }
+ else
+ {
+ BACKUP ();
+ result->type = CPP_COLON;
}
break;
+ case '*': IF_NEXT_IS ('=', CPP_MULT_EQ, CPP_MULT); break;
+ case '=': IF_NEXT_IS ('=', CPP_EQ_EQ, CPP_EQ); break;
+ case '!': IF_NEXT_IS ('=', CPP_NOT_EQ, CPP_NOT); break;
+ case '^': IF_NEXT_IS ('=', CPP_XOR_EQ, CPP_XOR); break;
+ case '#': IF_NEXT_IS ('#', CPP_PASTE, CPP_HASH); break;
+
case '~': result->type = CPP_COMPL; break;
case ',': result->type = CPP_COMMA; break;
case '(': result->type = CPP_OPEN_PAREN; break;
/* @ is a punctuator in Objective C. */
case '@': result->type = CPP_ATSIGN; break;
+ case '$':
+ if (CPP_OPTION (pfile, dollars_in_ident))
+ goto start_ident;
+ /* Fall through... */
+
random_char:
default:
result->type = CPP_OTHER;
return result;
}
-/* An upper bound on the number of bytes needed to spell a token,
+/* An upper bound on the number of bytes needed to spell TOKEN,
including preceding whitespace. */
unsigned int
cpp_token_len (token)
switch (TOKEN_SPELL (token))
{
default: len = 0; break;
+ case SPELL_NUMBER:
case SPELL_STRING: len = token->val.str.len; break;
case SPELL_IDENT: len = NODE_LEN (token->val.node); break;
}
- /* 1 for whitespace, 4 for comment delimeters. */
+ /* 1 for whitespace, 4 for comment delimiters. */
return len + 5;
}
}
break;
+ case SPELL_CHAR:
+ *buffer++ = token->val.c;
+ break;
+
+ spell_ident:
case SPELL_IDENT:
- spell_ident:
memcpy (buffer, NODE_NAME (token->val.node), NODE_LEN (token->val.node));
buffer += NODE_LEN (token->val.node);
break;
+ case SPELL_NUMBER:
+ memcpy (buffer, token->val.str.text, token->val.str.len);
+ buffer += token->val.str.len;
+ break;
+
case SPELL_STRING:
{
int left, right, tag;
case CPP_CHAR: left = '\''; right = '\''; tag = '\0'; break;
case CPP_WCHAR: left = '\''; right = '\''; tag = 'L'; break;
case CPP_HEADER_NAME: left = '<'; right = '>'; tag = '\0'; break;
- default: left = '\0'; right = '\0'; tag = '\0'; break;
+ default:
+ cpp_ice (pfile, "unknown string token %s\n", TOKEN_NAME (token));
+ return buffer;
}
if (tag) *buffer++ = tag;
- if (left) *buffer++ = left;
+ *buffer++ = left;
memcpy (buffer, token->val.str.text, token->val.str.len);
buffer += token->val.str.len;
- if (right) *buffer++ = right;
+ *buffer++ = right;
}
break;
- case SPELL_CHAR:
- *buffer++ = token->val.c;
- break;
-
case SPELL_NONE:
- cpp_ice (pfile, "Unspellable token %s", TOKEN_NAME (token));
+ cpp_ice (pfile, "unspellable token %s", TOKEN_NAME (token));
break;
}
return buffer;
}
-/* Returns a token as a null-terminated string. The string is
- temporary, and automatically freed later. Useful for diagnostics. */
+/* Returns TOKEN spelt as a null-terminated string. The string is
+ freed when the reader is destroyed. Useful for diagnostics. */
unsigned char *
cpp_token_as_text (pfile, token)
cpp_reader *pfile;
const cpp_token *token;
{
unsigned int len = cpp_token_len (token);
- unsigned char *start = _cpp_pool_alloc (&pfile->ident_pool, len), *end;
+ unsigned char *start = _cpp_unaligned_alloc (pfile, len), *end;
end = cpp_spell_token (pfile, token, start);
end[0] = '\0';
return start;
}
-/* Used by C front ends. Should really move to using cpp_token_as_text. */
+/* Used by C front ends, which really should move to using
+ cpp_token_as_text. */
const char *
cpp_type2name (type)
enum cpp_ttype type;
}
break;
+ case SPELL_CHAR:
+ putc (token->val.c, fp);
+ break;
+
spell_ident:
case SPELL_IDENT:
fwrite (NODE_NAME (token->val.node), 1, NODE_LEN (token->val.node), fp);
break;
+ case SPELL_NUMBER:
+ fwrite (token->val.str.text, 1, token->val.str.len, fp);
+ break;
+
case SPELL_STRING:
{
int left, right, tag;
case CPP_CHAR: left = '\''; right = '\''; tag = '\0'; break;
case CPP_WCHAR: left = '\''; right = '\''; tag = 'L'; break;
case CPP_HEADER_NAME: left = '<'; right = '>'; tag = '\0'; break;
- default: left = '\0'; right = '\0'; tag = '\0'; break;
+ default:
+ fprintf (stderr, "impossible STRING token %s\n", TOKEN_NAME (token));
+ return;
}
if (tag) putc (tag, fp);
- if (left) putc (left, fp);
+ putc (left, fp);
fwrite (token->val.str.text, 1, token->val.str.len, fp);
- if (right) putc (right, fp);
+ putc (right, fp);
}
break;
- case SPELL_CHAR:
- putc (token->val.c, fp);
- break;
-
case SPELL_NONE:
/* An error, most probably. */
break;
return (a->type != CPP_MACRO_ARG || a->val.arg_no == b->val.arg_no);
case SPELL_IDENT:
return a->val.node == b->val.node;
+ case SPELL_NUMBER:
case SPELL_STRING:
return (a->val.str.len == b->val.str.len
&& !memcmp (a->val.str.text, b->val.str.text,
accidental token paste for output. For simplicity, it is
conservative, and occasionally advises a space where one is not
needed, e.g. "." and ".2". */
-
int
cpp_avoid_paste (pfile, token1, token2)
cpp_reader *pfile;
hex_digit_value (c)
unsigned int c;
{
- if (c >= 'a' && c <= 'f')
- return c - 'a' + 10;
- if (c >= 'A' && c <= 'F')
- return c - 'A' + 10;
- if (c >= '0' && c <= '9')
- return c - '0';
- abort ();
+ if (hex_p (c))
+ return hex_value (c);
+ else
+ abort ();
}
/* Parse a '\uNNNN' or '\UNNNNNNNN' sequence. Returns 1 to indicate
return 1;
if (CPP_WTRADITIONAL (pfile))
- cpp_warning (pfile, "the meaning of '\\%c' varies with -traditional", c);
+ cpp_warning (pfile, "the meaning of '\\%c' is different in traditional C", c);
length = (c == 'u' ? 4: 8);
/* Interpret an escape sequence, and return its value. PSTR points to
the input pointer, which is just after the backslash. LIMIT is how
much text we have. MASK is a bitmask for the precision for the
- destination type (char or wchar_t). TRADITIONAL, if true, does not
- interpret escapes that did not exist in traditional C.
+ destination type (char or wchar_t).
Handles all relevant diagnostics. */
-
unsigned int
-cpp_parse_escape (pfile, pstr, limit, mask, traditional)
+cpp_parse_escape (pfile, pstr, limit, mask)
cpp_reader *pfile;
const unsigned char **pstr;
const unsigned char *limit;
unsigned HOST_WIDE_INT mask;
- int traditional;
{
int unknown = 0;
const unsigned char *str = *pstr;
case 'a':
if (CPP_WTRADITIONAL (pfile))
- cpp_warning (pfile, "the meaning of '\\a' varies with -traditional");
- if (!traditional)
- c = TARGET_BELL;
+ cpp_warning (pfile, "the meaning of '\\a' is different in traditional C");
+ c = TARGET_BELL;
break;
case 'e': case 'E':
case 'x':
if (CPP_WTRADITIONAL (pfile))
- cpp_warning (pfile, "the meaning of '\\x' varies with -traditional");
+ cpp_warning (pfile, "the meaning of '\\x' is different in traditional C");
- if (!traditional)
{
unsigned int i = 0, overflow = 0;
int digits_found = 0;
#endif
/* Interpret a (possibly wide) character constant in TOKEN.
- WARN_MULTI warns about multi-character charconsts, if not
- TRADITIONAL. TRADITIONAL also indicates not to interpret escapes
- that did not exist in traditional C. PCHARS_SEEN points to a
- variable that is filled in with the number of characters seen. */
+ WARN_MULTI warns about multi-character charconsts. PCHARS_SEEN points
+ to a variable that is filled in with the number of characters seen. */
HOST_WIDE_INT
-cpp_interpret_charconst (pfile, token, warn_multi, traditional, pchars_seen)
+cpp_interpret_charconst (pfile, token, warn_multi, pchars_seen)
cpp_reader *pfile;
const cpp_token *token;
int warn_multi;
- int traditional;
unsigned int *pchars_seen;
{
const unsigned char *str = token->val.str.text;
unsigned int width, max_chars, c;
unsigned HOST_WIDE_INT mask;
HOST_WIDE_INT result = 0;
+ bool unsigned_p;
#ifdef MULTIBYTE_CHARS
(void) local_mbtowc (NULL, NULL, 0);
/* Width in bits. */
if (token->type == CPP_CHAR)
- width = MAX_CHAR_TYPE_SIZE;
+ {
+ width = MAX_CHAR_TYPE_SIZE;
+ unsigned_p = CPP_OPTION (pfile, signed_char) == 0;
+ }
else
- width = MAX_WCHAR_TYPE_SIZE;
+ {
+ width = MAX_WCHAR_TYPE_SIZE;
+ unsigned_p = WCHAR_UNSIGNED;
+ }
if (width < HOST_BITS_PER_WIDE_INT)
mask = ((unsigned HOST_WIDE_INT) 1 << width) - 1;
#endif
if (c == '\\')
- c = cpp_parse_escape (pfile, &str, limit, mask, traditional);
+ c = cpp_parse_escape (pfile, &str, limit, mask);
#ifdef MAP_CHARACTER
if (ISPRINT (c))
chars_seen = max_chars;
cpp_warning (pfile, "character constant too long");
}
- else if (chars_seen > 1 && !traditional && warn_multi)
+ else if (chars_seen > 1 && warn_multi)
cpp_warning (pfile, "multi-character character constant");
- /* If char type is signed, sign-extend the constant. The
- __CHAR_UNSIGNED__ macro is set by the driver if appropriate. */
- if (token->type == CPP_CHAR && chars_seen)
+ /* If relevant type is signed, sign-extend the constant. */
+ if (chars_seen)
{
unsigned int nbits = chars_seen * width;
- unsigned int mask = (unsigned int) ~0 >> (HOST_BITS_PER_INT - nbits);
- if (pfile->spec_nodes.n__CHAR_UNSIGNED__->type == NT_MACRO
- || ((result >> (nbits - 1)) & 1) == 0)
+ mask = (unsigned HOST_WIDE_INT) ~0 >> (HOST_BITS_PER_WIDE_INT - nbits);
+ if (unsigned_p || ((result >> (nbits - 1)) & 1) == 0)
result &= mask;
else
result |= ~mask;
expansion. Also check the change in peak memory usage (NJAMD is a
good tool for this). */
#define MIN_BUFF_SIZE 8000
-#define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (8000 + (MIN_SIZE) * 3 / 2)
+#define BUFF_SIZE_UPPER_BOUND(MIN_SIZE) (MIN_BUFF_SIZE + (MIN_SIZE) * 3 / 2)
#define EXTENDED_BUFF_SIZE(BUFF, MIN_EXTRA) \
(MIN_EXTRA + ((BUFF)->limit - (BUFF)->cur) * 2)
+#if MIN_BUFF_SIZE > BUFF_SIZE_UPPER_BOUND (0)
+ #error BUFF_SIZE_UPPER_BOUND must be at least as large as MIN_BUFF_SIZE!
+#endif
+
struct dummy
{
char c;
of the buffer, so that buffer overflows will cause immediate chaos. */
static _cpp_buff *
new_buff (len)
- unsigned int len;
+ size_t len;
{
_cpp_buff *result;
- char *base;
+ unsigned char *base;
if (len < MIN_BUFF_SIZE)
len = MIN_BUFF_SIZE;
_cpp_buff *
_cpp_get_buff (pfile, min_size)
cpp_reader *pfile;
- unsigned int min_size;
+ size_t min_size;
{
_cpp_buff *result, **p;
for (p = &pfile->free_buffs;; p = &(*p)->next)
{
- unsigned int size;
+ size_t size;
if (*p == NULL)
return new_buff (min_size);
size = result->limit - result->base;
/* Return a buffer that's big enough, but don't waste one that's
way too big. */
- if (size >= min_size && size < BUFF_SIZE_UPPER_BOUND (min_size))
+ if (size >= min_size && size <= BUFF_SIZE_UPPER_BOUND (min_size))
break;
}
return result;
}
-/* Return a buffer chained on the end of BUFF. Copy to it the
- uncommitted remaining bytes of BUFF, with at least MIN_EXTRA more
- bytes. */
+/* Creates a new buffer with enough space to hold the uncommitted
+ remaining bytes of BUFF, and at least MIN_EXTRA more bytes. Copies
+ the excess bytes to the new buffer. Chains the new buffer after
+ BUFF, and returns the new buffer. */
_cpp_buff *
-_cpp_extend_buff (pfile, buff, min_extra)
+_cpp_append_extend_buff (pfile, buff, min_extra)
cpp_reader *pfile;
_cpp_buff *buff;
- unsigned int min_extra;
+ size_t min_extra;
{
- unsigned int size = EXTENDED_BUFF_SIZE (buff, min_extra);
+ size_t size = EXTENDED_BUFF_SIZE (buff, min_extra);
+ _cpp_buff *new_buff = _cpp_get_buff (pfile, size);
- buff->next = _cpp_get_buff (pfile, size);
- memcpy (buff->next->base, buff->cur, buff->limit - buff->cur);
- return buff->next;
+ buff->next = new_buff;
+ memcpy (new_buff->base, buff->cur, BUFF_ROOM (buff));
+ return new_buff;
+}
+
+/* Creates a new buffer with enough space to hold the uncommitted
+ remaining bytes of the buffer pointed to by BUFF, and at least
+ MIN_EXTRA more bytes. Copies the excess bytes to the new buffer.
+ Chains the new buffer before the buffer pointed to by BUFF, and
+ updates the pointer to point to the new buffer. */
+void
+_cpp_extend_buff (pfile, pbuff, min_extra)
+ cpp_reader *pfile;
+ _cpp_buff **pbuff;
+ size_t min_extra;
+{
+ _cpp_buff *new_buff, *old_buff = *pbuff;
+ size_t size = EXTENDED_BUFF_SIZE (old_buff, min_extra);
+
+ new_buff = _cpp_get_buff (pfile, size);
+ memcpy (new_buff->base, old_buff->cur, BUFF_ROOM (old_buff));
+ new_buff->next = old_buff;
+ *pbuff = new_buff;
}
/* Free a chain of buffers starting at BUFF. */
}
}
-static int
-chunk_suitable (chunk, size)
- cpp_chunk *chunk;
- unsigned int size;
-{
- /* Being at least twice SIZE means we can use memcpy in
- _cpp_next_chunk rather than memmove. Besides, it's a good idea
- anyway. */
- return (chunk && (unsigned int) (chunk->limit - chunk->base) >= size * 2);
-}
-
-/* Returns the end of the new pool. PTR points to a char in the old
- pool, and is updated to point to the same char in the new pool. */
+/* Allocate permanent, unaligned storage of length LEN. */
unsigned char *
-_cpp_next_chunk (pool, len, ptr)
- cpp_pool *pool;
- unsigned int len;
- unsigned char **ptr;
+_cpp_unaligned_alloc (pfile, len)
+ cpp_reader *pfile;
+ size_t len;
{
- cpp_chunk *chunk = pool->cur->next;
+ _cpp_buff *buff = pfile->u_buff;
+ unsigned char *result = buff->cur;
- /* LEN is the minimum size we want in the new pool. */
- len += POOL_ROOM (pool);
- if (! chunk_suitable (chunk, len))
+ if (len > (size_t) (buff->limit - result))
{
- chunk = new_chunk (POOL_SIZE (pool) * 2 + len);
-
- chunk->next = pool->cur->next;
- pool->cur->next = chunk;
+ buff = _cpp_get_buff (pfile, len);
+ buff->next = pfile->u_buff;
+ pfile->u_buff = buff;
+ result = buff->cur;
}
- /* Update the pointer before changing chunk's front. */
- if (ptr)
- *ptr += chunk->base - POOL_FRONT (pool);
-
- memcpy (chunk->base, POOL_FRONT (pool), POOL_ROOM (pool));
- chunk->front = chunk->base;
-
- pool->cur = chunk;
- return POOL_LIMIT (pool);
-}
-
-static cpp_chunk *
-new_chunk (size)
- unsigned int size;
-{
- unsigned char *base;
- cpp_chunk *result;
-
- size = POOL_ALIGN (size, DEFAULT_ALIGNMENT);
- base = (unsigned char *) xmalloc (size + sizeof (cpp_chunk));
- /* Put the chunk descriptor at the end. Then chunk overruns will
- cause obvious chaos. */
- result = (cpp_chunk *) (base + size);
- result->base = base;
- result->front = base;
- result->limit = base + size;
- result->next = 0;
-
+ buff->cur = result + len;
return result;
}
-void
-_cpp_init_pool (pool, size, align, temp)
- cpp_pool *pool;
- unsigned int size, align, temp;
-{
- if (align == 0)
- align = DEFAULT_ALIGNMENT;
- if (align & (align - 1))
- abort ();
- pool->align = align;
- pool->first = new_chunk (size);
- pool->cur = pool->first;
- if (temp)
- pool->cur->next = pool->cur;
-}
+/* Allocate permanent, unaligned storage of length LEN from a_buff.
+ That buffer is used for growing allocations when saving macro
+ replacement lists in a #define, and when parsing an answer to an
+ assertion in #assert, #unassert or #if (and therefore possibly
+ whilst expanding macros). It therefore must not be used by any
+ code that they might call: specifically the lexer and the guts of
+ the macro expander.
-void
-_cpp_free_pool (pool)
- cpp_pool *pool;
+ All existing other uses clearly fit this restriction: storing
+ registered pragmas during initialization. */
+unsigned char *
+_cpp_aligned_alloc (pfile, len)
+ cpp_reader *pfile;
+ size_t len;
{
- cpp_chunk *chunk = pool->first, *next;
+ _cpp_buff *buff = pfile->a_buff;
+ unsigned char *result = buff->cur;
- do
+ if (len > (size_t) (buff->limit - result))
{
- next = chunk->next;
- free (chunk->base);
- chunk = next;
+ buff = _cpp_get_buff (pfile, len);
+ buff->next = pfile->a_buff;
+ pfile->a_buff = buff;
+ result = buff->cur;
}
- while (chunk && chunk != pool->first);
-}
-
-/* Reserve LEN bytes from a memory pool. */
-unsigned char *
-_cpp_pool_reserve (pool, len)
- cpp_pool *pool;
- unsigned int len;
-{
- len = POOL_ALIGN (len, pool->align);
- if (len > (unsigned int) POOL_ROOM (pool))
- _cpp_next_chunk (pool, len, 0);
-
- return POOL_FRONT (pool);
-}
-
-/* Allocate LEN bytes from a memory pool. */
-unsigned char *
-_cpp_pool_alloc (pool, len)
- cpp_pool *pool;
- unsigned int len;
-{
- unsigned char *result = _cpp_pool_reserve (pool, len);
- POOL_COMMIT (pool, len);
+ buff->cur = result + len;
return result;
}
OSDN Git Service
