2 * TCC - Tiny C Compiler
4 * Copyright (c) 2001-2004 Fabrice Bellard
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2 of the License, or (at your option) any later version.
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
23 /********************************************************/
24 /* global variables */
26 ST_DATA
int tok_flags
;
27 ST_DATA
int parse_flags
;
29 ST_DATA
struct BufferedFile
*file
;
32 ST_DATA
const int *macro_ptr
;
33 ST_DATA CString tokcstr
; /* current parsed string, if any */
35 /* display benchmark infos */
36 ST_DATA
int total_lines
;
37 ST_DATA
int total_bytes
;
38 ST_DATA
int tok_ident
;
39 ST_DATA TokenSym
**table_ident
;
41 /* ------------------------------------------------------------------------- */
43 static TokenSym
*hash_ident
[TOK_HASH_SIZE
];
44 static char token_buf
[STRING_MAX_SIZE
+ 1];
45 static CString cstr_buf
;
46 static TokenString tokstr_buf
;
47 static unsigned char isidnum_table
[256 - CH_EOF
];
48 static int pp_debug_tok
, pp_debug_symv
;
50 static void tok_print(const char *msg
, const int *str
);
52 static struct TinyAlloc
*toksym_alloc
;
53 static struct TinyAlloc
*tokstr_alloc
;
54 static struct TinyAlloc
*cstr_alloc
;
56 /* isidnum_table flags: */
61 static TokenString
*macro_stack
;
63 static const char tcc_keywords
[] =
64 #define DEF(id, str) str "\0"
69 /* WARNING: the content of this string encodes token numbers */
70 static const unsigned char tok_two_chars
[] =
72 "<=\236>=\235!=\225&&\240||\241++\244--\242==\224<<\1>>\2+=\253"
73 "-=\255*=\252/=\257%=\245&=\246^=\336|=\374->\313..\250##\266";
94 '.','.', 0xa8, // C++ token ?
95 '#','#', TOK_TWOSHARPS
,
99 static void next_nomacro_spc(void);
101 ST_FUNC
void skip(int c
)
104 tcc_error("'%c' expected (got \"%s\")", c
, get_tok_str(tok
, &tokc
));
108 ST_FUNC
void expect(const char *msg
)
110 tcc_error("%s expected", msg
);
113 /* ------------------------------------------------------------------------- */
114 /* Custom allocator for tiny objects */
119 #define tal_free(al, p) tcc_free(p)
120 #define tal_realloc(al, p, size) tcc_realloc(p, size)
121 #define tal_new(a,b,c)
122 #define tal_delete(a)
124 #if !defined(MEM_DEBUG)
125 #define tal_free(al, p) tal_free_impl(al, p)
126 #define tal_realloc(al, p, size) tal_realloc_impl(&al, p, size)
127 #define TAL_DEBUG_PARAMS
130 //#define TAL_INFO 1 /* collect and dump allocators stats */
131 #define tal_free(al, p) tal_free_impl(al, p, __FILE__, __LINE__)
132 #define tal_realloc(al, p, size) tal_realloc_impl(&al, p, size, __FILE__, __LINE__)
133 #define TAL_DEBUG_PARAMS , const char *file, int line
134 #define TAL_DEBUG_FILE_LEN 15
137 #define TOKSYM_TAL_SIZE (768 * 1024) /* allocator for tiny TokenSym in table_ident */
138 #define TOKSTR_TAL_SIZE (768 * 1024) /* allocator for tiny TokenString instances */
139 #define CSTR_TAL_SIZE (256 * 1024) /* allocator for tiny CString instances */
140 #define TOKSYM_TAL_LIMIT 256 /* prefer unique limits to distinguish allocators debug msgs */
141 #define TOKSTR_TAL_LIMIT 128 /* 32 * sizeof(int) */
142 #define CSTR_TAL_LIMIT 1024
144 typedef struct TinyAlloc
{
150 struct TinyAlloc
*next
, *top
;
159 typedef struct tal_header_t
{
162 int line_num
; /* negative line_num used for double free check */
163 char file_name
[TAL_DEBUG_FILE_LEN
+ 1];
167 /* ------------------------------------------------------------------------- */
169 ST_FUNC TinyAlloc
*tal_new(TinyAlloc
**pal
, size_t limit
, size_t size
)
171 TinyAlloc
*al
= tcc_mallocz(sizeof(TinyAlloc
));
172 al
->p
= al
->buffer
= tcc_malloc(size
);
179 ST_FUNC
void tal_delete(TinyAlloc
*al
)
187 fprintf(stderr
, "limit=%5d, size=%5g MB, nb_peak=%6d, nb_total=%8d, nb_missed=%6d, usage=%5.1f%%\n",
188 al
->limit
, al
->size
/ 1024.0 / 1024.0, al
->nb_peak
, al
->nb_total
, al
->nb_missed
,
189 (al
->peak_p
- al
->buffer
) * 100.0 / al
->size
);
192 if (al
->nb_allocs
> 0) {
194 fprintf(stderr
, "TAL_DEBUG: mem leak %d chunks (limit= %d)\n",
195 al
->nb_allocs
, al
->limit
);
198 tal_header_t
*header
= (tal_header_t
*)p
;
199 if (header
->line_num
> 0) {
200 fprintf(stderr
, " file %s, line %u: %u bytes\n",
201 header
->file_name
, header
->line_num
, header
->size
);
203 p
+= header
->size
+ sizeof(tal_header_t
);
208 tcc_free(al
->buffer
);
214 ST_FUNC
void tal_free_impl(TinyAlloc
*al
, void *p TAL_DEBUG_PARAMS
)
219 if (al
->buffer
<= (uint8_t *)p
&& (uint8_t *)p
< al
->buffer
+ al
->size
) {
221 tal_header_t
*header
= (((tal_header_t
*)p
) - 1);
222 if (header
->line_num
< 0) {
223 fprintf(stderr
, "TAL_DEBUG: file %s, line %u double frees chunk from\n",
225 fprintf(stderr
, " file %s, line %u: %u bytes\n",
226 header
->file_name
, -header
->line_num
, header
->size
);
228 header
->line_num
= -header
->line_num
;
233 } else if (al
->next
) {
241 ST_FUNC
void *tal_realloc_impl(TinyAlloc
**pal
, void *p
, size_t size TAL_DEBUG_PARAMS
)
243 tal_header_t
*header
;
246 size_t adj_size
= (size
+ 3) & -4;
247 TinyAlloc
*al
= *pal
;
250 is_own
= (al
->buffer
<= (uint8_t *)p
&& (uint8_t *)p
< al
->buffer
+ al
->size
);
251 if ((!p
|| is_own
) && size
<= al
->limit
) {
252 if (al
->p
+ adj_size
+ sizeof(tal_header_t
) < al
->buffer
+ al
->size
) {
253 header
= (tal_header_t
*)al
->p
;
254 header
->size
= adj_size
;
256 { int ofs
= strlen(file
) - TAL_DEBUG_FILE_LEN
;
257 strncpy(header
->file_name
, file
+ (ofs
> 0 ? ofs
: 0), TAL_DEBUG_FILE_LEN
);
258 header
->file_name
[TAL_DEBUG_FILE_LEN
] = 0;
259 header
->line_num
= line
; }
261 ret
= al
->p
+ sizeof(tal_header_t
);
262 al
->p
+= adj_size
+ sizeof(tal_header_t
);
264 header
= (((tal_header_t
*)p
) - 1);
265 memcpy(ret
, p
, header
->size
);
267 header
->line_num
= -header
->line_num
;
273 if (al
->nb_peak
< al
->nb_allocs
)
274 al
->nb_peak
= al
->nb_allocs
;
275 if (al
->peak_p
< al
->p
)
282 ret
= tal_realloc(*pal
, 0, size
);
283 header
= (((tal_header_t
*)p
) - 1);
284 memcpy(ret
, p
, header
->size
);
286 header
->line_num
= -header
->line_num
;
293 TinyAlloc
*bottom
= al
, *next
= al
->top
? al
->top
: al
;
295 al
= tal_new(pal
, next
->limit
, next
->size
* 2);
303 ret
= tcc_malloc(size
);
304 header
= (((tal_header_t
*)p
) - 1);
305 memcpy(ret
, p
, header
->size
);
307 header
->line_num
= -header
->line_num
;
309 } else if (al
->next
) {
313 ret
= tcc_realloc(p
, size
);
322 /* ------------------------------------------------------------------------- */
323 /* CString handling */
324 static void cstr_realloc(CString
*cstr
, int new_size
)
328 size
= cstr
->size_allocated
;
330 size
= 8; /* no need to allocate a too small first string */
331 while (size
< new_size
)
333 cstr
->data
= tal_realloc(cstr_alloc
, cstr
->data
, size
);
334 cstr
->size_allocated
= size
;
338 ST_INLN
void cstr_ccat(CString
*cstr
, int ch
)
341 size
= cstr
->size
+ 1;
342 if (size
> cstr
->size_allocated
)
343 cstr_realloc(cstr
, size
);
344 ((unsigned char *)cstr
->data
)[size
- 1] = ch
;
348 ST_FUNC
void cstr_cat(CString
*cstr
, const char *str
, int len
)
352 len
= strlen(str
) + 1 + len
;
353 size
= cstr
->size
+ len
;
354 if (size
> cstr
->size_allocated
)
355 cstr_realloc(cstr
, size
);
356 memmove(((unsigned char *)cstr
->data
) + cstr
->size
, str
, len
);
360 /* add a wide char */
361 ST_FUNC
void cstr_wccat(CString
*cstr
, int ch
)
364 size
= cstr
->size
+ sizeof(nwchar_t
);
365 if (size
> cstr
->size_allocated
)
366 cstr_realloc(cstr
, size
);
367 *(nwchar_t
*)(((unsigned char *)cstr
->data
) + size
- sizeof(nwchar_t
)) = ch
;
371 ST_FUNC
void cstr_new(CString
*cstr
)
373 memset(cstr
, 0, sizeof(CString
));
376 /* free string and reset it to NULL */
377 ST_FUNC
void cstr_free(CString
*cstr
)
379 tal_free(cstr_alloc
, cstr
->data
);
383 /* reset string to empty */
384 ST_FUNC
void cstr_reset(CString
*cstr
)
390 static void add_char(CString
*cstr
, int c
)
392 if (c
== '\'' || c
== '\"' || c
== '\\') {
393 /* XXX: could be more precise if char or string */
394 cstr_ccat(cstr
, '\\');
396 if (c
>= 32 && c
<= 126) {
399 cstr_ccat(cstr
, '\\');
401 cstr_ccat(cstr
, 'n');
403 cstr_ccat(cstr
, '0' + ((c
>> 6) & 7));
404 cstr_ccat(cstr
, '0' + ((c
>> 3) & 7));
405 cstr_ccat(cstr
, '0' + (c
& 7));
410 /* ------------------------------------------------------------------------- */
411 /* allocate a new token */
412 static TokenSym
*tok_alloc_new(TokenSym
**pts
, const char *str
, int len
)
414 TokenSym
*ts
, **ptable
;
417 if (tok_ident
>= SYM_FIRST_ANOM
)
418 tcc_error("memory full (symbols)");
420 /* expand token table if needed */
421 i
= tok_ident
- TOK_IDENT
;
422 if ((i
% TOK_ALLOC_INCR
) == 0) {
423 ptable
= tcc_realloc(table_ident
, (i
+ TOK_ALLOC_INCR
) * sizeof(TokenSym
*));
424 table_ident
= ptable
;
427 ts
= tal_realloc(toksym_alloc
, 0, sizeof(TokenSym
) + len
);
429 ts
->tok
= tok_ident
++;
430 ts
->sym_define
= NULL
;
431 ts
->sym_label
= NULL
;
432 ts
->sym_struct
= NULL
;
433 ts
->sym_identifier
= NULL
;
435 ts
->hash_next
= NULL
;
436 memcpy(ts
->str
, str
, len
);
442 #define TOK_HASH_INIT 1
443 #define TOK_HASH_FUNC(h, c) ((h) + ((h) << 5) + ((h) >> 27) + (c))
446 /* find a token and add it if not found */
447 ST_FUNC TokenSym
*tok_alloc(const char *str
, int len
)
455 h
= TOK_HASH_FUNC(h
, ((unsigned char *)str
)[i
]);
456 h
&= (TOK_HASH_SIZE
- 1);
458 pts
= &hash_ident
[h
];
463 if (ts
->len
== len
&& !memcmp(ts
->str
, str
, len
))
465 pts
= &(ts
->hash_next
);
467 return tok_alloc_new(pts
, str
, len
);
470 /* XXX: buffer overflow */
471 /* XXX: float tokens */
472 ST_FUNC
const char *get_tok_str(int v
, CValue
*cv
)
477 cstr_reset(&cstr_buf
);
485 /* XXX: not quite exact, but only useful for testing */
487 sprintf(p
, "%u", (unsigned)cv
->i
);
489 sprintf(p
, "%llu", (unsigned long long)cv
->i
);
493 cstr_ccat(&cstr_buf
, 'L');
495 cstr_ccat(&cstr_buf
, '\'');
496 add_char(&cstr_buf
, cv
->i
);
497 cstr_ccat(&cstr_buf
, '\'');
498 cstr_ccat(&cstr_buf
, '\0');
502 return (char*)cv
->str
.data
;
504 cstr_ccat(&cstr_buf
, 'L');
506 cstr_ccat(&cstr_buf
, '\"');
508 len
= cv
->str
.size
- 1;
510 add_char(&cstr_buf
, ((unsigned char *)cv
->str
.data
)[i
]);
512 len
= (cv
->str
.size
/ sizeof(nwchar_t
)) - 1;
514 add_char(&cstr_buf
, ((nwchar_t
*)cv
->str
.data
)[i
]);
516 cstr_ccat(&cstr_buf
, '\"');
517 cstr_ccat(&cstr_buf
, '\0');
521 cstr_cat(&cstr_buf
, "<float>", 0);
524 cstr_cat(&cstr_buf
, "<double>", 0);
527 cstr_cat(&cstr_buf
, "<long double>", 0);
530 cstr_cat(&cstr_buf
, "<linenumber>", 0);
533 /* above tokens have value, the ones below don't */
542 return strcpy(p
, "...");
544 return strcpy(p
, "<<=");
546 return strcpy(p
, ">>=");
549 /* search in two bytes table */
550 const unsigned char *q
= tok_two_chars
;
556 return cstr_buf
.data
;
561 sprintf(cstr_buf
.data
, "<%02x>", v
);
562 return cstr_buf
.data
;
567 } else if (v
< tok_ident
) {
568 return table_ident
[v
- TOK_IDENT
]->str
;
569 } else if (v
>= SYM_FIRST_ANOM
) {
570 /* special name for anonymous symbol */
571 sprintf(p
, "L.%u", v
- SYM_FIRST_ANOM
);
573 /* should never happen */
578 return cstr_buf
.data
;
581 /* return the current character, handling end of block if necessary
583 ST_FUNC
int handle_eob(void)
585 BufferedFile
*bf
= file
;
588 /* only tries to read if really end of buffer */
589 if (bf
->buf_ptr
>= bf
->buf_end
) {
591 #if defined(PARSE_DEBUG)
596 len
= read(bf
->fd
, bf
->buffer
, len
);
603 bf
->buf_ptr
= bf
->buffer
;
604 bf
->buf_end
= bf
->buffer
+ len
;
605 *bf
->buf_end
= CH_EOB
;
607 if (bf
->buf_ptr
< bf
->buf_end
) {
608 return bf
->buf_ptr
[0];
610 bf
->buf_ptr
= bf
->buf_end
;
615 /* read next char from current input file and handle end of input buffer */
616 ST_INLN
void inp(void)
618 ch
= *(++(file
->buf_ptr
));
619 /* end of buffer/file handling */
624 /* handle '\[\r]\n' */
625 static int handle_stray_noerror(void)
632 } else if (ch
== '\r') {
646 static void handle_stray(void)
648 if (handle_stray_noerror())
649 tcc_error("stray '\\' in program");
652 /* skip the stray and handle the \\n case. Output an error if
653 incorrect char after the stray */
654 static int handle_stray1(uint8_t *p
)
659 if (p
>= file
->buf_end
) {
666 if (handle_stray_noerror()) {
667 if (!(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
668 tcc_error("stray '\\' in program");
669 *--file
->buf_ptr
= '\\';
676 /* handle just the EOB case, but not stray */
677 #define PEEKC_EOB(c, p)\
688 /* handle the complicated stray case */
694 c = handle_stray1(p);\
699 /* input with '\[\r]\n' handling. Note that this function cannot
700 handle other characters after '\', so you cannot call it inside
701 strings or comments */
702 ST_FUNC
void minp(void)
709 /* single line C++ comments */
710 static uint8_t *parse_line_comment(uint8_t *p
)
718 if (c
== '\n' || c
== CH_EOF
) {
720 } else if (c
== '\\') {
729 } else if (c
== '\r') {
747 ST_FUNC
uint8_t *parse_comment(uint8_t *p
)
756 if (c
== '\n' || c
== '*' || c
== '\\')
760 if (c
== '\n' || c
== '*' || c
== '\\')
764 /* now we can handle all the cases */
768 } else if (c
== '*') {
774 } else if (c
== '/') {
776 } else if (c
== '\\') {
781 tcc_error("unexpected end of file in comment");
783 /* skip '\[\r]\n', otherwise just skip the stray */
789 } else if (c
== '\r') {
806 /* stray, eob or eof */
811 tcc_error("unexpected end of file in comment");
812 } else if (c
== '\\') {
824 static inline void skip_spaces(void)
826 while (isidnum_table
[ch
- CH_EOF
] & IS_SPC
)
830 static inline int check_space(int t
, int *spc
)
832 if (t
< 256 && (isidnum_table
[t
- CH_EOF
] & IS_SPC
)) {
841 /* parse a string without interpreting escapes */
842 static uint8_t *parse_pp_string(uint8_t *p
,
843 int sep
, CString
*str
)
851 } else if (c
== '\\') {
857 /* XXX: indicate line number of start of string */
858 tcc_error("missing terminating %c character", sep
);
859 } else if (c
== '\\') {
860 /* escape : just skip \[\r]\n */
865 } else if (c
== '\r') {
868 expect("'\n' after '\r'");
871 } else if (c
== CH_EOF
) {
872 goto unterminated_string
;
875 cstr_ccat(str
, '\\');
881 } else if (c
== '\n') {
884 } else if (c
== '\r') {
888 cstr_ccat(str
, '\r');
904 /* skip block of text until #else, #elif or #endif. skip also pairs of
906 static void preprocess_skip(void)
908 int a
, start_of_line
, c
, in_warn_or_error
;
915 in_warn_or_error
= 0;
936 } else if (c
== '\\') {
937 ch
= file
->buf_ptr
[0];
938 handle_stray_noerror();
945 if (in_warn_or_error
)
947 p
= parse_pp_string(p
, c
, NULL
);
951 if (in_warn_or_error
)
958 p
= parse_comment(p
);
959 } else if (ch
== '/') {
960 p
= parse_line_comment(p
);
970 (tok
== TOK_ELSE
|| tok
== TOK_ELIF
|| tok
== TOK_ENDIF
))
972 if (tok
== TOK_IF
|| tok
== TOK_IFDEF
|| tok
== TOK_IFNDEF
)
974 else if (tok
== TOK_ENDIF
)
976 else if( tok
== TOK_ERROR
|| tok
== TOK_WARNING
)
977 in_warn_or_error
= 1;
978 else if (tok
== TOK_LINEFEED
)
980 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
981 p
= parse_line_comment(p
- 1);
982 } else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
983 p
= parse_line_comment(p
- 1);
996 /* ParseState handling */
998 /* XXX: currently, no include file info is stored. Thus, we cannot display
999 accurate messages if the function or data definition spans multiple
1002 /* save current parse state in 's' */
1003 ST_FUNC
void save_parse_state(ParseState
*s
)
1005 s
->line_num
= file
->line_num
;
1006 s
->macro_ptr
= macro_ptr
;
1011 /* restore parse state from 's' */
1012 ST_FUNC
void restore_parse_state(ParseState
*s
)
1014 file
->line_num
= s
->line_num
;
1015 macro_ptr
= s
->macro_ptr
;
1020 /* return the number of additional 'ints' necessary to store the
1022 static inline int tok_size(const int *p
)
1037 return 1 + ((sizeof(CString
) + ((CString
*)(p
+1))->size
+ 3) >> 2);
1043 return 1 + LDOUBLE_SIZE
/ 4;
1049 /* token string handling */
1051 ST_INLN
void tok_str_new(TokenString
*s
)
1055 s
->allocated_len
= 0;
1056 s
->last_line_num
= -1;
1059 ST_FUNC TokenString
*tok_str_alloc(void)
1061 TokenString
*str
= tal_realloc(tokstr_alloc
, 0, sizeof *str
);
1066 ST_FUNC
int *tok_str_dup(TokenString
*s
)
1070 str
= tal_realloc(tokstr_alloc
, 0, s
->len
* sizeof(int));
1071 memcpy(str
, s
->str
, s
->len
* sizeof(int));
1075 ST_FUNC
void tok_str_free_str(int *str
)
1077 tal_free(tokstr_alloc
, str
);
1080 ST_FUNC
void tok_str_free(TokenString
*str
)
1082 tok_str_free_str(str
->str
);
1083 tal_free(tokstr_alloc
, str
);
1086 ST_FUNC
int *tok_str_realloc(TokenString
*s
, int new_size
)
1090 size
= s
->allocated_len
;
1093 while (size
< new_size
)
1095 if (size
> s
->allocated_len
) {
1096 str
= tal_realloc(tokstr_alloc
, s
->str
, size
* sizeof(int));
1097 s
->allocated_len
= size
;
1103 ST_FUNC
void tok_str_add(TokenString
*s
, int t
)
1109 if (len
>= s
->allocated_len
)
1110 str
= tok_str_realloc(s
, len
+ 1);
1115 ST_FUNC
void begin_macro(TokenString
*str
, int alloc
)
1118 str
->prev
= macro_stack
;
1119 str
->prev_ptr
= macro_ptr
;
1120 macro_ptr
= str
->str
;
1124 ST_FUNC
void end_macro(void)
1126 TokenString
*str
= macro_stack
;
1127 macro_stack
= str
->prev
;
1128 macro_ptr
= str
->prev_ptr
;
1129 if (str
->alloc
== 2) {
1130 str
->alloc
= 3; /* just mark as finished */
1136 static void tok_str_add2(TokenString
*s
, int t
, CValue
*cv
)
1143 /* allocate space for worst case */
1144 if (len
+ TOK_MAX_SIZE
>= s
->allocated_len
)
1145 str
= tok_str_realloc(s
, len
+ TOK_MAX_SIZE
+ 1);
1154 str
[len
++] = cv
->tab
[0];
1161 /* Insert the string into the int array. */
1163 1 + (cv
->str
.size
+ sizeof(int) - 1) / sizeof(int);
1164 if (len
+ nb_words
>= s
->allocated_len
)
1165 str
= tok_str_realloc(s
, len
+ nb_words
+ 1);
1166 str
[len
] = cv
->str
.size
;
1167 memcpy(&str
[len
+ 1], cv
->str
.data
, cv
->str
.size
);
1174 #if LDOUBLE_SIZE == 8
1177 str
[len
++] = cv
->tab
[0];
1178 str
[len
++] = cv
->tab
[1];
1180 #if LDOUBLE_SIZE == 12
1182 str
[len
++] = cv
->tab
[0];
1183 str
[len
++] = cv
->tab
[1];
1184 str
[len
++] = cv
->tab
[2];
1185 #elif LDOUBLE_SIZE == 16
1187 str
[len
++] = cv
->tab
[0];
1188 str
[len
++] = cv
->tab
[1];
1189 str
[len
++] = cv
->tab
[2];
1190 str
[len
++] = cv
->tab
[3];
1191 #elif LDOUBLE_SIZE != 8
1192 #error add long double size support
1201 /* add the current parse token in token string 's' */
1202 ST_FUNC
void tok_str_add_tok(TokenString
*s
)
1206 /* save line number info */
1207 if (file
->line_num
!= s
->last_line_num
) {
1208 s
->last_line_num
= file
->line_num
;
1209 cval
.i
= s
->last_line_num
;
1210 tok_str_add2(s
, TOK_LINENUM
, &cval
);
1212 tok_str_add2(s
, tok
, &tokc
);
1215 /* get a token from an integer array and increment pointer
1216 accordingly. we code it as a macro to avoid pointer aliasing. */
1217 static inline void TOK_GET(int *t
, const int **pp
, CValue
*cv
)
1236 cv
->str
.size
= *p
++;
1238 p
+= (cv
->str
.size
+ sizeof(int) - 1) / sizeof(int);
1246 #if LDOUBLE_SIZE == 16
1248 #elif LDOUBLE_SIZE == 12
1250 #elif LDOUBLE_SIZE == 8
1253 # error add long double size support
1266 /* Calling this function is expensive, but it is not possible
1267 to read a token string backwards. */
1268 static int tok_last(const int *str0
, const int *str1
)
1270 const int *str
= str0
;
1275 TOK_GET(&tok
, &str
, &cval
);
1279 static int macro_is_equal(const int *a
, const int *b
)
1288 /* first time preallocate static cstr_buf, next time only reset position to start */
1289 cstr_reset(&cstr_buf
);
1290 TOK_GET(&t
, &a
, &cv
);
1291 cstr_cat(&cstr_buf
, get_tok_str(t
, &cv
), 0);
1292 TOK_GET(&t
, &b
, &cv
);
1293 if (strcmp(cstr_buf
.data
, get_tok_str(t
, &cv
)))
1299 /* defines handling */
1300 ST_INLN
void define_push(int v
, int macro_type
, int *str
, Sym
*first_arg
)
1305 s
= sym_push2(&define_stack
, v
, macro_type
, 0);
1307 s
->next
= first_arg
;
1308 table_ident
[v
- TOK_IDENT
]->sym_define
= s
;
1310 if (o
&& !macro_is_equal(o
->d
, s
->d
))
1311 tcc_warning("%s redefined", get_tok_str(v
, NULL
));
1314 /* undefined a define symbol. Its name is just set to zero */
1315 ST_FUNC
void define_undef(Sym
*s
)
1318 if (v
>= TOK_IDENT
&& v
< tok_ident
)
1319 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1322 ST_INLN Sym
*define_find(int v
)
1325 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1327 return table_ident
[v
]->sym_define
;
1330 /* free define stack until top reaches 'b' */
1331 ST_FUNC
void free_defines(Sym
*b
)
1333 while (define_stack
!= b
) {
1334 Sym
*top
= define_stack
;
1335 define_stack
= top
->prev
;
1336 tok_str_free_str(top
->d
);
1341 /* restore remaining (-D or predefined) symbols */
1344 if (v
>= TOK_IDENT
&& v
< tok_ident
) {
1345 Sym
**d
= &table_ident
[v
- TOK_IDENT
]->sym_define
;
1354 ST_FUNC Sym
*label_find(int v
)
1357 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1359 return table_ident
[v
]->sym_label
;
1362 ST_FUNC Sym
*label_push(Sym
**ptop
, int v
, int flags
)
1365 s
= sym_push2(ptop
, v
, 0, 0);
1367 ps
= &table_ident
[v
- TOK_IDENT
]->sym_label
;
1368 if (ptop
== &global_label_stack
) {
1369 /* modify the top most local identifier, so that
1370 sym_identifier will point to 's' when popped */
1372 ps
= &(*ps
)->prev_tok
;
1379 /* pop labels until element last is reached. Look if any labels are
1380 undefined. Define symbols if '&&label' was used. */
1381 ST_FUNC
void label_pop(Sym
**ptop
, Sym
*slast
)
1384 for(s
= *ptop
; s
!= slast
; s
= s1
) {
1386 if (s
->r
== LABEL_DECLARED
) {
1387 tcc_warning("label '%s' declared but not used", get_tok_str(s
->v
, NULL
));
1388 } else if (s
->r
== LABEL_FORWARD
) {
1389 tcc_error("label '%s' used but not defined",
1390 get_tok_str(s
->v
, NULL
));
1393 /* define corresponding symbol. A size of
1395 put_extern_sym(s
, cur_text_section
, s
->jnext
, 1);
1399 table_ident
[s
->v
- TOK_IDENT
]->sym_label
= s
->prev_tok
;
1405 /* eval an expression for #if/#elif */
1406 static int expr_preprocess(void)
1411 str
= tok_str_alloc();
1412 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1413 next(); /* do macro subst */
1414 if (tok
== TOK_DEFINED
) {
1419 c
= define_find(tok
) != 0;
1424 } else if (tok
>= TOK_IDENT
) {
1425 /* if undefined macro */
1429 tok_str_add_tok(str
);
1431 tok_str_add(str
, -1); /* simulate end of file */
1432 tok_str_add(str
, 0);
1433 /* now evaluate C constant expression */
1434 begin_macro(str
, 1);
1442 /* parse after #define */
1443 ST_FUNC
void parse_define(void)
1445 Sym
*s
, *first
, **ps
;
1446 int v
, t
, varg
, is_vaargs
, spc
;
1447 int saved_parse_flags
= parse_flags
;
1451 tcc_error("invalid macro name '%s'", get_tok_str(tok
, &tokc
));
1452 /* XXX: should check if same macro (ANSI) */
1455 /* '(' must be just after macro definition for MACRO_FUNC */
1456 parse_flags
|= PARSE_FLAG_SPACES
;
1459 /* must be able to parse TOK_DOTS (in asm mode '.' can be part of identifier) */
1460 parse_flags
&= ~PARSE_FLAG_ASM_FILE
;
1461 isidnum_table
['.' - CH_EOF
] = 0;
1464 if (tok
!= ')') for (;;) {
1468 if (varg
== TOK_DOTS
) {
1469 varg
= TOK___VA_ARGS__
;
1471 } else if (tok
== TOK_DOTS
&& gnu_ext
) {
1475 if (varg
< TOK_IDENT
)
1477 tcc_error("bad macro parameter list");
1478 s
= sym_push2(&define_stack
, varg
| SYM_FIELD
, is_vaargs
, 0);
1483 if (tok
!= ',' || is_vaargs
)
1489 parse_flags
|= (saved_parse_flags
& PARSE_FLAG_ASM_FILE
);
1490 isidnum_table
['.' - CH_EOF
] =
1491 (parse_flags
& PARSE_FLAG_ASM_FILE
) ? IS_ID
: 0;
1496 parse_flags
|= PARSE_FLAG_ACCEPT_STRAYS
| PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
;
1497 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1498 /* remove spaces around ## and after '#' */
1499 if (TOK_TWOSHARPS
== tok
) {
1505 } else if ('#' == tok
) {
1507 } else if (check_space(tok
, &spc
)) {
1510 tok_str_add2(&tokstr_buf
, tok
, &tokc
);
1515 parse_flags
= saved_parse_flags
;
1517 --tokstr_buf
.len
; /* remove trailing space */
1518 tok_str_add(&tokstr_buf
, 0);
1521 tcc_error("'##' cannot appear at either end of macro");
1522 define_push(v
, t
, tok_str_dup(&tokstr_buf
), first
);
1525 static CachedInclude
*search_cached_include(TCCState
*s1
, const char *filename
, int add
)
1527 const unsigned char *s
;
1533 s
= (unsigned char *) filename
;
1536 h
= TOK_HASH_FUNC(h
, toup(*s
));
1538 h
= TOK_HASH_FUNC(h
, *s
);
1542 h
&= (CACHED_INCLUDES_HASH_SIZE
- 1);
1544 i
= s1
->cached_includes_hash
[h
];
1548 e
= s1
->cached_includes
[i
- 1];
1549 if (0 == PATHCMP(e
->filename
, filename
))
1556 e
= tcc_malloc(sizeof(CachedInclude
) + strlen(filename
));
1557 strcpy(e
->filename
, filename
);
1558 e
->ifndef_macro
= e
->once
= 0;
1559 dynarray_add((void ***)&s1
->cached_includes
, &s1
->nb_cached_includes
, e
);
1560 /* add in hash table */
1561 e
->hash_next
= s1
->cached_includes_hash
[h
];
1562 s1
->cached_includes_hash
[h
] = s1
->nb_cached_includes
;
1564 printf("adding cached '%s'\n", filename
);
1569 static void pragma_parse(TCCState
*s1
)
1572 if (tok
== TOK_push_macro
|| tok
== TOK_pop_macro
) {
1576 if (next(), tok
!= '(')
1578 if (next(), tok
!= TOK_STR
)
1580 v
= tok_alloc(tokc
.str
.data
, tokc
.str
.size
- 1)->tok
;
1581 if (next(), tok
!= ')')
1583 if (t
== TOK_push_macro
) {
1584 while (NULL
== (s
= define_find(v
)))
1585 define_push(v
, 0, NULL
, NULL
);
1586 s
->type
.ref
= s
; /* set push boundary */
1588 for (s
= define_stack
; s
; s
= s
->prev
)
1589 if (s
->v
== v
&& s
->type
.ref
== s
) {
1595 table_ident
[v
- TOK_IDENT
]->sym_define
= s
->d
? s
: NULL
;
1597 tcc_warning("unbalanced #pragma pop_macro");
1598 pp_debug_tok
= t
, pp_debug_symv
= v
;
1600 } else if (tok
== TOK_once
) {
1601 search_cached_include(s1
, file
->filename
, 1)->once
= pp_once
;
1603 } else if (s1
->ppfp
) {
1604 /* tcc -E: keep pragmas below unchanged */
1606 unget_tok(TOK_PRAGMA
);
1608 unget_tok(TOK_LINEFEED
);
1610 } else if (tok
== TOK_pack
) {
1612 #pragma pack(1) // set
1613 #pragma pack() // reset to default
1614 #pragma pack(push,1) // push & set
1615 #pragma pack(pop) // restore previous */
1618 if (tok
== TOK_ASM_pop
) {
1620 if (s1
->pack_stack_ptr
<= s1
->pack_stack
) {
1622 tcc_error("out of pack stack");
1624 s1
->pack_stack_ptr
--;
1628 if (tok
== TOK_ASM_push
) {
1630 if (s1
->pack_stack_ptr
>= s1
->pack_stack
+ PACK_STACK_SIZE
- 1)
1632 s1
->pack_stack_ptr
++;
1635 if (tok
!= TOK_CINT
)
1638 if (val
< 1 || val
> 16 || (val
& (val
- 1)) != 0)
1642 *s1
->pack_stack_ptr
= val
;
1647 } else if (tok
== TOK_comment
) {
1657 file
= tcc_strdup((char *)tokc
.str
.data
);
1658 dynarray_add((void ***)&s1
->pragma_libs
, &s1
->nb_pragma_libs
, file
);
1664 if (s1
->warn_unsupported
)
1665 tcc_warning("#pragma %s is ignored", get_tok_str(tok
, &tokc
));
1670 tcc_error("malformed #pragma directive");
1674 /* is_bof is true if first non space token at beginning of file */
1675 ST_FUNC
void preprocess(int is_bof
)
1677 TCCState
*s1
= tcc_state
;
1678 int i
, c
, n
, saved_parse_flags
;
1682 saved_parse_flags
= parse_flags
;
1683 parse_flags
= PARSE_FLAG_PREPROCESS
1684 | PARSE_FLAG_TOK_NUM
1685 | PARSE_FLAG_TOK_STR
1686 | PARSE_FLAG_LINEFEED
1687 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
1696 pp_debug_symv
= tok
;
1702 pp_debug_symv
= tok
;
1703 s
= define_find(tok
);
1704 /* undefine symbol by putting an invalid name */
1709 case TOK_INCLUDE_NEXT
:
1710 ch
= file
->buf_ptr
[0];
1711 /* XXX: incorrect if comments : use next_nomacro with a special mode */
1716 } else if (ch
== '\"') {
1721 while (ch
!= c
&& ch
!= '\n' && ch
!= CH_EOF
) {
1722 if ((q
- buf
) < sizeof(buf
) - 1)
1725 if (handle_stray_noerror() == 0)
1733 /* eat all spaces and comments after include */
1734 /* XXX: slightly incorrect */
1735 while (ch1
!= '\n' && ch1
!= CH_EOF
)
1739 /* computed #include : either we have only strings or
1740 we have anything enclosed in '<>' */
1743 if (tok
== TOK_STR
) {
1744 while (tok
!= TOK_LINEFEED
) {
1745 if (tok
!= TOK_STR
) {
1747 tcc_error("'#include' expects \"FILENAME\" or <FILENAME>");
1749 pstrcat(buf
, sizeof(buf
), (char *)tokc
.str
.data
);
1755 while (tok
!= TOK_LINEFEED
) {
1756 pstrcat(buf
, sizeof(buf
), get_tok_str(tok
, &tokc
));
1760 /* check syntax and remove '<>' */
1761 if (len
< 2 || buf
[0] != '<' || buf
[len
- 1] != '>')
1762 goto include_syntax
;
1763 memmove(buf
, buf
+ 1, len
- 2);
1764 buf
[len
- 2] = '\0';
1769 if (s1
->include_stack_ptr
>= s1
->include_stack
+ INCLUDE_STACK_SIZE
)
1770 tcc_error("#include recursion too deep");
1771 /* store current file in stack, but increment stack later below */
1772 *s1
->include_stack_ptr
= file
;
1773 i
= tok
== TOK_INCLUDE_NEXT
? file
->include_next_index
: 0;
1774 n
= 2 + s1
->nb_include_paths
+ s1
->nb_sysinclude_paths
;
1775 for (; i
< n
; ++i
) {
1776 char buf1
[sizeof file
->filename
];
1781 /* check absolute include path */
1782 if (!IS_ABSPATH(buf
))
1786 } else if (i
== 1) {
1787 /* search in file's dir if "header.h" */
1790 path
= file
->filename
;
1791 pstrncpy(buf1
, path
, tcc_basename(path
) - path
);
1794 /* search in all the include paths */
1795 int j
= i
- 2, k
= j
- s1
->nb_include_paths
;
1796 path
= k
< 0 ? s1
->include_paths
[j
] : s1
->sysinclude_paths
[k
];
1797 pstrcpy(buf1
, sizeof(buf1
), path
);
1798 pstrcat(buf1
, sizeof(buf1
), "/");
1801 pstrcat(buf1
, sizeof(buf1
), buf
);
1802 e
= search_cached_include(s1
, buf1
, 0);
1803 if (e
&& (define_find(e
->ifndef_macro
) || e
->once
== pp_once
)) {
1804 /* no need to parse the include because the 'ifndef macro'
1805 is defined (or had #pragma once) */
1807 printf("%s: skipping cached %s\n", file
->filename
, buf1
);
1812 if (tcc_open(s1
, buf1
) < 0)
1815 file
->include_next_index
= i
+ 1;
1817 printf("%s: including %s\n", file
->prev
->filename
, file
->filename
);
1819 /* update target deps */
1820 dynarray_add((void ***)&s1
->target_deps
, &s1
->nb_target_deps
,
1822 /* push current file in stack */
1823 ++s1
->include_stack_ptr
;
1824 /* add include file debug info */
1826 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1827 tok_flags
|= TOK_FLAG_BOF
| TOK_FLAG_BOL
;
1828 ch
= file
->buf_ptr
[0];
1831 tcc_error("include file '%s' not found", buf
);
1838 c
= expr_preprocess();
1844 if (tok
< TOK_IDENT
)
1845 tcc_error("invalid argument for '#if%sdef'", c
? "n" : "");
1849 printf("#ifndef %s\n", get_tok_str(tok
, NULL
));
1851 file
->ifndef_macro
= tok
;
1854 c
= (define_find(tok
) != 0) ^ c
;
1856 if (s1
->ifdef_stack_ptr
>= s1
->ifdef_stack
+ IFDEF_STACK_SIZE
)
1857 tcc_error("memory full (ifdef)");
1858 *s1
->ifdef_stack_ptr
++ = c
;
1861 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1862 tcc_error("#else without matching #if");
1863 if (s1
->ifdef_stack_ptr
[-1] & 2)
1864 tcc_error("#else after #else");
1865 c
= (s1
->ifdef_stack_ptr
[-1] ^= 3);
1868 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1869 tcc_error("#elif without matching #if");
1870 c
= s1
->ifdef_stack_ptr
[-1];
1872 tcc_error("#elif after #else");
1873 /* last #if/#elif expression was true: we skip */
1877 c
= expr_preprocess();
1878 s1
->ifdef_stack_ptr
[-1] = c
;
1881 if (s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
+ 1)
1882 file
->ifndef_macro
= 0;
1891 if (s1
->ifdef_stack_ptr
<= file
->ifdef_stack_ptr
)
1892 tcc_error("#endif without matching #if");
1893 s1
->ifdef_stack_ptr
--;
1894 /* '#ifndef macro' was at the start of file. Now we check if
1895 an '#endif' is exactly at the end of file */
1896 if (file
->ifndef_macro
&&
1897 s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
) {
1898 file
->ifndef_macro_saved
= file
->ifndef_macro
;
1899 /* need to set to zero to avoid false matches if another
1900 #ifndef at middle of file */
1901 file
->ifndef_macro
= 0;
1902 while (tok
!= TOK_LINEFEED
)
1904 tok_flags
|= TOK_FLAG_ENDIF
;
1909 n
= strtoul((char*)tokc
.str
.data
, &q
, 10);
1913 if (tok
!= TOK_CINT
)
1915 tcc_error("wrong #line format");
1919 if (tok
!= TOK_LINEFEED
) {
1921 pstrcpy(file
->filename
, sizeof(file
->filename
), (char *)tokc
.str
.data
);
1922 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
1929 total_lines
+= file
->line_num
- n
;
1932 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1937 ch
= file
->buf_ptr
[0];
1940 while (ch
!= '\n' && ch
!= CH_EOF
) {
1941 if ((q
- buf
) < sizeof(buf
) - 1)
1944 if (handle_stray_noerror() == 0)
1951 tcc_error("#error %s", buf
);
1953 tcc_warning("#warning %s", buf
);
1961 /* ignore gas line comment in an 'S' file. */
1962 if (saved_parse_flags
& PARSE_FLAG_ASM_FILE
)
1964 if (tok
== '!' && is_bof
)
1965 /* '!' is ignored at beginning to allow C scripts. */
1967 tcc_warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok
, &tokc
));
1969 file
->buf_ptr
= parse_line_comment(file
->buf_ptr
- 1);
1972 /* ignore other preprocess commands or #! for C scripts */
1973 while (tok
!= TOK_LINEFEED
)
1976 parse_flags
= saved_parse_flags
;
1979 /* evaluate escape codes in a string. */
1980 static void parse_escape_string(CString
*outstr
, const uint8_t *buf
, int is_long
)
1995 case '0': case '1': case '2': case '3':
1996 case '4': case '5': case '6': case '7':
1997 /* at most three octal digits */
2002 n
= n
* 8 + c
- '0';
2006 n
= n
* 8 + c
- '0';
2011 goto add_char_nonext
;
2019 if (c
>= 'a' && c
<= 'f')
2021 else if (c
>= 'A' && c
<= 'F')
2031 goto add_char_nonext
;
2055 goto invalid_escape
;
2065 if (c
>= '!' && c
<= '~')
2066 tcc_warning("unknown escape sequence: \'\\%c\'", c
);
2068 tcc_warning("unknown escape sequence: \'\\x%x\'", c
);
2075 cstr_ccat(outstr
, c
);
2077 cstr_wccat(outstr
, c
);
2079 /* add a trailing '\0' */
2081 cstr_ccat(outstr
, '\0');
2083 cstr_wccat(outstr
, '\0');
2086 static void parse_string(const char *s
, int len
)
2088 uint8_t buf
[1000], *p
= buf
;
2091 if ((is_long
= *s
== 'L'))
2095 if (len
>= sizeof buf
)
2096 p
= tcc_malloc(len
+ 1);
2100 cstr_reset(&tokcstr
);
2101 parse_escape_string(&tokcstr
, p
, is_long
);
2107 /* XXX: make it portable */
2111 char_size
= sizeof(nwchar_t
);
2112 if (tokcstr
.size
<= char_size
)
2113 tcc_error("empty character constant");
2114 if (tokcstr
.size
> 2 * char_size
)
2115 tcc_warning("multi-character character constant");
2117 tokc
.i
= *(int8_t *)tokcstr
.data
;
2120 tokc
.i
= *(nwchar_t
*)tokcstr
.data
;
2124 tokc
.str
.size
= tokcstr
.size
;
2125 tokc
.str
.data
= tokcstr
.data
;
2133 /* we use 64 bit numbers */
2136 /* bn = (bn << shift) | or_val */
2137 static void bn_lshift(unsigned int *bn
, int shift
, int or_val
)
2141 for(i
=0;i
<BN_SIZE
;i
++) {
2143 bn
[i
] = (v
<< shift
) | or_val
;
2144 or_val
= v
>> (32 - shift
);
2148 static void bn_zero(unsigned int *bn
)
2151 for(i
=0;i
<BN_SIZE
;i
++) {
2156 /* parse number in null terminated string 'p' and return it in the
2158 static void parse_number(const char *p
)
2160 int b
, t
, shift
, frac_bits
, s
, exp_val
, ch
;
2162 unsigned int bn
[BN_SIZE
];
2173 goto float_frac_parse
;
2174 } else if (t
== '0') {
2175 if (ch
== 'x' || ch
== 'X') {
2179 } else if (tcc_ext
&& (ch
== 'b' || ch
== 'B')) {
2185 /* parse all digits. cannot check octal numbers at this stage
2186 because of floating point constants */
2188 if (ch
>= 'a' && ch
<= 'f')
2190 else if (ch
>= 'A' && ch
<= 'F')
2198 if (q
>= token_buf
+ STRING_MAX_SIZE
) {
2200 tcc_error("number too long");
2206 ((ch
== 'e' || ch
== 'E') && b
== 10) ||
2207 ((ch
== 'p' || ch
== 'P') && (b
== 16 || b
== 2))) {
2209 /* NOTE: strtox should support that for hexa numbers, but
2210 non ISOC99 libcs do not support it, so we prefer to do
2212 /* hexadecimal or binary floats */
2213 /* XXX: handle overflows */
2225 } else if (t
>= 'a') {
2227 } else if (t
>= 'A') {
2232 bn_lshift(bn
, shift
, t
);
2239 if (t
>= 'a' && t
<= 'f') {
2241 } else if (t
>= 'A' && t
<= 'F') {
2243 } else if (t
>= '0' && t
<= '9') {
2249 tcc_error("invalid digit");
2250 bn_lshift(bn
, shift
, t
);
2255 if (ch
!= 'p' && ch
!= 'P')
2262 } else if (ch
== '-') {
2266 if (ch
< '0' || ch
> '9')
2267 expect("exponent digits");
2268 while (ch
>= '0' && ch
<= '9') {
2269 exp_val
= exp_val
* 10 + ch
- '0';
2272 exp_val
= exp_val
* s
;
2274 /* now we can generate the number */
2275 /* XXX: should patch directly float number */
2276 d
= (double)bn
[1] * 4294967296.0 + (double)bn
[0];
2277 d
= ldexp(d
, exp_val
- frac_bits
);
2282 /* float : should handle overflow */
2284 } else if (t
== 'L') {
2286 #ifdef TCC_TARGET_PE
2291 /* XXX: not large enough */
2292 tokc
.ld
= (long double)d
;
2299 /* decimal floats */
2301 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2306 while (ch
>= '0' && ch
<= '9') {
2307 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2313 if (ch
== 'e' || ch
== 'E') {
2314 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2318 if (ch
== '-' || ch
== '+') {
2319 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2324 if (ch
< '0' || ch
> '9')
2325 expect("exponent digits");
2326 while (ch
>= '0' && ch
<= '9') {
2327 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2339 tokc
.f
= strtof(token_buf
, NULL
);
2340 } else if (t
== 'L') {
2342 #ifdef TCC_TARGET_PE
2344 tokc
.d
= strtod(token_buf
, NULL
);
2347 tokc
.ld
= strtold(token_buf
, NULL
);
2351 tokc
.d
= strtod(token_buf
, NULL
);
2355 unsigned long long n
, n1
;
2356 int lcount
, ucount
, must_64bit
;
2359 /* integer number */
2362 if (b
== 10 && *q
== '0') {
2369 /* no need for checks except for base 10 / 8 errors */
2379 tcc_error("invalid digit");
2382 /* detect overflow */
2383 /* XXX: this test is not reliable */
2385 tcc_error("integer constant overflow");
2388 /* Determine the characteristics (unsigned and/or 64bit) the type of
2389 the constant must have according to the constant suffix(es) */
2390 lcount
= ucount
= must_64bit
= 0;
2396 tcc_error("three 'l's in integer constant");
2397 if (lcount
&& *(p
- 1) != ch
)
2398 tcc_error("incorrect integer suffix: %s", p1
);
2400 #if !defined TCC_TARGET_X86_64 || defined TCC_TARGET_PE
2405 } else if (t
== 'U') {
2407 tcc_error("two 'u's in integer constant");
2415 /* Whether 64 bits are needed to hold the constant's value */
2416 if (n
& 0xffffffff00000000LL
|| must_64bit
) {
2424 /* Whether type must be unsigned to hold the constant's value */
2425 if (ucount
|| ((n1
>> 31) && (b
!= 10))) {
2426 if (tok
== TOK_CLLONG
)
2430 /* If decimal and no unsigned suffix, bump to 64 bits or throw error */
2431 } else if (n1
>> 31) {
2432 if (tok
== TOK_CINT
)
2435 tcc_error("integer constant overflow");
2441 tcc_error("invalid number\n");
2445 #define PARSE2(c1, tok1, c2, tok2) \
2456 /* return next token without macro substitution */
2457 static inline void next_nomacro1(void)
2459 int t
, c
, is_long
, len
;
2472 if (parse_flags
& PARSE_FLAG_SPACES
)
2473 goto keep_tok_flags
;
2474 while (isidnum_table
[*p
- CH_EOF
] & IS_SPC
)
2483 /* first look if it is in fact an end of buffer */
2484 c
= handle_stray1(p
);
2491 TCCState
*s1
= tcc_state
;
2492 if ((parse_flags
& PARSE_FLAG_LINEFEED
)
2493 && !(tok_flags
& TOK_FLAG_EOF
)) {
2494 tok_flags
|= TOK_FLAG_EOF
;
2496 goto keep_tok_flags
;
2497 } else if (!(parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2499 } else if (s1
->ifdef_stack_ptr
!= file
->ifdef_stack_ptr
) {
2500 tcc_error("missing #endif");
2501 } else if (s1
->include_stack_ptr
== s1
->include_stack
) {
2502 /* no include left : end of file. */
2505 tok_flags
&= ~TOK_FLAG_EOF
;
2506 /* pop include file */
2508 /* test if previous '#endif' was after a #ifdef at
2510 if (tok_flags
& TOK_FLAG_ENDIF
) {
2512 printf("#endif %s\n", get_tok_str(file
->ifndef_macro_saved
, NULL
));
2514 search_cached_include(s1
, file
->filename
, 1)
2515 ->ifndef_macro
= file
->ifndef_macro_saved
;
2516 tok_flags
&= ~TOK_FLAG_ENDIF
;
2519 /* add end of include file debug info */
2520 if (tcc_state
->do_debug
) {
2521 put_stabd(N_EINCL
, 0, 0);
2523 /* pop include stack */
2525 s1
->include_stack_ptr
--;
2534 tok_flags
|= TOK_FLAG_BOL
;
2537 if (0 == (parse_flags
& PARSE_FLAG_LINEFEED
))
2540 goto keep_tok_flags
;
2545 if ((tok_flags
& TOK_FLAG_BOL
) &&
2546 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2548 preprocess(tok_flags
& TOK_FLAG_BOF
);
2554 tok
= TOK_TWOSHARPS
;
2556 if (parse_flags
& PARSE_FLAG_ASM_FILE
) {
2557 p
= parse_line_comment(p
- 1);
2566 /* dollar is allowed to start identifiers when not parsing asm */
2568 if (!(isidnum_table
[c
- CH_EOF
] & IS_ID
)
2569 || (parse_flags
& PARSE_FLAG_ASM_FILE
))
2572 case 'a': case 'b': case 'c': case 'd':
2573 case 'e': case 'f': case 'g': case 'h':
2574 case 'i': case 'j': case 'k': case 'l':
2575 case 'm': case 'n': case 'o': case 'p':
2576 case 'q': case 'r': case 's': case 't':
2577 case 'u': case 'v': case 'w': case 'x':
2579 case 'A': case 'B': case 'C': case 'D':
2580 case 'E': case 'F': case 'G': case 'H':
2581 case 'I': case 'J': case 'K':
2582 case 'M': case 'N': case 'O': case 'P':
2583 case 'Q': case 'R': case 'S': case 'T':
2584 case 'U': case 'V': case 'W': case 'X':
2590 h
= TOK_HASH_FUNC(h
, c
);
2591 while (c
= *++p
, isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2592 h
= TOK_HASH_FUNC(h
, c
);
2597 /* fast case : no stray found, so we have the full token
2598 and we have already hashed it */
2599 h
&= (TOK_HASH_SIZE
- 1);
2600 pts
= &hash_ident
[h
];
2605 if (ts
->len
== len
&& !memcmp(ts
->str
, p1
, len
))
2607 pts
= &(ts
->hash_next
);
2609 ts
= tok_alloc_new(pts
, (char *) p1
, len
);
2613 cstr_reset(&tokcstr
);
2614 cstr_cat(&tokcstr
, p1
, len
);
2618 while (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2620 cstr_ccat(&tokcstr
, c
);
2623 ts
= tok_alloc(tokcstr
.data
, tokcstr
.size
);
2629 if (t
!= '\\' && t
!= '\'' && t
!= '\"') {
2631 goto parse_ident_fast
;
2634 if (c
== '\'' || c
== '\"') {
2638 cstr_reset(&tokcstr
);
2639 cstr_ccat(&tokcstr
, 'L');
2640 goto parse_ident_slow
;
2645 case '0': case '1': case '2': case '3':
2646 case '4': case '5': case '6': case '7':
2650 /* after the first digit, accept digits, alpha, '.' or sign if
2651 prefixed by 'eEpP' */
2653 cstr_reset(&tokcstr
);
2655 cstr_ccat(&tokcstr
, t
);
2656 if (!((isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2658 || ((c
== '+' || c
== '-')
2659 && (((t
== 'e' || t
== 'E')
2660 && !(parse_flags
& PARSE_FLAG_ASM_FILE
2661 /* 0xe+1 is 3 tokens in asm */
2662 && ((char*)tokcstr
.data
)[0] == '0'
2663 && toup(((char*)tokcstr
.data
)[1]) == 'X'))
2664 || t
== 'p' || t
== 'P'))))
2669 /* We add a trailing '\0' to ease parsing */
2670 cstr_ccat(&tokcstr
, '\0');
2671 tokc
.str
.size
= tokcstr
.size
;
2672 tokc
.str
.data
= tokcstr
.data
;
2677 /* special dot handling because it can also start a number */
2682 } else if ((parse_flags
& PARSE_FLAG_ASM_FILE
)
2683 && (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))) {
2685 goto parse_ident_fast
;
2686 } else if (c
== '.') {
2692 *--p
= '.'; /* may underflow into file->unget[] */
2703 cstr_reset(&tokcstr
);
2705 cstr_ccat(&tokcstr
, 'L');
2706 cstr_ccat(&tokcstr
, c
);
2707 p
= parse_pp_string(p
, c
, &tokcstr
);
2708 cstr_ccat(&tokcstr
, c
);
2709 cstr_ccat(&tokcstr
, '\0');
2710 tokc
.str
.size
= tokcstr
.size
;
2711 tokc
.str
.data
= tokcstr
.data
;
2720 } else if (c
== '<') {
2737 } else if (c
== '>') {
2755 } else if (c
== '=') {
2768 } else if (c
== '=') {
2781 } else if (c
== '=') {
2794 } else if (c
== '=') {
2797 } else if (c
== '>') {
2805 PARSE2('!', '!', '=', TOK_NE
)
2806 PARSE2('=', '=', '=', TOK_EQ
)
2807 PARSE2('*', '*', '=', TOK_A_MUL
)
2808 PARSE2('%', '%', '=', TOK_A_MOD
)
2809 PARSE2('^', '^', '=', TOK_A_XOR
)
2811 /* comments or operator */
2815 p
= parse_comment(p
);
2816 /* comments replaced by a blank */
2818 goto keep_tok_flags
;
2819 } else if (c
== '/') {
2820 p
= parse_line_comment(p
);
2822 goto keep_tok_flags
;
2823 } else if (c
== '=') {
2843 case '@': /* only used in assembler */
2849 if (c
>= 0x80 && c
<= 0xFF) /* utf8 identifiers */
2850 goto parse_ident_fast
;
2851 if (parse_flags
& PARSE_FLAG_ASM_FILE
)
2853 tcc_error("unrecognized character \\x%02x", c
);
2859 #if defined(PARSE_DEBUG)
2860 printf("token = %s\n", get_tok_str(tok
, &tokc
));
2864 /* return next token without macro substitution. Can read input from
2866 static void next_nomacro_spc(void)
2872 TOK_GET(&tok
, ¯o_ptr
, &tokc
);
2873 if (tok
== TOK_LINENUM
) {
2874 file
->line_num
= tokc
.i
;
2883 ST_FUNC
void next_nomacro(void)
2887 } while (tok
< 256 && (isidnum_table
[tok
- CH_EOF
] & IS_SPC
));
2891 static void macro_subst(
2892 TokenString
*tok_str
,
2894 const int *macro_str
,
2898 /* substitute arguments in replacement lists in macro_str by the values in
2899 args (field d) and return allocated string */
2900 static int *macro_arg_subst(Sym
**nested_list
, const int *macro_str
, Sym
*args
)
2912 TOK_GET(&t
, ¯o_str
, &cval
);
2917 TOK_GET(&t
, ¯o_str
, &cval
);
2920 s
= sym_find2(args
, t
);
2923 cstr_ccat(&cstr
, '\"');
2927 TOK_GET(&t
, &st
, &cval
);
2928 if (t
!= TOK_PLCHLDR
2930 && 0 == check_space(t
, &spc
)) {
2931 const char *s
= get_tok_str(t
, &cval
);
2933 if (t
== TOK_PPSTR
&& *s
!= '\'')
2934 add_char(&cstr
, *s
);
2936 cstr_ccat(&cstr
, *s
);
2942 cstr_ccat(&cstr
, '\"');
2943 cstr_ccat(&cstr
, '\0');
2945 printf("\nstringize: <%s>\n", (char *)cstr
.data
);
2948 cval
.str
.size
= cstr
.size
;
2949 cval
.str
.data
= cstr
.data
;
2950 tok_str_add2(&str
, TOK_PPSTR
, &cval
);
2954 expect("macro parameter after '#'");
2956 } else if (t
>= TOK_IDENT
) {
2957 s
= sym_find2(args
, t
);
2961 /* if '##' is present before or after, no arg substitution */
2962 if (*macro_str
== TOK_TWOSHARPS
|| t1
== TOK_TWOSHARPS
) {
2963 /* special case for var arg macros : ## eats the ','
2964 if empty VA_ARGS variable. */
2965 if (t1
== TOK_TWOSHARPS
&& t0
== ',' && gnu_ext
&& s
->type
.t
) {
2967 /* suppress ',' '##' */
2970 /* suppress '##' and add variable */
2977 TOK_GET(&t1
, &st
, &cval
);
2980 tok_str_add2(&str
, t1
, &cval
);
2986 /* NOTE: the stream cannot be read when macro
2987 substituing an argument */
2988 macro_subst(&str
, nested_list
, st
, 0);
2990 if (str
.len
== l0
) /* exanded to empty string */
2991 tok_str_add(&str
, TOK_PLCHLDR
);
2993 tok_str_add(&str
, t
);
2996 tok_str_add2(&str
, t
, &cval
);
3000 tok_str_add(&str
, 0);
3004 static char const ab_month_name
[12][4] =
3006 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
3007 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
3010 /* peek or read [ws_str == NULL] next token from function macro call,
3011 walking up macro levels up to the file if necessary */
3012 static int next_argstream(Sym
**nested_list
, int can_read_stream
, TokenString
*ws_str
)
3020 p
= macro_ptr
, t
= *p
;
3022 while (is_space(t
) || TOK_LINEFEED
== t
)
3023 tok_str_add(ws_str
, t
), t
= *++p
;
3025 if (t
== 0 && can_read_stream
) {
3027 /* also, end of scope for nested defined symbol */
3029 while (sa
&& sa
->v
== 0)
3038 while (is_space(ch
) || ch
== '\n' || ch
== '/') {
3041 uint8_t *p
= file
->buf_ptr
;
3044 p
= parse_comment(p
);
3045 file
->buf_ptr
= p
- 1;
3046 } else if (c
== '/') {
3047 p
= parse_line_comment(p
);
3048 file
->buf_ptr
= p
- 1;
3053 if (!(ch
== '\f' || ch
== '\v' || ch
== '\r'))
3054 tok_str_add(ws_str
, ch
);
3068 /* do macro substitution of current token with macro 's' and add
3069 result to (tok_str,tok_len). 'nested_list' is the list of all
3070 macros we got inside to avoid recursing. Return non zero if no
3071 substitution needs to be done */
3072 static int macro_subst_tok(
3073 TokenString
*tok_str
,
3076 int can_read_stream
)
3078 Sym
*args
, *sa
, *sa1
;
3079 int parlevel
, *mstr
, t
, t1
, spc
;
3086 /* if symbol is a macro, prepare substitution */
3087 /* special macros */
3088 if (tok
== TOK___LINE__
) {
3089 snprintf(buf
, sizeof(buf
), "%d", file
->line_num
);
3093 } else if (tok
== TOK___FILE__
) {
3094 cstrval
= file
->filename
;
3096 } else if (tok
== TOK___DATE__
|| tok
== TOK___TIME__
) {
3101 tm
= localtime(&ti
);
3102 if (tok
== TOK___DATE__
) {
3103 snprintf(buf
, sizeof(buf
), "%s %2d %d",
3104 ab_month_name
[tm
->tm_mon
], tm
->tm_mday
, tm
->tm_year
+ 1900);
3106 snprintf(buf
, sizeof(buf
), "%02d:%02d:%02d",
3107 tm
->tm_hour
, tm
->tm_min
, tm
->tm_sec
);
3114 cstr_cat(&cstr
, cstrval
, 0);
3115 cval
.str
.size
= cstr
.size
;
3116 cval
.str
.data
= cstr
.data
;
3117 tok_str_add2(tok_str
, t1
, &cval
);
3120 int saved_parse_flags
= parse_flags
;
3123 if (s
->type
.t
== MACRO_FUNC
) {
3124 /* whitespace between macro name and argument list */
3126 tok_str_new(&ws_str
);
3129 parse_flags
|= PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
3130 | PARSE_FLAG_ACCEPT_STRAYS
;
3132 /* get next token from argument stream */
3133 t
= next_argstream(nested_list
, can_read_stream
, &ws_str
);
3135 /* not a macro substitution after all, restore the
3136 * macro token plus all whitespace we've read.
3137 * whitespace is intentionally not merged to preserve
3139 parse_flags
= saved_parse_flags
;
3140 tok_str_add(tok_str
, tok
);
3141 if (parse_flags
& PARSE_FLAG_SPACES
) {
3143 for (i
= 0; i
< ws_str
.len
; i
++)
3144 tok_str_add(tok_str
, ws_str
.str
[i
]);
3146 tok_str_free_str(ws_str
.str
);
3149 tok_str_free_str(ws_str
.str
);
3151 next_nomacro(); /* eat '(' */
3153 /* argument macro */
3156 /* NOTE: empty args are allowed, except if no args */
3159 next_argstream(nested_list
, can_read_stream
, NULL
);
3160 } while (is_space(tok
) || TOK_LINEFEED
== tok
);
3162 /* handle '()' case */
3163 if (!args
&& !sa
&& tok
== ')')
3166 tcc_error("macro '%s' used with too many args",
3167 get_tok_str(s
->v
, 0));
3170 /* NOTE: non zero sa->t indicates VA_ARGS */
3171 while ((parlevel
> 0 ||
3173 (tok
!= ',' || sa
->type
.t
)))) {
3174 if (tok
== TOK_EOF
|| tok
== 0)
3178 else if (tok
== ')')
3180 if (tok
== TOK_LINEFEED
)
3182 if (!check_space(tok
, &spc
))
3183 tok_str_add2(&str
, tok
, &tokc
);
3184 next_argstream(nested_list
, can_read_stream
, NULL
);
3189 tok_str_add(&str
, 0);
3190 sa1
= sym_push2(&args
, sa
->v
& ~SYM_FIELD
, sa
->type
.t
, 0);
3194 /* special case for gcc var args: add an empty
3195 var arg argument if it is omitted */
3196 if (sa
&& sa
->type
.t
&& gnu_ext
)
3204 tcc_error("macro '%s' used with too few args",
3205 get_tok_str(s
->v
, 0));
3208 parse_flags
= saved_parse_flags
;
3210 /* now subst each arg */
3211 mstr
= macro_arg_subst(nested_list
, mstr
, args
);
3216 tok_str_free_str(sa
->d
);
3222 sym_push2(nested_list
, s
->v
, 0, 0);
3223 parse_flags
= saved_parse_flags
;
3224 macro_subst(tok_str
, nested_list
, mstr
, can_read_stream
| 2);
3226 /* pop nested defined symbol */
3228 *nested_list
= sa1
->prev
;
3231 tok_str_free_str(mstr
);
3236 static int paste_tokens(int t1
, CValue
*v1
, int t2
, CValue
*v2
)
3242 if (t1
!= TOK_PLCHLDR
)
3243 cstr_cat(&cstr
, get_tok_str(t1
, v1
), -1);
3245 if (t2
!= TOK_PLCHLDR
)
3246 cstr_cat(&cstr
, get_tok_str(t2
, v2
), -1);
3247 cstr_ccat(&cstr
, '\0');
3249 tcc_open_bf(tcc_state
, ":paste:", cstr
.size
);
3250 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3253 if (0 == *file
->buf_ptr
)
3257 tcc_warning("pasting \"%.*s\" and \"%s\" does not give a valid"
3258 " preprocessing token", n
, cstr
.data
, (char*)cstr
.data
+ n
);
3263 //printf("paste <%s>\n", (char*)cstr.data);
3268 /* handle the '##' operator. Return NULL if no '##' seen. Otherwise
3269 return the resulting string (which must be freed). */
3270 static inline int *macro_twosharps(const int *ptr0
)
3274 TokenString macro_str1
;
3275 int start_of_nosubsts
= -1;
3278 /* we search the first '##' */
3279 for (ptr
= ptr0
;;) {
3280 TOK_GET(&t
, &ptr
, &cval
);
3281 if (t
== TOK_TWOSHARPS
)
3287 tok_str_new(¯o_str1
);
3289 //tok_print(" $$$", ptr0);
3290 for (ptr
= ptr0
;;) {
3291 TOK_GET(&t
, &ptr
, &cval
);
3294 if (t
== TOK_TWOSHARPS
)
3296 while (*ptr
== TOK_TWOSHARPS
) {
3298 /* given 'a##b', remove nosubsts preceding 'a' */
3299 if (start_of_nosubsts
>= 0)
3300 macro_str1
.len
= start_of_nosubsts
;
3301 /* given 'a##b', remove nosubsts preceding 'b' */
3302 while ((t1
= *++ptr
) == TOK_NOSUBST
)
3304 if (t1
&& t1
!= TOK_TWOSHARPS
) {
3305 TOK_GET(&t1
, &ptr
, &cv1
);
3306 if (t
!= TOK_PLCHLDR
|| t1
!= TOK_PLCHLDR
) {
3307 if (paste_tokens(t
, &cval
, t1
, &cv1
)) {
3308 t
= tok
, cval
= tokc
;
3310 tok_str_add2(¯o_str1
, t
, &cval
);
3316 if (t
== TOK_NOSUBST
) {
3317 if (start_of_nosubsts
< 0)
3318 start_of_nosubsts
= macro_str1
.len
;
3320 start_of_nosubsts
= -1;
3322 tok_str_add2(¯o_str1
, t
, &cval
);
3324 tok_str_add(¯o_str1
, 0);
3325 //tok_print(" ###", macro_str1.str);
3326 return macro_str1
.str
;
3329 /* do macro substitution of macro_str and add result to
3330 (tok_str,tok_len). 'nested_list' is the list of all macros we got
3331 inside to avoid recursing. */
3332 static void macro_subst(
3333 TokenString
*tok_str
,
3335 const int *macro_str
,
3341 int t
, spc
, nosubst
;
3343 int *macro_str1
= NULL
;
3345 /* first scan for '##' operator handling */
3349 /* first scan for '##' operator handling */
3350 if (can_read_stream
& 1) {
3351 macro_str1
= macro_twosharps(ptr
);
3357 TOK_GET(&t
, &ptr
, &cval
);
3361 if (t
>= TOK_IDENT
&& 0 == nosubst
) {
3366 /* if nested substitution, do nothing */
3367 if (sym_find2(*nested_list
, t
)) {
3368 /* and mark it as TOK_NOSUBST, so it doesn't get subst'd again */
3369 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3375 str
.str
= (int*)ptr
;
3376 begin_macro(&str
, 2);
3379 macro_subst_tok(tok_str
, nested_list
, s
, can_read_stream
);
3381 if (str
.alloc
== 3) {
3382 /* already finished by reading function macro arguments */
3390 spc
= (tok_str
->len
&&
3391 is_space(tok_last(tok_str
->str
,
3392 tok_str
->str
+ tok_str
->len
)));
3396 if (t
== '\\' && !(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
3397 tcc_error("stray '\\' in program");
3400 if (!check_space(t
, &spc
))
3401 tok_str_add2(tok_str
, t
, &cval
);
3403 if (t
== TOK_NOSUBST
)
3408 tok_str_free_str(macro_str1
);
3412 /* return next token with macro substitution */
3413 ST_FUNC
void next(void)
3416 if (parse_flags
& PARSE_FLAG_SPACES
)
3422 if (tok
== TOK_NOSUBST
|| tok
== TOK_PLCHLDR
) {
3423 /* discard preprocessor markers */
3425 } else if (tok
== 0) {
3426 /* end of macro or unget token string */
3430 } else if (tok
>= TOK_IDENT
&& (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
3432 /* if reading from file, try to substitute macros */
3433 s
= define_find(tok
);
3435 Sym
*nested_list
= NULL
;
3437 macro_subst_tok(&tokstr_buf
, &nested_list
, s
, 1);
3438 tok_str_add(&tokstr_buf
, 0);
3439 begin_macro(&tokstr_buf
, 2);
3443 /* convert preprocessor tokens into C tokens */
3444 if (tok
== TOK_PPNUM
) {
3445 if (parse_flags
& PARSE_FLAG_TOK_NUM
)
3446 parse_number((char *)tokc
.str
.data
);
3447 } else if (tok
== TOK_PPSTR
) {
3448 if (parse_flags
& PARSE_FLAG_TOK_STR
)
3449 parse_string((char *)tokc
.str
.data
, tokc
.str
.size
- 1);
3453 /* push back current token and set current token to 'last_tok'. Only
3454 identifier case handled for labels. */
3455 ST_INLN
void unget_tok(int last_tok
)
3458 TokenString
*str
= tok_str_alloc();
3459 tok_str_add2(str
, tok
, &tokc
);
3460 tok_str_add(str
, 0);
3461 begin_macro(str
, 1);
3465 ST_FUNC
void preprocess_start(TCCState
*s1
)
3467 s1
->include_stack_ptr
= s1
->include_stack
;
3468 /* XXX: move that before to avoid having to initialize
3469 file->ifdef_stack_ptr ? */
3470 s1
->ifdef_stack_ptr
= s1
->ifdef_stack
;
3471 file
->ifdef_stack_ptr
= s1
->ifdef_stack_ptr
;
3474 pvtop
= vtop
= vstack
- 1;
3475 s1
->pack_stack
[0] = 0;
3476 s1
->pack_stack_ptr
= s1
->pack_stack
;
3478 isidnum_table
['$' - CH_EOF
] =
3479 s1
->dollars_in_identifiers
? IS_ID
: 0;
3480 isidnum_table
['.' - CH_EOF
] =
3481 (parse_flags
& PARSE_FLAG_ASM_FILE
) ? IS_ID
: 0;
3484 ST_FUNC
void tccpp_new(TCCState
*s
)
3489 /* might be used in error() before preprocess_start() */
3490 s
->include_stack_ptr
= s
->include_stack
;
3492 /* init isid table */
3493 for(i
= CH_EOF
; i
<128; i
++)
3494 isidnum_table
[i
- CH_EOF
]
3495 = is_space(i
) ? IS_SPC
3500 for(i
= 128; i
<256; i
++)
3501 isidnum_table
[i
- CH_EOF
] = IS_ID
;
3503 /* init allocators */
3504 tal_new(&toksym_alloc
, TOKSYM_TAL_LIMIT
, TOKSYM_TAL_SIZE
);
3505 tal_new(&tokstr_alloc
, TOKSTR_TAL_LIMIT
, TOKSTR_TAL_SIZE
);
3506 tal_new(&cstr_alloc
, CSTR_TAL_LIMIT
, CSTR_TAL_SIZE
);
3508 memset(hash_ident
, 0, TOK_HASH_SIZE
* sizeof(TokenSym
*));
3509 cstr_new(&cstr_buf
);
3510 cstr_realloc(&cstr_buf
, STRING_MAX_SIZE
);
3511 tok_str_new(&tokstr_buf
);
3512 tok_str_realloc(&tokstr_buf
, TOKSTR_MAX_SIZE
);
3514 tok_ident
= TOK_IDENT
;
3523 tok_alloc(p
, r
- p
- 1);
3528 ST_FUNC
void tccpp_delete(TCCState
*s
)
3532 /* free -D and compiler defines */
3535 /* cleanup from error/setjmp */
3541 n
= tok_ident
- TOK_IDENT
;
3542 for(i
= 0; i
< n
; i
++)
3543 tal_free(toksym_alloc
, table_ident
[i
]);
3544 tcc_free(table_ident
);
3547 /* free static buffers */
3548 cstr_free(&tokcstr
);
3549 cstr_free(&cstr_buf
);
3550 tok_str_free_str(tokstr_buf
.str
);
3552 /* free allocators */
3553 tal_delete(toksym_alloc
);
3554 toksym_alloc
= NULL
;
3555 tal_delete(tokstr_alloc
);
3556 tokstr_alloc
= NULL
;
3557 tal_delete(cstr_alloc
);
3561 /* ------------------------------------------------------------------------- */
3562 /* tcc -E [-P[1]] [-dD} support */
3564 static void tok_print(const char *msg
, const int *str
)
3570 fp
= tcc_state
->ppfp
;
3571 if (!fp
|| !tcc_state
->dflag
)
3574 fprintf(fp
, "%s ", msg
);
3576 TOK_GET(&t
, &str
, &cval
);
3579 fprintf(fp
,"%s", get_tok_str(t
, &cval
));
3584 static void pp_line(TCCState
*s1
, BufferedFile
*f
, int level
)
3586 int d
= f
->line_num
- f
->line_ref
;
3591 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_NONE
) {
3593 } else if (level
== 0 && f
->line_ref
&& d
< 8) {
3595 fputs("\n", s1
->ppfp
), --d
;
3596 } else if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_STD
) {
3597 fprintf(s1
->ppfp
, "#line %d \"%s\"\n", f
->line_num
, f
->filename
);
3599 fprintf(s1
->ppfp
, "# %d \"%s\"%s\n", f
->line_num
, f
->filename
,
3600 level
> 0 ? " 1" : level
< 0 ? " 2" : "");
3602 f
->line_ref
= f
->line_num
;
3605 static void define_print(TCCState
*s1
, int v
)
3611 if (NULL
== s
|| NULL
== s
->d
)
3615 fprintf(fp
, "#define %s", get_tok_str(v
, NULL
));
3616 if (s
->type
.t
== MACRO_FUNC
) {
3621 fprintf(fp
,"%s", get_tok_str(a
->v
& ~SYM_FIELD
, NULL
));
3628 tok_print("", s
->d
);
3631 static void pp_debug_defines(TCCState
*s1
)
3642 pp_line(s1
, file
, 0);
3643 file
->line_ref
= ++file
->line_num
;
3647 vs
= get_tok_str(v
, NULL
);
3648 if (t
== TOK_DEFINE
) {
3649 define_print(s1
, v
);
3650 } else if (t
== TOK_UNDEF
) {
3651 fprintf(fp
, "#undef %s\n", vs
);
3652 } else if (t
== TOK_push_macro
) {
3653 fprintf(fp
, "#pragma push_macro(\"%s\")\n", vs
);
3654 } else if (t
== TOK_pop_macro
) {
3655 fprintf(fp
, "#pragma pop_macro(\"%s\")\n", vs
);
3660 static void pp_debug_builtins(TCCState
*s1
)
3663 for (v
= TOK_IDENT
; v
< tok_ident
; ++v
)
3664 define_print(s1
, v
);
3667 /* Add a space between tokens a and b to avoid unwanted textual pasting */
3668 static int pp_need_space(int a
, int b
)
3670 return 'E' == a
? '+' == b
|| '-' == b
3671 : '+' == a
? TOK_INC
== b
|| '+' == b
3672 : '-' == a
? TOK_DEC
== b
|| '-' == b
3673 : a
>= TOK_IDENT
? b
>= TOK_IDENT
3677 /* maybe hex like 0x1e */
3678 static int pp_check_he0xE(int t
, const char *p
)
3680 if (t
== TOK_PPNUM
&& toup(strchr(p
, 0)[-1]) == 'E')
3685 /* Preprocess the current file */
3686 ST_FUNC
int tcc_preprocess(TCCState
*s1
)
3688 BufferedFile
**iptr
;
3689 int token_seen
, spcs
, level
;
3693 preprocess_start(s1
);
3694 ch
= file
->buf_ptr
[0];
3695 tok_flags
= TOK_FLAG_BOL
| TOK_FLAG_BOF
;
3696 parse_flags
= PARSE_FLAG_PREPROCESS
3697 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
3698 | PARSE_FLAG_LINEFEED
3700 | PARSE_FLAG_ACCEPT_STRAYS
3702 define_start
= define_stack
;
3704 /* Credits to Fabrice Bellard's initial revision to demonstrate its
3705 capability to compile and run itself, provided all numbers are
3706 given as decimals. tcc -E -P10 will do. */
3707 if (s1
->Pflag
== 1 + 10)
3708 parse_flags
|= PARSE_FLAG_TOK_NUM
, s1
->Pflag
= 1;
3711 /* for PP benchmarks */
3712 do next(); while (tok
!= TOK_EOF
); return 0;
3715 if (s1
->dflag
& 1) {
3716 pp_debug_builtins(s1
);
3720 token_seen
= TOK_LINEFEED
, spcs
= 0;
3721 pp_line(s1
, file
, 0);
3724 iptr
= s1
->include_stack_ptr
;
3728 level
= s1
->include_stack_ptr
- iptr
;
3731 pp_line(s1
, *iptr
, 0);
3732 pp_line(s1
, file
, level
);
3736 pp_debug_defines(s1
);
3741 if (token_seen
== TOK_LINEFEED
) {
3746 if (tok
== TOK_LINEFEED
) {
3750 pp_line(s1
, file
, 0);
3751 } else if (tok
== TOK_LINEFEED
) {
3754 spcs
= pp_need_space(token_seen
, tok
);
3758 fputs(" ", s1
->ppfp
), --spcs
;
3759 fputs(p
= get_tok_str(tok
, &tokc
), s1
->ppfp
);
3760 token_seen
= pp_check_he0xE(tok
, p
);;
3762 /* reset define stack, but keep -D and built-ins */
3763 free_defines(define_start
);
3767 /* ------------------------------------------------------------------------- */