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
)
1283 static CString localbuf
;
1289 /* first time preallocate static localbuf, next time only reset position to start */
1290 cstr_reset(&localbuf
);
1291 TOK_GET(&t
, &a
, &cv
);
1292 cstr_cat(&localbuf
, get_tok_str(t
, &cv
), 0);
1293 TOK_GET(&t
, &b
, &cv
);
1294 if (strcmp(localbuf
.data
, get_tok_str(t
, &cv
)))
1300 /* defines handling */
1301 ST_INLN
void define_push(int v
, int macro_type
, int *str
, Sym
*first_arg
)
1306 s
= sym_push2(&define_stack
, v
, macro_type
, 0);
1308 s
->next
= first_arg
;
1309 table_ident
[v
- TOK_IDENT
]->sym_define
= s
;
1311 if (o
&& !macro_is_equal(o
->d
, s
->d
))
1312 tcc_warning("%s redefined", get_tok_str(v
, NULL
));
1315 /* undefined a define symbol. Its name is just set to zero */
1316 ST_FUNC
void define_undef(Sym
*s
)
1319 if (v
>= TOK_IDENT
&& v
< tok_ident
)
1320 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1323 ST_INLN Sym
*define_find(int v
)
1326 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1328 return table_ident
[v
]->sym_define
;
1331 /* free define stack until top reaches 'b' */
1332 ST_FUNC
void free_defines(Sym
*b
)
1334 while (define_stack
!= b
) {
1335 Sym
*top
= define_stack
;
1336 define_stack
= top
->prev
;
1337 tok_str_free_str(top
->d
);
1342 /* restore remaining (-D or predefined) symbols */
1345 if (v
>= TOK_IDENT
&& v
< tok_ident
) {
1346 Sym
**d
= &table_ident
[v
- TOK_IDENT
]->sym_define
;
1355 ST_FUNC Sym
*label_find(int v
)
1358 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1360 return table_ident
[v
]->sym_label
;
1363 ST_FUNC Sym
*label_push(Sym
**ptop
, int v
, int flags
)
1366 s
= sym_push2(ptop
, v
, 0, 0);
1368 ps
= &table_ident
[v
- TOK_IDENT
]->sym_label
;
1369 if (ptop
== &global_label_stack
) {
1370 /* modify the top most local identifier, so that
1371 sym_identifier will point to 's' when popped */
1373 ps
= &(*ps
)->prev_tok
;
1380 /* pop labels until element last is reached. Look if any labels are
1381 undefined. Define symbols if '&&label' was used. */
1382 ST_FUNC
void label_pop(Sym
**ptop
, Sym
*slast
)
1385 for(s
= *ptop
; s
!= slast
; s
= s1
) {
1387 if (s
->r
== LABEL_DECLARED
) {
1388 tcc_warning("label '%s' declared but not used", get_tok_str(s
->v
, NULL
));
1389 } else if (s
->r
== LABEL_FORWARD
) {
1390 tcc_error("label '%s' used but not defined",
1391 get_tok_str(s
->v
, NULL
));
1394 /* define corresponding symbol. A size of
1396 put_extern_sym(s
, cur_text_section
, s
->jnext
, 1);
1400 table_ident
[s
->v
- TOK_IDENT
]->sym_label
= s
->prev_tok
;
1406 /* eval an expression for #if/#elif */
1407 static int expr_preprocess(void)
1412 str
= tok_str_alloc();
1413 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1414 next(); /* do macro subst */
1415 if (tok
== TOK_DEFINED
) {
1420 c
= define_find(tok
) != 0;
1425 } else if (tok
>= TOK_IDENT
) {
1426 /* if undefined macro */
1430 tok_str_add_tok(str
);
1432 tok_str_add(str
, -1); /* simulate end of file */
1433 tok_str_add(str
, 0);
1434 /* now evaluate C constant expression */
1435 begin_macro(str
, 1);
1443 /* parse after #define */
1444 ST_FUNC
void parse_define(void)
1446 Sym
*s
, *first
, **ps
;
1447 int v
, t
, varg
, is_vaargs
, spc
;
1448 int saved_parse_flags
= parse_flags
;
1452 tcc_error("invalid macro name '%s'", get_tok_str(tok
, &tokc
));
1453 /* XXX: should check if same macro (ANSI) */
1456 /* '(' must be just after macro definition for MACRO_FUNC */
1457 parse_flags
|= PARSE_FLAG_SPACES
;
1460 /* must be able to parse TOK_DOTS (in asm mode '.' can be part of identifier) */
1461 parse_flags
&= ~PARSE_FLAG_ASM_FILE
;
1462 isidnum_table
['.' - CH_EOF
] = 0;
1465 if (tok
!= ')') for (;;) {
1469 if (varg
== TOK_DOTS
) {
1470 varg
= TOK___VA_ARGS__
;
1472 } else if (tok
== TOK_DOTS
&& gnu_ext
) {
1476 if (varg
< TOK_IDENT
)
1478 tcc_error("bad macro parameter list");
1479 s
= sym_push2(&define_stack
, varg
| SYM_FIELD
, is_vaargs
, 0);
1484 if (tok
!= ',' || is_vaargs
)
1490 parse_flags
|= (saved_parse_flags
& PARSE_FLAG_ASM_FILE
);
1491 isidnum_table
['.' - CH_EOF
] =
1492 (parse_flags
& PARSE_FLAG_ASM_FILE
) ? IS_ID
: 0;
1497 parse_flags
|= PARSE_FLAG_ACCEPT_STRAYS
| PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
;
1498 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1499 /* remove spaces around ## and after '#' */
1500 if (TOK_TWOSHARPS
== tok
) {
1506 } else if ('#' == tok
) {
1508 } else if (check_space(tok
, &spc
)) {
1511 tok_str_add2(&tokstr_buf
, tok
, &tokc
);
1516 parse_flags
= saved_parse_flags
;
1518 --tokstr_buf
.len
; /* remove trailing space */
1519 tok_str_add(&tokstr_buf
, 0);
1522 tcc_error("'##' cannot appear at either end of macro");
1523 define_push(v
, t
, tok_str_dup(&tokstr_buf
), first
);
1526 static CachedInclude
*search_cached_include(TCCState
*s1
, const char *filename
, int add
)
1528 const unsigned char *s
;
1534 s
= (unsigned char *) filename
;
1537 h
= TOK_HASH_FUNC(h
, toup(*s
));
1539 h
= TOK_HASH_FUNC(h
, *s
);
1543 h
&= (CACHED_INCLUDES_HASH_SIZE
- 1);
1545 i
= s1
->cached_includes_hash
[h
];
1549 e
= s1
->cached_includes
[i
- 1];
1550 if (0 == PATHCMP(e
->filename
, filename
))
1557 e
= tcc_malloc(sizeof(CachedInclude
) + strlen(filename
));
1558 strcpy(e
->filename
, filename
);
1559 e
->ifndef_macro
= e
->once
= 0;
1560 dynarray_add((void ***)&s1
->cached_includes
, &s1
->nb_cached_includes
, e
);
1561 /* add in hash table */
1562 e
->hash_next
= s1
->cached_includes_hash
[h
];
1563 s1
->cached_includes_hash
[h
] = s1
->nb_cached_includes
;
1565 printf("adding cached '%s'\n", filename
);
1570 static void pragma_parse(TCCState
*s1
)
1573 if (tok
== TOK_push_macro
|| tok
== TOK_pop_macro
) {
1577 if (next(), tok
!= '(')
1579 if (next(), tok
!= TOK_STR
)
1581 v
= tok_alloc(tokc
.str
.data
, tokc
.str
.size
- 1)->tok
;
1582 if (next(), tok
!= ')')
1584 if (t
== TOK_push_macro
) {
1585 while (NULL
== (s
= define_find(v
)))
1586 define_push(v
, 0, NULL
, NULL
);
1587 s
->type
.ref
= s
; /* set push boundary */
1589 for (s
= define_stack
; s
; s
= s
->prev
)
1590 if (s
->v
== v
&& s
->type
.ref
== s
) {
1596 table_ident
[v
- TOK_IDENT
]->sym_define
= s
->d
? s
: NULL
;
1598 tcc_warning("unbalanced #pragma pop_macro");
1599 pp_debug_tok
= t
, pp_debug_symv
= v
;
1601 } else if (tok
== TOK_once
) {
1602 search_cached_include(s1
, file
->filename
, 1)->once
= pp_once
;
1604 } else if (s1
->ppfp
) {
1605 /* tcc -E: keep pragmas below unchanged */
1607 unget_tok(TOK_PRAGMA
);
1609 unget_tok(TOK_LINEFEED
);
1611 } else if (tok
== TOK_pack
) {
1613 #pragma pack(1) // set
1614 #pragma pack() // reset to default
1615 #pragma pack(push,1) // push & set
1616 #pragma pack(pop) // restore previous */
1619 if (tok
== TOK_ASM_pop
) {
1621 if (s1
->pack_stack_ptr
<= s1
->pack_stack
) {
1623 tcc_error("out of pack stack");
1625 s1
->pack_stack_ptr
--;
1629 if (tok
== TOK_ASM_push
) {
1631 if (s1
->pack_stack_ptr
>= s1
->pack_stack
+ PACK_STACK_SIZE
- 1)
1633 s1
->pack_stack_ptr
++;
1636 if (tok
!= TOK_CINT
)
1639 if (val
< 1 || val
> 16 || (val
& (val
- 1)) != 0)
1643 *s1
->pack_stack_ptr
= val
;
1648 } else if (tok
== TOK_comment
) {
1658 file
= tcc_strdup((char *)tokc
.str
.data
);
1659 dynarray_add((void ***)&s1
->pragma_libs
, &s1
->nb_pragma_libs
, file
);
1665 if (s1
->warn_unsupported
)
1666 tcc_warning("#pragma %s is ignored", get_tok_str(tok
, &tokc
));
1671 tcc_error("malformed #pragma directive");
1675 /* is_bof is true if first non space token at beginning of file */
1676 ST_FUNC
void preprocess(int is_bof
)
1678 TCCState
*s1
= tcc_state
;
1679 int i
, c
, n
, saved_parse_flags
;
1683 saved_parse_flags
= parse_flags
;
1684 parse_flags
= PARSE_FLAG_PREPROCESS
1685 | PARSE_FLAG_TOK_NUM
1686 | PARSE_FLAG_TOK_STR
1687 | PARSE_FLAG_LINEFEED
1688 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
1697 pp_debug_symv
= tok
;
1703 pp_debug_symv
= tok
;
1704 s
= define_find(tok
);
1705 /* undefine symbol by putting an invalid name */
1710 case TOK_INCLUDE_NEXT
:
1711 ch
= file
->buf_ptr
[0];
1712 /* XXX: incorrect if comments : use next_nomacro with a special mode */
1717 } else if (ch
== '\"') {
1722 while (ch
!= c
&& ch
!= '\n' && ch
!= CH_EOF
) {
1723 if ((q
- buf
) < sizeof(buf
) - 1)
1726 if (handle_stray_noerror() == 0)
1734 /* eat all spaces and comments after include */
1735 /* XXX: slightly incorrect */
1736 while (ch1
!= '\n' && ch1
!= CH_EOF
)
1740 /* computed #include : either we have only strings or
1741 we have anything enclosed in '<>' */
1744 if (tok
== TOK_STR
) {
1745 while (tok
!= TOK_LINEFEED
) {
1746 if (tok
!= TOK_STR
) {
1748 tcc_error("'#include' expects \"FILENAME\" or <FILENAME>");
1750 pstrcat(buf
, sizeof(buf
), (char *)tokc
.str
.data
);
1756 while (tok
!= TOK_LINEFEED
) {
1757 pstrcat(buf
, sizeof(buf
), get_tok_str(tok
, &tokc
));
1761 /* check syntax and remove '<>' */
1762 if (len
< 2 || buf
[0] != '<' || buf
[len
- 1] != '>')
1763 goto include_syntax
;
1764 memmove(buf
, buf
+ 1, len
- 2);
1765 buf
[len
- 2] = '\0';
1770 if (s1
->include_stack_ptr
>= s1
->include_stack
+ INCLUDE_STACK_SIZE
)
1771 tcc_error("#include recursion too deep");
1772 /* store current file in stack, but increment stack later below */
1773 *s1
->include_stack_ptr
= file
;
1774 i
= tok
== TOK_INCLUDE_NEXT
? file
->include_next_index
: 0;
1775 n
= 2 + s1
->nb_include_paths
+ s1
->nb_sysinclude_paths
;
1776 for (; i
< n
; ++i
) {
1777 char buf1
[sizeof file
->filename
];
1782 /* check absolute include path */
1783 if (!IS_ABSPATH(buf
))
1787 } else if (i
== 1) {
1788 /* search in file's dir if "header.h" */
1791 path
= file
->filename
;
1792 pstrncpy(buf1
, path
, tcc_basename(path
) - path
);
1795 /* search in all the include paths */
1796 int j
= i
- 2, k
= j
- s1
->nb_include_paths
;
1797 path
= k
< 0 ? s1
->include_paths
[j
] : s1
->sysinclude_paths
[k
];
1798 pstrcpy(buf1
, sizeof(buf1
), path
);
1799 pstrcat(buf1
, sizeof(buf1
), "/");
1802 pstrcat(buf1
, sizeof(buf1
), buf
);
1803 e
= search_cached_include(s1
, buf1
, 0);
1804 if (e
&& (define_find(e
->ifndef_macro
) || e
->once
== pp_once
)) {
1805 /* no need to parse the include because the 'ifndef macro'
1806 is defined (or had #pragma once) */
1808 printf("%s: skipping cached %s\n", file
->filename
, buf1
);
1813 if (tcc_open(s1
, buf1
) < 0)
1816 file
->include_next_index
= i
+ 1;
1818 printf("%s: including %s\n", file
->prev
->filename
, file
->filename
);
1820 /* update target deps */
1821 dynarray_add((void ***)&s1
->target_deps
, &s1
->nb_target_deps
,
1823 /* push current file in stack */
1824 ++s1
->include_stack_ptr
;
1825 /* add include file debug info */
1827 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1828 tok_flags
|= TOK_FLAG_BOF
| TOK_FLAG_BOL
;
1829 ch
= file
->buf_ptr
[0];
1832 tcc_error("include file '%s' not found", buf
);
1839 c
= expr_preprocess();
1845 if (tok
< TOK_IDENT
)
1846 tcc_error("invalid argument for '#if%sdef'", c
? "n" : "");
1850 printf("#ifndef %s\n", get_tok_str(tok
, NULL
));
1852 file
->ifndef_macro
= tok
;
1855 c
= (define_find(tok
) != 0) ^ c
;
1857 if (s1
->ifdef_stack_ptr
>= s1
->ifdef_stack
+ IFDEF_STACK_SIZE
)
1858 tcc_error("memory full (ifdef)");
1859 *s1
->ifdef_stack_ptr
++ = c
;
1862 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1863 tcc_error("#else without matching #if");
1864 if (s1
->ifdef_stack_ptr
[-1] & 2)
1865 tcc_error("#else after #else");
1866 c
= (s1
->ifdef_stack_ptr
[-1] ^= 3);
1869 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1870 tcc_error("#elif without matching #if");
1871 c
= s1
->ifdef_stack_ptr
[-1];
1873 tcc_error("#elif after #else");
1874 /* last #if/#elif expression was true: we skip */
1878 c
= expr_preprocess();
1879 s1
->ifdef_stack_ptr
[-1] = c
;
1882 if (s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
+ 1)
1883 file
->ifndef_macro
= 0;
1892 if (s1
->ifdef_stack_ptr
<= file
->ifdef_stack_ptr
)
1893 tcc_error("#endif without matching #if");
1894 s1
->ifdef_stack_ptr
--;
1895 /* '#ifndef macro' was at the start of file. Now we check if
1896 an '#endif' is exactly at the end of file */
1897 if (file
->ifndef_macro
&&
1898 s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
) {
1899 file
->ifndef_macro_saved
= file
->ifndef_macro
;
1900 /* need to set to zero to avoid false matches if another
1901 #ifndef at middle of file */
1902 file
->ifndef_macro
= 0;
1903 while (tok
!= TOK_LINEFEED
)
1905 tok_flags
|= TOK_FLAG_ENDIF
;
1910 n
= strtoul((char*)tokc
.str
.data
, &q
, 10);
1914 if (tok
!= TOK_CINT
)
1916 tcc_error("wrong #line format");
1920 if (tok
!= TOK_LINEFEED
) {
1922 pstrcpy(file
->filename
, sizeof(file
->filename
), (char *)tokc
.str
.data
);
1923 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
1930 total_lines
+= file
->line_num
- n
;
1933 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1938 ch
= file
->buf_ptr
[0];
1941 while (ch
!= '\n' && ch
!= CH_EOF
) {
1942 if ((q
- buf
) < sizeof(buf
) - 1)
1945 if (handle_stray_noerror() == 0)
1952 tcc_error("#error %s", buf
);
1954 tcc_warning("#warning %s", buf
);
1962 /* ignore gas line comment in an 'S' file. */
1963 if (saved_parse_flags
& PARSE_FLAG_ASM_FILE
)
1965 if (tok
== '!' && is_bof
)
1966 /* '!' is ignored at beginning to allow C scripts. */
1968 tcc_warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok
, &tokc
));
1970 file
->buf_ptr
= parse_line_comment(file
->buf_ptr
- 1);
1973 /* ignore other preprocess commands or #! for C scripts */
1974 while (tok
!= TOK_LINEFEED
)
1977 parse_flags
= saved_parse_flags
;
1980 /* evaluate escape codes in a string. */
1981 static void parse_escape_string(CString
*outstr
, const uint8_t *buf
, int is_long
)
1996 case '0': case '1': case '2': case '3':
1997 case '4': case '5': case '6': case '7':
1998 /* at most three octal digits */
2003 n
= n
* 8 + c
- '0';
2007 n
= n
* 8 + c
- '0';
2012 goto add_char_nonext
;
2020 if (c
>= 'a' && c
<= 'f')
2022 else if (c
>= 'A' && c
<= 'F')
2032 goto add_char_nonext
;
2056 goto invalid_escape
;
2066 if (c
>= '!' && c
<= '~')
2067 tcc_warning("unknown escape sequence: \'\\%c\'", c
);
2069 tcc_warning("unknown escape sequence: \'\\x%x\'", c
);
2076 cstr_ccat(outstr
, c
);
2078 cstr_wccat(outstr
, c
);
2080 /* add a trailing '\0' */
2082 cstr_ccat(outstr
, '\0');
2084 cstr_wccat(outstr
, '\0');
2087 static void parse_string(const char *s
, int len
)
2089 uint8_t buf
[1000], *p
= buf
;
2092 if ((is_long
= *s
== 'L'))
2096 if (len
>= sizeof buf
)
2097 p
= tcc_malloc(len
+ 1);
2101 cstr_reset(&tokcstr
);
2102 parse_escape_string(&tokcstr
, p
, is_long
);
2108 /* XXX: make it portable */
2112 char_size
= sizeof(nwchar_t
);
2113 if (tokcstr
.size
<= char_size
)
2114 tcc_error("empty character constant");
2115 if (tokcstr
.size
> 2 * char_size
)
2116 tcc_warning("multi-character character constant");
2118 tokc
.i
= *(int8_t *)tokcstr
.data
;
2121 tokc
.i
= *(nwchar_t
*)tokcstr
.data
;
2125 tokc
.str
.size
= tokcstr
.size
;
2126 tokc
.str
.data
= tokcstr
.data
;
2134 /* we use 64 bit numbers */
2137 /* bn = (bn << shift) | or_val */
2138 static void bn_lshift(unsigned int *bn
, int shift
, int or_val
)
2142 for(i
=0;i
<BN_SIZE
;i
++) {
2144 bn
[i
] = (v
<< shift
) | or_val
;
2145 or_val
= v
>> (32 - shift
);
2149 static void bn_zero(unsigned int *bn
)
2152 for(i
=0;i
<BN_SIZE
;i
++) {
2157 /* parse number in null terminated string 'p' and return it in the
2159 static void parse_number(const char *p
)
2161 int b
, t
, shift
, frac_bits
, s
, exp_val
, ch
;
2163 unsigned int bn
[BN_SIZE
];
2174 goto float_frac_parse
;
2175 } else if (t
== '0') {
2176 if (ch
== 'x' || ch
== 'X') {
2180 } else if (tcc_ext
&& (ch
== 'b' || ch
== 'B')) {
2186 /* parse all digits. cannot check octal numbers at this stage
2187 because of floating point constants */
2189 if (ch
>= 'a' && ch
<= 'f')
2191 else if (ch
>= 'A' && ch
<= 'F')
2199 if (q
>= token_buf
+ STRING_MAX_SIZE
) {
2201 tcc_error("number too long");
2207 ((ch
== 'e' || ch
== 'E') && b
== 10) ||
2208 ((ch
== 'p' || ch
== 'P') && (b
== 16 || b
== 2))) {
2210 /* NOTE: strtox should support that for hexa numbers, but
2211 non ISOC99 libcs do not support it, so we prefer to do
2213 /* hexadecimal or binary floats */
2214 /* XXX: handle overflows */
2226 } else if (t
>= 'a') {
2228 } else if (t
>= 'A') {
2233 bn_lshift(bn
, shift
, t
);
2240 if (t
>= 'a' && t
<= 'f') {
2242 } else if (t
>= 'A' && t
<= 'F') {
2244 } else if (t
>= '0' && t
<= '9') {
2250 tcc_error("invalid digit");
2251 bn_lshift(bn
, shift
, t
);
2256 if (ch
!= 'p' && ch
!= 'P')
2263 } else if (ch
== '-') {
2267 if (ch
< '0' || ch
> '9')
2268 expect("exponent digits");
2269 while (ch
>= '0' && ch
<= '9') {
2270 exp_val
= exp_val
* 10 + ch
- '0';
2273 exp_val
= exp_val
* s
;
2275 /* now we can generate the number */
2276 /* XXX: should patch directly float number */
2277 d
= (double)bn
[1] * 4294967296.0 + (double)bn
[0];
2278 d
= ldexp(d
, exp_val
- frac_bits
);
2283 /* float : should handle overflow */
2285 } else if (t
== 'L') {
2287 #ifdef TCC_TARGET_PE
2292 /* XXX: not large enough */
2293 tokc
.ld
= (long double)d
;
2300 /* decimal floats */
2302 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2307 while (ch
>= '0' && ch
<= '9') {
2308 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2314 if (ch
== 'e' || ch
== 'E') {
2315 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2319 if (ch
== '-' || ch
== '+') {
2320 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2325 if (ch
< '0' || ch
> '9')
2326 expect("exponent digits");
2327 while (ch
>= '0' && ch
<= '9') {
2328 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2340 tokc
.f
= strtof(token_buf
, NULL
);
2341 } else if (t
== 'L') {
2343 #ifdef TCC_TARGET_PE
2345 tokc
.d
= strtod(token_buf
, NULL
);
2348 tokc
.ld
= strtold(token_buf
, NULL
);
2352 tokc
.d
= strtod(token_buf
, NULL
);
2356 unsigned long long n
, n1
;
2357 int lcount
, ucount
, must_64bit
;
2360 /* integer number */
2363 if (b
== 10 && *q
== '0') {
2370 /* no need for checks except for base 10 / 8 errors */
2380 tcc_error("invalid digit");
2383 /* detect overflow */
2384 /* XXX: this test is not reliable */
2386 tcc_error("integer constant overflow");
2389 /* Determine the characteristics (unsigned and/or 64bit) the type of
2390 the constant must have according to the constant suffix(es) */
2391 lcount
= ucount
= must_64bit
= 0;
2397 tcc_error("three 'l's in integer constant");
2398 if (lcount
&& *(p
- 1) != ch
)
2399 tcc_error("incorrect integer suffix: %s", p1
);
2401 #if !defined TCC_TARGET_X86_64 || defined TCC_TARGET_PE
2406 } else if (t
== 'U') {
2408 tcc_error("two 'u's in integer constant");
2416 /* Whether 64 bits are needed to hold the constant's value */
2417 if (n
& 0xffffffff00000000LL
|| must_64bit
) {
2425 /* Whether type must be unsigned to hold the constant's value */
2426 if (ucount
|| ((n1
>> 31) && (b
!= 10))) {
2427 if (tok
== TOK_CLLONG
)
2431 /* If decimal and no unsigned suffix, bump to 64 bits or throw error */
2432 } else if (n1
>> 31) {
2433 if (tok
== TOK_CINT
)
2436 tcc_error("integer constant overflow");
2442 tcc_error("invalid number\n");
2446 #define PARSE2(c1, tok1, c2, tok2) \
2457 /* return next token without macro substitution */
2458 static inline void next_nomacro1(void)
2460 int t
, c
, is_long
, len
;
2473 if (parse_flags
& PARSE_FLAG_SPACES
)
2474 goto keep_tok_flags
;
2475 while (isidnum_table
[*p
- CH_EOF
] & IS_SPC
)
2484 /* first look if it is in fact an end of buffer */
2485 c
= handle_stray1(p
);
2492 TCCState
*s1
= tcc_state
;
2493 if ((parse_flags
& PARSE_FLAG_LINEFEED
)
2494 && !(tok_flags
& TOK_FLAG_EOF
)) {
2495 tok_flags
|= TOK_FLAG_EOF
;
2497 goto keep_tok_flags
;
2498 } else if (!(parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2500 } else if (s1
->ifdef_stack_ptr
!= file
->ifdef_stack_ptr
) {
2501 tcc_error("missing #endif");
2502 } else if (s1
->include_stack_ptr
== s1
->include_stack
) {
2503 /* no include left : end of file. */
2506 tok_flags
&= ~TOK_FLAG_EOF
;
2507 /* pop include file */
2509 /* test if previous '#endif' was after a #ifdef at
2511 if (tok_flags
& TOK_FLAG_ENDIF
) {
2513 printf("#endif %s\n", get_tok_str(file
->ifndef_macro_saved
, NULL
));
2515 search_cached_include(s1
, file
->filename
, 1)
2516 ->ifndef_macro
= file
->ifndef_macro_saved
;
2517 tok_flags
&= ~TOK_FLAG_ENDIF
;
2520 /* add end of include file debug info */
2521 if (tcc_state
->do_debug
) {
2522 put_stabd(N_EINCL
, 0, 0);
2524 /* pop include stack */
2526 s1
->include_stack_ptr
--;
2535 tok_flags
|= TOK_FLAG_BOL
;
2538 if (0 == (parse_flags
& PARSE_FLAG_LINEFEED
))
2541 goto keep_tok_flags
;
2546 if ((tok_flags
& TOK_FLAG_BOL
) &&
2547 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2549 preprocess(tok_flags
& TOK_FLAG_BOF
);
2555 tok
= TOK_TWOSHARPS
;
2557 if (parse_flags
& PARSE_FLAG_ASM_FILE
) {
2558 p
= parse_line_comment(p
- 1);
2567 /* dollar is allowed to start identifiers when not parsing asm */
2569 if (!(isidnum_table
[c
- CH_EOF
] & IS_ID
)
2570 || (parse_flags
& PARSE_FLAG_ASM_FILE
))
2573 case 'a': case 'b': case 'c': case 'd':
2574 case 'e': case 'f': case 'g': case 'h':
2575 case 'i': case 'j': case 'k': case 'l':
2576 case 'm': case 'n': case 'o': case 'p':
2577 case 'q': case 'r': case 's': case 't':
2578 case 'u': case 'v': case 'w': case 'x':
2580 case 'A': case 'B': case 'C': case 'D':
2581 case 'E': case 'F': case 'G': case 'H':
2582 case 'I': case 'J': case 'K':
2583 case 'M': case 'N': case 'O': case 'P':
2584 case 'Q': case 'R': case 'S': case 'T':
2585 case 'U': case 'V': case 'W': case 'X':
2591 h
= TOK_HASH_FUNC(h
, c
);
2592 while (c
= *++p
, isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2593 h
= TOK_HASH_FUNC(h
, c
);
2598 /* fast case : no stray found, so we have the full token
2599 and we have already hashed it */
2600 h
&= (TOK_HASH_SIZE
- 1);
2601 pts
= &hash_ident
[h
];
2606 if (ts
->len
== len
&& !memcmp(ts
->str
, p1
, len
))
2608 pts
= &(ts
->hash_next
);
2610 ts
= tok_alloc_new(pts
, (char *) p1
, len
);
2614 cstr_reset(&tokcstr
);
2615 cstr_cat(&tokcstr
, p1
, len
);
2619 while (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2621 cstr_ccat(&tokcstr
, c
);
2624 ts
= tok_alloc(tokcstr
.data
, tokcstr
.size
);
2630 if (t
!= '\\' && t
!= '\'' && t
!= '\"') {
2632 goto parse_ident_fast
;
2635 if (c
== '\'' || c
== '\"') {
2639 cstr_reset(&tokcstr
);
2640 cstr_ccat(&tokcstr
, 'L');
2641 goto parse_ident_slow
;
2646 case '0': case '1': case '2': case '3':
2647 case '4': case '5': case '6': case '7':
2651 /* after the first digit, accept digits, alpha, '.' or sign if
2652 prefixed by 'eEpP' */
2654 cstr_reset(&tokcstr
);
2656 cstr_ccat(&tokcstr
, t
);
2657 if (!((isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2659 || ((c
== '+' || c
== '-')
2660 && (((t
== 'e' || t
== 'E')
2661 && !(parse_flags
& PARSE_FLAG_ASM_FILE
2662 /* 0xe+1 is 3 tokens in asm */
2663 && ((char*)tokcstr
.data
)[0] == '0'
2664 && toup(((char*)tokcstr
.data
)[1]) == 'X'))
2665 || t
== 'p' || t
== 'P'))))
2670 /* We add a trailing '\0' to ease parsing */
2671 cstr_ccat(&tokcstr
, '\0');
2672 tokc
.str
.size
= tokcstr
.size
;
2673 tokc
.str
.data
= tokcstr
.data
;
2678 /* special dot handling because it can also start a number */
2683 } else if ((parse_flags
& PARSE_FLAG_ASM_FILE
)
2684 && (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))) {
2686 goto parse_ident_fast
;
2687 } else if (c
== '.') {
2693 *--p
= '.'; /* may underflow into file->unget[] */
2704 cstr_reset(&tokcstr
);
2706 cstr_ccat(&tokcstr
, 'L');
2707 cstr_ccat(&tokcstr
, c
);
2708 p
= parse_pp_string(p
, c
, &tokcstr
);
2709 cstr_ccat(&tokcstr
, c
);
2710 cstr_ccat(&tokcstr
, '\0');
2711 tokc
.str
.size
= tokcstr
.size
;
2712 tokc
.str
.data
= tokcstr
.data
;
2721 } else if (c
== '<') {
2738 } else if (c
== '>') {
2756 } else if (c
== '=') {
2769 } else if (c
== '=') {
2782 } else if (c
== '=') {
2795 } else if (c
== '=') {
2798 } else if (c
== '>') {
2806 PARSE2('!', '!', '=', TOK_NE
)
2807 PARSE2('=', '=', '=', TOK_EQ
)
2808 PARSE2('*', '*', '=', TOK_A_MUL
)
2809 PARSE2('%', '%', '=', TOK_A_MOD
)
2810 PARSE2('^', '^', '=', TOK_A_XOR
)
2812 /* comments or operator */
2816 p
= parse_comment(p
);
2817 /* comments replaced by a blank */
2819 goto keep_tok_flags
;
2820 } else if (c
== '/') {
2821 p
= parse_line_comment(p
);
2823 goto keep_tok_flags
;
2824 } else if (c
== '=') {
2844 case '@': /* only used in assembler */
2850 if (c
>= 0x80 && c
<= 0xFF) /* utf8 identifiers */
2851 goto parse_ident_fast
;
2852 if (parse_flags
& PARSE_FLAG_ASM_FILE
)
2854 tcc_error("unrecognized character \\x%02x", c
);
2860 #if defined(PARSE_DEBUG)
2861 printf("token = %s\n", get_tok_str(tok
, &tokc
));
2865 /* return next token without macro substitution. Can read input from
2867 static void next_nomacro_spc(void)
2873 TOK_GET(&tok
, ¯o_ptr
, &tokc
);
2874 if (tok
== TOK_LINENUM
) {
2875 file
->line_num
= tokc
.i
;
2884 ST_FUNC
void next_nomacro(void)
2888 } while (tok
< 256 && (isidnum_table
[tok
- CH_EOF
] & IS_SPC
));
2892 static void macro_subst(
2893 TokenString
*tok_str
,
2895 const int *macro_str
,
2899 /* substitute arguments in replacement lists in macro_str by the values in
2900 args (field d) and return allocated string */
2901 static int *macro_arg_subst(Sym
**nested_list
, const int *macro_str
, Sym
*args
)
2913 TOK_GET(&t
, ¯o_str
, &cval
);
2918 TOK_GET(&t
, ¯o_str
, &cval
);
2921 s
= sym_find2(args
, t
);
2924 cstr_ccat(&cstr
, '\"');
2928 TOK_GET(&t
, &st
, &cval
);
2929 if (t
!= TOK_PLCHLDR
2931 && 0 == check_space(t
, &spc
)) {
2932 const char *s
= get_tok_str(t
, &cval
);
2934 if (t
== TOK_PPSTR
&& *s
!= '\'')
2935 add_char(&cstr
, *s
);
2937 cstr_ccat(&cstr
, *s
);
2943 cstr_ccat(&cstr
, '\"');
2944 cstr_ccat(&cstr
, '\0');
2946 printf("\nstringize: <%s>\n", (char *)cstr
.data
);
2949 cval
.str
.size
= cstr
.size
;
2950 cval
.str
.data
= cstr
.data
;
2951 tok_str_add2(&str
, TOK_PPSTR
, &cval
);
2955 expect("macro parameter after '#'");
2957 } else if (t
>= TOK_IDENT
) {
2958 s
= sym_find2(args
, t
);
2962 /* if '##' is present before or after, no arg substitution */
2963 if (*macro_str
== TOK_TWOSHARPS
|| t1
== TOK_TWOSHARPS
) {
2964 /* special case for var arg macros : ## eats the ','
2965 if empty VA_ARGS variable. */
2966 if (t1
== TOK_TWOSHARPS
&& t0
== ',' && gnu_ext
&& s
->type
.t
) {
2968 /* suppress ',' '##' */
2971 /* suppress '##' and add variable */
2978 TOK_GET(&t1
, &st
, &cval
);
2981 tok_str_add2(&str
, t1
, &cval
);
2987 /* NOTE: the stream cannot be read when macro
2988 substituing an argument */
2989 macro_subst(&str
, nested_list
, st
, 0);
2991 if (str
.len
== l0
) /* exanded to empty string */
2992 tok_str_add(&str
, TOK_PLCHLDR
);
2994 tok_str_add(&str
, t
);
2997 tok_str_add2(&str
, t
, &cval
);
3001 tok_str_add(&str
, 0);
3005 static char const ab_month_name
[12][4] =
3007 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
3008 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
3011 /* peek or read [ws_str == NULL] next token from function macro call,
3012 walking up macro levels up to the file if necessary */
3013 static int next_argstream(Sym
**nested_list
, int can_read_stream
, TokenString
*ws_str
)
3021 p
= macro_ptr
, t
= *p
;
3023 while (is_space(t
) || TOK_LINEFEED
== t
)
3024 tok_str_add(ws_str
, t
), t
= *++p
;
3026 if (t
== 0 && can_read_stream
) {
3028 /* also, end of scope for nested defined symbol */
3030 while (sa
&& sa
->v
== 0)
3039 while (is_space(ch
) || ch
== '\n' || ch
== '/') {
3042 uint8_t *p
= file
->buf_ptr
;
3045 p
= parse_comment(p
);
3046 file
->buf_ptr
= p
- 1;
3047 } else if (c
== '/') {
3048 p
= parse_line_comment(p
);
3049 file
->buf_ptr
= p
- 1;
3054 if (!(ch
== '\f' || ch
== '\v' || ch
== '\r'))
3055 tok_str_add(ws_str
, ch
);
3069 /* do macro substitution of current token with macro 's' and add
3070 result to (tok_str,tok_len). 'nested_list' is the list of all
3071 macros we got inside to avoid recursing. Return non zero if no
3072 substitution needs to be done */
3073 static int macro_subst_tok(
3074 TokenString
*tok_str
,
3077 int can_read_stream
)
3079 Sym
*args
, *sa
, *sa1
;
3080 int parlevel
, *mstr
, t
, t1
, spc
;
3087 /* if symbol is a macro, prepare substitution */
3088 /* special macros */
3089 if (tok
== TOK___LINE__
) {
3090 snprintf(buf
, sizeof(buf
), "%d", file
->line_num
);
3094 } else if (tok
== TOK___FILE__
) {
3095 cstrval
= file
->filename
;
3097 } else if (tok
== TOK___DATE__
|| tok
== TOK___TIME__
) {
3102 tm
= localtime(&ti
);
3103 if (tok
== TOK___DATE__
) {
3104 snprintf(buf
, sizeof(buf
), "%s %2d %d",
3105 ab_month_name
[tm
->tm_mon
], tm
->tm_mday
, tm
->tm_year
+ 1900);
3107 snprintf(buf
, sizeof(buf
), "%02d:%02d:%02d",
3108 tm
->tm_hour
, tm
->tm_min
, tm
->tm_sec
);
3115 cstr_cat(&cstr
, cstrval
, 0);
3116 cval
.str
.size
= cstr
.size
;
3117 cval
.str
.data
= cstr
.data
;
3118 tok_str_add2(tok_str
, t1
, &cval
);
3121 int saved_parse_flags
= parse_flags
;
3124 if (s
->type
.t
== MACRO_FUNC
) {
3125 /* whitespace between macro name and argument list */
3127 tok_str_new(&ws_str
);
3130 parse_flags
|= PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
3131 | PARSE_FLAG_ACCEPT_STRAYS
;
3133 /* get next token from argument stream */
3134 t
= next_argstream(nested_list
, can_read_stream
, &ws_str
);
3136 /* not a macro substitution after all, restore the
3137 * macro token plus all whitespace we've read.
3138 * whitespace is intentionally not merged to preserve
3140 parse_flags
= saved_parse_flags
;
3141 tok_str_add(tok_str
, tok
);
3142 if (parse_flags
& PARSE_FLAG_SPACES
) {
3144 for (i
= 0; i
< ws_str
.len
; i
++)
3145 tok_str_add(tok_str
, ws_str
.str
[i
]);
3147 tok_str_free_str(ws_str
.str
);
3150 tok_str_free_str(ws_str
.str
);
3152 next_nomacro(); /* eat '(' */
3154 /* argument macro */
3157 /* NOTE: empty args are allowed, except if no args */
3160 next_argstream(nested_list
, can_read_stream
, NULL
);
3161 } while (is_space(tok
) || TOK_LINEFEED
== tok
);
3163 /* handle '()' case */
3164 if (!args
&& !sa
&& tok
== ')')
3167 tcc_error("macro '%s' used with too many args",
3168 get_tok_str(s
->v
, 0));
3171 /* NOTE: non zero sa->t indicates VA_ARGS */
3172 while ((parlevel
> 0 ||
3174 (tok
!= ',' || sa
->type
.t
)))) {
3175 if (tok
== TOK_EOF
|| tok
== 0)
3179 else if (tok
== ')')
3181 if (tok
== TOK_LINEFEED
)
3183 if (!check_space(tok
, &spc
))
3184 tok_str_add2(&str
, tok
, &tokc
);
3185 next_argstream(nested_list
, can_read_stream
, NULL
);
3190 tok_str_add(&str
, 0);
3191 sa1
= sym_push2(&args
, sa
->v
& ~SYM_FIELD
, sa
->type
.t
, 0);
3195 /* special case for gcc var args: add an empty
3196 var arg argument if it is omitted */
3197 if (sa
&& sa
->type
.t
&& gnu_ext
)
3205 tcc_error("macro '%s' used with too few args",
3206 get_tok_str(s
->v
, 0));
3209 parse_flags
= saved_parse_flags
;
3211 /* now subst each arg */
3212 mstr
= macro_arg_subst(nested_list
, mstr
, args
);
3217 tok_str_free_str(sa
->d
);
3223 sym_push2(nested_list
, s
->v
, 0, 0);
3224 parse_flags
= saved_parse_flags
;
3225 macro_subst(tok_str
, nested_list
, mstr
, can_read_stream
| 2);
3227 /* pop nested defined symbol */
3229 *nested_list
= sa1
->prev
;
3232 tok_str_free_str(mstr
);
3237 static int paste_tokens(int t1
, CValue
*v1
, int t2
, CValue
*v2
)
3243 if (t1
!= TOK_PLCHLDR
)
3244 cstr_cat(&cstr
, get_tok_str(t1
, v1
), -1);
3246 if (t2
!= TOK_PLCHLDR
)
3247 cstr_cat(&cstr
, get_tok_str(t2
, v2
), -1);
3248 cstr_ccat(&cstr
, '\0');
3250 tcc_open_bf(tcc_state
, ":paste:", cstr
.size
);
3251 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3254 if (0 == *file
->buf_ptr
)
3258 tcc_warning("pasting \"%.*s\" and \"%s\" does not give a valid"
3259 " preprocessing token", n
, cstr
.data
, (char*)cstr
.data
+ n
);
3264 //printf("paste <%s>\n", (char*)cstr.data);
3269 /* handle the '##' operator. Return NULL if no '##' seen. Otherwise
3270 return the resulting string (which must be freed). */
3271 static inline int *macro_twosharps(const int *ptr0
)
3275 TokenString macro_str1
;
3276 int start_of_nosubsts
= -1;
3279 /* we search the first '##' */
3280 for (ptr
= ptr0
;;) {
3281 TOK_GET(&t
, &ptr
, &cval
);
3282 if (t
== TOK_TWOSHARPS
)
3288 tok_str_new(¯o_str1
);
3290 //tok_print(" $$$", ptr0);
3291 for (ptr
= ptr0
;;) {
3292 TOK_GET(&t
, &ptr
, &cval
);
3295 if (t
== TOK_TWOSHARPS
)
3297 while (*ptr
== TOK_TWOSHARPS
) {
3299 /* given 'a##b', remove nosubsts preceding 'a' */
3300 if (start_of_nosubsts
>= 0)
3301 macro_str1
.len
= start_of_nosubsts
;
3302 /* given 'a##b', remove nosubsts preceding 'b' */
3303 while ((t1
= *++ptr
) == TOK_NOSUBST
)
3305 if (t1
&& t1
!= TOK_TWOSHARPS
) {
3306 TOK_GET(&t1
, &ptr
, &cv1
);
3307 if (t
!= TOK_PLCHLDR
|| t1
!= TOK_PLCHLDR
) {
3308 if (paste_tokens(t
, &cval
, t1
, &cv1
)) {
3309 t
= tok
, cval
= tokc
;
3311 tok_str_add2(¯o_str1
, t
, &cval
);
3317 if (t
== TOK_NOSUBST
) {
3318 if (start_of_nosubsts
< 0)
3319 start_of_nosubsts
= macro_str1
.len
;
3321 start_of_nosubsts
= -1;
3323 tok_str_add2(¯o_str1
, t
, &cval
);
3325 tok_str_add(¯o_str1
, 0);
3326 //tok_print(" ###", macro_str1.str);
3327 return macro_str1
.str
;
3330 /* do macro substitution of macro_str and add result to
3331 (tok_str,tok_len). 'nested_list' is the list of all macros we got
3332 inside to avoid recursing. */
3333 static void macro_subst(
3334 TokenString
*tok_str
,
3336 const int *macro_str
,
3342 int t
, spc
, nosubst
;
3344 int *macro_str1
= NULL
;
3346 /* first scan for '##' operator handling */
3350 /* first scan for '##' operator handling */
3351 if (can_read_stream
& 1) {
3352 macro_str1
= macro_twosharps(ptr
);
3358 TOK_GET(&t
, &ptr
, &cval
);
3362 if (t
>= TOK_IDENT
&& 0 == nosubst
) {
3367 /* if nested substitution, do nothing */
3368 if (sym_find2(*nested_list
, t
)) {
3369 /* and mark it as TOK_NOSUBST, so it doesn't get subst'd again */
3370 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3376 str
.str
= (int*)ptr
;
3377 begin_macro(&str
, 2);
3380 macro_subst_tok(tok_str
, nested_list
, s
, can_read_stream
);
3382 if (str
.alloc
== 3) {
3383 /* already finished by reading function macro arguments */
3391 spc
= (tok_str
->len
&&
3392 is_space(tok_last(tok_str
->str
,
3393 tok_str
->str
+ tok_str
->len
)));
3397 if (t
== '\\' && !(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
3398 tcc_error("stray '\\' in program");
3401 if (!check_space(t
, &spc
))
3402 tok_str_add2(tok_str
, t
, &cval
);
3404 if (t
== TOK_NOSUBST
)
3409 tok_str_free_str(macro_str1
);
3413 /* return next token with macro substitution */
3414 ST_FUNC
void next(void)
3417 if (parse_flags
& PARSE_FLAG_SPACES
)
3423 if (tok
== TOK_NOSUBST
|| tok
== TOK_PLCHLDR
) {
3424 /* discard preprocessor markers */
3426 } else if (tok
== 0) {
3427 /* end of macro or unget token string */
3431 } else if (tok
>= TOK_IDENT
&& (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
3433 /* if reading from file, try to substitute macros */
3434 s
= define_find(tok
);
3436 Sym
*nested_list
= NULL
;
3438 macro_subst_tok(&tokstr_buf
, &nested_list
, s
, 1);
3439 tok_str_add(&tokstr_buf
, 0);
3440 begin_macro(&tokstr_buf
, 2);
3444 /* convert preprocessor tokens into C tokens */
3445 if (tok
== TOK_PPNUM
) {
3446 if (parse_flags
& PARSE_FLAG_TOK_NUM
)
3447 parse_number((char *)tokc
.str
.data
);
3448 } else if (tok
== TOK_PPSTR
) {
3449 if (parse_flags
& PARSE_FLAG_TOK_STR
)
3450 parse_string((char *)tokc
.str
.data
, tokc
.str
.size
- 1);
3454 /* push back current token and set current token to 'last_tok'. Only
3455 identifier case handled for labels. */
3456 ST_INLN
void unget_tok(int last_tok
)
3459 TokenString
*str
= tok_str_alloc();
3460 tok_str_add2(str
, tok
, &tokc
);
3461 tok_str_add(str
, 0);
3462 begin_macro(str
, 1);
3466 ST_FUNC
void preprocess_start(TCCState
*s1
)
3468 s1
->include_stack_ptr
= s1
->include_stack
;
3469 /* XXX: move that before to avoid having to initialize
3470 file->ifdef_stack_ptr ? */
3471 s1
->ifdef_stack_ptr
= s1
->ifdef_stack
;
3472 file
->ifdef_stack_ptr
= s1
->ifdef_stack_ptr
;
3475 pvtop
= vtop
= vstack
- 1;
3476 s1
->pack_stack
[0] = 0;
3477 s1
->pack_stack_ptr
= s1
->pack_stack
;
3479 isidnum_table
['$' - CH_EOF
] =
3480 s1
->dollars_in_identifiers
? IS_ID
: 0;
3481 isidnum_table
['.' - CH_EOF
] =
3482 (parse_flags
& PARSE_FLAG_ASM_FILE
) ? IS_ID
: 0;
3483 if (s1
->nb_cmd_include_files
) {
3487 for (i
= 0; i
< s1
->nb_cmd_include_files
; i
++) {
3488 cstr_cat(&cstr
, "#include \"", -1);
3489 cstr_cat(&cstr
, s1
->cmd_include_files
[i
], -1);
3490 cstr_cat(&cstr
, "\"\n", -1);
3492 *s1
->include_stack_ptr
++ = file
;
3493 tcc_open_bf(s1
, "<command line>", cstr
.size
);
3494 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3499 ST_FUNC
void tccpp_new(TCCState
*s
)
3504 /* might be used in error() before preprocess_start() */
3505 s
->include_stack_ptr
= s
->include_stack
;
3507 /* init isid table */
3508 for(i
= CH_EOF
; i
<128; i
++)
3509 isidnum_table
[i
- CH_EOF
]
3510 = is_space(i
) ? IS_SPC
3515 for(i
= 128; i
<256; i
++)
3516 isidnum_table
[i
- CH_EOF
] = IS_ID
;
3518 /* init allocators */
3519 tal_new(&toksym_alloc
, TOKSYM_TAL_LIMIT
, TOKSYM_TAL_SIZE
);
3520 tal_new(&tokstr_alloc
, TOKSTR_TAL_LIMIT
, TOKSTR_TAL_SIZE
);
3521 tal_new(&cstr_alloc
, CSTR_TAL_LIMIT
, CSTR_TAL_SIZE
);
3523 memset(hash_ident
, 0, TOK_HASH_SIZE
* sizeof(TokenSym
*));
3524 cstr_new(&cstr_buf
);
3525 cstr_realloc(&cstr_buf
, STRING_MAX_SIZE
);
3526 tok_str_new(&tokstr_buf
);
3527 tok_str_realloc(&tokstr_buf
, TOKSTR_MAX_SIZE
);
3529 tok_ident
= TOK_IDENT
;
3538 tok_alloc(p
, r
- p
- 1);
3543 ST_FUNC
void tccpp_delete(TCCState
*s
)
3547 /* free -D and compiler defines */
3550 /* cleanup from error/setjmp */
3556 n
= tok_ident
- TOK_IDENT
;
3557 for(i
= 0; i
< n
; i
++)
3558 tal_free(toksym_alloc
, table_ident
[i
]);
3559 tcc_free(table_ident
);
3562 /* free static buffers */
3563 cstr_free(&tokcstr
);
3564 cstr_free(&cstr_buf
);
3565 tok_str_free_str(tokstr_buf
.str
);
3567 /* free allocators */
3568 tal_delete(toksym_alloc
);
3569 toksym_alloc
= NULL
;
3570 tal_delete(tokstr_alloc
);
3571 tokstr_alloc
= NULL
;
3572 tal_delete(cstr_alloc
);
3576 /* ------------------------------------------------------------------------- */
3577 /* tcc -E [-P[1]] [-dD} support */
3579 static void tok_print(const char *msg
, const int *str
)
3585 fp
= tcc_state
->ppfp
;
3586 if (!fp
|| !tcc_state
->dflag
)
3589 fprintf(fp
, "%s ", msg
);
3591 TOK_GET(&t
, &str
, &cval
);
3594 fprintf(fp
,"%s", get_tok_str(t
, &cval
));
3599 static void pp_line(TCCState
*s1
, BufferedFile
*f
, int level
)
3601 int d
= f
->line_num
- f
->line_ref
;
3606 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_NONE
) {
3608 } else if (level
== 0 && f
->line_ref
&& d
< 8) {
3610 fputs("\n", s1
->ppfp
), --d
;
3611 } else if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_STD
) {
3612 fprintf(s1
->ppfp
, "#line %d \"%s\"\n", f
->line_num
, f
->filename
);
3614 fprintf(s1
->ppfp
, "# %d \"%s\"%s\n", f
->line_num
, f
->filename
,
3615 level
> 0 ? " 1" : level
< 0 ? " 2" : "");
3617 f
->line_ref
= f
->line_num
;
3620 static void define_print(TCCState
*s1
, int v
)
3626 if (NULL
== s
|| NULL
== s
->d
)
3630 fprintf(fp
, "#define %s", get_tok_str(v
, NULL
));
3631 if (s
->type
.t
== MACRO_FUNC
) {
3636 fprintf(fp
,"%s", get_tok_str(a
->v
& ~SYM_FIELD
, NULL
));
3643 tok_print("", s
->d
);
3646 static void pp_debug_defines(TCCState
*s1
)
3657 pp_line(s1
, file
, 0);
3658 file
->line_ref
= ++file
->line_num
;
3662 vs
= get_tok_str(v
, NULL
);
3663 if (t
== TOK_DEFINE
) {
3664 define_print(s1
, v
);
3665 } else if (t
== TOK_UNDEF
) {
3666 fprintf(fp
, "#undef %s\n", vs
);
3667 } else if (t
== TOK_push_macro
) {
3668 fprintf(fp
, "#pragma push_macro(\"%s\")\n", vs
);
3669 } else if (t
== TOK_pop_macro
) {
3670 fprintf(fp
, "#pragma pop_macro(\"%s\")\n", vs
);
3675 static void pp_debug_builtins(TCCState
*s1
)
3678 for (v
= TOK_IDENT
; v
< tok_ident
; ++v
)
3679 define_print(s1
, v
);
3682 /* Add a space between tokens a and b to avoid unwanted textual pasting */
3683 static int pp_need_space(int a
, int b
)
3685 return 'E' == a
? '+' == b
|| '-' == b
3686 : '+' == a
? TOK_INC
== b
|| '+' == b
3687 : '-' == a
? TOK_DEC
== b
|| '-' == b
3688 : a
>= TOK_IDENT
? b
>= TOK_IDENT
3692 /* maybe hex like 0x1e */
3693 static int pp_check_he0xE(int t
, const char *p
)
3695 if (t
== TOK_PPNUM
&& toup(strchr(p
, 0)[-1]) == 'E')
3700 /* Preprocess the current file */
3701 ST_FUNC
int tcc_preprocess(TCCState
*s1
)
3703 BufferedFile
**iptr
;
3704 int token_seen
, spcs
, level
;
3708 preprocess_start(s1
);
3709 ch
= file
->buf_ptr
[0];
3710 tok_flags
= TOK_FLAG_BOL
| TOK_FLAG_BOF
;
3711 parse_flags
= PARSE_FLAG_PREPROCESS
3712 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
3713 | PARSE_FLAG_LINEFEED
3715 | PARSE_FLAG_ACCEPT_STRAYS
3717 define_start
= define_stack
;
3719 /* Credits to Fabrice Bellard's initial revision to demonstrate its
3720 capability to compile and run itself, provided all numbers are
3721 given as decimals. tcc -E -P10 will do. */
3722 if (s1
->Pflag
== 1 + 10)
3723 parse_flags
|= PARSE_FLAG_TOK_NUM
, s1
->Pflag
= 1;
3726 /* for PP benchmarks */
3727 do next(); while (tok
!= TOK_EOF
);
3728 free_defines(define_start
);
3732 if (s1
->dflag
& 1) {
3733 pp_debug_builtins(s1
);
3737 token_seen
= TOK_LINEFEED
, spcs
= 0;
3738 pp_line(s1
, file
, 0);
3741 iptr
= s1
->include_stack_ptr
;
3745 level
= s1
->include_stack_ptr
- iptr
;
3748 pp_line(s1
, *iptr
, 0);
3749 pp_line(s1
, file
, level
);
3753 pp_debug_defines(s1
);
3758 if (token_seen
== TOK_LINEFEED
) {
3763 if (tok
== TOK_LINEFEED
) {
3767 pp_line(s1
, file
, 0);
3768 } else if (tok
== TOK_LINEFEED
) {
3771 spcs
= pp_need_space(token_seen
, tok
);
3775 fputs(" ", s1
->ppfp
), --spcs
;
3776 fputs(p
= get_tok_str(tok
, &tokc
), s1
->ppfp
);
3777 token_seen
= pp_check_he0xE(tok
, p
);;
3779 /* reset define stack, but keep -D and built-ins */
3780 free_defines(define_start
);
3784 /* ------------------------------------------------------------------------- */